Pretty much confirmed: Pixar to switch [Archive]

agreenster

Jul 31, 2003, 01:42 PM

I was browsing www.cgtalk.com today, reading up on Siggraph 2003 and came across this snippet aparently taken from Maccentral, though I cannot find where exactly:

I'll give you guys the **BIG NEWS** that was announced yesterday at Apple's Shake Users' Event in San Diego.

First, Pixar announced that they're bringing prMan over to OS X and will have a beta sometime in August or September.

Second, Pixar announced that they'll be migrating over to... guess what... the Mac!

They based their decision on issues such as performance, code portability, X11 integration, development culture, vender dedication to the fim industry, app availability, audio/video friendlyness and capabilities.

In 1980 Pixar made its first short on a Vax running VMS. Then in the mid '80s they migrated over to Sun workstations running Solaris. In the late '80s - '90s they were running entirely SGI Irix workstations, and in 2000 they moved over to Intels running Linux. An NT migration would've meant exiting the Unix market. But apparantly they were also having issues with Linux such as universal cut & paste, issues synching sound with picture, I think he said color management and other problems. He (a Pixar VP) said that while none of the issues were show stoppers or significant on their own, and they could all be dealt with or worked around, taken as a whole they presented real problems for the company.

Moving to OS X let's them essentially recompile the core functionality of their entire code base without even bothering with higher level UI development. Once they're settled in, then they could port the UI from X to Aqua.

During his slide show, the Pixar VP of something was about to present a list of why they'll be switiching to the Mac. But RIGHT at that time Keynote crashed out, drawing sarcastic applause from the audience. The timing just couldn't be worse. I think it may have been some kind of data corruption in one of the elements in the frame that crashed Keynote if you went BACK to the page from the next page. So from then on the presenters made sure not to go BACK to that particular frame.

Then a Hollywood-based motion design studio called Yu+Co showed their reel and the type of work they do on Shake. Their projects included The Hulk, Italian Job, X-Men2, and many others.

Then another company whose name I don't remember showed some shots of the sequel to Starship Troopers. This particular demonstration of Shake's expansion capabilities was quite impressive. They exported footage from Maya maintaining all the "DOD" data (if anyone knows what this is, let me know... a buncha visual fx artists from work had no idea today) when they used in Shake to create 3D ambient & spot lights. The guy, who had previously used Shake only on a dual AMD box, created a light which he could position anywhere in the scene, including in front, behind, or within the insect. The light behaved just as it would in a 3D environment! He had control over its color temperature, spread, intensity, diffussness, etc... it was quite impressive. He even made a little comment "I've only used this on my dual AMD, this is the first time I've used Shake on a Mac and it's quick [as he's moving the light around the insect]. Yes, my box doesn't do it nearly this fast."

Plus a nice RenderMan OSX screenshot:http://home.comcast.net/~zeio/sig/3.jpg

All sounding pretty sweet if you ask me!

patrick0brien

Jul 31, 2003, 01:45 PM

-All

Well it only makes sense.

I wonder if they have a preorder on a batch of G5 xServe Nodes on the books as well.

Though I think those Dell nodes they have will last them a while longer.

idea_hamster

Jul 31, 2003, 01:45 PM

This could be a huge coup for Apple and I think that it bodes well for the performance of the G5 -- regardless of any benchmark argument that people may want to get into.

Regardless of how those bar charts look, Pixar won't switch to G5 unless the performance advantage is real -- and they'll get others thinking about the switch.

GPTurismo

Jul 31, 2003, 02:12 PM

If they moved to OSX it would only be for workstations. XServes for RenderFarms would be way to expensive, especially after they bought all those intel blades.

Oh no, OSX on Intel Rumors again :|

Mr. Anderson

Jul 31, 2003, 02:21 PM

Give them some time - in a few years they'll be updating their renderfarm and who knows what they'll use then.

Anyway - this is good news and further anchors the Apple as a high end machine - something they've needed for a long time....

D

Nermal

Jul 31, 2003, 02:22 PM

Does anyone know what the Renderman logo looks like? I've seen an icon that I think was Renderman in a screenshot somewhere, sitting in the dock. I just don't want to embarrass myself by describing the icon to a completely different app!

Edit: I just looked at Renderman on Pixar's site, and I've definitely seen that logo sitting in a dock. It was yellow, and I think it was slightly 3D.

Another Edit: I just looked at the screenshot linked above (didn't see if before) and I don't see a yellow Renderman there. Maybe I'm just imagining I ever saw one :eek:

Longey Nowze

Jul 31, 2003, 02:33 PM

Originally posted by Nermal
Does anyone know what the Renderman logo looks like? I've seen an icon that I think was Renderman in a screenshot somewhere, sitting in the dock. I just don't want to embarrass myself by describing the icon to a completely different app!

Edit: I just looked at Renderman on Pixar's site, and I've definitely seen that logo sitting in a dock. It was yellow, and I think it was slightly 3D.

Another Edit: I just looked at the screenshot linked above (didn't see if before) and I don't see a yellow Renderman there. Maybe I'm just imagining I ever saw one :eek:

lol my kitty (http://homepage.mac.com/mat_lock) is called nermal lol

anyway there's a video that they showed when they released the G5 and there was a guy from Pixar talking about how fast the G5 the icon was there I suppose? I'm to lazy to check

BOW TO THE KITTY!
MaT

VIREBEL661

Jul 31, 2003, 02:39 PM

Great news overall, and long overdue... Because of the Unix BSD base, there's no reason they shouldn't be using Macs for this kind of stuff... It already dominates the top end professional digital recording market (ProTools)...

jadariv

Jul 31, 2003, 03:01 PM

There's going to be a sequel to Starship Troopers?

Cool.

Mr. Anderson

Jul 31, 2003, 03:08 PM

Originally posted by jadariv
There's going to be a sequel to Starship Troopers?

Cool.

Oh please let that not be true - that movie was total crap.

The FX were ok, but the story was crap, along with the acting.....blah

D

MacFan25

Jul 31, 2003, 03:10 PM

This is good news for Apple and Pixar. I remember watching 60 Minutes a couple months ago, and Pixar was featured, and I saw Dell machines in the background.

Frobozz

Jul 31, 2003, 03:11 PM

Originally posted by idea_hamster
This could be a huge coup for Apple and I think that it bodes well for the performance of the G5 -- regardless of any benchmark argument that people may want to get into.

Regardless of how those bar charts look, Pixar won't switch to G5 unless the performance advantage is real -- and they'll get others thinking about the switch.

Yeah, I agree. I think the biggest point is that that Mac has all the speed (and then some) AND it has all the gui goodness. NO other platform has both.

LimeLite

Jul 31, 2003, 03:12 PM

Originally posted by GPTurismo
If they moved to OSX it would only be for workstations. XServes for RenderFarms would be way to expensive, especially after they bought all those intel blades.

Oh no, OSX on Intel Rumors again :|
It couldn't be *that* expensive. I'm sure Pixar would get a price discount. I mean, I hear the CEO of Pixar and the CEO of Apple are good friends. :D ;)

Powerbook G5

Jul 31, 2003, 03:26 PM

I am pretty sure you saw that Renderman icon from the G5 promotional video shown both at the WWDC keynote and on the PowerMac G5 page, when you decribed it, it just brought that back to memory.

agreenster

Jul 31, 2003, 03:31 PM

Originally posted by Mr. Anderson
Oh please let that not be true - that movie was total crap.

The FX were ok, but the story was crap, along with the acting.....blah

D

I think it was supposed to be that was. Believe it or not the director (slipped my mind) has a history of doing films where there is cheesy acting. He is kinda anti-people and always likes to satire that. I thought the movie was good, if you look at through the eyes of someone who sees people as destructive and good for nothing.

Mr. Anderson

Jul 31, 2003, 03:36 PM

DeusOmnis

Jul 31, 2003, 03:38 PM

this is... sweet... lol

jaedreth

Jul 31, 2003, 03:40 PM

So, basically, those rumors of High End workstations could be true, and if so, I know who their first customer will be. Pixar.

Apple can start marketing and producing products for people like Pixar, and start targeting that market. Yes! Finally! Replace SGI!

LOL I used to play on a VMS system on VAX. Taught LISP how to do Algebra. What a pain.

BTW, Starship Troopers, the movie, will be pretty much lost on anyone who hasn't read the book by Robert Heinlein. And yes, the movie *was* supposed to be like that. Sure they trimmed it up and toned it down a little, but if they hadn't, the movie would have been R or worse...

I'm sure most Heinlein fans enjoyed the movie, except those who feel the movies should be just like the books, and I don't see that as realistic.

But for those who never read Heinlein, he was as much a Social Philosopher as a scifi writer. The movie was supposed to sound like one big ad for the military. He's painting a picture of how different society and the miltary might be if you had to serve to become a citizen. Historically, this has been done many places.

And in this movie, it demonstrates how non-civilians honor the committment and dedication of the civilians, especially since they're fighting a losing battle against the superior bug race...

But that is all history. Can't wait to see the next one.

Can't wait to get my hands on a G5 either.

Jaedreth

patrick0brien

Jul 31, 2003, 03:44 PM

-Anyway, to get back on topic.

I sense two good things from the Pixar switch (did we really think they wouldn't switch eventually?).

1. Validation. Pixar is riding high in the minds of young and old. What an advertisement "This movie made with Macintosh".

2. Some of the highest-end testors and user group to the man himself. I'm sure we'd all like to have Steve's ear to whisper sweet upgrade user request nothings into, but I think these guys won't mince words, and thereby be great advocates as well.

Mr. Anderson

Jul 31, 2003, 03:48 PM

Originally posted by patrick0brien
-Anyway, to get back on topic.

1. Validation. Pixar is riding high in the minds of young and old. What an advertisement "This movie made with Macintosh".

Sorry about that - I really loath the movie. ;)

But like you said - seeing 'Made on a Mac' at the end of a Pixar movie would be a coup.....we'll have to wait a while for that, though.

D

GPTurismo

Jul 31, 2003, 03:56 PM

The Animated show Roughnecks: Starship Troopers was superb, and it was closer to the book than the movie. But it was cancled due to such a high cost per episode. :(

Nermal

Jul 31, 2003, 03:59 PM

Originally posted by Longey Nowze
lol my kitty (http://homepage.mac.com/mat_lock) is called nermal lol

anyway there's a video that they showed when they released the G5 and there was a guy from Pixar talking about how fast the G5 the icon was there I suppose? I'm to lazy to check

BOW TO THE KITTY!
MaT

Yes! That's where I saw it. Good to know I wasn't dreaming it all up :rolleyes:

cc bcc

Jul 31, 2003, 04:15 PM

his particular demonstration of Shake's expansion capabilities was quite impressive. They exported footage from Maya maintaining all the "DOD" data (if anyone knows what this is, let me know... a buncha visual fx artists from work had no idea today) when they used in Shake to create 3D ambient & spot lights.

I think "DOD" should read "DOF", which is Depth Of Field, also called depth channel. It is a standard type of image containing depth info in black&white, white being close and fading to black further away. This in itself is not spectacular, most 3D apps and compositing apps can handle this data to make fog, or used for depth of field blurring/defocusing.
But what he is doing with it is amazing!

(I could be wrong about DOD actually being DOF, but depth images could perhaps be used for such fx in conjuction with other data about specular maps etc.)

cc bcc

Jul 31, 2003, 04:26 PM

Originally posted by agreenster
I think it was supposed to be that was. Believe it or not the director (slipped my mind) has a history of doing films where there is cheesy acting. He is kinda anti-people and always likes to satire that. I thought the movie was good, if you look at through the eyes of someone who sees people as destructive and good for nothing.

It was Paul Verhoeven, dutch guy. I like him a lot, and he's done some great movies: Starship Troopers, Hollow Man, Showgirls, Basic Instinct, Total Recall, RoboCop, Flesh + Blood.
Turks Fruit (Turkish Delight) a dutch spoken movie was true genious.
He doesn't hate people, but he's very critical about society, and starship troopers is all about that, and I thought it was great although many people I know think it sucked.. Ah well.
And then Robocop, modern Jezus, he even walks on water after his resurrection. All his movies have a couple of layers, different meanings.

unsane1

Jul 31, 2003, 04:30 PM

First off, DOD is Domain of Depth, it is similar to Depth of Field, but contains even more information if I remember correctly.

Also, if you think about it, you can see the logic. apple has developed a new video encoder that allows you to display full frame images on a desktop workstation in real time? Hunh, what could that be useful for? Would it be a good way for animators to preview the work while creating it?

Apple has the creative market by the balls, and no one really understands how. Adobe is pissed off, Microsoft is ignorant of the market, I think Apple is coming up with a pretty good plan on how to take control of more of the market, and how to do it in very key places.

All the cool people will be using them, so who cares about everyone else? ;)

cc bcc

Jul 31, 2003, 04:36 PM

Originally posted by unsane1
First off, DOD is Domain of Depth, it is similar to Depth of Field, but contains even more information if I remember correctly.

Ah thanks for the info.

All the cool people will be using them

Me for instance :D

solvs

Jul 31, 2003, 04:37 PM

The guy that made Starship Troopers (which, even as a parody, sucked) was the same guy who made Showgirls. Also Robocop, Total Recall, Hollow Man, and a bunch of other movies that had great ideas but not so great execution. Lets hope someone else does the sequel.

BTW, it was rated R.

*Back on subject*
Weren't the Intels leased? I heard Pixar had a bunch of Dells, too, but I doubt they buy everything. Wonder if they could get a discount on all those G5s. Even if Steve has to keep the companies separate, for fairness and in the best interest of both, you'd think they'd at least get a discount for buying (or leasing)bulk.

agreenster

Jul 31, 2003, 04:52 PM

Originally posted by jaedreth
Yes! Finally! Replace SGI!

SGI has been dead for a while. It should be an easy take-over. Most studios are using Linux/x86 boxes for their workstations. Hopefully the tide will shift to the G5...starting with Pixar

jaedreth

Jul 31, 2003, 04:57 PM

Yeah, I know they've fallen into obscurity... But they've fallen, Mac's haven't surpassed them, yet...

If Apple makes those workstations though, and designs them for exactly what Pixar needs, leaves out all the stuff Pixar doesn't, I'm sure such machines would be a shoe-in for a lot of pro markets, including the sciences...

I've been waiting on Macs to get to this level for a very very long time. *grins*

Jaedreth

jettredmont

Jul 31, 2003, 05:06 PM

Originally posted by GPTurismo
If they moved to OSX it would only be for workstations. XServes for RenderFarms would be way to expensive, especially after they bought all those intel blades.

