saphires speed test Combustion vs. flame -Update 14avril-

[luc]

Hello

i have made the same test as Peter Webb with the saphirres plugins. The result is here and some more result whit hd footage is at the bottom of the post of peter at this link: http://www.fxguide.com/article183.html&mode=&order=0&thold=0

_____________/ mac G4 ________/ pc ________/ flame _____/ G5 ________/ Tezro ????? _____

RDefocus || 1 min 40 sec || 0 min 8.5sec || 0 min 25 sec || 3 sec ||

EdgeRays || 1 min 48 sec || 0 min 20 sec || 1 min 26 sec || 4 sec ||

WarpDrops || 5 min 06 sec || 0 min 31 sec || 1 min 09 sec || 21 sec ||

—Glows— || 1 min 04 sec || 0 min 7.5sec || 0 min 18 sec || 4 sec ||

AutoPaint || 1 min 43 sec || 0 min 10.5sec || 0 min 34 sec || 9 sec ||
_________________________________________________________

MAC: macintosh G4 17″, powerbook 1ghz G4 CPU with 1GB of RAM.
MAC: macintosh G5 dual 2ghz CPU with 1.5GB of RAM.and a ATI Radion 9600
PC: Dual Xeon 3.06ghz Hypertrading with 2GB of RAM and a nvidia quadro4 980XGL.
Flame 8.3.1 on an Octane dual 400mhz with 2GB of RAM.

thanks Peter .

luc julien

[Anonymous]

Interesting results
Just one point we asked Peter to test on mid range gear – and not on a Tezro, 8 proc Inferno, or a dual G5… so I guess we should look at doing some benchmarking on the latest gear as well to get a full picture.

mike

[luc]

Hi

if any one want to participate to this saphires speed test, feel free to send your result in reply of this post. Mostly if you run a dual G5 or a tezro.

Quote:

This is the setup of Peter for the test:

I used a 50 frame PAL sequence of images … tiffs on a local framestore … and loaded the Sparks named in the comparison as operators. I used the default settings for them all except for the rack defocus.

For that I animated it from a .4 value to a 1.2 value.

Thanks for your participation !

luc julien

[Anonymous]

i agree. quite interesting.
however, this is like looking at tests of home user magazines just focusing some peaks.
i hope pros know better.

but i think most important is the interactive working and i guess there’s no doubt… 😀

[eltopo]

The mac test is very unfare compared to the PC and SGI machine.it’s a laptop with less memory and only one processor. however it still manages to male very decent times.

Ibvously, the saphire test on a G5 optimized (64bit) for it, would be more accurate.

Also, on it is important to state on what application were tthe saphire applied from

[John Montgomery]

Quote:

The mac test is very unfare compared to the PC and SGI machine.it’s a laptop with less memory and only one processor.

C’mon eltopo, your bias towards the mac platform seems to be clearly clouding your judgement and really seems to take away from the value of all of your other posts. Don’t get me wrong — I love the mac –all of the personal systems I own are Macintoshes. However, I’ll be adding a PC to the mix soon…just makes sense from a professional standpoint.

But unfare (sic)? What are you talking about? The tests clearly state what hardware was used — nothing is hidden at all. And the powerbook test is really useful, considering the idea that it is a great device for on-set work:

MAC: macintosh G4 17″, powerbook 1ghz G4 CPU with 1GB of RAM.
PC: Dual Xeon 3.06ghz Hypertrading with 2GB of RAM and a nvidia quadro4 980XGL.
Flame 8.3.1 on an Octane dual 400mhz with 2GB of RAM.

Quote:

however it still manages to male very decent times.

Yeah…the times in general are not bad and I’m glad to have the resource. ,But 8X, 6X 10X, 6X, and 9X slower than the (very fast dual PC) aren’t that stellar.Not as bad vs. the flame, but stil…..

