The G5

[psyjax]

quote:Originally posted by Viper:
Oh yeah. It is fair for Apple to test a G5 with all of it's optimizations but test the P4 without SSE2.

Riiiiggghhtt.   :rolleyes:  

The testing wasn't done in-house first off, and at the time of testing I don't thing the extreme existed, but I hardly doubt if that would make much of a dent considering the overal remarkable performance of the G5. Not to mention that it's due for a 3ghz speedboost by the end of the year.

Don't forget, IBM is backing apple here, that's like a freight traing pushing a baby carrige or something equally powerfull  

[Zombie9920]

This is interesting.

Here's a complete normalized PS7Bench results list.

   

quote:2x 2000 G5 OSX 10.2.7             555 (energy settings highest perf)
2x 2000 G5 OSX 10.2.7             497 (energy settings auto bus slewing)
2x 3060 Xeon (no L3) HT enabled   490
2x 3060 Xeon (OC'd 2400)          488
2x 2930 Xeon (OC'd 2400)          471
  3200 P4 (800MHz)               427
  3000 P4 (800MHz)               405
  3495 P4 (OC'd 3.06)            386
  3060 P4 XP Pro (533 FSB)       358 HT
2x 2200 Xeon PC 800 RDRAM CPQ Evo 357 HT
2x 3000+Athlon (2166)             355 (provisional Utwig)
2x 1500 G4 (OC'd 1420)            348
2x 1333 G4 DDR OS9.2 (oc'd 1.25)  346
  1800 G5 OSX 10.2.7w/G5plugin   344 (energy settings highest perf)
2x 1420 G4 OSX 10.2.4             338
2x 2400+Athlon MP                 338
2x 1250 G4 OS 9.2.2j              337
  3200+Athlon XP                 332
  1800 Opteron(dual-chnlDDR 333) 332
2x 1333 G4 DDR OSX10.2.2(oc 1.25) 326
  1800 OPteron(singl-chnlDDR333) 320
  3000+Athlon XP                 318
2x 1250 G4 OSX 10.2.5             318
2x 1250 G4 DDR OSX 10.2.1         316
2x 1800 Athlon MP                 312
  2800+Athlon XP Barton          298
2x 2000 P4 Xeon                   286
2x 1200 G4Powerlogix(867MHzG4/QS) 285 upgraded
2x 1533 Athlon MP                 285
2x 1533 Athlon MP                 283
  2530 P4 mobile (OC'd 1400)     282
  2700 P4B (OC 2400, 600 MHz FSB)280
2x 1466 Athlon XP                 279
  1600 G5 OSX 10.2.7w/G5 Plugin  276 *MacNNscores (energy settings on auto)
  2666 P4 (DDR 333)              269
2x 1000 G4 DDR 10.2               267
  2400+Athlon XP                 262
2x 1000 G4 OS9                    260
2x 1000 G4 OSX 10.1.5             254
  2400+Athlon                    252
  2400 P4B (800MHz)              251
  2400b (sis 648 DDR400)         251
  1600 Centrino IBM T40          250
  2400 P4 (533MHz bus)           249
  2400 P4 B                      241
  2340 P4 (overclock)            239
  1600 Centrino Dell D800        236
  2400 P4                        234
  1800+Athlon XP (1533 MHz)      226
  1577 oc'd Athlon XP (Lestat)   221
2x 1000 G4 OSX 10.2.2 (upgraded)  218 ?!(dual 533 logic board)
  1548 Athlon XP                 214
  1670 Athlon XP (2000+)         213
  1667 Athlon XP                 211
  1400 Athlon XP 1600+ xp pro    200
1x 1533 Athlon MP                 197
  1300 Centrino Sony VAIO Z1A    196
  1000 G4 17" Powrbk OSX 10.2.6  196
  2000 P4 Xeon                   194
  1400 Athlon XP 1600+'98SE      191
  1000 G4 OSX TiPbk 10.2.2       185
2x  533 G4 OSX 10.1.5             175
2x  533 G4 OS 9.2.2               174
  1800 P4                        173
  1200 AthlonMP                  168
  1508 Celeron (overclock)       167
  1400 PIII Tualatin             160 **?
2x  550 G4 OSX 10.2.3 (OC Cube)   160 **?
2x  500 G4 OSX                    152
2x  450 G4 OS9                    151
  1333 Athlon TBird              147
2x  450 G4 OSX 10.1.5             143
   800 G4 Pbook OSX  1MB L3      135
   733 G4 (miro7)                134
   667 G4 PBk OS9 noL3           127
   667 G4 PBk OSX 10.2.3 no L3   125
   466 G4 OS9                    123
   667 G4 OSX TiPBk 10.1.5 noL3  121
   866 PIII                      114
   466 G4 OSX 133 MHz bus        112
   550 G4 Powrbk OS9*            104
   500 G4 Pbook (OC'd 400)       103
1x  450 G4 OSX 100 MHz bus        101
  1000 Athlon TBird (PS6.01)     100
   550 G4 Powrbk OSX*             95
   933 Transmeta Crusoe Sony      78
   700 G3 iBook                   74
   600 G3 iBook OS 9.2.2j         70
   233 PII                        30

Now, to get an idea of overall performance let's compare some more benchmark results between the platforms.