Oh no, OSX on Intel Rumors again :|

When you are Pixar, you are replacing your hardware at such a rate that PowerPC vs Intel doesn't matter. All that matters is that your code compiles and runs fast as bejesus on this year's hardware.

Pixar is a damned far cry from your "typical corporate client" where boxes originally designed for Windows 3.1 are still "in the system".

cc bcc

Jul 31, 2003, 05:16 PM

Originally posted by solvs
The guy that made Starship Troopers (which, even as a parody, sucked) was the same guy who made Showgirls. Also Robocop, Total Recall, Hollow Man, and a bunch of other movies that had great ideas but not so great execution. Lets hope someone else does the sequel.

Offtopic:
I was reading Paul Verhoeven's (.net) website, he recieved a "Worst Director" award from Razzie Awards (?):
Director Paul Verhoeven came to the Razzie Awards ceremony and accepted his award in person, the first winner of this prize ever to do so.
:D

ricktowers

Jul 31, 2003, 05:52 PM

In the world of Shake DOD stands for DOMAIN OF DEFINITION. Basically when you turn this feature on Shake only process a defined aera in your viewer. For example the search region in the tracker.

BTW: Starship Troopers 2 is only coming out on video. And that part of the presentation was done by Tippet Studios.

fpnc

Jul 31, 2003, 06:22 PM

Seems like no one noticed or has yet to comment on the fact that the screen shot showing RenderMan results indicates that the 2.0 GHz G5 was only slightly faster than the Xeon running at 2.8GHz (also, I guess, a dual processor). What happened to the 2X advantage that Apple demoed at WWDC?

Actually, I think this is just closer to the truth. The G5 only brings Apple back to rough parity with the high-end Intel and AMD offerings. In any case, if Pixar is moving more of their production to PowerMac G5s then that is a very good thing for Apple and the "rest of us."

peterh

Jul 31, 2003, 06:28 PM

sfoalex

Jul 31, 2003, 06:29 PM

NASA was able to do Single 2.0 GHz G5 benchmarks, don't you think Pixar, being 57% owned by Steve Jobs has that same capability?

Hello?? Anyone home....

what makes you even think both of the systems discussed were not either Single or Dual on both sides of the platform fence..

Use your head...

cc bcc

Jul 31, 2003, 06:29 PM

The screenshot was just added, I just saw it for the first time.

I think the performance of the G5 is about on par with Apple's SPEC benchmarks, a small win in raw speed.

P-Worm

Jul 31, 2003, 06:40 PM

Originally posted by cc bcc
The screenshot was just added, I just saw it for the first time.

I think the performance of the G5 is about on par with Apple's SPEC benchmarks, a small win in raw speed.

Right, and I doubt that at this stage there is much going on in the AltiVec department. No sources, jsut a guess.

P-Worm

jaedreth

Jul 31, 2003, 06:49 PM

Pixar is going to optimize the fool out of Renderman, take advantage of the 64 bit hardware and the altivec, and when optimization is complete, Pixar will get a *lot* of work done quickly on these machines. Wait til they get their hands on higher end workstation macs... 64 bit optimized with SMP support on 8, 16, 32, and 64 proc machines. I notice the much much earlier rumor said *nothing* about Quad processor Workstations. (The rumor actually said servers, but the design they described are more like workstations, though high end...) Perhaps if the PowerBooks ever make it to Dual G5, that low end PowerMacs may be dual and high end quad... If Apple could do that in a year or two, and keep the price down, we'd toast the market...

Jaedreth

Mac Kiwi

Jul 31, 2003, 07:10 PM

Originally posted by peterh
While not as good as Apple hype it still is 20% faster. If Renderman scaled linearly with clock, a 3.2GHz Xenon would only be 14% faster. Still pretty impressive.

Ya but the G5 is running gcc and the Xeon is runnng icc :)

That Starship troopers 3d char animation series was made with LW,how do I know?,one day the soldiers were having a day off and one was wearing a LW t shirt.

This is incredible news,it will make more studios look at the possibility of a Mac pipeline.I noticed on Boxxs site that the mention they have of the G5 is a link to the guy from the digital video editing site who contacted the Apple engineers in order to set them up and make them look stupid.

Damn I hope Apple stops poking Adobe with a stick soon,because if we lose Photoshop Apple will be in big trouble...

Might be wrong but I think Digital Domain did the majority of the FX in the first Starship troopers movie.

Stu.

ricktowers

Jul 31, 2003, 07:15 PM

Sony Imageworks and Tippet Studios were the main houses that worked on the original Starship Troopers feature. It was actually Sony's first big VFX movie. I don't think DD did any work on it.

Mac Kiwi

Jul 31, 2003, 07:26 PM

Ah ok I was wrong :) ....must have been because I was reading a Digital Domain interview recently and it has gotten stuck in my head :D

Stu.

whooleytoo

Jul 31, 2003, 07:55 PM

Originally posted by P-Worm
Right, and I doubt that at this stage there is much going on in the AltiVec department. No sources, jsut a guess.

Whereas the PC version is compiled using icc, the current speed champ. I've heard the advantage is largely due to the fact it auto-vectorizes code, which gcc doesn't.

Obviously we don't know much about how much Renderman uses Altivec, or how optimized it is for the G5, but I'd guess not very.

It's still very good news for Apple.

Mike.

Abstract

Jul 31, 2003, 08:12 PM

Originally posted by patrick0brien

I sense two good things from the Pixar switch (did we really think they wouldn't switch eventually?).

1. Validation. Pixar is riding high in the minds of young and old. What an advertisement "This movie made with Macintosh".

The problem is that some Windows lemming will look at this news, and all of the "Pixar Chooses G5 Mac" advertising hoopla from Apple and think, "Bah, Pixar only switched to Macs because Jobsy is the CEO of both companies!!". Then he'll start yelling expletives about that big stupid nerd, Steve Jobs, and continue french kissing the Bill Gates poster he has hanging on his ceiling. Then he'll make it sticky. I can already see the cynics typing out their "Jobs and Pixar: Jeez, Go Figure" articles and stories for their magazines.

Not that I'm a pessimist or anything, but why can't we win!!

jaedreth

Jul 31, 2003, 08:18 PM

The icc does have extensive optimization, and the gcc is basically just running on the G5. It's not optimized for even G4's altivec. They did have to make minor alterations to it so it would recognize the hardware properly. I read an indepth article on this elsewhere. Forgot where. But basically, just enough changes to make it run right. No altivec or anything...

Jaedreth

paulc

Jul 31, 2003, 08:32 PM

ultrafiel

Jul 31, 2003, 09:26 PM

Well Pixar having an Intel render farm makes sense. There are no G5 xServes out yet, and still it is probably quite a bit cheaper buying 2000 Xeons or whatnot than G5s. Not a big deal, but for workstations G5s would make sense. Also Finding Nemo just became the highest grossing animated feature ever, so Pixar still rides high (and saves Disney's hide currently) as the animation king. And the typical PC user will say anyway, who cares it's only Pixar using G5s, Jobs probably gave them for Apple advertising. It doesn't matter what gains Apple makes, PC users will ignore and dismiss it as nothing. Take benchmarks for example. For Pixar's next animated feature though I want to see some Panthers and Jaguars tear apart Longhorns... That would be great, if just for that plot alone!!!

Forgot to add, Adobe won't stop developing Photoshop for Mac. That would be stupid. That want to make money. Photoshop is very profitable. I use it every day at work, and most design professionals do. Adobe would be stupid to do that.

TMay

Jul 31, 2003, 09:32 PM

Originally posted by paulc
Still a tad confused. Pixar obviously knew all about the G5 when they decided their render farm was going to be intel blades. And as far as I can tell, they are still goling that way for rendering. Sort of implying that while they may choose G5's for workstations, when flat out constant speed is needed, intel is still the way to go.

In case anyone missed this, IBM will be building 970 boxen for Linux. I don't expect that Pixar would have a problem uprating the renderfarm to these in the future, though they might not be all that much less than Xserve and most likely more expensive than Xeon based servers. A nice tip to IBM I would think, in lieu of the potentially more costly Xserve, while still maintaining a fairly consistent code base.

kansast

Jul 31, 2003, 10:19 PM

with all the hype around Pixar's latest movie Nemo .. would have been nice if they could have claimed it was all done on G5s... maybe the next big Pixar release ?

macphoria

Jul 31, 2003, 10:27 PM

Are we going to start seeing Apple logos at the beginning of Pixar movies?

DISNEY PICTURES logo

cuts to

PIXAR logo

cuts to

APPLE logo "Made with Apple"

then people jump up and applaud.

;)

Mr. Anderson

Jul 31, 2003, 10:31 PM

Originally posted by kansast
with all the hype around Pixar's latest movie Nemo .. would have been nice if they could have claimed it was all done on G5s... maybe the next big Pixar release ?

Even the next release, The Incredibles, will have used the current render farm if they plan on getting it out next year. Although some G5s might end up getting put to use in some aspects of production.

I'm guessing that the next one after The Incredibles will have more of a G5 and Apple influence. :D

D

sentinal

Jul 31, 2003, 10:34 PM

Originally posted by unsane1
First off, DOD is Domain of Depth, it is similar to Depth of Field, but contains even more information if I remember correctly.

No, DOD, is a Shake specific term, Domain of Definition. A more generic term is crop window- ie, it tells the app that there's nothing outside of the area that it has to be concerned with, other than the geometry of the frame size.

nagromme

Jul 31, 2003, 11:10 PM

Regarding how the Wintel crowd see the Pixar change: many will indeed claim that Pixar is choosing a platform that will harm their business, just becuse their CEO cares more about his other company than about Pixar. Absurd, but I agree, it will be said.

But it doesn't matter that much. Others IN the film/TV industry (and in other industries too) will take the Mac more seriously when they see Pixar's lead. THEY won't think Pixar is stupid. And when THEY start adopting Macs, Wintel users won't have the Steve Jobs excuse to explain it.

Regarding render farms... Apple doesn't MAKE a blade server. Now, if an Xserve Blade appears, that's a different story. (And rendering can typically be done across a mix of OS's anyway. Keep the Dells and add Xserves too.)

Minor tangent:

Originally posted by jaedreth
The icc does have extensive optimization, and the gcc is basically just running on the G5. It's not optimized for even G4's altivec. They did have to make minor alterations to it so it would recognize the hardware properly. I read an indepth article on this elsewhere. Forgot where. But basically, just enough changes to make it run right. No altivec or anything...

Jaedreth

I've seen Windows devotees lately criticizing Apple for changing gcc to skew the test results when the G5s were demoed. They insist that Apple did not merely make minor alterations so gcc ran on the G5. Rather, they say, they tweaked it unfairly, using a speed-optimized gcc on the G5, but using plain old gcc on the Xeon. A cheat. (Not my opinion, just what I have seen stated.)

Any thorough, reputable, non-biased refutation of that fact--showing that Apple did NOT optimize gcc for their SPEC tests--would be a link I'd like to see, if anyone has a URL. (Otherwise, probably best to ignore this post rather than dig into the SPEC nonsense again!)

ogun7

Jul 31, 2003, 11:41 PM

Originally posted by unsane1
Apple has the creative market by the balls, and no one really understands how. Adobe is pissed off, Microsoft is ignorant of the market, I think Apple is coming up with a pretty good plan on how to take control of more of the market, and how to do it in very key places.

All the cool people will be using them, so who cares about everyone else? ;)

A few influential recording industry friends I have here in NY, (One on the Developer A-list @ Apple) have said that Uncle Steve wants to corner the content CREATION software market on one end, and slide into the enterprise intro to Unix network (read:migration from WinNT to Linux) on the other. Apple has a corps of evangelists from their marketing department that goes to these Fortune 500 companies and finds out what UNIX apps would make them buy a Mac network, then courts the developers to port to the Mac.
Microsoft would have left them to it, my friends said, if their DOJ woes took a bad turn. Now that the politicos are in their pocket$$, Microsoft is OFFICIALLY licensing Unix from SCO!
How's that for biting off the Mac platform!

kcmac

Jul 31, 2003, 11:42 PM

How much did the Intel renderfarm cost Pixar? So far, they are splitting $320M dollars with Disney on Nemo. This will only go much higher when the DVD's go public.

I can't see that a renderfarm is more than a drop in the bucket, a tax write off, a business expense. Heck, they could probably donate it and write the whole thing off. Since I am not an accountant or knowledgeable about costs, I could be sooooo wrong. But come on! :p

Wonder Boy

Aug 1, 2003, 12:51 AM

I just assumed they always used apple computers and software. Since they didn't, this is good news.

stingerman

Aug 1, 2003, 01:02 AM

Originally posted by kcmac
How much did the Intel renderfarm cost Pixar? So far, they are splitting $320M dollars with Disney on Nemo. This will only go much higher when the DVD's go public.

I can't see that a renderfarm is more than a drop in the bucket, a tax write off, a business expense. Heck, they could probably donate it and write the whole thing off. Since I am not an accountant or knowledgeable about costs, I could be sooooo wrong. But come on! :p

Not when one dual server can replace 4 Dells. You guys are making much about nothing. Pixar will rip out the whole render farm and put in a new one if it means they will double their productivity. They will not even blink an eye. What they care about is building the very best flick.

The G5 is an awesome processor, new generation and all, but what Pixar is really drooling over is the entire system. The dual 1GHz FSB , one each for each processor. 8GB of the fastest RAM, etc, etc. The G5 was built from the ground up to be a Pixar engineers dream machine. And for that matter a dream machine for anyone in the creative business.

stingerman

Aug 1, 2003, 01:21 AM

Originally posted by fpnc
Seems like no one noticed or has yet to comment on the fact that the screen shot showing RenderMan results indicates that the 2.0 GHz G5 was only slightly faster than the Xeon running at 2.8GHz (also, I guess, a dual processor). What happened to the 2X advantage that Apple demoed at WWDC?

Actually, I think this is just closer to the truth. The G5 only brings Apple back to rough parity with the high-end Intel and AMD offerings. In any case, if Pixar is moving more of their production to PowerMac G5s then that is a very good thing for Apple and the "rest of us."

Read the whole report, this was only one particular test. Here is another quote from a Siggraph attendee:

Also of note, the Pixar booth was showing off Renderman on the new G5, rendering out film-res frames of Finding Nemo. The dual 2.0ghz G5 was rendering signifcantly faster than a dual 3.06ghz Xeon, which was interesting to see, and let's just leave it at that (please start another thread if you want to bash processors/OSes. Preferably on another web site entirely. Thanks!)

