PowerMac Dual 2.0GHz PS7Bench Benchmarks

[3.1416]

Originally posted by mvc
It's inconceiveable that, over time, the mainstream PCs will not naturally be superior in speed that a niche player like the Mac

I disagree. For most of the lifetime of the PowerPC processors, they have been faster than the contemporary x86 competitors. The G4 debacle is the exception, not the rule. The PPC architecture is a fundamentally better design than the legacy-encrusted x86, so it is quite possible to make superior chips with fewer resources.

If money were all it took, Windows would be light years ahead of Mac OS X.

 

[Mav451]

Pretty good results--as i've said b4, this is one of Apple's strongpoints--and it shows

What i'm disappointed about the list is that it does not show the most CURRENT PC side processors overclocked.

The 845 chipset is VERY last year. I'd like to see the 800fsb cpu/chipset OCed.

Take for example the 2.4C. Cost effective at a palty 170bucks, coupled with your 150buck Abit IC7, FSB overclocks of 900-950 are easily achieved. 1000fsb will take a bit more voltage/cooling, but has been done.

Ditto for the 3200+. Take the 2500+ instead, bring up the fsb to 220 (which is literally guaranteed on any ultra 400 board, but particularly the gigabyte and nf7-s). A decent overclock with the nf7-s is around 220fsb * 10.5x = 2310mhz.

Not the highest (i.e. 240 fsb), and 2310 is definitely not the best overclock (2600-2700mhz is common on water); but it should provide a decent impression of the "overclocked" barton.

Still, the bottom line is that the G5 wins NOT OVERCLOCKED--and that is a good thing.

(i'm ALMOST running 200fsb, but i'm running an old board I'm pretty sure it's ram limited as i could run my current settings @ 1.675 vcore, with TIGHT timings, while 9.5*200 wouldn't even run with 2.8vdimm, 11-4-4-3 timings, or 1.725 vcore.)

 

[solvs]

OC'ing doesn't count.

But I'm sure some people will contest that this test is inaccurate. They will run other tests showing the G5 losing. Then people will complain that the softwares not optimised. And it will go back and forth like that forever.

BUT WHO CARES! In the real world, its fast. That's all there is to it. Benchmarks, schmenchmarks.

 

[socbyset]

Originally posted by mvc
...

It's inconceiveable that, over time, the mainstream PCs will not naturally be superior in speed that a niche player like the Mac - this current good result is really a sweet spot due to a good new architecture competing against either old / inefficient / overstretched / poorly implemented or undeveloped ones.

Logically, it cannot stay this way when so much more money overall is being poured into the competitions R&D, no matter how clever IBM's engineers are. Only if we somehow gain a more major marketshare (say over 10%), will there be any real hope of sustaining this sort of result.

Ask yourself, why did Motorola go from being a successful fast chip maker to a laggard? (apart from general incompetence)?

Because they don't care about the Powerpc as a standalone Computer platform, because it doesn't make enough money. IBM has a more vested interest with their servers using related Power4 architecture, but there are no guarantees they will advance the consumer spinoff 970-9X0 chips as fast.

...

It's true that intel has an advantage in r&d and sheer volume. Ever since the powerpc came out people have been predicting that soon Apple will pull away as the inherent advantages of the RISC platform became evident, but what we have had was more or less a game of leapfrog.

But I do think that things are looking brighter for the Mac performancewise with their new relationship with IBM. From what have read Mot could not keep good engineering talent. They seem to be a badly run company, and only halfheartedly commited to being in the chip business. Besides which they were aiming the G4 at embedded applications.

IBM on the other hand has state of the art fabs, and have committed to staying on the cutting edge of process technology. How many breakthroughs in process technology have you ever heard of Motorola ever having? And the new 9xx series utilizes the R&D from their top of the line workstation chips. Whats more if they put 9xx into their own workstations and blades, they have all the more incentive to keep them cutting edge and prices will benefit from the economies of scale.

 

[Mav451]

Originally posted by solvs
OC'ing doesn't count.

But I'm sure some people will contest that this test is inaccurate. They will run other tests showing the G5 losing. Then people will complain that the softwares not optimised. And it will go back and forth like that forever.

