Itanium 2 and P4, could they beat the G4?

[gjohns01]

Re: Re: Itanium 2 and P4, could they beat the G4?

Originally posted by mc68k


Even if a P4 beat the G4 at 10GHz, the fact remains:

It makes absolutely no difference if your machine is "faster" if you are blankly looking at the screen, trying to figure out what to do next. For most users in most tasks, a consistent interface, ease of use, and easy to learn software are much more important than raw horsepower.

Windows is a poor OS, so end users spend less time utilizing the CPU, and more time fooling around.

The MacOS is a superior OS where you might not have as much power, but you spend more time utilizing the CPU due to the above mentioned facts.

Here is an excellent article about Mac vs. PC performance:

http://homepage.mac.com/mac_vs_pc/6.html

I learned something, I hope you do to.

I use both OSes. Why is Windows a poor OS? Because you don't know how to use it? Have you used it? I experience no loss in production switching between the two OSes. You can't figure out how to open an application? Can't make the switch between Command+Q and Alt+F4? You should try not to make blanket statements like that. What makes MacOS so superior? I could point you to a few computer users who would try the MacOS and not have a clue. So much for ease of use. Is it easier for you to read an analog clock or a digital one? The OSes do the same thing; it's all a matter of what you're comfortable with.

I wouldn't read too much into a Mac vs. PC article listed on mac.com. Numbers be damned. "Real World" performance will show you a P4, Athlon, Xeon, Itanium matches or flat out whips the G4 at any task. It will only get worse as they push ahead with clock speed. Btw, Apple should benchmark more than just Photoshop and they should throw their dual against a Dual Athlon (the never use athlons) or a Dual Xeon (not a single processor machine). Then see how many photoshop filter tests they win. You do realize that Intel will be going from 2ghz to 3ghz in a years time. How superior is that G4 again? Seems to have growing pains. Does the G5 even exist? We are talking Motorola here. They can’t manufacture for s**t.

 

[jefhatfield]

Re: Re: Re: Itanium 2 and P4, could they beat the G4?

Originally posted by gjohns01


I use both OSes. Why is Windows a poor OS? Because you don't know how to use it? Have you used it? I experience no loss in production switching between the two OSes. You can't figure out how to open an application? Can't make the switch between Command+Q and Alt+F4? You should try not to make blanket statements like that. What makes MacOS so superior? I could point you to a few computer users who would try the MacOS and not have a clue. So much for ease of use. Is it easier for you to read an analog clock or a digital one? The OSes do the same thing; it's all a matter of what you're comfortable with.

I wouldn't read too much into a Mac vs. PC article listed on mac.com. Numbers be damned. "Real World" performance will show you a P4, Athlon, Xeon, Itanium matches or flat out whips the G4 at any task. It will only get worse as they push ahead with clock speed. Btw, Apple should benchmark more than just Photoshop and they should throw their dual against a Dual Athlon (the never use athlons) or a Dual Xeon (not a single processor machine). Then see how many photoshop filter tests they win. You do realize that Intel will be going from 2ghz to 3ghz in a years time. How superior is that G4 again? Seems to have growing pains. Does the G5 even exist? We are talking Motorola here. They can’t manufacture for s**t.

i am a microsoft certified professional and windows is how i make my make my living...but i used to be a mac only person who hated microsoft...but if windows was a product as good as the mac os and as easy to use, then i would have to find a different way to make a living

so many professional and intelligent people still do not know how to do the simplest things with windows and even when i charge them less to troubleshoot and tutor my clients on windows than my competition, working with microsoft's operating system is an unlimited source of wealth for people in my field