Search on Pixar to find the threads on the CGTALK.com site.

fpnc

Aug 1, 2003, 02:49 AM

Originally posted by peterh
While not as good as Apple hype it still is 20% faster. If Renderman scaled linearly with clock, a 3.2GHz Xenon would only be 14% faster. Still pretty impressive.

Well, this is a minor issue and we're only talking about a single benchmark result, but we should be accurate. The G5 was only 12% faster in the RenderMan benchmark, not 20%. Therefore, a 3.2 GHz Xeon would scale to being just slightly faster than the 2 GHz G5. However, I don't believe that Xeons are available at 3.2 GHz (they still max out at 3.06 GHz). The 3.2 GHz parts are Pentium 4's.

Do the math:

G5 @ 2.0 GHz: 216 seconds
Xeon @ 2.8 GHz: 246 seconds

(246 - 216)
--------------- = 0.12
246

or

(1.00 - 0.12) x 246 = 216

Nothing to get real excited about, but we should try to be accurate in our posts.

Hugin777

Aug 1, 2003, 04:03 AM

Originally posted by fpnc
Well, this is a minor issue and we're only talking about a single benchmark result, but we should be accurate. The G5 was only 12% faster in the RenderMan benchmark, not 20%.

While it is correct that the G5 uses 12% less seconds for the task, it is 14% faster:

G5 speed: N/216 work-units/sec.
Xeon speed: N/246 WU/s

set F=216*246

(246N/F-216N/F) / (N/246) = (30/F)*246 = 30/216 = 13,9 %

Originally posted by fpnc
Nothing to get real excited about, but we should try to be accurate in our posts.

:D

bluedalmatian

Aug 1, 2003, 04:14 AM

excuse me for being ignorant but what exactly is a blade server?

AndreHAL

Aug 1, 2003, 04:34 AM

sorry i didn't think through it.

read next post

Hugin777

Aug 1, 2003, 04:41 AM

Originally posted by AndreHAL
Wow! :eek:
Why not just do one simple division:
246/216 = 13,9%

Very impressive though!

- because I wanted to make clear that you have to divide by 216 and not by 246... You have to divide by the Xeon, but as speed is the inverse of "number of seconds" you end up with 30/216 instead of 30/246...

And now, after this extremely important aside, back on topic... ;)

tychay

Aug 1, 2003, 04:58 AM

Originally posted by bluedalmatian
excuse me for being ignorant but what exactly is a blade server?

Typical servers are measured in height of units (1.75"). A thin server is now 1U (such as the XServe (http://www.apple.com/xserve/)) which means that you can fit up to 48 of them in a rack (48U)--in practice this is much less because of cabling, room for routers/switches, KVM switches, and general inconvenience. Blade servers are around 0.5U (thus up to 192 processors per rack) by via vertical mounting, using some notebook components, removing some unnecessary components (like a graphics card and PCI expansion), and sharing systems such as keyboard/video/mouse, ethernet and fibrechannel switches, redundant power supplies, etc. I should add that the AGP graphics card is not needed for any "scale out" applications such as a renderfarm.

Right now the most impressive blade on the market is the IBM Bladecenter (http://www-1.ibm.com/servers/eserver/bladecenter/) which uses Dual P4 XEON systems and 14 blades in a 7U enclosure. RLX (a spinoff of Compaq developers) was the first, followed by Compaq and HP and recently IBM, Dell and Sun (the last one is a very nice blade system and is also worth a look). I may be biased because we use the IBM Bladecenter at work. IBM and Intel have some general goal to standardize the components of blade servers across vendors which is another reason I'm partial to them.

If you look at the Bladecenter design, you'll see that its ventillation system may have been some influence on the PowerMac G5.

IBM has demo'd a bladecenter using the 970 (http://www.osnews.com/story.php?news_id=2932) (G5) blades running Linux and AIX (I assume no MacOS X) later this year. Note that since the BladeCenter is modular, you should be able to use these new blades "mix and match" within an existing blade center.

It's not likely that Apple will have an offering here since a blade gives up certain things like a graphics card which MacOS X relies on intensely and the 1U servers have dual use as a SOHO server and is the server of choice for SME.

By the way, to my knowledge, the Intel servers that Pixar uses are probably leased (http://news.zdnet.com/business/0,39020645,2130152,00.htm) from Rackspace (http://www.rackspace.com/index.php) which is why you see all those Dells in photos of Pixars renderfarm. So Pixar can switch when the lease expires or if it has some out clause. I doubte they are using Dell blades as they are rather new and have a generally inferior design (P3, slow bus, RAM limitations), which, while fine for simple web applications, may have performance implications for renderfarms; I'm more inclined to believe the report that they use RackSaver (http://www.redriveros.com/solutions/publish/article16.shtml) servers and explain away the Dells in photographs as being what RackSpace is known to be partial to.

However, I don't see Pixar switching to Mac OS X for their renderfarm since the price/performance won't be able to keep up with a Linux on 970 (G5), or Linux on x86, Linux on Itanium, or Linux on whatever is best at the time. And yes, if this quote from Pixar's president of technology (http://www.polyhedron.com/intel/Reseller_ProductPage_C++_FORTRAN_Compilers.htm) is to believed, Renderman uses ICC to compile, which auto-vectorizates for MMX/SSE. Apparently, IBM will be submitting something along those lines for acceptance into GCC for the 970/G5. It is suspect that Renderman on the G5 has been similarly optimized yet. Certainly Pixar will benefit if either GCC or Apple make such patches available.

Honestly ask yourself what a Mac gives you, and you see when you take away the GUI and ease-of-use (neither of which you need when you are going to remotely provision your application across 1000 CPUs), you can get the same thing on Linux without any licensing. It doesn't make good business sense to switch to OS X.

Note, I was talking about the renderfarm not their desktops. It is a foregone conclusion that Pixar is migrating to Mac OS X G5 desktop workstations or is this conference (http://conferences.oreillynet.com/cs/macosx2003/view/e_sess/4478) just a really big typo? Here, Mac OS X really solves a problem. That is simply good business sense.

I think it is generally understood that the folks at Pixar seem to be running things so smoothly that Jobs hardly ever shows up at his office there. I don't think he'd be demanding that they use Macs. After all, we are talking about a guy who used an IBM Thinkpad with NeXTSTEP/86 for years after becoming iCEO at Apple.

DarkNovaMatter

Aug 1, 2003, 05:15 AM

If I remember right Pixar rents the server farm they now have, from another company that sells them (not makes them). They did this because of the time crunch they had with Nemo. Well its nice to hear that Renderman is coming to X. Its also interesting that (I am sapposing here) that the G5 isn't running anything but 10.2.7 (the renderfarm at Pixar is using ICC? This might be verry interesting!) and renderman for X is still beta? If thats true then with Panther, some good altivec with new GCC and then some finalization of the software and this is going to be a nice year for Apple!

Quila

Aug 1, 2003, 05:40 AM

Originally posted by jaedreth
They did have to make minor alterations to it so it would recognize the hardware properly. I read an indepth article on this elsewhere. Forgot where. But basically, just enough changes to make it run right. No altivec or anything...

Jaedreth
Probably the Ars Technica interview with one of the PPC 970 engineers. They had to lie to the compiler to get decent performance since gcc just isn't built to handle the PPC 970's architecture.

Quila

Aug 1, 2003, 05:43 AM

Originally posted by Mr. Anderson
I'm guessing that the next one after The Incredibles will have more of a G5 and Apple influence. :D

D
Now that would be wonderful. Pixar's first post-Disney release (yeah!) made with Macs (yeah again!).

Geetar

Aug 1, 2003, 08:08 AM

And that'll be...Toy Story 3? They've been itching to do that, but their Disney distribution deal doesn't make that a clever film for Pixar to make right now. But it'll be a sure-fire success, and it'll be made on the Mac.

unsane1

Aug 1, 2003, 09:00 AM

Apologies on the Domain of depth thing, I was getting it confused with a Maya thing, and when I went back to check I realized my mistake. Ahh how cool the lighting effects in Shake are.

Where are these photos of the Dell renderfarm that Pixar is using?

Last time I saw (and admittedly it was a while ago) Pixar was using an all Sun renderfarm. I'll have to see if I can find that picture I have of it and post it.

If they have changed to an all intel renderfarm, then all that does is go to show that it really isn't that hard to do a platform/renderfarm shift.

Lanbrown

Aug 1, 2003, 09:19 AM

The article seems to have some facts wrong. Pixar has several different platforms in use, and when they "switched" to Intel, they switched one application to linsux on Intel. They still sue Sun systems. Even after the switch the OS X, they will still have different platforms. If you look at dual CPU configurations, The USIII is faster in fp then the G5 at 2Ghz.
Base fp according to Apple for a dual 2GHz: 15.7
Sun Blade 2000 with one 1.2GHz USIII: 11.1
Sun Blade 2000 with two 1.2GHz USIII: 19.7
Sun Blade 2000 with two 1.05GHz USIII: 14.6

Pixar is going to use what is economically feasible, and not all of their equipment gets updated at the same time. They still use Sun equipment as well.

illumin8

Aug 1, 2003, 09:31 AM

Originally posted by agreenster
SGI has been dead for a while. It should be an easy take-over. Most studios are using Linux/x86 boxes for their workstations. Hopefully the tide will shift to the G5...starting with Pixar
Actually, SGI's biggest customer is the Department of Defense (the real DOD) ;) Anyway, I don't think they're in danger of going out of business any time soon. They fill a niche market that nobody else seems to do very well. Ever try to hook 16 graphics cards up to a single computer and make a huge video wall with each screen displaying different 3d modeled real-time data? You can say they are dead, antiquated dinosaurs of computers, or whatever. But let me remind you of another company a lot of pundits like to say is dead. It's a company run by Steve Jobs that starts with an A and ends with an e. ;) So before you go spouting your mouth off about companies that are dead, research the facts.

nickgold

Aug 1, 2003, 09:40 AM

Originally posted by Mr. Anderson
Oh please let that not be true - that movie was total crap.

The FX were ok, but the story was crap, along with the acting.....blah

D

*OT*

Kinda killed the intent of the original Heinlein novel, in many ways. It's a good read, in a freaky kind of way.

I mean heck, in the book they had mech suits with nuclear bomb launchers and stuff, it wasn't like modern day GIs going against bugs one-on-one and getting totally mauled. In the book they were pretty much unstoppable, and that was kind of the point.

whooleytoo

Aug 1, 2003, 10:48 AM

Originally posted by nagromme

Any thorough, reputable, non-biased refutation of that fact--showing that Apple did NOT optimize gcc for their SPEC tests--would be a link I'd like to see, if anyone has a URL. (Otherwise, probably best to ignore this post rather than dig into the SPEC nonsense again!)

The two main points I've seen, are they turned off a feature called software pre-fetch (no idea what it does!), but they claimed it will be off for production as well, so it's valid for benchmarking.

The other is they used a special, single thread only version of malloc (the function used to allocate memory), according to Apple they haven't decided whether or not this will be used in production.

If anything, this all just shows the folly in benchmarking machines over two months before release. If Apple were "cheating", then how were their 2.0GHz benchmarks lower than IBM's 1.8GHz benchmarks?

Mike.

whooleytoo

Aug 1, 2003, 10:55 AM

Originally posted by nagromme
Regarding how the Wintel crowd see the Pixar change: many will indeed claim that Pixar is choosing a platform that will harm their business, just becuse their CEO cares more about his other company than about Pixar. Absurd, but I agree, it will be said.

Another point, Jobs has been CEO for both companies for a while now and they haven't made the switch, indeed they made a high-publicized switch to Intel from Sun. So if they make the switch now, in the light of the arrival of the G5's and OSX finally reaching maturity (in my opinion), it sends a more powerful signal.

Mike.

Quila

Aug 1, 2003, 11:01 AM

Originally posted by whooleytoo
The two main points I've seen, are they turned off a feature called software pre-fetch (no idea what it does!), but they claimed it will be off for production as well, so it's valid for benchmarking.

Read the interview (http://www.arstechnica.com/cpu/03q2/ppc970-interview/ppc970-interview-1.html) with the PPC 970's chief designer. He talks about compiler problems.

Lanbrown

Aug 1, 2003, 11:32 AM

Originally posted by illumin8
Actually, SGI's biggest customer is the Department of Defense (the real DOD) ;) Anyway, I don't think they're in danger of going out of business any time soon. They fill a niche market that nobody else seems to do very well. Ever try to hook 16 graphics cards up to a single computer and make a huge video wall with each screen displaying different 3d modeled real-time data? You can say they are dead, antiquated dinosaurs of computers, or whatever. But let me remind you of another company a lot of pundits like to say is dead. It's a company run by Steve Jobs that starts with an A and ends with an e. ;) So before you go spouting your mouth off about companies that are dead, research the facts.

SGI is also moving towards the Itanic, so they will just be another peecee manufacturer. It won't be hard for them to find a market; they should actually hurt the other peecee companies. If you want Itanic, you can go with a company that can build large systems and can do clustering, or you can go with someone that only does cluster setups and have little real world experience.

If you don't want Itanic, SGI will be out.

SGI is also much smaller then what it once was. Their current CPU is really showing its age and has not been updated for sometime now, just a faster in terms of MHz.

visor

Aug 1, 2003, 12:08 PM

Abstract

Aug 1, 2003, 12:20 PM

Originally posted by Geetar
And that'll be...Toy Story 3? They've been itching to do that, but their Disney distribution deal doesn't make that a clever film for Pixar to make right now. But it'll be a sure-fire success, and it'll be made on the Mac.

Disney is riding Pixar's coattails. Pixar knows that, and so does Disney. Right now Disney gets 50% of the profits (or something like that), but Pixar is now big enough to survive on their own, and they only owe Disney another 2-3 movies (its contractual). They're making a movie called "Cars" and that should be the last of the Pixar-Disney movies before Pixar starts making the movies completely without Disneys unnecessary presence.

agreenster

Aug 1, 2003, 01:10 PM

Actually, its Disney who wants to capitalize on the Toy Story 3 idea, and Pixar doesnt want to do it because it lacks creativity. They just dont want to do another sequel (I wouldnt either)

Pixar has moved on, and wants to push the creative envelope (as they did with A Bugs Life, Monsters, Nemo) but Disney is just interested in the cash cow.

Lanbrown

Aug 1, 2003, 01:16 PM

Originally posted by visor
I can't remember anyone saying the 2ghz g5 was twice as fast as the 3 gig xeon.

