Developer Transition Kit Benchmarks

[ailleur]

Based on Apple pricing, they actually do not need to worry about Dell specs. They will need to worry about AlienWare specs. Apple will be a high end X86 box maker, they need to compete with the high end makers. They really don't need to compete with Dell products, which also use pretty old technology on the low end.

I dont think they have to worry about anyone. Alienware pcs cant run osx, and intel-mac pcs may or may not run windows. I dont think using an intel chip instantly changes apple into a direct competitor to any of the pc-in-a-box manufacturer. Or at least, not any more than they were before.

 

[CubaTBird]

my ibook gets higher scores than that intel p4 configured mac.. go figure.. i get at least like 105 on my ibook.. with performance set to high

 

[ailleur]

my ibook gets higher scores than that intel p4 configured mac.. go figure.. i get at least like 105 on my ibook.. with performance set to high

Fantastic, now try running benches of windows through VPC on your powerbook and see how it fares against a real windows box. Those 2 sets of benchmarks will be about as valid.

 

[jimbobb24]

I dont think they have to worry about anyone. Alienware pcs cant run osx, and intel-mac pcs may or may not run windows. I dont think using an intel chip instantly changes apple into a direct competitor to any of the pc-in-a-box manufacturer.

I think you are partially correct. But, when I go to buy a computer and can now compare the exact same hardware significant price differences become a problem. However, it should be pointed out, that most Macs are pretty competatively priced right now. The is an Apple tax, but its not as high as people imagine.

 

[TreeHugger]

maybe the G5 is just faster than a P4.

just like apple has been advertising for the past few years....

 

[eurrythmic]

“After Jobs' presentation, Apple Senior Vice President Phil Schiller addressed the issue of running Windows on Macs, saying there are no plans to sell or support Windows on an Intel-based Mac. 'That doesn't preclude someone from running it on a Mac. They probably will,' he said. 'We won't do anything to preclude that.'"

"However, Schiller said the company does not plan to let people run Mac OS X on other computer makers' hardware. 'We will not allow running Mac OS X on anything other than an Apple Mac.'"

 

[jauh]

“After Jobs' presentation, Apple Senior Vice President Phil Schiller addressed the issue of running Windows on Macs, saying there are no plans to sell or support Windows on an Intel-based Mac. 'That doesn't preclude someone from running it on a Mac. They probably will,' he said. 'We won't do anything to preclude that.'"

"However, Schiller said the company does not plan to let people run Mac OS X on other computer makers' hardware. 'We will not allow running Mac OS X on anything other than an Apple Mac.'"

The only problem is how much more can you charge for the priviledge of running macos x... I guess most home users would go with the (much?) cheaper wintel config, and gamers would (possibly?) prefer amd...

 

[iMeowbot]

if you dont want to buy new softwares, fine.
Apple have rosetta to support u guys and taking u back to the speed of a G3 and softwares for G3 (yes, G3 not G4, without Altivec)

The people who spend thousands on new hardware every year and never, ever upgrade their software are still running ancient pre-Altivec versions of their programs anyway, and will never notice the difference.

 

[eurrythmic]

The machine Steve used on stage to demo OS X and apps on Intel was SO fast, as I commented earlier. Before we even knew it was an Intel machine we thought that the machine seemed to be responding exceptionally fast. Steve was in fact using a 3.6GHz Pentium 4, single core, just like everyone else.

From XLR8yourmac: First, the thing is fast. Native apps readily beat a single 2.7 G5, and sometimes beat duals. Really.
(I asked about real-world apps - if any were already available in native code-Mike)
All the iLife apps other than iTunes, plus all the other apps that come with the OS are already universal binaries....

 

[matznentosh]

I've read these forums for years but never registered until today.

My question is this...

Could/would it be possible for Apple/Intel to actually create an Altivec friendly version of the x86 based processor? After all, IBM had to do this with the G5, right?

Couldn't Intel and Apple be working on such a solution that would make the ease of the transition easier and the processor that much more powerful?

Thanks!

Answer, YES. This is precisely what IBM did. If you look back at Ars Technica's original descriptions of the G5, it clearly described the Altivec unit on the G5 being bolted on as opposed to the integrated version in the G4. So sure, Intel could do this too.

Having said that, my prediction is they won't for two reasons:

1. Intel won't want to spend the money on a small run of specialized CPU's, since Apple will only be buying a small percentage of Intel's product. No reason to make a separate Pentium M + Altivec when no one else will want it. They will want Apple to just take their pick of what they have available. Having said that, I could imagine Intel being influenced by Apple's innovation pressure to come up with further new ideas, but likely those would be used across the board, not just for Apple.

2. Unfortunately, we can all point to great technology that Apple developed and pushed hard, then just walked away from (mostly due to Steve's business decisions). Apple has been pushing developers more and more to take advantage of Altivec lately, and now will just say "too bad, get over it".