...i have a friend, self taught in windows and troubleshooting, that bought a house with the money made from his tech business he built around troubleshooting windows and the problems people had with it...before that, he was a broke musician and young theatre artist/actor

and he made that money to get a house in america's most expensive real estate market in 4 years

to be dipomatic, i say windows is good but mac os is better

if windows was truly horrible, then it would not exist in this market or any other

and what the heck do you think the whole government vs microsoft thing was about...besides microsoft shafting netscape with a free ie, the next most common government objection to microsoft was how they copied apple's interface with windows

while i don't consider "pirates of silicon valley" the definitive history of apple and microsoft, the movie did make a good point in showing bill's ripoff of apple's os

in conclusion, windows is "bad" enough to confuse enough people that there is money in tutoring and troubleshooting windows otherwise people would not get rich off of working in and around microsoft's operating system

if i only worked on the mac os, i would not make that much money...so thank god for microsoft's fallible operating system...it pays our bills and 1.5 million microsoft certified professionals love you, bill gates

 

[gjohns01]

Re: Re: Re: Re: Itanium 2 and P4, could they beat the G4?

Originally posted by jefhatfield


i am a microsoft certified professional and windows is how i make my make my living...but i used to be a mac only person who hated microsoft...but if windows was a product as good as the mac os and as easy to use, then i would have to find a different way to make a living

so many professional and intelligent people still do not know how to do the simplest things with windows and even when i charge them less to troubleshoot and tutor my clients on windows than my competition, working with microsoft's operating system is an unlimited source of wealth for people in my field

...i have a friend, self taught in windows and troubleshooting, that bought a house with the money made from his tech business he built around troubleshooting windows and the problems people had with it...before that, he was a broke musician and young theatre artist/actor

and he made that money to get a house in america's most expensive real estate market in 4 years

to be dipomatic, i say windows is good but mac os is better

if windows was truly horrible, then it would not exist in this market or any other

and what the heck do you think the whole government vs microsoft thing was about...besides microsoft shafting netscape with a free ie, the next most common government objection to microsoft was how they copied apple's interface with windows

while i don't consider "pirates of silicon valley" the definitive history of apple and microsoft, the movie did make a good point in showing bill's ripoff of apple's os

in conclusion, windows is "bad" enough to confuse enough people that there is money in tutoring and troubleshooting windows otherwise people would not get rich off of working in and around microsoft's operating system

if i only worked on the mac os, i would not make that much money...so thank god for microsoft's fallible operating system...it pays our bills and 1.5 million microsoft certified professionals love you, bill gates

True. Keeps the bills paid. But I just wanted to throw something in there about MS copying Apple's interface. WinXP and OS X in particular. I used to work for a company that was contracted to design the next Windows interface. They were one of many companies that wer contracted to do so, but I had a chance to look at the interface that we did. First, the design was done on Macs (Blue and White G3s if I remember correctly). Secondly, our IT guy was a big mac head and he had OS X on a machine very close to the designers. Needless to say, there is a reason there are similarities between Windows XP and Mac OS X. MS used a lot of what my company designed. Who knows, same thing could have happened with prior versions of windows.

 

[jefhatfield]

Re: Re: Re: Re: Re: Itanium 2 and P4, could they beat the G4?

Originally posted by gjohns01


True. Keeps the bills paid. But I just wanted to throw something in there about MS copying Apple's interface. WinXP and OS X in particular. I used to work for a company that was contracted to design the next Windows interface. They were one of many companies that wer contracted to do so, but I had a chance to look at the interface that we did. First, the design was done on Macs (Blue and White G3s if I remember correctly). Secondly, our IT guy was a big mac head and he had OS X on a machine very close to the designers. Needless to say, there is a reason there are similarities between Windows XP and Mac OS X. MS used a lot of what my company designed. Who knows, same thing could have happened with prior versions of windows.

wow, lawsuit

 

[spuncan]

Sue Happy Nation

Yeah thats defintely evidence for a lawsuit against M$ (since the contract ends this summer) But as for Windows, any GUI OS that u have to go right click then go down to the Rename option to rename is just horrible (May I remind u that u just press enter on a Mac). As for the chips I am really scared I may build my own X86 computer useing the "Hammer" or a DP Athlon if the G5 falls behind. Hmm whats ur fav. Linux flavor (I'd hate to use 98 SE (My Winblows fav.))