Actually the xeon single proc was tested faster on specint, and slightly slower on specfp.

It's still interesting that a proc running 1 ghz slower, is still 15% faster in getting the job done.

Here is how they stack up in the SPECfp_rate2000:
Apple G5 2GHz PowerPC 970 x 2: 15.7
Dell Precession 650 3.06GHz Xeon x 1: 13.6
Dell Precession 650 3.06GHz Xeon x 2: 17.3
Dell 1750 3.06GHz Xeon x 2: 19.6
IBM Power 4 1450MHz x 2: 19.6
Sun Blade 2000 1.2GHz USIII x 2: 19.7
Opteron 144 1.8GHz x 2: 24.7
Intel Itanium 2 1.5GHz x 2: 37.3

jettredmont

Aug 1, 2003, 01:20 PM

Originally posted by kcmac
How much did the Intel renderfarm cost Pixar? So far, they are splitting $320M dollars with Disney on Nemo. This will only go much higher when the DVD's go public.

I can't see that a renderfarm is more than a drop in the bucket, a tax write off, a business expense. Heck, they could probably donate it and write the whole thing off. Since I am not an accountant or knowledgeable about costs, I could be sooooo wrong. But come on! :p

This is bad business.

You don't ignore an expense just because it is smaller than your revenues.

The real question is this:

given their existing renderfarm, how much would it cost to replace that renderfarm with this year's hardware (and, possibly, a new underlying architecture ... ie, PPC instead of IA-32) (yes, this figure includes the tax deduction from donating the old hardware or the income from selling it to someone else)? Compare that to how much you will lose in productivity and time-to-market by staying with last-year's hardware.

That is the sole business question. If it is better profits-wise to stick with a fleet of Dells and an Intel-based render farm, then that is what Pixar should do. If Pixar moves to OS X on their workstations *and* their render farm, it will be because the increased productivity of their engineers on the workstation and power-per-dollar of the G5 in the farm offsets the hardware and migration costs.

JBracy

Aug 1, 2003, 01:21 PM

Originally posted by Lanbrown
SGI is also moving towards the Itanic, so they will just be another peecee manufacturer. It won't be hard for them to find a market; they should actually hurt the other peecee companies. If you want Itanic, you can go with a company that can build large systems and can do clustering, or you can go with someone that only does cluster setups and have little real world experience.

If you don't want Itanic, SGI will be out.

SGI is also much smaller then what it once was. Their current CPU is really showing its age and has not been updated for sometime now, just a faster in terms of MHz.

What do you mean they are moving towards the Itanic? SGI have 1 Itanium based Linux server. That's it. The rest of their product line is all MIPS/IRIX. While it's true a few years ago they started making Intel based Irix workstations, thay have recognized the error of their ways and gone back to MIPS/IRIX. I'll bet that soon the Itanium will be gone as well.

If you think the MIPS processor is slow YOU'RE INSANE! The reason it hasn't been updated much is that it doesn't need to be. It's a server/workstation chip like the IBM Power4. These chips are made to have a extremely long life. They are aimed at industries that upgrade every 10+ years and who need to know that their equipment will still be supported for that long.

SGI (http://www.sgi.com)
MIPS (http://www.mips.com)

agreenster

Aug 1, 2003, 01:34 PM

EVERYONE is FREAKING out about this rumor. The truth is, we dont know if Apple is going to replace their intel servers with G5-powered X-serves, and we dont know how extensive they will even be using the G5.

The only thing we really know is that they have ported RenderMan to OSX, and are considering (via a rumor from a random MacCentral poster) switching to the G5 as a workstation, or maybe more.

Maybe it's on a limited basis, or maybe it over the entire production pipeline. Bottom line is, just enjoy the fact that Apple has FINALLY made a piece of hardware worthy of Pixars consideration and interest.

Regardless of who their CEO is, Pixar's only concern is productivity and fast turnaround, not their loyalties to the hardware/software they are using. Therefore, if its Apple, no one should be able to accuse Pixar of doing this as a marketing ploy. They've done JUST FINE without using Apple for years and years.

Cubeboy

Aug 1, 2003, 01:35 PM

Originally posted by visor
I can't remember anyone saying the 2ghz g5 was twice as fast as the 3 gig xeon.

Actually the xeon single proc was tested faster on specint, and slightly slower on specfp.

It's still interesting that a proc running 1 ghz slower, is still 15% faster in getting the job done.

How about the Xeon being 30% slower in SPECfp and 5% faster in SPECint. Besides, I think he was referring to the "real world" benchmarks.

The Photoshop comparison showed the Dual 2 GHz G5 being *exactly* twice as fast as a Dual 3.06 GHz Xeon, the BLAST comparison showed it being three times as fast, and finally we have HMMer in which the Dual 2 GHz G5 was at least four times as fast and the Single 1.6 GHz G5 was well over twice as fast as a Dual Xeon.

Cubeboy

Aug 1, 2003, 01:39 PM

whooleytoo

Aug 1, 2003, 01:46 PM

Originally posted by Quila
Read the interview (http://www.arstechnica.com/cpu/03q2/ppc970-interview/ppc970-interview-1.html) with the PPC 970's chief designer. He talks about compiler problems.

True, but I don't think the were the "optimizations" that people were talking about when they called Apple cheats - though I could be wrong: there were so many accusations of cheating it's hard to keep track! :-)

By the way, I certainly wouldn't consider those compiler changes cheating, if those changes make into the released gcc. It's hardly surprising a transition from the G4 series to the 970 would require a similar transition for its compilers.

Mike.

AidenShaw

Aug 1, 2003, 02:20 PM

Originally posted by whooley
By the way, I certainly wouldn't consider those compiler changes cheating, if those changes make into the released gcc. It's hardly surprising a transition from the G4 series to the 970 would require a similar transition for its compilers.

I agree, using the best compiler and switches is important for every platform.

The underhanded tactic of Apple's was to use such a compiler for the G5, yet not use a similarly optimized compiler for the Xeon.

If they wanted to be fair, they could have used the best compiler on each platform. Or conversely, they could have used GCC 3.2 - which had sub-par optimizations for both x86 and PPC.

But no, they used GCC 3.3 which has a ton of special tweaks written by chip vendor IBM -- but they didn't use the other compiler with a ton of tweaks by chip vendor Intel.

That's the issue of fairness - using IBM's tweaks but not Intel's....

Cubeboy

Aug 1, 2003, 02:24 PM

Originally posted by fpnc
Well, this is a minor issue and we're only talking about a single benchmark result, but we should be accurate. The G5 was only 12% faster in the RenderMan benchmark, not 20%. Therefore, a 3.2 GHz Xeon would scale to being just slightly faster than the 2 GHz G5. However, I don't believe that Xeons are available at 3.2 GHz (they still max out at 3.06 GHz). The 3.2 GHz parts are Pentium 4's.

Do the math:

G5 @ 2.0 GHz: 216 seconds
Xeon @ 2.8 GHz: 246 seconds

(246 - 216)
--------------- = 0.12
246

or

(1.00 - 0.12) x 246 = 216

Nothing to get real excited about, but we should try to be accurate in our posts.

The new 3.06 GHz Xeon with 1 meg on die cache actually performs slightly better than a Pentium 4 3.2 GHz with a 800 MHz FSB in SPECint and slightly worse in SPECfp if it is any indication as to how it would perform in renderman.

Xeon DP 3.06, 1 MB On-Die L3 Cache
SPECint Base:1242
SPECfp Base: 1173

Xeon DP 2.80
SPECint Base: 1022
SPECfp Base: 1010

All in all a 21% improvement in SPECint base and a 16% improvement in SPECfp base. Looking at how newer ICC compilers affected performance*, it is doubtful that the newer compiler played a significant role in improving that score. Intel is planning to come out with a 3.2 GHz version of the Xeon with L3 cache soon.

I don't know if either of these is indicative of how a 3.06 GHz Xeon would perform in Renderman.

*:link below
http://forums.macrumors.com/showthread.php?s=&threadid=31775&perpage=&pagenumber=2

Lanbrown

Aug 1, 2003, 02:49 PM

Originally posted by JBracy
What do you mean they are moving towards the Itanic? SGI have 1 Itanium based Linux server. That's it. The rest of their product line is all MIPS/IRIX. While it's true a few years ago they started making Intel based Irix workstations, thay have recognized the error of their ways and gone back to MIPS/IRIX. I'll bet that soon the Itanium will be gone as well.

If you think the MIPS processor is slow YOU'RE INSANE! The reason it hasn't been updated much is that it doesn't need to be. It's a server/workstation chip like the IBM Power4. These chips are made to have a extremely long life. They are aimed at industries that upgrade every 10+ years and who need to know that their equipment will still be supported for that long.

SGI (http://www.sgi.com)
MIPS (http://www.mips.com)

Those MIPS processors can be found inside communication equipment as well.

SGI Origin 200 with 32 R14000A CPU's rates a 153 on the SPECfp-rate2000. Their Altix 3000 with 32 1GHz Itanium 2's gets 443. Almost three times the performance.

SPECfp2000:
SGI Origin 3200 with one 600MHz R14k gets a 529
SGI Altix 3000 (1500MHz, Itanium 2) gets a 2055

SPECint2000:
SGI Origin 3200 1X 600MHz R14k gets a 500
SGI Altix 3000 (1500MHz, Itanium 2) gets a 1077

Selling the same chip for that long is INSANE. The Power4 is compatible with the Power5 and they are compatible with the Power3. They are not the same chip though; just the like Ultra SPARC III is not the same as the US III.

JBracy

Aug 1, 2003, 03:15 PM

Originally posted by Lanbrown
Those MIPS processors can be found inside communication equipment as well.

SGI Origin 200 with 32 R14000A CPU's rates a 153 on the SPECfp-rate2000. Their Altix 3000 with 32 1GHz Itanium 2's gets 443. Almost three times the performance.

SPECfp2000:
SGI Origin 3200 with one 600MHz R14k gets a 529
SGI Altix 3000 (1500MHz, Itanium 2) gets a 2055

SPECint2000:
SGI Origin 3200 1X 600MHz R14k gets a 500
SGI Altix 3000 (1500MHz, Itanium 2) gets a 1077

Selling the same chip for that long is INSANE. The Power4 is compatible with the Power5 and they are compatible with the Power3. They are not the same chip though; just the like Ultra SPARC III is not the same as the US III.

Yes, the Itanium is faster, but that was not my point if you read my post. The MIPS is not a slouch.

Also, I think you will find that the Sun Ultra Sparc III is exactly the same as the US III. :-)

JBracy

Aug 1, 2003, 03:23 PM

Also don't forget that SGI doesn't have all of it's hopes in Hardware. They also own Alias (Maya) and are the driving force behind OpenGL.

SGI - like Apple - is a niche player, and they are the best at what they do. At a company I worked for in London (A busy content creation / pre-press house), we replaced about 10 of our SUN Servers with 2 SGI's. We went from about 2 crashes per month on average to never being down in over 2 years (maybe more, but I left after 2 years)!

agreenster

Aug 1, 2003, 03:40 PM

Originally posted by JBracy
Also don't forget that SGI doesn't have all of it's hopes in Hardware. They also own Alias (Maya) and are the driving force behind OpenGL.

Alias is a subsidiary of SGI

JBracy

Aug 1, 2003, 03:49 PM

Originally posted by agreenster
Alias is a subsidiary of SGI

Was that a question or a comment?

Alias is a wholly owned subsidiary of SGI.

Cubeboy

Aug 1, 2003, 03:50 PM

What do you mean they are moving towards the Itanic? SGI have 1 Itanium based Linux server. That's it. The rest of their product line is all MIPS/IRIX. While it's true a few years ago they started making Intel based Irix workstations, thay have recognized the error of their ways and gone back to MIPS/IRIX. I'll bet that soon the Itanium will be gone as well.

He does have a point there, both SGI and HP have been trying to migrate the customers of their "house brand" RISC chips to IA64.

The customer base for both cores (PA-RISC and R10K) have been shrinking significantly in recent years and the only reason both companies continue "recycling" these half decade old cores (rather than spend the resources to develop a new core), is because the remaining customer base is still economically important to them. In other words, they don't want to lose any of their current customers (to another vendor) but they don't want any new customers either unless it's for IA64.

JBracy

Aug 1, 2003, 04:10 PM

Originally posted by Cubeboy
In other words, they don't want to lose their old customers (to another vendor) but they don't want any more customers for their respective RISC chips either..

Then why did SGI stop selling Intel/Linux based workstations and just released new MIPS/IRIX based ones? The SGI Tezro (http://www.sgi.com/workstations/tezro/)

If they don't want new customers for their MIPS/IRIX platform then why spend the R&D on new systems?

If they want to move them to a new platform then they need to actually sell a platform to move them to!

It's like saying that Apple doesn't want new customers they only want to support the ones they already have!

agreenster

Aug 1, 2003, 04:21 PM

What I'm really interested in seeing is how well Pixlet performs on my 1gHz TiBook. Fcheck seems to play full frame animation pretty well, but it drops a frame here and there to maintain framerate. I would love to be able to fire up quicktime, and open a frame sequence and it play back in real time.........

Can you say aaaaaaaaaaaaaaaaaaaahhhhhhhhhhhhh.........

tychay

Aug 1, 2003, 05:01 PM

Originally posted by Lanbrown
Here is how they stack up in the SPECfp_rate2000:
Apple G5 2GHz PowerPC 970 x 2: 15.7
Dell Precession 650 3.06GHz Xeon x 1: 13.6
Dell Precession 650 3.06GHz Xeon x 2: 17.3
Dell 1750 3.06GHz Xeon x 2: 19.6
IBM Power 4 1450MHz x 2: 19.6
Sun Blade 2000 1.2GHz USIII x 2: 19.7
Opteron 144 1.8GHz x 2: 24.7
Intel Itanium 2 1.5GHz x 2: 37.3

Let's be a bit careful here. Apple's listing here is done using gcc/NAGWare Fortran90 compiler virtually "out-of-box", this is not standard practice (all the other listings) in the industry. Standard practice is to report the best result using any compiler/tweak/OS necessary to achieve it--you cannot modify the hardware though since SPEC is a system test, not a CPU one.

Apple's result should be not listed, or listed with an asterix since the actual number would be much higher.

Take care,

terry

Smurfman

Aug 1, 2003, 05:02 PM

Originally posted by paulc
Still a tad confused. Pixar obviously knew all about the G5 when they decided their render farm was going to be intel blades. And as far as I can tell, they are still goling that way for rendering. Sort of implying that while they may choose G5's for workstations, when flat out constant speed is needed, intel is still the way to go.