Come to think of it, doesn't Tiger automatically transfer appropriate calls to Altivec, so developers wouldn't have to recode their programs specifically to take advantage of it? I bet Apple did that because they already realized Altivec is a dead end, no reason to keep pushing for Altivec to be explicitly used by programmers.

 

[ruud]

Let's just hope this isn't anything like Windows running on top of DOS.

I don't like Windows as much as the next guy, but saying that Windows runs on top of DOS hasn't been true for the last 5 years.

 

[jauh]

I could imagine Intel being influenced by Apple's innovation pressure to come up with further new ideas, but likely those would be used across the board, not just for Apple.

sse4?

 

[tjwett]

I just hope that Apple will maintain their top-of-the-line Mactels with the fastest Intel processors. Apple has already been known to have "last generation" graphics cards. Apple will have all sorts of problems if they don't keep up with companies like Dell, HP, etc and offer processors of equal performance.

I can easily see Apple being slow to update the processors and new chips are released. Dell, for example, wastes no time releasing PCs with the latest Intel chips on-board. In fact, Dell releases computers with the latest chips on the day they are made available by Intel. This won't matter much for the hardcore Mac users, but it will become generally known that Macs are always a step behind in performance.

Apple has a habit of releasing major updates and not simple processor speed increases. This strategy won't work well when they have Intel chips inside.

-rich

this is a good thing to think about. we've grown so used to Apple updating a machine about once every 6 months or so, it would be weird for it to be any other way. i really wonder how they'll handle this. the one problem i see is that a Mac is considerably more expensive to manufacture than any HP/Dell/Compaq etc. most PCs are built of crappy thin steel and screws and are then sloppily spray painted on the outside and outfitted with useless plastic doors and the latest piece of crap Creative Labs soundcard. Firewire, USB, Ethernet etc are often not part of the motherboard. Apple hardware is very different and i think part of Apple's slow updates are to give them time to build a bunch of machines and then actually clear the stock as much as possible before moving on. if Intel come out with a new speedbump every month i wonder how Apple will rearrange its inventory and/or catalog. i suppose they could have a ton of cases with mobos inside on hand and pop in the newest processors as they come down the pipe. that would be nice to see.

 

[hayesk]

The only BS is between the average Mac Religionist's ears. Watch the damn keynote. He said the performance hit would be minimal. I guess it depends on how you define "performance hit", "the" and so forth (Steve and Billba are buddies, aren't they? They seem to have the same speach coach). Just to get out of the Mac Fanatic echo chamber a little more...

Before you go calling BS on everyone else, think about it a bit. You are seeing the results running, in emulation, a tool designed to run "once of everything" designed for a different processor.

Real world software does not do this. It does several things over and over again, to varying degrees. Much of which is OS system calls which won't be emulated.

You are kidding yourself if you believe that these benchmarks in any way resemble real-world performance of a PPC app running under Rosetta. Back in the 68K->PPC transition, you may remember that 68K MacBench didn't run so hot either. But most 68K software ran quite well on a PPC Mac. Far better than benchmark software would have you believe.

 

[deputy_doofy]

My girlfriend's parents have a 2GHz celeron - it runs like a 400mhz pentium 2 (even after a clean install of Windows)

I would never choose a Celeron over a G4. The G4 has its problems and yet it still wipes the floor with every Celeron I've ever used.

I hope Macs don't use any Celerons at all. What a mistake that would be...

 

[deputy_doofy]

Trust me, this is a "New Coke" class F'up by IBM (if anyone around here is old enough to know what I'm talking about).

I remember.
If Apple is smart, they should never remove the ability to create universal binaries and, of course, keep OS X built for both processors. Honestly, I would have no problem with Intel AND PPC based Macs in the future.

Long live my dual 2.3GHz G5!!

 

[matznentosh]

...I bet Apple did that because they already realized Altivec is a dead end, no reason to keep pushing for Altivec to be explicitly used by programmers.

quoting myself to finish my thought: this technique for automatically routing code to the Altivec was an attempt to wring every last bit of processing power out of the G4 and G5, since the speed of the chip couldn't increase much. Looking back, it does seem like Apple was getting desperate about IBM's poor development cycle.

 

[Spazmodius]

Before you go calling BS on everyone else, think about it a bit. You are seeing the results running, in emulation, a tool designed to run "once of everything" designed for a different processor.

Real world software does not do this. It does several things over and over again, to varying degrees. Much of which is OS system calls which won't be emulated.

You are kidding yourself if you believe that these benchmarks in any way resemble real-world performance of a PPC app running under Rosetta. Back in the 68K->PPC transition, you may remember that 68K MacBench didn't run so hot either. But most 68K software ran quite well on a PPC Mac. Far better than benchmark software would have you believe.