First let's start with the benchmarks C't magazine ran in a recent issue (all the systems are using a Radeon 9600). They used the G5 optimized version of Cinebench which isn

[hm_murdock]

I was under the impression that in GL got accelerated through DirectX in version 9.

I officially retract my claim. I was mistaken as to their 3D architecture!   :eek:

As for what I said about Linux... no, for what I do (audio recording/editing with realtime effects) Linux can't do it. It's not because of the OS, but because of the rather poor, inconsistant implementation of sound drivers.

And lack of apps for high end recording and editing.

I'm sorry. I don't come and flame you out when you say something negative about Mac OS. Linux isn't cut out for media production (although I was just informed that video editing is quite good). It can't do it all... audio isn't that easy.

I'm sorry... it just isn't. does that make you sad?    

[ September 28, 2003: Message edited by: Jimmy the Shyster ]

[Zombie9920]

I'm going to throw this in the mix as well.

Here are some Lightwave 7.5 scores from AnandTech and PCMagazine that show some Single Processor PCs versus a Dual G5:


 

quote:Rad. Ref. Scene
P4 EE 3.2GHz           42.5 sec
P4 3.2C                46.0 sec
P4 3.0C                48.4 sec
Athlon 64 FX 51 2.2GHz 49.3 sec
Athlon 64 2.2GHz       50.6 sec
Dual G5 2GHz           51.1 sec
Athlon 64 3200+        54.5 sec

Raytrace Scene
Athlon 64 FX 51 2.2GHz 87.9 sec
Athlon 64 2.2GHz       88.3 sec
P4 EE 3.2GHz           89.3 sec
P4 3.2C                93.1 sec
Athlon 64 2.2GHz       96.4 sec
P4 3.0C                99.1 sec
Dual G5 2GHz          112.0 sec

[psyjax]

If you think running a couple of video games, and an unspecified rendertest is the gamut of rigorus testing, you sir are an idiot. Notice, that it's a 1.8G5 against a 3Ghz P!V in the latter tests and in teh first tests were they use 2 Ghz dual G5's you, naturaly, get better performance. The 1.8 is out by a hair.

Of course this is irelivant, because here you are dealing with software optimization issues, run some apps well optomized for both platforms and you get a diffrent story, likewise run the stander floating point, integer tests, and you get a diffrent story.

Apple has done their part in making the computers faster, now the ball is in the developers court to start optomizing their shit.

WOW! Sorry Viper, That was a major mistake! I hit the edit button instead of the reply button. Please correct your post above. heh... perils of having Admin privaleges  

[ September 28, 2003: Message edited by: psyjax: plain 'ol psyjax ]

[Zombie9920]

I don't appreciate you editing my post Psyjax. Those tests were not gaming benchmarks. They were Cinebench scores. Cinebench tests the true performance of a CPU. You edited my post because the x86 rigs own the G5. In most cases a 2x G5 can't even beat a 1x3.2ghz P4.

Don't edit my posts asshole.

(EDIT) I see it was a mistake. My bad. Sorry for calling you an asshole.  

[ September 28, 2003: Message edited by: Viper ]

[psyjax]

quote:Originally posted by Viper:

Don't edit my posts asshole.

I told you, it was an accident, you don't need to call me an asshole cunt wad  

EDIT: applogy accepted  

[ September 28, 2003: Message edited by: psyjax: plain 'ol psyjax ]

[Zombie9920]

Post re-edited

[psyjax]

quote:Originally posted by Viper:
Post re-edited

Well, at this point, it's a war about one guys word again'st the other. Cinebench marks the Intel faster, and SPEC marks the G5 faster.

Whichever you choose to belive, I think you will note that very rarely does the Xeon rate significantly faster than the G5. And if Apple makes good on it's promiss for a 3Ghz by the end of the year, not to mention better (64Bit) optimization by developers, the speed issue will be moot.

On another point, you may notice that the G5's are actually very well priced compared to PC's in the same range. So you must admit, that if anythin, the G5 is a comparable, and competative product.

Personaly, I trust the SPEC marks  

[Zombie9920]

Oh the SPEC marks are real for the G5. However, the SPEC marks that Apple lists for the P4 are not accurate. In reality a 3.2ghz P4 w/Dual Channel DDR400 scores well over 1,200 Spec marks(Apple listed it in the 700-800 mark range). The G5 if I remember right was in the high 800 to low 900 range according to Apples benchmarks. That means that the P4 in normal conditions actually outperforms the G5 in highly optimized conditions(just like how the P4 does alot better in Quake III than Apple wants people to think).

The G5 is a competitive product, but it is not the best nor the fastest.