I would guess it's a matter of cost, scalability, and availability. Pixar needed more rendering speed ASAP. We're talking around 1000+ CPU's. When Pixar bought the Intel-based blade servers, Apple was probably 2 years away from offering them something comparable in speed AND cost.

HSMM Imaging Studio (http://www.hsmm.com/3danim)

jaedreth

Aug 1, 2003, 05:19 PM

Apple is the only one currently not using whatever compiler that gives its processors the best performance rating.

*Everybody* else uses what makes them look best.

Apple, in the past, did use what made them look better.

But now they're using GCC. And GCC is not reporting the G5 anywhere near it's true speed, it's not optimized. But the SIMDs for the other chips aren't omptimized in GCC either, though they are in ICC.

If you looked carefully, and also listen to windows customers, Windows pcs *in use* are almost never as fast as their specs claim.

There was an extensive article about this posted here, and I doubt many people read it.

Specs these days are very much like statistics.

"There are three kinds of lies. Lied, damned lies, and statistics."
-- Samuel Clemens

Jaedreth

tychay

Aug 1, 2003, 05:21 PM

Originally posted by AidenShaw
The underhanded tactic of Apple's was to use such a compiler for the G5, yet not use a similarly optimized compiler for the Xeon.

If they wanted to be fair, they could have used the best compiler on each platform. Or conversely, they could have used GCC 3.2 - which had sub-par optimizations for both x86 and PPC.

But no, they used GCC 3.3 which has a ton of special tweaks written by chip vendor IBM -- but they didn't use the other compiler with a ton of tweaks by chip vendor Intel.

Not true. Show me that "using the best compiler" is "fair"--most people think that doing so is part of what is wrong with benchmarking in general. Show me the evidence that any version of GCC, is better on PPC than x86--most experts think the reverse is true and that, at best, GCC3.3 is a little closer to platform parity than earlier versions. Show me the "ton of tweaks" Apple used--there were only two, both may have been necessary to get the thing to compile at all.

You are confusing IBMs intention to submit tweaks to GCC (most of which involve the fact that the 970 has an interesting way of grouping code that is unique to that processor and a rumor that they may submit autovectorization a la ICC) none of which were actually submitted and none of which would necessarily be accepted due to GCC's design goals (portability), and Apple's misguided intent to "normalize out" the compiler (whatever that means).

The only thing underhanded here is how so many news sites spread such ill-informed B.S. about Apple "juicing" their benchmark that now it is accepted conventional wisdom.

jaedreth

Aug 1, 2003, 05:47 PM

Yes, IBM did redo GCC a little. Why? Because the G4 version just wouldn't compile on a G5. Period.

It *does* handle some data forms properly.

And guess what? They had to *rig* it. The only way to make it run properly according to traditional proper programming conventions would have required a complete rewrite of GCC for G5. They didn't have the time.

So they used some coding tricks that allowed the compiler to guestimate how the G5 was handling the data, but they had to get damned inventive. Their method was horribly inefficient and far from optimized.

But it made GCC run, and run "properly".

So if anything, Gcc on G5 is running like a G4 with the left hand tied behind it's back...

And the G5 still kicked ass.

Jaedreth

Cubeboy

Aug 1, 2003, 05:47 PM

Originally posted by JBracy
Then why did SGI stop selling Intel/Linux based workstations and just released new MIPS/IRIX based ones? The SGI Tezro (http://www.sgi.com/workstations/tezro/)

If they don't want new customers for their MIPS/IRIX platform then why spend the R&D on new systems?

If they want to move them to a new platform then they need to actually sell a platform to move them to!

It's like saying that Apple doesn't want new customers they only want to support the ones they already have!

SGI stopped selling Intel/Linux based workstations because sales were very poor, simple as that. SGI has had MIPS based workstations before, they just made the mistake of trying to impose a Intel based solution on their customers too quickly and too soon (and this goes back to my original point). Apparently the customers weren't very happy with this, and responded by moving to other vendors as a result.

If you look at the specifications of the Tecra, theirs really nothing new about it, it still uses the same MIPS R-16K processor (yawn), DDR memory, SCSI hard drives. It's certainly not going to get SGI any new customers considering the performance and price. I'd say it's just SGI's attempt to woo back some of it's old customers.

Cubeboy

Aug 1, 2003, 06:03 PM

Originally posted by Tychay
Not true. Show me that "using the best compiler" is "fair"--most people think that doing so is part of what is wrong with benchmarking in general. Show me the evidence that any version of GCC, is better on PPC than x86--most experts think the reverse is true and that, at best, GCC3.3 is a little closer to platform parity than earlier versions. Show me the "ton of tweaks" Apple used--there were only two, both may have been necessary to get the thing to compile at all.

You are confusing IBMs intention to submit tweaks to GCC (most of which involve the fact that the 970 has an interesting way of grouping code that is unique to that processor and a rumor that they may submit autovectorization a la ICC) none of which were actually submitted and none of which would necessarily be accepted due to GCC's design goals (portability), and Apple's misguided intent to "normalize out" the compiler (whatever that means).

The only thing underhanded here is how so many news sites spread such ill-informed B.S. about Apple "juicing" their benchmark that now it is accepted conventional wisdom.

Actually, we already know that the MD file used was modified over default (which were wrong). So apparently, the "no tweaks over default during testing" statement is false. This leads me to doubt that their weren't any tweaks, especially considering Veritest specifically listed the version of GCC as a specific build as opposed to just GCC.

From what I understand, IBM and Apple spent a considerable amount of time optimizing the scheduler for the PPC970.

"The gcc scheduler is not really designed ideally for a processor like the 970 and the Power4 and others, and that's a lot of what the IBM and Apple teams have worked on".

GCC is not able to schedule for the Pentium 4 at all, this is really quite important mind you considering nearly all floating point code needs to be well scheduled.

Many of the commands in GCC such as -march (which was used in the test) are normalized towards RISC chips, this means that the Pentium 4 (and most other x86 cpus for that matter), with it's "measily" eight registers, well probably be better off without them. I know this for a fact with -march=pentium 4 which is normalized for a 32 register RISC chip.

Do you know exactly how much autovectorization improves Pentium 4 ICC performance over x87 only code in SPECfp?

Can you actually give *ANY* evidence of exactly *HOW* Intel might cheat with ICC?

Cubeboy

Aug 1, 2003, 06:45 PM

Originally posted by tychay
Standard practice is to report the best result using any compiler/tweak/OS necessary to achieve it--you cannot modify the hardware though since SPEC is a system test, not a CPU one.

Apple's result should be not listed, or listed with an asterix since the actual number would be much higher.

terry

SPEC's benchmark suite consists of lots of kernels and all the programs have a memory footprint of between 100 and 200 MB so it's more about cpu/memory subsystem/compiler than anything else. Base score allows for only a single set of flags and limits the number of flags to four. Assertion flags also may not be used. Optimizations must generate correct code and improve performance on a class of programs, SPEC's source code cannot be changed (making it impossible to add hints) and contrary to popular belief, you can't manually change the machine output. Need I go on?

Considering IBM's preliminary SPEC scores for the PPC970, I doubt the scores would be much higher.

jettredmont

Aug 1, 2003, 07:09 PM

Originally posted by tychay
Not true. Show me that "using the best compiler" is "fair"--most people think that doing so is part of what is wrong with benchmarking in general.

The whole idea of SPEC is to say:

"You (who are about to spend $20-30k+ on a server and thus have the time and money to buy the best compiler out there and use the right switches): this is the best performance you can get on this particular hardware."

No, SPEC benchmarks have absolutely nothing to do with how fast MS Word runs or how fast Safari/IE runs, or even how fast Photoshop might run. SPEC is very server-oriented, and is based on the assumption that whatever you are going to be running on this investment will either be your own code compiled on the best compiler out there, or it will be someone else's code compiled on the best compiler out there.

Now, that's the crux of the problem for Apple: there's really no "great" compiler out there for PPC, certainly not for the G5! Yes, CodeWarrior traditionally bests gcc in performance, and perhaps that would hold true for the G5 as well (no guarantees), but even if CW could get a good 10% gain on gcc's best settings, the G5 will still end up beneath the high-end P4 boxes on SPEC.

So, Apple decided to go the "same compiler" route. At least, gcc sucks on Intel too (relative to Intel's compiler).

Now, yes, ICC does contain SPEC-specific optimization by most accounts. And, yes, that's what most folks would call "cheating". However, there's no denying that the P4 can and will (with Intel's compiler) run through the SPEC benchmarks really darned fast (a good bit faster than Apple was able to get it through using gcc).

Now, to give a bit of help to Apple:

1) Most production software is NOT compiled on ICC (it is generally compiled on MS's compiler, which still beats gcc in my experience but not as soundly as ICC can on SPEC). GCC is what most folks on Mac will be using (if not Code Warrior, which gets better results generally).

2) ICC isn't cheap. IBM's G5 compiler (were they to create one for public consumption) wouldn't be cheap either. GCC is really cheap.

3) While gcc has a heck of a lot of room for improvement wrt the G5, ICC ain't getting much better than it already is for the P4. In other words, the next generation of gcc might actually make the hardware run as fast as it should.

Is this what is wrong with benchmarks? Yeah, kind of. Even SPEC really falls apart on cross-platform comparisons. It's better than most, though.

On the other hand, while many deride "bake-offs", they do tend to be closer to what consumers should be looking at. How fast can I get my work done on Machine A vs Machine B? Hands-on testing is obviously the best way to judge this, but bakeoffs administered and defended by the authors of cross-platform software get pretty close (as opposed to bake-offs like we've seen in the past where Steve picks ten Photoshop filters where the G4 shines and shows how well the G4 does on those specific filters ...)

Show me the evidence that any version of GCC, is better on PPC than x86--most experts think the reverse is true and that, at best, GCC3.3 is a little closer to platform parity than earlier versions.

Which is exactly the problem: while gcc on PPC isn't more optimized for the hardware than gcc for Intel (I strongly agree with you there!), gcc for PPC is, well, pretty much all we've got. Aside from Code Warrior, of course, which isn't nearly as significantly getter as, say ICC is than gcc on Intel-based SPECs.

The only thing underhanded here is how so many news sites spread such ill-informed B.S. about Apple "juicing" their benchmark that now it is accepted conventional wisdom.

Ah, but conventional wisdom is never either conventional nor wisdom. Well, maybe conventional ... :)

tychay

Aug 1, 2003, 07:11 PM

Originally posted by Cubeboy
From what I understand, IBM and Apple spent a considerable amount of time optimizing the scheduler for the PPC970.

Do you know exactly how much autovectorization improves Pentium 4 ICC performance over x87 only code in SPECfp?

Can you actually give *ANY* evidence of exactly *HOW* Intel might cheat with ICC?

No, I don't know the exact amount of how much autovectorization improves ICC. However, the latest GCC performance was roughly comparable with ICC until the point where SSE replaces the FPU. Also, the head of Pixar technology is quoted on Intel's website (see my earlier post that was actually on topic) as saying Renderman received a 50% gain vs. GCC (http://www.polyhedron.com/intel/Reseller_ProductPage_C++_FORTRAN_Compilers.htm). Since Renderman is highly FP-intensive I'd say that almost all the evidence points to the fact that the improvement is "significant" for floating point tasks and benchmarks.

The reason GCC doesn't autovectorize P4 code has either to do with the fact that no such optimization was submitted, or because the GCC team felt that the optimizations that were submitted went against the chief goal of portability. The latter seems as equally valid as the former because the FPU unit goes "off" when the SIMD unit turns "on" in x86 because of a physical limitation without generic design reason behind it.

BTW, how is Intel using ICC cheating, and when have I ever said so? It is perfectly legal and the accepted practice to use the best compiler/best optimization/custom OS to produce the highest SPEC when reporting. This is not cheating. It is, however, still misleading to report Apple's benchmark (which isn't true to SPEC's accepted practice) next to true SPEC results, just as it is misleading not to mention how Veritest is not doing a true SPECmark when benchmarking for Apple--that doesn't mean that Apple's benchmarks are "juiced", evidence points to the opposite being true.

This means that the G5 could have looked significantly better had IBM IntelliAge or a hacked version of CodeWarrior been used. As much of a gain as ICC on x86? No way! One is the definition of a mature platform; the other doesn't even have a system out the door! That is Apple's deception. As the person you quoted said after the WWDC keynote, "Anybody who purchased a computer based on benchmarks deserves to be taken for a few grand."

The quote you mention actually deals with the machine description file which describes "what the CPU looks like" to the compiler and the issue involved in the quote was how the numbers in it for the G5/970 weren't exactly in line with the actual numbers in the processor. The reason, I gathered, is that the grouping of 5 model used by the 970 to keep track of so many instructions in flight does not mesh well with the way GCC models what a CPU actually looks like. Hence the numbers have been tweaked to deceive the compiler to schedule more efficiently. I'd be hard pressed to believe that such a tweak won't be accepted by the GCC team since it is the CPU specific file being provided for the existing GCC compiler model. (This was along a different thread in which a number of rabid anti-PPC people were accusing IBM of catering to Apple's irrational obsession with secrecy by misrepresenting numbers in their preliminary MD files provided to GCC, which turned out not to be the case.)

This is a far cry from tweaking the GCC compiler model itself which is what the original poster implied! IBM has announced plans to introduce some optimizations which is up to the GCC team to accept or not. Since GCC's design goal is portability, the acceptance of such patches depends on how the patch is framed. Apple, in any case, will probably supply those patches with Developer Tools/XCode even if they are not accepted so there is little practical difference to the Mac developer whichever case turns out.

I don't think there is any evidence that the two tweaks that Apple actually put into GCC during the oft-quoted, much-maligned Veritest benchmark are related to changing the compiler model and would not be accepted as standard "what is necessary to make GCC portable to the G5" patches. However, it seems a lot of press would have you believe otherwise and further cloud the issue by misinterpreting quotes such as the one you give.

In any case, this example as well as the measily register space of the x86 FPU unit are both examples which prove the rule: the compiler plays a huge role in the benchmark and that different compilers have different design goals.