 

[Rower_CPU]

Back to SETI

I was intrigued by the SETI@home comparison done above, and since I have the luxury of having both OSX and XP machines here at home I decided to perform my own test. I set them up last night and let them have a go at it, with no background programs running. So far here's what I've got:

PowerBook G4:
G4 400 MHz
640 MB PC100
Airport
1 unit ~ 18 hours and 20 minutes

Athlon rig:
1.33 GHz T-bird
768 MB PC1600 DDR
Hardwired to network
90% ~ 19 hours and 10 minutes, and counting

So, what do you think about this one? The G4 is 1/3 of the speed of the Athlon, and yet it manages to beat it out in a CPU test...

Say it with me now: "The megahertz myth is a reality."

 

[Rower_CPU]

Here's a screenie of the G4. I'll post the PC results once it finishes.

 

[mc68k]

Re: Back to SETI

Originally posted by Rower_CPU
So, what do you think about this one? The G4 is 1/3 of the speed of the Athlon, and yet it manages to beat it out in a CPU test...

Say it with me now: "The megahertz myth is a reality."

I agree. There's not too many cross-platform CPU taxing apps that you can truly say "test the CPU's performance". This I think is one of those few apps that you can use as a benchmark. Not like the Photoshop test that are skewed in the macs favor.

This is just pure, unadulterated number crunching that's the same as it's counterpart on the PC. The only real variable is the system (moslty CPU) that it's running on.

Your Mac machine would seem "way slower" because of the high MHz ratio, but the fact that your TiBook beat out your XP system is amusing, to say the least. Or even reassuring.

 

[Catfish_Man]

The G4's Altivec...

...gives it the advantage. For that type of repetetive task Altivec is amazing. It's a G4 400 so it only has one Altivec pipeline ( I think, it might be two), that's still the equivalent of running a normal pipeline at 1.6GHz, also (I'm guessing here) the stuff it's working on might be able to fit into its oversize cache, but not into the Athlon's smaller one. A DP1GHz (if it could keep all it's Altivec pipelines fed) would have the equivalent of 8 4GHz pipelines. Also, Altivec has special intructions that make it ideal for signal processing.

 

[Rower_CPU]

Re: The G4's Altivec...

Originally posted by Catfish_Man
...gives it the advantage. For that type of repetetive task Altivec is amazing. It's a G4 400 so it only has one Altivec pipeline ( I think, it might be two), that's still the equivalent of running a normal pipeline at 1.6GHz, also (I'm guessing here) the stuff it's working on might be able to fit into its oversize cache, but not into the Athlon's smaller one. A DP1GHz (if it could keep all it's Altivec pipelines fed) would have the equivalent of 8 4GHz pipelines. Also, Altivec has special intructions that make it ideal for signal processing.

So Altivec allows a chip 1/3 as fast to beat one without it? Interesting...
Please provide some tech docs that show how the programs differ on each platform.

Here's the pic of the PC, which just finished:

18 hours and 20 minutes on the Mac
20 hours and 45 minutes on the PC

 

[TypeR389]

Re: Back to SETI

Originally posted by Rower_CPU
So, what do you think about this one? The G4 is 1/3 of the speed of the Athlon, and yet it manages to beat it out in a CPU test...

Say it with me now: "The megahertz myth is a reality."

Excellent, thanks for doing a comparison, my LC580 I have isn't really a fair test with my DP933 ya think?

I was looking around at the Seti@Home website, and they sorta track this, but looks like there is a ton of weird data. Anoyone out there care to run a test on a 667 TiBook? I am REALLY itching to get a Tibook when this next release goes out, but since I don't live in photoshop or places where there are tons of optimazations, I just want to make sure that the 'perceptible speed' of using the OS to write code, doing basic office stuffis close to what my 1 year old desktop PC is. If a 400 is roughly half the speed of my current desktop rig, that means that the 800 (or maybe faster ) would theoretically be about the same, OS not withholding.