BUT WHO CARES! In the real world, its fast. That's all there is to it. Benchmarks, schmenchmarks.

dude did you even READ my post? I mention OCing b/c they display results for the OCed Xeon (2.4 @ 3.06) and "old" 3.06 @ 349.

2ndly, where the heck do i say that this test is inaccurate? Did you come up with that misconception by yourself?

Don't stereotype and lookdown on overclocking just because you do not understand it.

Look at my sig--it is 100% safe b/c the only issue with overclocking is too much voltage (i.e. 1.8v +, i'm not in that range), PCI bus overclock? (nope, nforce2 and all modern Intel boards lock them at 33, agp @ 66 respectively).

And actually the #1 cause of failure is heat. Most newbies do not understand that they cannot overclock with "stock" equipment. Spending 30bucks for that Swiftech or Thermalright SLK will pay HEAVY HEAVY dividends in increased stability, more headroom, and overall lower case temperatures.

 

[raynegus]

Cool pic series. Feels like I just bought one. Kid on x-mas morning.

 

[moby1]

Way to go IBM!

I think it's slowly sinking into the population just how impressive the G5 is.

Looking at the G5 brochure the bench for BLAST is quite impressive vs. PC arch. running Red Hat. Does anyone have more information on BLAST scores. I'm intersted because we have so much BioTech near where I live and I'd love to see some all-G5 labs.

 

[dongmin]

just the beginning...

1. Panther should speed things up even more. According to Barefeats, it's around 40% faster than Jag in certain key Xbench tests. Can't wait.

2. 50% (3 ghz) by next August, as Jobs reaffirmed. The way things are going with IBM, I'd say that's a conservative estimate. And with the 90-nano process 970s, it'll run even cooler. So we're talking more powerful, but with less heat and noise.

3. And third-party software optimizations should take things even further.

 

[Mav451]

Haha i wish UMD had that. UMD is under so many budget cuts we're just hoping we don't lose more dining services, or even basic services. The convenience stores hours have been ridiculously cut, so i doubt new computers would even be feasible

 

[pjhornak]

06/24/2003
08:40 AM PST XXXXXXXXX • POWER MAC G5 2.0GHZDP CTO
Shipped

POWER MAC G5 2.0GHZDP CTO


Buyers Name:
Web Order Number:
Apple Sales Order Number:
Apple Delivery Number:
Apple Customer Number:
PO Number:


Dear Apple Customer,

The following products shipped on 09/17/2003. Transit time will
depend upon whether you have chosen standard or premium freight
options. If your order is shipping standard freight, it should arrive
within 2 - 5 business days of shipment.

Product # Product Description Qty Ext Price
__________ ________________________________________ ____ ________________

_________________________________________________________________________
Z07K POWER MAC G5 2.0GHZDP CTO 1 3,349.00
With the following configuration:

Processor 065-4227 Dual 2GHz PowerPC G5
Memory 065-4356 512MB DDR400 (PC3200) - 2x256
Hard Drive 065-4376 160GB Serial ATA - 7200rpm
Optical Drive 065-4168 Super Drive (DVD-R/CD-RW)
Graphics Support 065-4288 ATI Radeon 9800 Pro
Fibre Chl PCI 065-4491 None
Modem 065-4166 56k V.92 Internal Modem
BlueTooth 065-4292 BlueTooth Module
Airport 065-1899 None
Keyboard Language 065-4161 Apple Keyboard & Mouse
Mac OS Language 065-4160 Mac OS X
Country Kit 065-4159 Country Kit

Looks like I better get my camera ready....

I almost fell off my chair when I saw shipped...as you can see above it was CTO ordered the day after at an Apple Store in NJ...

Hopefully I will have pictures of me chasing the FedEx truck down the street....

Good luck to everyone else hope they come soon...

My est ship date was 22 September but the Apple care person said it would go out on the 20th....

Ok now I can go back to loving Apple and Steve again.... hopefully the rest of you will too...
_________________________________________________________________________

 

[solvs]

Originally posted by Mav451
dude did you even READ my post? I mention OCing b/c they display results for the OCed Xeon (2.4 @ 3.06) and "old" 3.06 @ 349.

2ndly, where the heck do i say that this test is inaccurate? Did you come up with that misconception by yourself?

Don't stereotype and lookdown on overclocking just because you do not understand it.