For someone like me, I only care about the GCC benchmark as a programmer/research (or a CodeWarrior vs. Visual Studio one if I'm running apps). A Java developer has different criteria. Obviously SPECmark is not for either of us. Pixar is the opposite because they can choose to compile their products on whatever platform they want, can actually afford to purchase ICC, and their codebase is similar to many of the operations in the SPECfp suite. The fact (not rumor! (http://conferences.oreillynet.com/cs/macosx2003/view/e_sess/4478)) that Pixar is migrating to Mac OS X workstations and the preliminary RenderMan benchmarks coupled with the reality that the 970 is an immature and unoptimized platform bode well for the future of the G5 PowerMac as a graphics workstation and the IBM PowerPC 970 blades (not Macs!) in the renderfarm.

But, to bring this discussion back on topic. There seem to me to be two Pixars: the Pixar that is a movie studio and has a huge renderfarm and workflow that involves internally-developed and externally-purchased products; the Pixar that sells RenderMan. The former is migrating to Mac OS X on the desktop and probably going to keep Intel P4 with Linux (http://cgw.pennnet.com/Articles/Article_Display.cfm?Section=OnlineArticles&SubSection=Display&PUBLICATION_ID=18&ARTICLE_ID=168408) for rendering for quite some time; the latter is testing the waters to see if anyone else is interested in a Mac OS X version of their product.

Take care,

jettredmont

Aug 1, 2003, 07:21 PM

Originally posted by Cubeboy
Actually, we already know that the MD file used was modified over default (which were wrong). So apparently, the "no tweaks over default during testing" statement is false. This leads me to doubt that their weren't any tweaks, especially considering Veritest specifically listed the version of GCC as a specific build as opposed to just GCC.

GCC tested Apple's GCC code. No, it wasn't pulled down off the Web because Apple couldn't submit their G5 gcc tweaks to the main branch until the G5 "existed"!

"Special tweaks" would be code that was specifically introduced for the testing, which will never make it into the mainline gcc branch. There is no evidence of such tweaks, and a flat denial from Apple that any such tweaks were made. Thus, your assertion that Apple must have made such tweaks anyways says a bit more about yourself than about Apple.

From what I understand, IBM and Apple spent a considerable amount of time optimizing the scheduler for the PPC970.

"The gcc scheduler is not really designed ideally for a processor like the 970 and the Power4 and others, and that's a lot of what the IBM and Apple teams have worked on".

GCC is not able to schedule for the Pentium 4 at all, this is really quite important mind you considering nearly all floating point code needs to be well scheduled.

WHAT??? If gcc were unable to schedule for the P4 you would not be able to run P4 code. That's just plain stupid.

Now, back to the amount of time Apple and IBM spent fine-tuning the G5 scheduling code: that amount of time absolutely pales in comparison to the amount of time spent (by IBM amongst others) fine-tuning the gcc scheduling for the P3 and P4!

Do you know exactly how much autovectorization improves Pentium 4 ICC performance over x87 only code in SPECfp?

Can you actually give *ANY* evidence of exactly *HOW* Intel might cheat with ICC?

Evidence? Do your homework. Intel's compiler picks up a few specific patterns for auto-vectorization. Such patterns just happen to be in bottleneck portions of SPECfp. And, they aren't abundant in user code.

You can call that a fortunate coincidence, or you can call it a cheat.

Compare the results of a gcc vs icc bakeoff on SPEC (massive differences) with the results of a gcc vs icc bakeoff on application-based server benchmarks (where they come out pretty much in a dead heat, only a slight margin of victory for icc), and you have yet another bit of evidence that Intel's compiler is geared towards the SPEC benchmark, not towards its users.

Cubeboy

Aug 1, 2003, 09:00 PM

Originally posted by jettredmont
GCC tested Apple's GCC code. No, it wasn't pulled down off the Web because Apple couldn't submit their G5 gcc tweaks to the main branch until the G5 "existed"!

"Special tweaks" would be code that was specifically introduced for the testing, which will never make it into the mainline gcc branch. There is no evidence of such tweaks, and a flat denial from Apple that any such tweaks were made. Thus, your assertion that Apple must have made such tweaks anyways says a bit more about yourself than about Apple.

Did I ever say "special tweaks"? NO, you just assumed I did and made a circular argument out of it right? That the MD file was modified obviously demonstrates that their was at least some modification during and after WWDC that weren't their before WWDC which was the whole point of my statement. You should at least understand that.

WHAT??? If gcc were unable to schedule for the P4 you would not be able to run P4 code. That's just plain stupid.

Since *WHEN* did not having a scheduler ever make it impossible to run code? My god, this is just so wrong.

Have you EVER looked at GCC's source code? Let's see, we have scheduling for Pentium, Pentium Pro, Pentium 2, Pentium 3, K6, K7, and K8 BUT NOT THE PENTIUM 4. Now it might be that Intel did not release enough information to write a remotely good scheduler description but THAT DOES NOT CHANGE THE FACT THAT THE PENTIUM 4 DOES NOT HAVE A SCHEDULER!

Now, back to the amount of time Apple and IBM spent fine-tuning the G5 scheduling code: that amount of time absolutely pales in comparison to the amount of time spent (by IBM amongst others) fine-tuning the gcc scheduling for the P3 and P4!

Okay this just flat out wrong, almost all the time, gcc is tuned by people from the CPU vendor for a particular CPU, Intel does NOT do this, they would rather spend their time tuning ICC and for good reason.

Honestly have you *EVEN* bothered to look at the ChangeLog? Let's see, we have a bunch of K7 only tuning, a bunch of K8 only tuning, some PIII only tuning, ALMOST NO P4 ONLY TUNING and trust me the P4 depends on optimizations ALOT MORE than any of these other cpus. Honestly where do you come up with this sheer and utter BS?

Evidence? Do your homework. Intel's compiler picks up a few specific patterns for auto-vectorization. Such patterns just happen to be in bottleneck portions of SPECfp. And, they aren't abundant in user code.

You can call that a fortunate coincidence, or you can call it a cheat.

I did do my homework, perhaps you should do yours.

http://www.aceshardware.com/Spades/read.php?article_id=25000196

http://developer.intel.com/technology/itj/q12001/articles/art_2.htm

Looking over the documentation, on ICC/IFC 5.0, SSE2 only improved performance 5% over x87 only code in SPECfp. Consider this, a 2 GHz Pentium 4a with code compiled by ICC 7.0 scored 4% better than the same Pentium 4 with code compiled by ICC/IFC 5.0, which according to the documentation would score 5% better than the same processor running x87 only code produced by the same compiler. Thus a 3 GHz Pentium 4 running ICC/IFC 5.0 thats not producing ANY packed/scalar SSE2 code would STILL score over 1000 in SPECfp. Now can you honestly tell me exactly HOW much auto-vectorization improves SPEC?

Compare the results of a gcc vs icc bakeoff on SPEC (massive differences) with the results of a gcc vs icc bakeoff on application-based server benchmarks (where they come out pretty much in a dead heat, only a slight margin of victory for icc), and you have yet another bit of evidence that Intel's compiler is geared towards the SPEC benchmark, not towards its users.

My god, have your EVER seen how a Pentium 4 performs on ICC and GCC, I suppose not, you pulled this little statement out of thin air just like the rest of your "FUD". Here, why don't I just end this little demonstration by showing you some benchies, and than you can **** okay? No really, I'm sick of all this crap thats being posted today.

http://www.willus.com/ccomp_benchmark.shtml?p1

http://www.coyotegulch.com/reviews/almabench.html

http://www.polyhedron.com/
(your going to have to browse around with this one.)

http://www.coyotegulch.com/reviews/intel_comp/intel_gcc_bench2.html

http://www.intel.com/software/produ...er_gnu_perf.pdf

http://www.aceshardware.com/read_news.jsp?id=75000387 (again, NO GCC DOESN'T SCHEDULE FOR THE P4)

GregA

Aug 1, 2003, 09:22 PM

How does a renderfarm work?

I would have thought a renderfarm has an application running on each computer in the farm, with a controlling application that distributes the work for each frame to an available system.

If so, couldn't the renderfarm have their App running on Linux/Intel, Sun, Mac, Linux/970 all simultaneously?

Does anyone know? Does the render farm have to be the same systems or just the same application running on different systems?

Greg
ps. If they had multiple systems, you'd get some good info on the comparative performance of all those systems - which ones render more frames per hour etc, versus hardware cost, versus support cost.

nagromme

Aug 2, 2003, 02:06 AM

Yes--a renderfarm can easily be a mix of several OS's. (As long as the render client software exists for all the computers.)

I've rendered to a mixed batch of Macs with a few PCs mixed in before.

sanjef

Aug 2, 2003, 02:51 AM

If current news about developers' trends is any indicator, we'll most likely find them dumping OS X in favor of something else (Linux?).

Fitzcaraldo

Aug 2, 2003, 03:55 AM

Renderman uses an app called "Alfred"TM to detect free resources and distribute render tasks.

Alfred talks to "Nimby"TM wich is used to limit and or deny resource use on specific workstations.

The biggest obsticle I see to Pixar and the Mac is the lack of Maya Unlimited...

The Pixar render farm is a Mix Match of various bits on Industrial shelving...

Go to:

http://www.pixar.com/howwedoit/index.html

Pull the lever and you will see the farm

You can also see the Renderman logo in B/W which I have seen 'Somewhere' it was the B&W in red on a yellow organic shape.

websterphreaky

Aug 2, 2003, 06:59 PM

This is the most rediculous BS Rumor I've seen recently! Like a business as cut throat as the movie industry would allow a company to dump hundreds of thousands, maybe millions of dollas worth of expensive Intel hardware and server software, only to spend 3 times as much (because Apple hardware is 3 times overpriced) on new Apple servers and server software!

I do professional video and multimedia editing, and there is no way our company would do such an irresponcible financial move as this after one year in this economy!

Perhaps, it's the CEO got a new Apple 17" Ironing board PowerBook and he needs to migrate for his chipped up, scratched and flaking paint G4 Titanium. Now that's reality! :p

tychay

Aug 2, 2003, 07:08 PM

I know you weren't replying to me, but I am reading your comment at the end to imply that I am responsible for "some of the crap that is being posted today".

Originally posted by Cubeboy
Did I ever say "special tweaks"? NO, you just assumed I did and made a circular argument out of it right?

Actually, A different poster referred to "special tweaks". If you look at the previous postings you will note the exact term he used was that GCC has "a ton of special tweaks written by chip vendor IBM." (a pretty damning quote!) I disputed that, you jumped on me, a benchmark war ensued, the guy you're flaming defended my statements and got confused as to who said what.

This mistake is understandable as you took what I said out of context when flaming me.

That the MD file...
My position on the machine description file has already been set. The original arguments I've seen took IBM to task on an incorrect machine description file because it was believed that changes in the file were deliberate FUD on behalf of Apple, it had nothing to do IBMs right, as a chip vendor, to modify a MD file to improve performance. This was later misinterpreted when the "oft-cited, much-maligned" Veritest benchmark war. It turned out to be that the file needed to be tweaked because gcc's CPU model is not a good model for the 970. IBMs explanation for the late change sounds very reasonable--their explanation was roughly equivalent to "we were tweaking it because a MD file isn't as easy as looking at a spec sheet as these people would have you believe."

(You and I both know that there is a world of difference between tweaking a machine description file and tweaking the GCC compiler's CPU model itself. We both know that the P4's MD file has far more commits than the 970's.)

scheduling stuff deleted

I ignored this stuff because it is true and I've never disputed this. We don't know why an optimal scheduler isn't available for the P4 in GCC. There is no need to shout about this and drive an honest discussion into the uncivil.

Even if such a thing is submitted we don't know what the improvement will be. I'm inclined to believe that there are a number of factors in addition to the lack of documentation coming out of Intel. Remember, the P4 of today is a completely different beast inside than the earlier models (hyperthreading, etc): these might not map well to the GCC CPU model either.

The difference is that IBM is committed to convincing GCC to improve the model and Intel isn't. This may mean that in the future GCC may become biased toward PPC, but that doesn't refute the present reality that GCC is a better compiler for x86 than PPC. (I'm on record, many times, as believing that using GCC to "normalize out" the compiler for SPEC is misguided: SPEC is a system benchmark that tests CPU, memory, and compiler (and to little extent: bus, OS and other components)). It always bothered me that the average computer user has even heard of it when making a purchase.

Okay this just flat out wrong, almost all the time, gcc is tuned by people from the CPU vendor for a particular CPU, Intel does NOT do this, they would rather spend their time tuning ICC and for good reason.

Actually, his statement is correct. The amount of time spent tuning that Apple/IBM have done for the 970 does "pale" in comparison to the amount of tuning done on for the x86. As evidence, note how good the performance is of gcc3.3 vs. ICC for the P3. In the PowerPC/POWER world, gcc has not achieved close to parity with CodeWarrior or IntelliAge and there are a lot less developers working on and with gcc for the PowerPC. Again, this will change because IBM's stated commitment to open-source and Apple's obvious dependance on gcc as the only Objective-C compiler around (weighed against the GCC team's biases against accepting any changes which affect portability).

I'd even bet that more time was spent tuning for the P4 specifically than the 970 in gcc. The P4 has been out for a while and is the probably the second most used CPU (after the P3) for Linux. The problem here is a noticeable lack of documentation on how to go about doing such tuning. The trick of having hyperthreaded double the apparent number CPUs available to the kernel alone must have taken a good bit of time. It was a hack that has since been fixed, but that doesn't mean it didn't take a lot of time.

Looking over the documentation, on ICC/IFC 5.0, SSE2 only improved performance 5% over x87 only code in SPECfp.

Whoa, you (and AMDZone) are misreading your own cites. First, the 5% performance gain is specifically due to instructions added when they jumped between MMX/MMX2 (SIMD in Pentium II and III) and SSE/SSE2 (SIMD in Pentium IV). Second, the fact that there is a gain at all points to autovectorization being done. (Now I agree with you that I've been guilty of referring to autovectorization when I generically mean autovectorization, autoparallelization, and other CPU modelling optimzations.) Third, this performance gain will increase in later versions of ICC as the Intel folks figure out more places the new instructions create benefits.

My god, have your EVER seen how a Pentium 4 performs on ICC and GCC, I suppose not, you pulled this little statement out of thin air just like the rest of your "FUD".

Those are more examples that reinforce the Pixar statement (50% speed gain on P4 with ICC vs. GCC). The optimizer in ICC is really good (by those statements) and really mature (by the fact that Intel's compiler shows better results than GCC with even AMD's chips). I should note for the others not willing to sift through all your cites that there are a couple tests where GCC benchmarked in rough parity or better than ICC.