Question for you recent OSX converts, my LC580 is the last mac I bought, so I am still running 8.1 on it (yea I know...) but how long did it take you to to get use to the OS either coming from traditional MacOS or from a windoze side (been mainly running that for a few years now) But from playing with a 667 running OSX at the local apple store, I WANT ONE, just curious how long people think it took to 'think in OSX' I do use Solaris at work, so at least I am pretty familiar with UNIX...

Thanks for listening to my babble

 

[TypeR389]

Seti is not G4 or SMP aware

I was just reading the FAQ at http://setiathome.ssl.berkley.edu/fax.html at in there is goes on that Seti does not have any specific Altivec,MMX,3dNow etc and is also not SMP aware, so it will only run on one processor...

Just curious that it doesn't inherently use Altivec and the 400 G4 is still doin pretty good again a fast athalon...

Somebody going to write Steve a email? Looks like good marketing to me!

 

[bobindashadows]

Re: Re: Mac OS X wins

Originally posted by TypeR389


Just out of curiousity, how long does an avg set take you on the 867?

Just curious as you were saying that the mac is 2-6 hours faster than the PC, but my PC only takes about 8 hours for the entire set. Just happened that my current dataset is 99% complete

Dual 933 PIII
512MB PC800 Rambus
Windoze XP

Actually, it takes you about 16 hours for each one. I ran the numbers, and I got an average of 16.88 hours per unit. Now, seti@home is a difficult program to benchmark because the data units come in different sizes, however you have completed over 400 units so the average is going to be pretty accurate. I finished one in 16 minutes (no joke) and that's why in my benchmark we're using many, many units, and not just one.

 

[Rower_CPU]

Re: Re: Re: Mac OS X wins

Originally posted by bobindashadows
Actually, it takes you about 16 hours for each one. I ran the numbers, and I got an average of 16.88 hours per unit. Now, seti@home is a difficult program to benchmark because the data units come in different sizes, however you have completed over 400 units so the average is going to be pretty accurate. I finished one in 16 minutes (no joke) and that's why in my benchmark we're using many, many units, and not just one.

Good point. Since it would take me months to get that kind of comparison, anybody have a lot of Mac hours logged?

 

[mc68k]

Re: Seti is not G4 or SMP aware

Originally posted by TypeR389
I was just reading the FAQ at http://setiathome.ssl.berkley.edu/fax.html

Here's the FAQ link.

 

[TypeR389]

Re: Re: Seti is not G4 or SMP aware

Originally posted by mc68k


Here's the FAQ link.

Oops! Cant type

 

[TypeR389]

Re: Re: Re: Mac OS X wins

Originally posted by bobindashadows

Actually, it takes you about 16 hours for each one. I ran the numbers, and I got an average of 16.88 hours per unit. Now, seti@home is a difficult program to benchmark because the data units come in different sizes, however you have completed over 400 units so the average is going to be pretty accurate. I finished one in 16 minutes (no joke) and that's why in my benchmark we're using many, many units, and not just one.

That is also because I have been running this program for over 2 years, my old machine took over a day to process a aset, but the last 4 units or so have been averaging about 8.5 hours or so. I know what my numbers are now, I am going to see how it avg's out over the next week...My last machine was a PII 266, so no real spead demon...