Look at my sig--it is 100% safe b/c the only issue with overclocking is too much voltage (i.e. 1.8v +, i'm not in that range), PCI bus overclock? (nope, nforce2 and all modern Intel boards lock them at 33, agp @ 66 respectively).

And actually the #1 cause of failure is heat. Most newbies do not understand that they cannot overclock with "stock" equipment. Spending 30bucks for that Swiftech or Thermalright SLK will pay HEAVY HEAVY dividends in increased stability, more headroom, and overall lower case temperatures.

Dude, relax. That wasn't a personal attack. When I hit reply, your post wasn't even there. I just noticed it as I began typing the comment that you might want to read again. Perhaps it would have been better if I tossed in a or a to better make my point. You know, "yeah but overclocking doesn't count". Ha ha, you know. It's cool, we're all friends here.

And I could tell you about my custom built PCs, like the one I'm currently running (sorry, but VPC is slow, and it was cheap, and I had lots of parts laying around, and for some stuff... I just need a PC). An AMD running at about 30 - 40 degrees load, custom cooling. I know all about OC'ing, I've done it, but it wasn't worth it to me. I prefer it to be quiet and fairly cool because what little I do with it doesn't require a lot of speed.

Apparently, you do, which is cool with me. I know all about OC'ing, the goods and the bads. And I'm very familiar with PC motherboards thank you very much. Intel and AMD. It's what I do.

My point was, that people are going to be arguing over these tests. It's good to see them, but even "real world" tests like these don't mean much. The G5s are fast. Yippy.

I apologize for making it seem like that was a slam against you. It wasn't.

 

[Mav451]

ok cool. I think OCing is getting your money's worth.

I paid 90bucks for my processor. It is reaching the same speeds as a 400 dollar one.

And cooling? It is not loud. All it took was a Thermalright SLK800 (30-40ish bucks) and a quiet Panaflo H1A (considerably quieter than its competitors.

39cfm for the same dba (32) as most case fans? Insane.

I replaced 4 of my case fans with panaflos and the sound literally dropped 20-30%. And the temperatures stayed the same, or got even lower (probably with the SLK800 that replaced the Volcano6cu).

Overclocking doesn't have to be loud
It just has to be done right--and with the right equipment. You GET WHAT YOU PAY for. (i.e. don't expect Coolermaster case quality for only 50 bucks).

 

[ddbean]

good to see...

good to see real folks receiving their orders.

I'm updating from a 1994 6100/60 PPC w/233 Mhz G3 card, 2 gig hd, 72 MB ram, 4x cd, borrowed 15" monitor running OS8.6 to a Dual 2Ghz G5 w/ ATI Radeon 9800 Pro and 20" Apple Display. I ordered through local ind. Apple dealer and all they tell me is maybe before end of month. Paid check in full on July 27.

I want my G5 now!!!

 

[pjhornak]

Re: good to see...

Originally posted by ddbean
good to see real folks receiving their orders.

I'm updating from a 1994 6100/60 PPC w/233 Mhz G3 card, 2 gig hd, 72 MB ram, 4x cd, borrowed 15" monitor running OS8.6 to a Dual 2Ghz G5 w/ ATI Radeon 9800 Pro and 20" Apple Display. I ordered through local ind. Apple dealer and all they tell me is maybe before end of month. Paid check in full on July 27.

I want my G5 now!!!

brah... that is hardcore oldschool....do you ride a longboard...

 

[Genie]

Very exciting!

I feel like I did when I got my first G4, the first G4 model made, which is still performing very well for me.

 

[sososowhat]

Re: Re: So where can I find that benchmark to try it myself?

Originally posted by nospleen
Someone give this man a link. Have you xbenched yet? My stock 1.8 was 154, I would like to see what the ram does to the score.

Ok, here you go: XBench on my 2.5Gig 1.8. Note: I don't know why it says I've got 2.0Gig memory. "About This Mac" confirms I've got 2.5.