I never claimed that the ICC optimizations only benefit SPEC (others may have). My guess is the biggest gains are not AV or AP at all but are the use of a lookup table for trigonometric functions in ICC. LibMoto (A math library Motorola made for the PowerPC) used to do the same thing and would pump up the FP marks in old Mac benchmarks by 80%, but because of incompleteness of the tables, it would affect the stability of some video games which depended on the accuracy of the numbers. I find it doubtful that the GCC team would accept such changes even if they were offered.

Some of your cites actually reinforce a significant speed gain with ICC vs. GCC regarding P4 SPEC2000. That does reinforce my statement that 1) You cannot "normalize out" the compiler in SPEC benchmarking as Apple claims and 2) it is misleading to report Apple's SPEC numbers side-by-side with standard SPEC2000 benchmarks.

tychay

Aug 2, 2003, 07:29 PM

Originally posted by websterphreaky
This is the most rediculous BS Rumor I've seen recently!

Too bad for you, it's true.

Like a business as cut throat as the movie industry would allow a company to dump hundreds of thousands, maybe millions of dollas worth

The rumor never said that. You, like many others, don't understand the difference between desktop workstations and a renderfarm.

...of expensive Intel hardware and server software...

The new server (1024 Xeon CPUs in their renderfarm) hardware is leased. The server software is currently Linux and free so I hardly think of it as costing "hundreds of thousands" of dollars to migrate, which they're not.

Their desktop hardware is either Linux or NT. If the latter, it is definitely "not free". Also their workflow scripts were originally written for IRIX and porting to NT must have caused a big headache if that were the case. Porting to Mac OS X is relatively painless and the painful parts make a subject of their talk at Mac OS X Developer Conference in two months.

...only to spend 3 times as much (because Apple hardware is 3 times overpriced) on new Apple servers and server software!

Doh! Obviously this is flame by a Wintel idiot. I've been punked!

The rumor never claimed that they will be buying any Apple servers or Mac OS X Server licenses (their server software).

I do professional video and multimedia editing, and there is no way our company would do such an irresponcible financial move as this after one year in this economy!

Read as: I'm losing money now that all those iMovie/iDVD 'doits are undercutting what I overcharge for wedding videos I put together with Premiere on my PeeCee.

Perhaps, it's the CEO got a new Apple 17" Ironing board PowerBook and he needs to migrate for his chipped up, scratched and flaking paint G4 Titanium. Now that's reality!

The same person who used a IBM Thinkpad his first two years as CEO of Apple? The same person who is nowhere to be seen at Pixar's offices just across the Bay that the Pixar folk joke about it in their DVD extras?

(And yes, I happen to own a scratched and flaking paint G4 Titanium Powerbook like the one you mention. If, during the life of it, the only complaint I have with this thing is cosmetic, I don't know whether to be happy or pissed off that I bothered to buy AppleCare.)

Cubeboy

Aug 2, 2003, 10:37 PM

Originally posted by tychay
I know you weren't replying to me, but I am reading your comment at the end to imply that I am responsible for "some of the crap that is being posted today".

My original thought was that the final statement would be the perfect clincher to a otherwise mediocre post. Please excuse my ignorance to the otherwise readily obvious. :D

Actually, A different poster referred to "special tweaks". If you look at the previous postings you will note the exact term he used was that GCC has "a ton of special tweaks written by chip vendor IBM." (a pretty damning quote!) I disputed that, you jumped on me, a benchmark war ensued, the guy you're flaming defended my statements and got confused as to who said what.

This mistake is understandable as you took what I said out of context when flaming me.

Personally, I consider my original reply to be quite mild in light of my more recent posts, most of it was just stating my views on the matter, the last two parts were actually more related to a one of our previous disputes than what you might call "flaming". I agree with you that the mistake is quite understandable, I just don't believe that (falsely) accusing someone in a particularly condescending tone should be held in the same light as a "understandable" mistake. Keeping that in mind, would you consider my response to be too strong?

My position on the machine description file has already been set. The original arguments I've seen took IBM to task on an incorrect machine description file because it was believed that changes in the file were deliberate FUD on behalf of Apple, it had nothing to do IBMs right, as a chip vendor, to modify a MD file to improve performance. This was later misinterpreted when the "oft-cited, much-maligned" Veritest benchmark war. It turned out to be that the file needed to be tweaked because gcc's CPU model is not a good model for the 970. IBMs explanation for the late change sounds very reasonable--their explanation was roughly equivalent to "we were tweaking it because a MD file isn't as easy as looking at a spec sheet as these people would have you believe."

(You and I both know that there is a world of difference between tweaking a machine description file and tweaking the GCC compiler's CPU model itself. We both know that the P4's MD file has far more commits than the 970's.)

I think you misunderstood what I have said in my previous two post. The entire point of me including the MD file in the debate at all was to prove that there were indeed modifications in GCC that were there during and after WWDC that weren't there before. The intent was to prove a particular statement false which is why I never discussed the MD file in detail.

I ignored this stuff because it is true and I've never disputed this. We don't know why an optimal scheduler isn't available for the P4 in GCC. There is no need to shout about this and drive an honest discussion into the uncivil.

Even if such a thing is submitted we don't know what the improvement will be. I'm inclined to believe that there are a number of factors in addition to the lack of documentation coming out of Intel. Remember, the P4 of today is a completely different beast inside than the earlier models (hyperthreading, etc): these might not map well to the GCC CPU model either.

Well, I never shouted about the scheduler in the first place so I don't really know exactly what your getting at:

"From what I understand, IBM and Apple spent a considerable amount of time optimizing the scheduler for the PPC970."

I did type in CAPITALIZED LETTERS in my response to Mr Redmont's flames and false accusations, if thats what you mean, I assumed it was acceptable for the occassion.

Again, the reason I believe a "optimal scheduler" isn't available for the Pentium 4 goes back to my original point, which I will be discussing shortly. Having well scheduled code is really quite important performance-wise, this is especially true for floating point code which ironically (or coincidentally), is where the Pentium 4 took the biggest performance hit over ICC.

The difference is that IBM is committed to convincing GCC to improve the model and Intel isn't. This may mean that in the future GCC may become biased toward PPC, but that doesn't refute the present reality that GCC is a better compiler for x86 than PPC. (I'm on record, many times, as believing that using GCC to "normalize out" the compiler for SPEC is misguided: SPEC is a system benchmark that tests CPU, memory, and compiler (and to little extent: bus, OS and other components)). It always bothered me that the average computer user has even heard of it when making a purchase.

Once again, this goes back to my original point, Intel is not commited to "improving the model" for GCC and hasn't been since it released ICC 5.0 which came out in the same time frame as the Pentium 4. I will be discussing this more in detail in the paragraph below. I don't believe most average computer users have heard of SPEC, at least not until WWDC where it was put on display.

Note that I segmented my response to your post into two parts as a single large post would exceed the word count.

Cubeboy

Aug 3, 2003, 10:08 AM

Actually, his statement is correct. The amount of time spent tuning that Apple/IBM have done for the 970 does "pale" in comparison to the amount of tuning done on for the x86. As evidence, note how good the performance is of gcc3.3 vs. ICC for the P3. In the PowerPC/POWER world, gcc has not achieved close to parity with CodeWarrior or IntelliAge and there are a lot less developers working on and with gcc for the PowerPC. Again, this will change because IBM's stated commitment to open-source and Apple's obvious dependance on gcc as the only Objective-C compiler around (weighed against the GCC team's biases against accepting any changes which affect portability).

I'd even bet that more time was spent tuning for the P4 specifically than the 970 in gcc. The P4 has been out for a while and is the probably the second most used CPU (after the P3) for Linux. The problem here is a noticeable lack of documentation on how to go about doing such tuning. The trick of having hyperthreaded double the apparent number CPUs available to the kernel alone must have taken a good bit of time. It was a hack that has since been fixed, but that doesn't mean it didn't take a lot of time.

Yes, more time was spent tuning for x86 (note that GCC can schedule for almost every x86 processor) but that doesn't mean more time was spent tuning the Pentium 4. As I've stated in my previous post, almost all GCC optimizations are submitted by people from the CPU vendor for a particular CPU. You've agreed that Intel isn't very commited to submitting improvements to GCC. I pointed out they haven't been commited since releasing the Pentium 4 and ICC 5.0 and this neglect resulted in the Pentium 4 being very poorly optimized for GCC. This was proven from my reference to the ChangeLog, which shows lots of tuning for other x86 processors like the Opteron and Athlon (which might explain why a Opteron running GCC compiled code scores within 5% of the same chip running ICC compiled code in SPEC) but almost no tuning for the Pentium 4.

Regarding hyperthreading, it's not just modified kernels that "sees" two logical processors on a Pentium 4, most software will "see" the same thing. SMT or hyperthreading (parallel set of registers and logic) basically "creates" two "virtual" cores that run independently on a single chip. Now, these "virtual" cores do share alot of the same cpu components (cache, ALU(s), FPU, SIMD unit, code-decode unit, hence the serious resource contention problems), but again, two seperate independent cores.

Whoa, you (and AMDZone) are misreading your own cites. First, the 5% performance gain is specifically due to instructions added when they jumped between MMX/MMX2 (SIMD in Pentium II and III) and SSE/SSE2 (SIMD in Pentium IV). Second, the fact that there is a gain at all points to autovectorization being done. (Now I agree with you that I've been guilty of referring to autovectorization when I generically mean autovectorization, autoparallelization, and other CPU modelling optimzations.) Third, this performance gain will increase in later versions of ICC as the Intel folks figure out more places the new instructions create benefits.

Here's the quotes:
"For SPECfp2000 the new SSE/SSE2 instructions offer about a 5% performance gain compared to an x87-only version."

"As the compiler improves over time the gain from these new instructions will increase."

I don't really see the point of this dispute, x87 code is regarded as standard when benching these CPUs (Athlons, Pentium 4s and Opterons all have x87 FPUs), the main concern was that the Pentium 4 could run packed SSE2 code produced by ICC much faster than than x87 only code (and this is one of the more widely believed reasons as to why the P4 running GCC compiled code fared so poorly in SPECfp), which I've just proven to be false.

I never doubted that ICC didn't have auto-vectorization (I made this clear when I asked you how much autovectorization boosted performance in my first response). I was questioning if it could account for a significant boost in SPECfp scores.

I've already shown in previous threads and my previous post that the latest version of ICC (7.0) only improved overall SPECfp score 4% over ICC 5.0, and this is from official SPEC submissions. Either way, the gain is not that significant.

Those are more examples that reinforce the Pixar statement (50% speed gain on P4 with ICC vs. GCC). The optimizer in ICC is really good (by those statements) and really mature (by the fact that Intel's compiler shows better results than GCC with even AMD's chips). I should note for the others not willing to sift through all your cites that there are a couple tests where GCC benchmarked in rough parity or better than ICC.

I never claimed that the ICC optimizations only benefit SPEC (others may have). My guess is the biggest gains are not AV or AP at all but are the use of a lookup table for trigonometric functions in ICC. LibMoto (A math library Motorola made for the PowerPC) used to do the same thing and would pump up the FP marks in old Mac benchmarks by 80%, but because of incompleteness of the tables, it would affect the stability of some video games which depended on the accuracy of the numbers. I find it doubtful that the GCC team would accept such changes even if they were offered.

Some of your cites actually reinforce a significant speed gain with ICC vs. GCC regarding P4 SPEC2000. That does reinforce my statement that 1) You cannot "normalize out" the compiler in SPEC benchmarking as Apple claims and 2) it is misleading to report Apple's SPEC numbers side-by-side with standard SPEC2000 benchmarks.

True, there were a few statements that support evidence of a trigonometric functions lookup table and auto-vectorization (which I've already acknowledged from the start) but the general feeling and conclusions of all the cites I've listed was that ICC produced significantly faster code for the Pentium 4 than GCC, which was the main reason I included them.

Regarding ICC's lookup table, would a Opteron/Athlon recieve the same benefits by using ICC? This is considering a Opteron running ICC compiled code offers only a marginal increase in SPEC score to the same cpu running GCC compiled code.

To what degree would it change the machine output? I know that SPEC is very strict about making sure that the output for their benchmark suite remains unchanged.

sparkplug

Aug 5, 2003, 10:50 PM

If they moved to OSX it would only be for workstations.

This is highly unlikely, what front end hardware are they going to run? The only 3D gfx cards available are gaming cards, IE geforces and Radeons. That is all that is available for the mac, their are no pro 3d gfx for the mac period. In particular no Quadro FX, no quadro at all for that matter, so no Cg shaders. They would make decent render engines however, but no pro animator is going to want to cripple there workflow by using a "workstation" that can only use a game card.

Maya 5 with real time rendering/Cg shaders on a Quadro fx is a thing of joy. You cant get that on the G5, for any amount of money. Basically there are three things needed for a professional 3d workstation

1/ a stable multitasking operating system
2/ fast cpu with plenty of memory, the more the merrier
3/ fast high quality/accurate graphics subsytem

The G5 has the first two.......

patrick0brien

Aug 6, 2003, 09:32 AM

Originally posted by sparkplug
The G5 has the first two.......

-sparkplug

That's a very good point.

Write to this link: http://www.apple.com/macosx/feedback/

Perhaps we can get Apple to write us the drivers so we could use such cards.

Mac Kiwi

Aug 6, 2003, 09:55 AM

I am almost convinced that Pixar already has these drivers to tell you the truth.As to when the rest of us can get them would be nice.It would also be nice to know who writes them Apple or Nvidia etc.A nvidia driver writer tells me Apple do,but I have a report that an ATI guy told someone they do,so who knows.I dont really care who writes them I want a Quadro.

I was also told by a friend today that works for a game company that Pixar have ordered hundreds of G5s.I cant back this up with any facts other then a friend in a game company,but he seemed fairly sure.I am a little skeptical myself but I suppose we will see soon enough.If they have ordered hundreds it might explain some of the delay for the rest of us as well.

Stu.

capitalhood

Aug 6, 2003, 10:28 AM

are you all saying that pixar doesnt use macs?! but how can that be... pixar has steve so whyt dont they use macs... plus there always doing **** with apple... the jaquar logo... the g5 video... all kinds o' stuff... woow.

patrick0brien

Aug 6, 2003, 12:04 PM

Originally posted by capitalhood
are you all saying that pixar doesnt use macs?! but how can that be... pixar has steve so whyt dont they use macs... plus there always doing **** with apple... the jaquar logo... the g5 video... all kinds o' stuff... woow.

-capitalhood

At the risk of sounding blunt: Have you read the entire thread? All of the information to answer your questions is here.

agreenster

Aug 6, 2003, 12:41 PM

Originally posted by Fitzcaraldo
The biggest obsticle I see to Pixar and the Mac is the lack of Maya Unlimited...