I'm not kidding myself about the claims made of this software's potential performance, which are there for anyone who listens, and is what you would see is the point I'm making, if you and others had a modicum of reading comprehension. The evidence extant clearly indicates it is no better than any emulator on any other platform. Telling me this is to be expected is restating the obvious. I am not arguing that point (though you and other apologists appear to feel reaepatedly and persistently providing me redundant or non-information substitutes for making a valid argument).

 

[madmaxmedia]

I'm not kidding myself about the claims made of this software's potential performance, which are there for anyone who listens, and is what you would see is the point I'm making, if you and others had a modicum of reading comprehension. The evidence extant clearly indicates it is no better than any emulator on any other platform. Telling me this is to be expected is restating the obvious. I am not arguing that point (though you and other apologists appear to feel reaepatedly and persistently providing me redundant or non-information substitutes for making a valid argument).

His point is just that the evidence is not so clear, and that when you are actually running PPC apps under Rosetta, the general speed and performance will be better than these benchmarks indicate.

How much better, I don't know. Just clarifying the discussion-

 

[ImAlwaysRight]

These XBench results whilst interesting, should be taken as a pitch of salt as they will not reflect the end result of production Intel on a Mac. This is partly why Apple did not want benchmarks to be released - they are somewhat misleading in the large picture.

Yeah, well Apple has never had any problem misleading people in the past with their benchmarks when trying to sell PPC. I know it is all marketing, but still. Let Apple sit on the wrong end of the benchmarks this time.

 

[Azurael]

One thing is for sure, the disparity in performance between a G5 and P4 in these benchmarks explains why Apple didn't want benchmarks to be posted.

Quite the opposite, IMHO. Rosetta doesn't do AltiVec, which explains the low VecLib score. Have you ever used PearPC? Even on my 2.4GHz Athlon64 it was like using an early PPC machine with framebuffer video to run OS X - the Xbench scores are rarely above 10. If Apple can achieve effectively the performance of an 800MHz-1GHz G3 with an emulator doing most of the work (though I appreciate that it has an advantage over things like PearPC because system calls are handled natively), a) the emulator is bloody fast (but we already knew Transcript's dynamic recompilation is fast) and b) The chip it's running on is incredibly fast. Think what OS X would run like on a top of the range Athlon64 X2

Native apps will fly on this thing! Did you see how fast iPhoto opened with 4k+ photos in Steve's keynote?! And Im sure if XBench was compiled for x86, it would outpace the G5 in almost every benchmark (though possibly not vector processing because that sucks on ALL x86 CPUs compared to G4s/G5s)

 

[SiliconAddict]

>>Does Apple really think they're going to shoehorn the P4 into a slim 1 inch Powerbook case...?
---
No-but they WILL be able to shoehorn the coming Pentium M, however, as part of the plan...

Heck. I bet if they wanted too they could "shoehorn" (Do you guys even know what that means?) 2 1.5Ghz Pentium M's into a 1" PowerBook and still get about 4 hours of battery life. The M's are just that good.

 

[Neerazan]

Post Deleted By Poster

 

[Peace]

Heck. I bet if they wanted too they could "shoehorn" (Do you guys even know what that means?) 2 1.5Ghz Pentium M's into a 1" PowerBook and still get about 4 hours of battery life. The M's are just that good.

I know what a shoehorn is so I'm guessin it means a tight fit..

 

[GuyClinch]

Yeah, well Apple has never had any problem misleading people in the past with their benchmarks when trying to sell PPC. I know it is all marketing, but still. Let Apple sit on the wrong end of the benchmarks this time.

They weren't misleading you. The PPC was faster at photoshop filters however it's slower at 3D games, media encoding and some other operations.
When a company designs a chip they make trade-offs. They have a certain number of transistors to work with and thus they allocate only so many for certain functions.

That being said Job's is rightly worried. Sure some people run Maya and the like but you can't base a PC business on just graphic design programs. You need to worry about things like interger operations.

For example a 2.2 GHZ G5 scored 986 on Spec Int 2000. Even if we assume that a 2.7 GHZ machine is 30% faster your only talking a score of 1300 or so.

http://www.spec.org/cpu2000/results/res2004q4/

A P4 right now though can do hit 1800 on the same bench mark.

http://www.spec.org/cpu2000/results/res2005q1/

And it's the same thing with the "cell" it's not suited to be a general purpose CPU. It will die on "branchy" code and do really well on say game code.

And for those "people" who demand a link to the Doom results. Well I thought you read your own magazine.

http://www.macworld.com/news/2005/03/16/doom3/index.php

http://www.nzone.com/object/nzone_doom3_benchmarks.html

G5 gets smoked in this benchmark. And I have heard the same thing about World of Warcraft. Before you get in tizzy ALOT more people are interested in playing games then Maya. <g>

Get over your cognitive dissonace the G5 isn't some uber speed demon of a chip. It was just marketing hype.

Pete