Results 159.65
System Info
Xbench Version 1.1.1
System Version 10.2.7 (G5) (6S80)
Physical RAM 2048 MB
Model PowerMac7,2
Processor PowerPC 970 @ 1.80 GHz
L1 Cache 64K (instruction), 32K (data)
L2 Cache 512K @ 1.80 GHz
Bus Frequency 900 MHz
Video Card GeForce FX 5200
Drive Type ST3160023AS
CPU Test 169.67
GCD Loop 100.74 3.93 Mops/sec
Floating Point Basic 291.19 1.05 Gflop/sec
AltiVec Basic 123.40 3.58 Gflop/sec
vecLib FFT 192.86 2.99 Gflop/sec
Floating Point Library 354.60 14.19 Mops/sec
Thread Test 105.67
Computation 75.49 604.43 Kops/sec, 4 threads
Lock Contention 176.03 2.21 Mlocks/sec, 4 threads
Memory Test 275.95
System 279.79
Allocate 465.03 313.58 Kalloc/sec
Fill 193.70 1541.84 MB/sec
Copy 293.31 1466.53 MB/sec
Stream 272.21
Copy 230.57 1685.45 MB/sec [G5]
Scale 237.68 1754.10 MB/sec [G5]
Add 315.74 2020.71 MB/sec [G5]
Triad 335.27 2048.48 MB/sec [G5]
Quartz Graphics Test 188.71
Line 211.66 5.39 Klines/sec [50% alpha]
Rectangle 188.92 13.29 Krects/sec [50% alpha]
Circle 206.49 4.76 Kcircles/sec [50% alpha]
Bezier 174.37 1.89 Kbeziers/sec [50% alpha]
Text 169.48 2.76 Kchars/sec
OpenGL Graphics Test 212.53
Spinning Squares 212.53 148.73 frames/sec
User Interface Test 179.32
Elements 179.32 57.68 refresh/sec
Disk Test 107.72
Sequential 113.23
Uncached Write 131.10 52.19 MB/sec [4K blocks]
Uncached Write 125.60 49.02 MB/sec [256K blocks]
Uncached Read 85.70 13.57 MB/sec [4K blocks]
Uncached Read 123.92 50.07 MB/sec [256K blocks]
Random 102.72
Uncached Write 100.42 1.44 MB/sec [4K blocks]
Uncached Write 100.22 22.60 MB/sec [256K blocks]
Uncached Read 99.70 0.66 MB/sec [4K blocks]
Uncached Read 111.43 22.93 MB/sec [256K blocks]

 

[Genie]

How soon do you think PC3200 DDR400 Gigabyte Ram sticks will be cost effective? (around $200, which is what two 512's cost).

 

[mrwalker]

Originally posted by Kid Red
...i can't figure out why this article has 2 negative votes...

Well, this IS negative for those favoring the PC... Some people do, y´know...

 

[macphoria]

This is definitely a promising news, of G5's capabilities. I hope this is still the case when 3GHz G5 comes out next August.

I haven't heard much about AMD or Intel's development. Anything new going on with those guys?

 

[elmimmo]

Re: PowerMac Dual 2.0GHz PS7Bench Benchmarks

Originally posted by Macrumors
The new results have been added to the previous PS7Bench Benchmark chart. The PowerMac G5 Dual 2.0GHz results are the fastest in 15/21 of the tests run.

Why is everyone so surprised about? It has 2 frigging 2 GHz processors in it. It had BETTER be faster that one single Athlon 3000+, which does not even run at 3 Ghz.

The real winner, IMHO, is the G5 1,6. Compare figure by figure to the Athlon 3000+ and you will see that both computers are pretty much on pair.

 

[jettredmont]

Originally posted by mvc
Well, I'm a true mac zealot, but I don't expect we will be faster in everyway and every test. We have always done well in Photoshop tests.

Quite true. As is always the case, we'll win some and we'll lose some. I do think that we'll see ourselves winning a lot more frequently in the next couple of years than in the last couple.

It's inconceiveable that, over time, the mainstream PCs will not naturally be superior in speed that a niche player like the Mac - this current good result is really a sweet spot due to a good new architecture competing against either old / inefficient / overstretched / poorly implemented or undeveloped ones.

The chief advantage of the PPC platform, as others have pointed out, is that the architecture itself doesn't have an unweildy legacy to support (80086 instruction set support). The architecture was designed for the current state of the art, designed for today, for 64-bit computing.

The primary advantage of Intel is that its instruction set embodies the vast majority of non-custom software in the world. This is also its disadvantage, as it is next to impossible for Intel to innovate (Itanium has some good ideas in it, really) without bolting on the clumsy 1970's-designed 8086 compliance layer (IMHO the primary downfall of Itanium).