No, not really. Pixar only uses Maya for modeling, not animation. They have proprietary software called 'Marionette" (spelling?) for animation. Anything that Unlimited would do (fur, liquid) is all done in house via renderman and their effects department.

Originally posted by sparkplug Maya 5 with real time rendering/Cg shaders on a Quadro fx is a thing of joy

Are you speaking from experience? If so, I am totally jealous.

Actually, check out this (http://www.uemedia.com/CPC/printer_9010.shtml) website--it hints that nVidia is working on developing their vidCard and all its corresponding coding for OSX...

sparkplug

Aug 8, 2003, 01:27 AM

Thanks for the link, yes thats a good overview. Yes I am speaking from experience, I've been using Cg shaders since the beta and am completely in love with there integration in maya5 and with the FX. This is "must have" technology for me, and indeed I would think for many many people once they have used it and seen what can be done, It wasnt very long ago when this level of graphics could only be purchased as a small fridge sized unit minimum. I have some of that stuff knocking around collecting dust. :rolleyes:

I wrote the author of that article, hopefully he can tell me something concrete because there is no word from any of my other sources on the existance of such support, Id be very happy to learn that there was.

MacRonin

Aug 15, 2003, 10:54 AM

Bump?!?

shiftysands

Dec 2, 2004, 05:25 AM

Could the unnamed client be Pixar? :rolleyes:
Does anyone know what a "hybrid grid" means?

Hi Miguel,
We are currently deciding between a high end NVidia card, a established hardware raytracing solution and a hybrid grid rendering solution. The likelihood is that I will ship with all of these since they each have their good points.

Fireball 999 is developing an imaging solution for a private client - while retaining design rights to the system and software for sale to other imaging companies (This resale facility becomes available in mid 2006 to give the main client time to capitalise on their investment in the design and software).

The problem dynamic is one of integrating studio technology under a unified grid. This grid will enable a company of 800 employees to work seamlessly on the production of a 90min animated and live action hybrid film. The client wants to keep a common format and workflow between all stages of the production environment and is leveraging the best in consumer level hardware to achieve professional calibre work (this maintains a symbiotic relationship between the mass production and low cost of the consumer market and the high demand top fidelity requirements of my client).

The initial imaging solution will ship for around $12million as a suite of several computers each with a CRT monitor and TV. These computers link together to form the skeleton of the rendering grid in each studio and link directly to an offsite render farm (Provided by Fireball 999 and paid for by the client on a per TeraFragment rate). I would be looking to include a couple of your highend nvidia cards in each computer along with the hardware raytracing card and the grid rendering software.

Other features of the package are optional and include motion capture, laser scanning booths and camera rigs, but each of these options dramatically increases the cost of the imaging solution.

We am hoping that falling technology prices will bring this imaging solution in the reach of the high-end production TV studio's budgets by 2006 at a price tag of under $3million per studio, and within the decade we would like to see this package become the norm for game studios at the merry price of $0.2million.

Ideally one card would output straight to DVD resolution approx 800x1200, 16xAA with bump mapping, spherical harmonics and specular highlights (implimented as HDRI map) (24fps) and the other would output to TV resolution with interlaced fields 50fps.

Please recommend which cards may be appropriate (to ship in May 2005).

The current business plan has a target of 50 sales over the next 5 years, but each sale would be resourced and built to the customers exact requirements - at this stage I am only putting together the prototype systems to test the software on (The software has finished architecture and initial infrastructure but we have not licensed the rendering or modelling software yet (probably Mental Ray and Maya)).

I want a quote from you per system for the dual output system (DVD and TV resolutions). If this is not something NVidia has an interest in please recommend a specialist engineering firm that may be able to help with an original hardware solution.

many thanks,

Tabitha Ben
Biz Dev
Fireball 999

JFreak

Dec 2, 2004, 07:41 AM

Still a tad confused. Pixar obviously knew all about the G5 when they decided their render farm was going to be intel blades. And as far as I can tell, they are still goling that way for rendering. Sort of implying that while they may choose G5's for workstations, when flat out constant speed is needed, intel is still the way to go.

pixar knew about the cutting-edge G5 before we did, that's for sure, but pixar is a kind of company that cannot be the first to adopt such tech. now they know enough about it and have had time to play with it, so if they indeed have recognized its capabilities and have begun porting their SOFTWARE to the osx platform, they will have it ready and tweaked by the time they buy new HARDWARE next time.

last time they weren't ready for the osx switch, next time they will be if they want to. only time will tell what they are really doing...

JFreak

Dec 2, 2004, 07:44 AM

it is probably quite a bit cheaper buying 2000 Xeons or whatnot than G5s

hardware purchase is peanuts for this kind of company. it is the software that decides what will be required to run it. last time the software wasn't ready for G5 hardware, so it was impossible to consider apple at all.

(and even if they had it "ready", it wasn't properly tested and optimized, which is really just as important.)

Mr. Anderson

Dec 2, 2004, 08:49 AM

so why the necromancy? there's nothing new here and no new news....

D

Potus

Dec 3, 2004, 08:04 AM

'
Other than the news that IBM is selling its hardware co.

davegoody

Dec 3, 2004, 10:58 AM

TranceClubMusic

Dec 4, 2004, 03:30 PM

I was sooooo excited about the whole "MAC OS X" Switch, Steve Jobs owns Apple and Pixar, the "Renderman" Program, etc....
I also just got a New 2.5 G5 (switched from the PC World) and all my family and friends are in shock over the change.
Of course, I did ALL this bragging about Apple, Pixar, MAC OS X, etc...
Only to have them all go to see The Incredibles at the movies (of course I told them - stay through ALL the credits)....Now? I have HUGE Egg on my face. Laughing Stock of the Neighborhood and everyone tells me (had to also see it for myslef) the HUGE INTEL LOGO at the end of the credits!
:eek: So much for bragging rights :rolleyes:
Now all I hear is.......So how does Steve Jobs like using PeeCees with Intel Processors? :mad:

PS - Is it me or is Apple more full of hype then anything else.

PSS - 20th Anniversary was the WORST I have ever experienced. :rolleyes:

asif786

Dec 4, 2004, 03:45 PM

Now? I have HUGE Egg on my face. Laughing Stock of the Neighborhood and everyone tells me (had to also see it for myslef) the HUGE INTEL LOGO at the end of the credits!
:eek: So much for bragging rights :rolleyes:
Now all I hear is.......So how does Steve Jobs like using PeeCees with Intel Processors? :mad:

PS - Is it me or is Apple more full of hype then anything else.

PSS - 20th Anniversary was the WORST I have ever experienced. :rolleyes:

Lol, that is bad. Dont forget, this is a rumour site.. :p

/asif

Blue Velvet

Dec 4, 2004, 03:49 PM

I like him a lot, and he's done some great movies...

Showgirls.

?!

asif786

Dec 4, 2004, 04:59 PM

?!

Sometimes it's better just not to ask... :rolleyes:

/asif

Dr. No

Dec 4, 2004, 11:07 PM

So Pixar never switched????

angelneo

Dec 5, 2004, 01:22 AM

Mac Kiwi

Dec 5, 2004, 03:21 AM

asif786

Dec 5, 2004, 04:22 AM

It takes time for them to switch, they are probably halfway through the production of the The Incredibles when they reach a decision of porting over to G5 last year.

Exactly. I think I recall steve saying each Pixar film takes 4 years to make - they wouldnt interrupt that schedule just to switch computers. Perhaps 'Cars' will have an Apple logo at the end :)

/asif

Mac Kiwi

Dec 5, 2004, 05:12 AM

The render Farm is Intel,so that will be why the big Intel logo,but to change the render farm over would cost mega bucks,even for Pixar.

TranceClubMusic

Dec 5, 2004, 10:09 AM

Exactly. I think I recall steve saying each Pixar film takes 4 years to make - they wouldnt interrupt that schedule just to switch computers. Perhaps 'Cars' will have an Apple logo at the end :)

/asif

Ummmmm, 4 years ago - didnt Apple have PowerMac G4's? Tells me ALOT about Apple when Steve Jobs doesnt and hasnt even been using HIS VERY OWN PRODUCTS!!!! The ones that are soooooooo much better than the "dark side"?? Goes to show that if a huge movie production company never felt comfortable using Apple products.......what makes "comsumers" believe the switch is sooooo important?
To think, so many people on this forum (still to this day) compare the G4 processor (still in Powerbooks I might add) to Intel!! Yet the very person who owns Apple would never use it in his OWN company pixar??
Very interesting to me.......sounds like lies and hype. :(

TranceClubMusic

Dec 5, 2004, 10:11 AM

The render Farm is Intel,so that will be why the big Intel logo,but to change the render farm over would cost mega bucks,even for Pixar.

Why would the render farm even of had STARTED with PeeCee's when the owner had the ALMIGHTY G4?? :rolleyes:

TranceClubMusic

Dec 5, 2004, 10:12 AM

Actually you were half right,they do use G5s as workstations for modeling,and the PC users dont like the slower Open GL we have.

Feel better now? :)

Sorry dont feel better now - just goes to show how Steve Jobs lies. Hey Steve Jobs.....Products are sooooooo great - why dont you use them! :p

asif786

Dec 5, 2004, 11:18 AM

Why would the render farm even of had STARTED with PeeCee's when the owner had the ALMIGHTY G4?? :rolleyes:

Pixar is a business. The fact that steve runs pixar and apple makes no difference - it doesn't mean they are going to replace their renderfarm when thy dont even need it.

You seem to think that Steve can walk in one day and say 'Let's use Apple products, cos I run the company'. It doesn't happen like that. It takes research. Steve never said G4's were supercomputer quality. But G5's are, and thats why Pixars switching.

To be honest, the tone and attitude of your posts haven't impressed me. I look forward to your insight when your running a multi-billion dollar animation studio.

TranceClubMusic

Dec 6, 2004, 08:32 AM

Pixar is a business

WOW! Correct you are! Notice the word "Business" and notice the use of Intel in "Business" not Apple Products :p

The fact that steve runs pixar and apple makes no difference

:eek: Thats a shocking thing to say! The Owner of Apple makes NO difference that he owns Pixar? Thats like Bill Gates using an Apple G5 on his Desk at home. I mean based on that thinking it makes no difference that he owns Microsoft but instead uses the competitions products at home! :rolleyes: Come on, think about it....Steve Jobs every single year in January touts his products and compares them to Intel!! :cool:

It doesn't mean they are going to replace their renderfarm when thy dont even need it.

Another wonderful statement. Yes why replace it! I mean they dont need it they already have the best. :p

You seem to think that Steve can walk in one day and say 'Let's use Apple products, cos I run the company'. It doesn't happen like that.

Yes it does and Yes it can! :D

It takes research. Steve never said G4's were supercomputer quality.

LMFAO - Go back about 4 years and listen to some Keynotes. G4's back then (and still are in many ways) Super Computer Quality. If you made that statement 4 years ago on the Forum you would be eaten alive!!!

But G5's are, and thats why Pixars switching.

I hope so - really waiting for that transition. I really wanna see Apple Products in more Business. :D

To be honest, the tone and attitude of your posts haven't impressed me. I look forward to your insight when your running a multi-billion dollar animation studio.

Ditto!! :)

asif786

Dec 6, 2004, 08:44 AM

TranceClubMusic

Dec 6, 2004, 08:50 AM

Oh dear, I think you need to relx and have a cup of coffee. Perhaps even take a chill pill.

You can let me know by PM when you're feeling better ;)

Lord knows why this oooold thread was dragged up anyway.

I still loves ya man! :D

misfit68

Dec 6, 2004, 04:26 PM

The other studio showing Starship troopers was most likely Tippett Studios out of Berkeley, Ca.

The movie is available on DVD.

I was browsing www.cgtalk.com today, reading up on Siggraph 2003 and came across this snippet aparently taken from Maccentral, though I cannot find where exactly:

I'll give you guys the **BIG NEWS** that was announced yesterday at Apple's Shake Users' Event in San Diego.

First, Pixar announced that they're bringing prMan over to OS X and will have a beta sometime in August or September.

Second, Pixar announced that they'll be migrating over to... guess what... the Mac!

They based their decision on issues such as performance, code portability, X11 integration, development culture, vender dedication to the fim industry, app availability, audio/video friendlyness and capabilities.

In 1980 Pixar made its first short on a Vax running VMS. Then in the mid '80s they migrated over to Sun workstations running Solaris. In the late '80s - '90s they were running entirely SGI Irix workstations, and in 2000 they moved over to Intels running Linux. An NT migration would've meant exiting the Unix market. But apparantly they were also having issues with Linux such as universal cut & paste, issues synching sound with picture, I think he said color management and other problems. He (a Pixar VP) said that while none of the issues were show stoppers or significant on their own, and they could all be dealt with or worked around, taken as a whole they presented real problems for the company.

Moving to OS X let's them essentially recompile the core functionality of their entire code base without even bothering with higher level UI development. Once they're settled in, then they could port the UI from X to Aqua.

During his slide show, the Pixar VP of something was about to present a list of why they'll be switiching to the Mac. But RIGHT at that time Keynote crashed out, drawing sarcastic applause from the audience. The timing just couldn't be worse. I think it may have been some kind of data corruption in one of the elements in the frame that crashed Keynote if you went BACK to the page from the next page. So from then on the presenters made sure not to go BACK to that particular frame.

Then a Hollywood-based motion design studio called Yu+Co showed their reel and the type of work they do on Shake. Their projects included The Hulk, Italian Job, X-Men2, and many others.

Then another company whose name I don't remember showed some shots of the sequel to Starship Troopers. This particular demonstration of Shake's expansion capabilities was quite impressive. They exported footage from Maya maintaining all the "DOD" data (if anyone knows what this is, let me know... a buncha visual fx artists from work had no idea today) when they used in Shake to create 3D ambient & spot lights. The guy, who had previously used Shake only on a dual AMD box, created a light which he could position anywhere in the scene, including in front, behind, or within the insect. The light behaved just as it would in a 3D environment! He had control over its color temperature, spread, intensity, diffussness, etc... it was quite impressive. He even made a little comment "I've only used this on my dual AMD, this is the first time I've used Shake on a Mac and it's quick [as he's moving the light around the insect]. Yes, my box doesn't do it nearly this fast."

Plus a nice RenderMan OSX screenshot:http://home.comcast.net/~zeio/sig/3.jpg

All sounding pretty sweet if you ask me!