 

[Rower_CPU]

Thanks to GeeYouEye we now have a pretty definitive table of system performance based on OS:
http://setiathome.berkeley.edu/stats/platforms.html

Windows = 21 hr 03 min 21.3 sec
Mac OS X = 18 hr 10 min 17.5 sec

 

[Choppaface]

SETI times
dual G4 500, 1 gig ram, radeon, 16gb cheetah 15k rpm, OSX= 10-11 hours per workunit per command line client, using one GUI client 11 to 12 hours

dual athlon 1900+ XP, 1 gig DDR RAM, geforce ti 4400, 100gb WD w/ jumbo buffer, win XP pro= 4.5 hours per work unit using GUI client, have not tried running two command line clients

I also ran cinibench once on the pc but I forgot the scores. it was a lot faster than when I've run it on my dual 500. one thing I do remember is that my dual 500 got 1.7 I think on the dual processor factor. the PC got 1.6

its not exactly a fair comparison, but I'd estimate my PC is far more than 2 to 3 times faster doing normal tasks than my G4. IE is especially fast, probably because its so closely integrated into the OS, and I'm getting near-instantaneous page rendering even with SETI running in the background. there's a really big difference between the two running hi-end flash sites. i've been meaning to do a framerate test to see the difference between the two, but the PC really screams through like mjau-mjau.com and other sites listed on styleboost.com.
photoshop seems a little bit faster, but it's really hard to switch over to the win keystrokes...I keep on hitting the alt key thinking its command
although I havent put the stopwatch to it, overally PS doesnt seem that much quicker, at least not fast enough to make me start using the PC just for the speed.
so far I've had a lot of problems with XP, starting with numerous crashes after putting in bad CD burner drivers, and now I still am having some problems with drop down menus and things after I fixed it. one thing I have to say: XP is hella fugly >_<

 

[alex_ant]

Re: Seti is not G4 or SMP aware

Originally posted by TypeR389
I was just reading the FAQ at http://setiathome.ssl.berkley.edu/fax.html at in there is goes on that Seti does not have any specific Altivec,MMX,3dNow etc and is also not SMP aware, so it will only run on one processor...

As far as SMP goes, the client does not support more than 1 CPU, but you can run more than 1 client at the same time, thus taking advantage of more than one CPU. At least that's how I understood it.

Where on the FAQ page does it say Seti@home does not use AltiVec, MMX, SSE, etc.? I couldn't see that question listed... If it does not use MMX or SSE, the seti@home developers must have gone to extra trouble to make it that way, or else used a crap compiler, since (AFAIK) most common Windows compilers out there optimize for both of these automatically. I don't see why the seti@home people would be interested in having people use poorly-optimized clients, especially when it's so easy to make them much faster (e.g. by using AltiVec)...

The seti@home client HAS to use AltiVec. It just has to. There is simply no other explanation for the performance results Rower_CPU got. I'm sorry, I don't care if you're checking your e-mail or doing protein analysis, but if you're not taking advantage of AltiVec, there is no way in hell a 400MHz G4 can beat or even come close to a 1.33GHz Athlon at ANYTHING. None at all.

Alex

 

[alex_ant]

Re: Back to SETI

Originally posted by Rower_CPU
Say it with me now: "The megahertz myth is a reality."

The megahertz myth is a reality... but only when running AltiVec-enabled software.

I wonder what's keeping Intel and AMD from implementing a vector processor into their chips. It doesn't seem like it would be a very tall order, plus Intel certainly has the expertise to write a decent compiler that could auto-vectorize code, unlike whoever is in charge of writing Apple's compiler, which would make said VPU transparent to software developers. Can you imagine what would happen to Apple if Intel implemented PentiVec? A very dark day for Mac users everywhere... I don't even want to think about that.

Alex

 

[TypeR389]

Re: Re: Seti is not G4 or SMP aware

Originally posted by alex_ant
Where on the FAQ page does it say Seti@home does not use AltiVec, MMX, SSE, etc.? I couldn't see that question listed... Alex [/B]

I didn't mean to say that it didn't use, but there is not specific enhancements for a G4,MMX, SMP etc. Here is a snip of the FAQ I was referring to:

Are you planning on optimizing the SETI@home client for the G4/3DNow/etc instruction set?