[/quote]

[Rayk]

eltopo, john,

maybe, both of your vision is a bit clouded 🙂

first, I’m no mac-fan, at all -for the most part of it, because of the attitude of the hard-core-mac-fans and because I consider the mac as overpriced for its performance.
second. as john pointed out, the hardware used was clearly stated. so why not have a look at it: there’s a powerbook utilizing a 1GHz G4 with 1MB ram and on the other side there’s a dual Xeon with 3.06 GHz each and 2MB of ram. The later one has about 5 TIMES the processing power of the powerbook (that is, if you take some overhead into account and assume a similar performance on a GHz by GHz basis). so, if you look at it from that perspective, the mac DOES NOT perform so bad compared to the pc as john indicated.
also, I could do a price performance comparition. but that would be fair at all 😀 (in case you wander, the pc cost 4 times less per second render time compared to the powerbook, but as I said, the comparition isn’t fair at all)

-rayk

[gcapps]

I just had to jump in on this one.. I use both Mac’s and PC’s at work each day, and hate when similar boxes aren’t compared during these types of ‘tests’… So… a dual Xeon 3.2 with 2 gigs of ram and a Quatro Gfx card costs how much these days? last time I checked (which honestly has been a few months) it was comparable to a G5 with 2 gigs of ram etc.

We have a dual G5 with 1.5 gigs of ram, so I decided to see how it fares against the dual intel box.. here are my results

I downloaded the sapphire demo for AE, and ran it on 50 frames of PAL video.. My results are as follows

RDefocus 3 seconds
EdgeRays 4 seconds
D_Warpdrop 21 seconds
Glows 4 seconds
Autopaint 9 seconds

All my times were applying the default settings of the plug-ins to the clip…

let me know if I missed something in the test setup

p.s. just checked.. from BoxxTech, a dual 3.2 Xeon with 2 gigs and a quatro 980 goes for $4,466

A Dual 2ghz G5 from Apple, only has a Radion 9600 in it, but I couldn’t configure it with a Quatro (wonder if they will write drivers), sells for $3,449..

RDefocus || 1 min 40 sec || 0 min 8.5sec || 0 min 25 sec

EdgeRays || 1 min 48 sec || 0 min 20 sec || 1 min 26 sec

WarpDrops || 5 min 06 sec || 0 min 31 sec || 1 min 09 sec

—Glows— || 1 min 04 sec || 0 min 7.5sec || 0 min 18 sec

AutoPaint || 1 min 43 sec || 0 min 10.5sec || 0 min 34 sec

[renderizer]

Interesting figures indeed.

And just to add my $0.02 to the whole apple vs. pc war (crap!): some friends of mine have just bought a G5, and I have to say that I am quite impressed by it, yet I am so used to the PC side of things that I’d have some diffculties with MACs…and in my book, mice should have two legs…er…buttons. 😉

I’d love to go with dual 64-bit AMDs, even though there is no optimized software yet…

Anyway: I just got rid of my job, so I’m not going to buy any new hardware in the near future. 😥

[gcapps]

I have a 1.42ghz G4 on my desk with a Belkin 3 button mouse, and a dual 1.7 P4 with a Logitec Mouseman 3 button mouse (I like the PC’s mouse much better btw)… KVM switch is built into my monitor, so I can pop back and forth between boxes… makes long renders great when you need to composite as well… I run Maya , AE, etc. on the PC mostly because it’s faster than my G4.. but we have a G5 here that is faster than both (just not on my desk 🙁 ) they’re all just tools.. whatever helps you go home on time at night is the box I want to use 🙂

p.s. I just realized that I didn’t use Combustion as the host platform.. I don’t know if there is additional overhead with the AE plug -> Combustion translation… might explain the time delta?

[luc]

Hi folks,

finaly i have make the bench of the sapphires on a brand new tezro. The result is not very impressive…

So… The G5 win the sapphires bench.

luc julien

_______________/ pc ________/ flame _____/ G5 ___/ Tezro _____

RDefocus || 0 min 8.5sec || 0 min 25 sec || 3 sec || 5.0 sec ||

EdgeRays || 0 min 20 sec || 1 min 26 sec || 4 sec || 30.5 sec ||

WarpDrops || 0 min 31sec || 1 min 09sec || 21sec || 26 sec ||

—Glows— || 0 min 7.5sec || 0 min 18 ec || 4 sec || 7.0 sec ||

AutoPaint || 0 min 10.5sec || 0 min 34sec || 9 sec || 12.0 sec ||
________________________________________________________________

I have also test a dual amd 64bits opteron and the result is quite similar to the dual Xeon

[Rayk]

really very interesting. thank you for testing.
would be interesting to see the gap on 2k or 4k image sequences.
also, as far as i heard it, the sgi never were the fasted in rendering, but were/are alsways/still unbeatable in interactivity with high-res footage.
what is your impression?

-rayk

[Author]

Posts