[ September 28, 2003: Message edited by: Viper ]

[Zombie9920]

Here are accurate Spec scores.

http://www.spec.org/cpu2000/results/res2003q3/

As you can see, the P4 does much better than what Apple claims. The P4 does better than the G5(I'm sure you all have already seen the G5 results at their best@apples' web site). Apple lies, end of story.

(EDIT)So The highest end P4 outperforms the highest end G5 in everything but Photoshop in the real-world. I can really say that doesn't surprise me at all.

[ September 28, 2003: Message edited by: Viper ]

[psyjax]

Here is the register article on the G5 benchmarks:

http://www.theregister.co.uk/content/39/32498.html

At the bottom you will find an article arguing about Apple's version of the SPEC scores, which are well defended in the following article:

http://www.theregister.co.uk/content/39/31416.html

Infact, Apple's benchmarks are on the concervative end. They are lower than the G5 chip in other tests. So, saying the x86 performed 1200, isn't taking into account the methodology of the test.

As we all know, tests are never 100 percent, and you can allways weigh the data one way or another, but by all accounts apple did a fair test, and their results are acurate. Again, see the above articles.

[Zombie9920]

Really, I don't believe for a minute that Apple did a fair test simply because of their published results for Quake III on a Pentium 4. Everybody knows that a Pentium 4 scores way higher than what Apple posted in Quake III.

Disabling SSE2 optimizations on the P4 and leaving G5 optimizations enabled is not fair testing. When you test a CPU you test it with everything it has to offer.

If you want to disable optimizations for 1 CPU you have to disable the optimizations on the other CPU for it to be considered a fair test.

Nonetheless, Every benchmark aside from Apples' benchmarks shows that the P4 is indeed faster than what Apple claims and is indeed faster than the G5 in everything but Photoshop. Apples' CPU's have always excelled@photoshop(even the G3's).

Apple can not fool me because I know better than to trust hardware manufacturers' benchmarks. They all lie, Apple is no different.

The G5 is indeed competitive, I won't hesitate to admit that, but it simply does not have x86 beat. The Prescott is right around the corner as well. Next year we will be seeing the Tejas(Pentium V). IBM is going to have to do some major work to their CPU's if they want them to remain competitve to the Athlon FX, Pentium 4 Prescott and the upcoming Pentium V. If they don't Apple will be left in the dust in terms of performance once again(like how the P4/Athlon XP left the G4's in the dust as they got faster and faster).

[mushrooomprince]

quote:   Really, I don't believe for a minute that Apple did a fair test simply because of their published results for Quake III on a Pentium 4. Everybody knows that a Pentium 4 scores way higher than what Apple posted in Quake III.

As of now it doesn't really matter if the pentium 4 is faster than the dual G5.  The G5 can make use of twice as much memory as the Pentium 4.  

On top of that, with Suns new chip technology on the horizon I don't  think Pentium 4's have much of a future in anything.

Not that I'm biased to anyone.  But I'm convinced that for price/performance right now the best cpu on the market is the Athlon 2200.  

And if your some poor farmer that lives in Ecuador.  Your most likely going to get a nice 1.3 ghz Duron.  

Intel has no future.  Theyre pushing the clockspeed on those P4's as much as they can.  


And as for the G5 ?  Well what makes the G5 better than the Pentium 4 isn't how well it can run Quake 3 but how well it can run OS X.

[Zombie9920]

You don't get it man. This isn't about Quake III. The main reason why I beleive SPEC results published on SPEC's page for the P4 over Apples' SPEC results posted for a P4 is because 1. SPEC doesn't post biased results and 2. Apple clearly lied about the P4's Quake III performance(if hey are going to lie about 1 thing they are going to lie about everything). It isn't about how fast the P4 runs Quake III, it is about Apple lying about it's true performance. Please understand that. Please understand that SPEC in no way has anything to do with Quake III. Furthermore, the P4 is faster at Lightwave, Cinebench(a CPU intensive benchmark), UT2k3, etc.  

Actually, the AXP 2500+ Barton offers the best price/performance ration in the AMD camp. Nothing in the AMD camp can beat the P4 2.4c ghz's price/performance ratio though. That chip is a little more expensive than the Athlon 2500+ Barton(it is $170) but it can be overclocked to 3.2ghz with stock air-cooling with no hitch. In the end, you get a high-end chip for the price of a budget level chip. I think I'd rather spend $300 on a P4 2.4c and a good i865 motherboard for it and run it @ 3.2ghz(which beats out an Athlon XP 3200+ with no trouble at all) than spend about $200 on a Athlon 2200+(like you suggested) and a decent motherboard for it and only have it able to overclock to a little over 2300+ specs.

The speed potential differnce between what you suggested and that p4 2.4c is worth the extra $100. $300 for 3.2ghz w/800mhz FSB and Hyperthreading is a steal...it really is.

[ September 29, 2003: Message edited by: Viper ]