Logically, it cannot stay this way when so much more money overall is being poured into the competitions R&D, no matter how clever IBM's engineers are. Only if we somehow gain a more major marketshare (say over 10%), will there be any real hope of sustaining this sort of result.

How do the R&D budgets compare? I don't have exact figures, but IIRC the IBM microproc division's R&D budget is very close to if not equal to Intel's microproc R&D budget.

The key here is that IBM makes serious money off its server processors, far moreso than Intel.

Ask yourself, why did Motorola go from being a successful fast chip maker to a laggard? (apart from general incompetence)?

Motorola made fundamental business decisions which ruined them. Without going into a hundred-page analysis of what went wrong when, I think the primary decision was based on the supposition that the embedded devices market was going to be more lucrative and stable than the desktop processor market. Moto starved its microproc division of R&D funding, and made the fundamental decision that such was no longer their focus.

I believe the more important moral here is that no measure of dominance is permanent, and that at the end of the day it is more often business decisions that destroy technical dominance rather than technical failings destroying business dominance.

Such a cautionary tale is as applicable to Intel as it is to IBM.

Because they don't care about the Powerpc as a standalone Computer platform, because it doesn't make enough money. IBM has a more vested interest with their servers using related Power4 architecture, but there are no guarantees they will advance the consumer spinoff 970-9X0 chips as fast.

Correct. There are never any guarantees. However, it is far easier to spin off a 9x0 chip from a POWERx chip than it is for Intel to apply its Itanium research dollars to its Pentium line. This gives IBM a distinct advantage in consumer-level hardware research: it is a low-cost by-product of a strong and stable business. IBM might of course decide that this business is not in its best interests, but I think that of all the contenders for Intel's throne, IBM stands the tallest (yes, moreso than AMD).

In other words, if IBM's model for consumer chip development can't compete with Intel's dominance then I don't think anyone will be able to.

 

[jettredmont]

Originally posted by Raiden
hey guys lay off the 600 ibook! Its not as bad as the pitiful benchmark scores say. I mean the one im on right now runs jag great.

and yeah, those dual G5 scores are insane!! w00t for apple!

Note, of course, that the ibook fares so poorly because it does not support Altivec. It is somewhat remarkable, in fact, that the ibook without SIMD bests the P2 with MMX and, I believe SSE. Of course, I'm not positive that Photoshop takes advantage of MMX; it may require SSE or SSE2 on the Intel side ...

 

[jettredmont]

Originally posted by Mav451
Don't stereotype and lookdown on overclocking just because you do not understand it.

Look at my sig--it is 100% safe b/c the only issue with overclocking is too much voltage (i.e. 1.8v +, i'm not in that range), PCI bus overclock? (nope, nforce2 and all modern Intel boards lock them at 33, agp @ 66 respectively).

And actually the #1 cause of failure is heat. Most newbies do not understand that they cannot overclock with "stock" equipment. Spending 30bucks for that Swiftech or Thermalright SLK will pay HEAVY HEAVY dividends in increased stability, more headroom, and overall lower case temperatures.

Okay, show of hands.

How many people here would overclock their work machines?

How many IT folks here are ready to overclock their servers?

Overclocking is all well and good for non-mission-critical machines, but it does introduce consequences (less room for error, significantly increased reliance upon lack of failure in heat dissipation devices, and of course reliance upon a CPU operating under conditions for which it was never tested or certified). This is fine for the 1337 gamrz, but not exactly commonplace in the world of Photoshop business users.

That having been said, overclocking current CPUs does often give performance insight as to upcoming CPUs (as is the case with the Xeon). In the "currently available" benchmarking world, however, overclocked results are to be ignored as they are currently irrelevant.

 

[Titian]

PS Benchmarks

I am a little bit confuse:
is PS the only application we can run on a Mac?

 

[Analog Kid]

Re: good to see...

Originally posted by ddbean
good to see real folks receiving their orders.

I'm updating from a 1994 6100/60 PPC w/233 Mhz G3 card, 2 gig hd, 72 MB ram, 4x cd, borrowed 15" monitor running OS8.6 to a Dual 2Ghz G5 w/ ATI Radeon 9800 Pro and 20" Apple Display.

Sweet! Welcome to modernity!