Lightwave benchmarks at blanos?

[SponZz]

Why are there so little G5 Lightwave benchmarks at blanos.com???
There must be some Lightwave owners with G5 machines out there!

[Ade]

pc mag (http://www.pcmag.com/article2/0,4149,1274637,00.asp)

Keep in mind LW for G5 isnt even optimised yet. It does excellent against a chip thats 1ghz more.

[Lightwolf]

Originally posted by Ade
Keep in mind LW for G5 isnt even optimised yet. It does excellent against a chip thats 1ghz more.
Yeah, but don't the GHz have nothing to do with performance? ;)
Cheers,
Mike - Sorry Ade, couldn't resist.

[Meshbuilder]

Originally posted by Ade
pc mag (http://www.pcmag.com/article2/0,4149,1274637,00.asp)

Keep in mind LW for G5 isnt even optimised yet. It does excellent against a chip thats 1ghz more.

Also keep in mind that the test isn't done right..

From the PC Mag..

Raytrace test scene:
The new G5: 112 s
Dual 1,42 GHz G4: 170,5 s

My G4 dual 1,25 GHz "ONLY" at work:
104 s

Ok so the G4 at work with "ONLY" 1GB Ram, not 2 GB as the computer in the test is faster then "the faster" dual 1,4 GHz G4 and the new G5?????

Radiosity reflective things test scene:

G5: 51.10 s
G4 dual 1,42 Ghz: 72.30 s
My "slower" G4 dual 1,25 GHz: 47 s

[wapangy]

Yeah, all the scores were low, they probably had it set to only 1 thread.

[Ge4-ce]

maybe they didn't set the treads right?

maybe they used 1 thread??

Ahh... I'm sick of all this!! I'll just wait until I really can test a Dual G5 on my own..

PC-mag did not even tested the normal Benchmark scenes provided by newtek!!

And what about the photoshop stuff?? Apple claims to outperform the dual Xeon by 2 times! I seriously doubt that a company would lie about results and simply double the performance of a system!! That would be great ANTI PR!! I've seen those duals at Mac Expo in Paris, and even had the chance to talk to some people from adobe and other company's.. Believe me: These things rock! they did a lot of test there showing that the machine was very fast!

PC-Mag did not give us the technical details about the machines either.. how much ram? what cards?...

I will get my hands on a dual 2 ghz system soon, and test it intensive!

[toby]

Originally posted by Lightwolf
Yeah, but don't the GHz have nothing to do with performance? ;)
Cheers,
Mike - Sorry Ade, couldn't resist.

oh, that's not what we mean when we talk about the 'megahertz myth'

btw, why are all the benchmarks low-poly? How 'bout something with 250K polys?

[Ade]

The benchmarks are flawed, i dont think they set more threads?
MHZ myth..Intel betetr start getting used to it with their itaniums.

[DaveW]

Originally posted by Ge4-ce
maybe they didn't set the treads right?

maybe they used 1 thread??

They only used one thread, the scores make that pretty clear, and they said they rendered using default settings which would mean one thread. So it wasn't benchmarked correctly, but it is consistent since the Xeon was also set to one thread.


PC-mag did not even tested the normal Benchmark scenes provided by newtek!!

They tested a couple of the normal benchmark scenes and a couple other scenes in the content directory. You can load those scenes up and test them yourself if you want.


And what about the photoshop stuff?? Apple claims to outperform the dual Xeon by 2 times! I seriously doubt that a company would lie about results and simply double the performance of a system!! That would be great ANTI PR!! I've seen those duals at Mac Expo in Paris, and even had the chance to talk to some people from adobe and other company's.. Believe me: These things rock! they did a lot of test there showing that the machine was very fast!

Yes, the machine is fast but it isn't the first time Apple has fudged benchmarks to make themselves look better. Even Adobe claimed the dual 2ghz G5 was 1.5-2x faster than the fastest G4 system, a dual 1.42ghz. Do you think a dual 1.42ghz G4 is as fast as a Dual 3ghz Xeon? I don't. You can either believe Adobe, PC Mag, and the PSBench scores, or you believe Apples scores which nobody seems to be able to duplicate. The G5 is very fast, but Steve isn't going to get Mac-heads into a frenzy if he just says they finally caught up to the PC crowd.


PC-Mag did not give us the technical details about the machines either.. how much ram? what cards?...

Yes they did. Look at the first page of the test.

I will get my hands on a dual 2 ghz system soon, and test it intensive!

Looking forward to seeing your scores since nobody wants to believe the scores on the Internet. A coworker should be getting a dual 2ghz system within the next week or so. I'll run some LW benchmarks and post them at blanos.com.

