NASA Benchmarks Power Mac G5

[amigamad]

NASA recently benchmarked Apple's dual 2GHz Power Mac G5 at its Langley Research Center in Virginia. The main purpose of the tests was to compare the G5 to the G4 for "computational fluid dynamics applications" however they also compare it to the Pentium 4. the link is here osnews

[Fot]

The 94 page PowerMac G5 Developer Note PDF is an interesting read.

[Ilwrath]

Hmm....  33% faster than a P4 clock-for-clock on floating point.  Not that impressive.  That'd put the 2.0 ghz G5 almost even with a 2.8 ghz P4 on the G5's strongest mark (FP) while everything else lags back even further...  Considering 3.2 ghz P4s are out in consumer level, while the G5 isn't yet...  Hmm...  I was hoping for a little stronger showing from the spunky underdog.  Hopefully they can ramp up the clock on the G5 some more...

[dammy]

I'm still waiting for a head to head 64 bit vs 64 bit CPU comparison on like (technology wise) mobos.

Dammy

[mikeymike]

If leading the hardware market was that easy, there wouldn't be just the players who have been in it a long time.

Perhaps the computer hardware market is in need of an entirely new platform?  It's usually how leapfrogging occurs, look at the Amiga and the PS2 for example.

There's a big difference between the circumstances of the two platforms though, the PS2 had Sony's full backing behind it, which means a lot of money.

[Melaure]

Read the NASA test until the end ( http://members.cox.net/craig.hunter/g5/ ).

The Apps were not compiled for G5 but for G4 (monoproc) and without Altivec. With a G5 optimisation, the G5/2Ghz should match the P4/3.2 Ghz or even beat it by 20%. These tests consist only in floating point calculation.

In a second phase, they have done Vector Benchmarks. In this case the Altivec is used and their ratio Mflops/mhz had explosed !!!
There is no bench for P4 which does not have a vector unit and I'm sure that the result would have been very poor ...


The G5 rocks, believe it ! ;-)

[dammy]

Poster: Melaure Date: 2003/7/5 11:59:05

The Apps were not compiled for G5 but for G4 (monoproc) and without Altivec. With a G5 optimisation, the G5/2Ghz should match the P4/3.2 Ghz or even beat it by 20%. These tests consist only in floating point calculation.

In a second phase, they have done Vector Benchmarks. In this case the Altivec is used and their ratio Mflops/mhz had explosed !!!
There is no bench for P4 which does not have a vector unit and I'm sure that the result would have been very poor ...


The G5 rocks, believe it !

970 is a good CPU, no doubt about it. Let's look at what the report actually said:

Based on raw scalar floating point performance in Jet3D, a 2GHz G5 system can match a 2.66GHz P4 system, and this is a dramatic improvement from earlier tests where G4 systems lagged behind higher clock speed P4 systems. Based on an extrapolation of current P4 results, the 2GHz G5 would lag newly announced 3.2GHz P4 systems in Jet3D scalar floating point performance by about 20%, but this kind of comparison is best deferred until G5-aware compiler tools become available (since a 20% performance gain is well within the potential of compiler optimization).

Says the G5 would lag newly announced 3.2 GHz in Jet3D scalar floating point.  It's a great increase over G4, but let's keep Jobs' smoke and mirrors confined to Neverneverland.

Again, next month's release of Athlon-64 on same technology mobos as G5 should prove a realistic benchmark comparisons.

Dammy

[Waccoon]

Hmm... well, given that the same software is available on both machines, and these systems are, at the very least running neck-in-neck, the cheaper system wins.

You also have to factor in long-term upgrades.  Mobo and CPU upgrades for Macs cost a fortune, and that's not good for pure number-crunching applications.  Hell, I've already swapped my motherboard and CPU three times using the same case, cables, sound card, and DVD-Rom & CD-R.  Upgrades are Apples's real heel.

[JoannaK]

/me wonders if here is any upgrade to my old Absoft Amiga Fortran? Having one  PPC native with full Altivec support for Pega+Mos would be nice to have. (ok.. Aone+Os4 too, why not)

[Melaure]

Well but look closely at what Nasa testers wrote :

Note that the higher level of optimization (-O2) and SSE/SSE2 options in the Portland compiler degraded Jet3D performance on the P4 system, and were therefore not used.

Again, note that the vector benchmark does not include P4 systems because the AltiVec instruction set is only available on G4 and G5 systems.  Consistent with earlier Jet3D tests, the vector version of Jet3D runs an order of magnitude faster than the scalar version (speedups of 10X-13X are typical).

It means that SSE 1 or 2 is degrading performances and that Altivec accelerate the performance by 10 to 13. It's remarquable, isn't it ?

We can really count on this wonderful G5 processor, especially with dev tools optimized !!! :-D

[Panthro]

Mmmm I wasnt going to say it but now I have too...

I want this chip "panther" in my amiga!!! :-D  :crazy:

[DethKnight]

first ,Im in no way a maczealot. Actually despised them.  The G5 has at least raised a curiousity in me.

Hell, I've already swapped my motherboard and CPU three times...

bottlenecks in the subsystem usually *force* x86'ers to upgrade, (this is by market design for x86 imho)
IIRC, the G5 has more-or-less reduced/removed bottlenecks in the subsystem.

OTOH, you could probably upgrade an x86 to the next three revisions (which may have better subsytems then) for similar costs.

[Waccoon]

bottlenecks in the subsystem usually *force* x86'ers to upgrade

I really hate that remark.  Nobody forces me to upgrate.  I upgrade to get extra performance, not because the system is too slow due to bottlenecks.

I mean, if a G3 and a P4 are the same speed, but the G3 costs twice as much, is it a better long-term purchase?  No, because when a faster P4 becomes available, I can upgrade, and then have a much faster machine than the G3 without really spending more money.

If you have a G3 tower and want to upgrade, you have two choices:  Buy a G4 upgrade card for an obscene amount of money that isn't well matched to your motherboard and chipset, or throw it out and buy a new machine from scratch, which *forces* you to get a new case/CD/HD/Ram/etc. when you don't need it.  My case is several years old and still works just fine.  Why fix what's not broken?

That's why I hate Macs.  The software is cool, but the hardware is a ripoff.

I just really hate to see "alternative" OS developers use the same tactics as Apple.  I never wanted to see OS4 on proprietary hardware, because I used to use/sysadmin Macs in the past, and they are slow and expensive due to closed architecture.  To me, it makes no difference if the G5 is "slightly" faster than a PC.  For number crunching, closed architectures make no sense.

[bloodline]

Since computer subsystems (graphics, sound, netwoking...etc..) became commodities (is that the right word?) in the late 90's closed hardware became pointless...

Even the Amiga Designers in the early 90's started to plan for third party gfx cards (read some of the comment about the AAA chips set Dave Haynie has made). let us not forget that even the Hallowed Mac's use Third party sound and graphics... Tthe CPU will eventaully be come commodified, although a common CPU interface would be needed if that is to be come a totally true satement... Hypertransport anyone?

The point is that now you can get a great system with everything you can need (ie mini-itx) for the same price now as a Ram expansion would have cost you 10 years ago...  :-o

Computers are no longer special, they just are... everybody has one and they all do the same thing. All you have to choose is the OS and the price...

[Hammer]

It seems that NASA didn’t use the latest i875 chipsets (i.e. dual channel PC3200 DDR SDRAM and 800Mhz QDR FSB)....