As of yet, there are no plans to optimize the SETI@home client for any particular platform or processor. Future versions of the client may be optimized, but there are no plans for doing so in the immediate future.

The compiler is going to do some stuff, and if FF transforms use matrix operations, which they do, the compiler might natvily do this, but there are no specific enhancments. I agree though, Altivec has to be used for a 500Mhz G4 to keep up with a much faster pc processor (Mhz-wise anyway)

 

[jefhatfield]

thanks for the interesting data!

do you think there really is life out there?

...actually, that sounds like a good thread for when things slow down again

 

[barkmonster]

Realtime performance is the only REAL judge of speed.

Photoshop can gain reasonable speed increases if you give it 100s of MBs of RAM and fast RAID drive for the scratch disk. It's not realtime, it's no test of a computer's true potential, it's just number crunching, no photoshop user including myself looks for a new computer to render the filter previews, it's the amount of time spent staring at progress bars that matters, same with seti, Bryce, Cinema 4D and any other application that renders or otherwise processes digital information in a non realtime sense.

Audio applications such as protools and logic are where the realworld tests show off a computer in it's true light, emulation. in particular arcade emulation is another field that the mac falls way behind the wintel boxes at.

Just check out a few REAL benchmarks for yourself:

This test from the Digidesign User Conference is based on the number of tracks (audio & aux) loaded with plug-ins that can be played back without running out of CPU power in protools LE. The plug-ins used were Compressor, 4 Band EQ, Slap delay, medium delay & long delay. All the test settings were the same so no one could cheat and use larger buffers and other tweaks to swing the results in their favour. I wouldn't change from mac os to windows regardless of how fast PCs are personally, but a little truth about cpu speed isn't going to hurt.

733Mhz G4: 26 tracks.
800Mhz G4: 27 tracks.

The PC benchmarks arn't really all that impressive when you concidering they're spinning a lot faster with those GHz speeds but only work out at most 30-40% faster than the macs.

P3 1GHz : 22 tracks.
P4 1.6Ghz : 18 tracks.
P4 1.8Ghz : 25 tracks.


Athlon 1.2Ghz : 26 tracks.
Athlon 1.4Ghz : 32 tracks.
Athlon XP1800+ : 40 tracks.

I put more PC results than mac results because there were more tests in the PC thread and it shows how trivial clock speed is even within PC vs PC tests.

I don't want to be flamed for this, but I know a lot of people are more interested in the less than everyday use benchmarks where you sit watching progress bars for minutes or hours and the computer that bores you the least is the winner

For realtime number crunching on just 1 CPU these protools test results are perfectly valid and one reason why I won't stop saving for a new mac till I can get at least a used dual 800 model later in the year.

Here's the most recent results thread for the macs and the other thread with results from both Macs and PCs is here.

Just for people who arn't into protools, there's a site I found with logic benchmarks but the mac results arn't from anything more recent than a 500Mhz G4 so I didn't bother putting them in the message.

 

[Firestormf]

Originally posted by Rower_CPU
Windows = 21 hr 03 min 21.3 sec
Mac OS X = 18 hr 10 min 17.5 sec

Good God, what is wrong with your computers? An Athlon XP taking 21 hours to do a workunit?? Even your G4 numbers look a little high if the G4 scales the same as a G3 when running SETI.

Here are MY numbers (approx)

PC running GUI SETI: 6.5-8 hours
PC running cmd line SETI: 6 hours

Mac running GUI SETI OS 9: 28 hours
Mac running GUI SETI OS X: 50+ hours
Mac running cmd line SETI OS X: 24-25 hours

Specs:
Micron PC
Athlon Thunderbird 1.2 ghz
384mb 266mhz DDR
Gigabye GA7DX motherboard
40GB 7200 RPM HD
Win 2000 SP 2

Blueberry iBook (original)
G3 300mhz
160mb SDRAM


note: woohoo 2 posts!