[tallscot]

But the PSBench 7 scores shows the G5 is faster than the dual Xeon:

http://arstechnica.infopop.net/OpenTopic/page?a=tpc&s=50009562&f=8300945231&m=1150967585

I don't believe PCMagazine's numbers. They didn't use PSBench7 and excluded three filters that always do much better on the Mac. Their excuse for not using the PSBench 50 meg file was very lame - it was too fast to clock? Are they just counting "one one thousand, two one thousand"? They didn't normalize their results. They used a very large file that probably hit the scratch disk. They didn't include the price of the Dell. They jacked up the RAM to make the price of the G5 look high.

They don't know how to use Lightwave.

Acrobat, Photoshop, Squeeze, Avid Express, After Effects, and Bryce on the G5 spank the dual Xeon PC.

Lightwave on the G5 should be as fast as the dual Xeon when benchmarked properly, but we don't have definitive results either.

[DaveW]

Yes, the PSbench scores show the G5 being faster, but not twice as fast like the "Jobsbench" :)

And as for the larger file, that should give the G5 an advantage due to the faster bus and drives. A lot of places will order their RAM preinstalled by the manufacturer instead of having their IT guys waste time installing it all. Dell also charges way too much for their RAM, the price of the Dell system I configured was $4,607; PC Mag didn't say what price they got from Dell but they did say it was similarly priced. If you put your own RAM in both systems will be cheaper but the G5 will still cost less.

I'm not sure the G5 will beat the Xeon in LW though; the test may have been done wrong but it's not invalid. Those scores are what you'd see from a single cpu G5 or Xeon. I doubt the G5 will get a massive speed increase from using multiple threads. I think we're gonna have to wait for a G5 optimized version of LW.

[tallscot]

The price of the Dell configured similarly comes to $4,800 for me.

The dual 2 Ghz G5 starts at $2,999. Does the Dell? Not even close.

Lightwave 7 results are invalid. Those aren't the scores you would see from single CPU systems. Not even close. A single 3Ghz Pentium 4 does the raytrace in 85 seconds. Look at PCMag's raytrace for the Xeon. They are invalid. They don't tell us anything.

[DaveW]

There are also P4 3ghz sores for raytrace as high as 98 seconds at blanos.com. That is very close to the PC Mag score of 99.5 seconds. It tells us plenty. Anyway, Darth Mole posted his dual 2ghz G5 scores and they confirm my previous post. Good, but LW still needs to be optimized for the G5 before it's going to pull ahead of the Xeon.

I don't know why the Dell you priced was $200 more, but it doesn't matter. Everyone knows the dual 2ghz G5 is cheaper.

edit: forgot to add, the PC Mag scores would also have Show Rendering in Progress at 320x240. That adds a couple seconds to the rendertime.

[tallscot]

There are also P4 3ghz sores for raytrace as high as 98 seconds at blanos.com.

Not with 2 gigs of RAM, there aren't.

Darth Mole posted his dual 2ghz G5 scores and they confirm my previous post. Good, but LW still needs to be optimized for the G5 before it's going to pull ahead of the Xeon.

63 seconds for raytrace is very good for an application that is written for a different architecture.

I never disagreed that LW needs to be optimized. I only disagree with the notion that the G5 isn't a much faster system. Maybe that wasn't your implication at all.

Other applications, like After Effects, Squeeze, Acrobat, and Avid Xpress are incredibly fast on the G5, and again, with no optimizations.

There's no doubt that at $2,999, the dual 2 Ghz G5 is an incredibly value.

[DaveW]

Originally posted by tallscot

Not with 2 gigs of RAM, there aren't.

True, but that scene is hardly RAM dependent. A 3ghz with 2GB got 85 seconds, and a 3ghz with 512MB got 87 seconds. And a 3ghz with 1GB that got 91 seconds. And unlike the PC Mag tests, these presumably have show render progress turned off. Since PC Mag had that turned on it will add a few seconds to the render time. So like I said, it's incorrect but at least it's consistent and does provide a good indication of how the G5 compares in LW.


I never disagreed that LW needs to be optimized. I only disagree with the notion that the G5 isn't a much faster system. Maybe that wasn't your implication at all.


I think you just misunderstood my post. I'm not suggesting the G5 isn't a faster system, just not 2-3x faster than the Xeon like the Apple benchmarks show. Ge4-ce seemed to think the Apple benchmarks were valid, and anyone who thinks that is going to be disappointed. The PC Mag benchmarks might have been biased, but at least they are more realistic than Apple's claims.

[SponZz]

The first real benchmarks are posted on blanos, pretty good too.
I think I can finaly order my G5!