PDA

View Full Version : MacBook Pro and Win XP! Render test.



sonofmickel
11-12-2006, 02:03 PM
I just did an unscientific test on the Virus scene located in the Hypervoxels in the LW9 content.

MacBook Pro Core dou 2 Ghz, 1 GB Ram. LW 9. Tiger 10.4.
render time: 1M 50 sec

Same MacBook Pro on the Win XP partition.
render time: 36 sec!

Holy cannoli!

I wonder how the quads(and new 8 procs) would do in XP? The speed makes OSX look silly. I say OSX looks silly because it is purely Apples own doing(you know, 3 layers of operating system vs XP's 2 layers). Maybe the new OSX will rectify this.

Scazzino
11-12-2006, 02:30 PM
Unless you're running a universal binary of LightWave, then it's running under Rosetta emulation which is why it's so slow in Mac OS X. You won't be able to do a proper speed test until the Universal Binary version of LightWave is available...

sonofmickel
11-12-2006, 03:50 PM
True that!
I just tested on my Powermac G5 dual 2.3 with 4 GB ram.

render time 41.7 seconds!

Still my slower MacBook Pro is faster than my G5.

I guess Apple was right= the megahertz myth.
A slower PeeCee chip is faster than a more powerful G5.
Strange daze these are.

sonofmickel
11-16-2006, 10:27 AM
Asper request, screen grabs.

Scazzino
11-16-2006, 10:32 AM
Not quite sure what you're trying to test...

You won't be able to get a real Mac/Win speed test until the Universal Binary is out, then you can test LightWave Mac/Win on the exact same machine, both running natively...

When you do, you'll also want to set the preview to the same size for both tests, since the larger redraw will also slow it down further.

Scazzino
11-16-2006, 10:53 AM
Even then, I'm not sure if it will be a completely fair test of the hardware speed itself, since the Windows version may have some optimizations that the Mac UB may lack...

Just by being developed on Windows tends to give Windows a leg-up since all the "cross-platform" code will be written and tested on Windows before it ever gets ported to the Mac... so certain choices may be made during the Windows development that may give Windows a built-in speed advantage...

If ALL the developers were developing and testing the software equally on both platforms ALL the way through the process, then I'd be more confident that it would be a completely fair test of the differences in hardware speeds.

So if you're after seeing which hardware is faster itself (to test things like the megahertz myth) then you'd be better off using a truly cross-platform benchmark tester, rather than any software that is developed primarily on one platform and ported to the other...

If you're only after which platform runs LightWave faster, then the Universal Binary should indeed be able to tell us that when it's ready... ;)

byte_fx
11-16-2006, 12:11 PM
Granted there's really no way to compare speed among OS's.

However - my Mac Pro running LW under XP is noticeably faster than my faster cpus dual Xeon accross the board. Both running eight threads.

Difference most likely due to an extra core being more efficient than HyperThreading, the faster, dedicated bus and core cpu design differences.

Don't quote me because my memory may be lax but seem to recall the virus scene default frame was something like 8 seconds on the Mac and 12 seconds on the dualie.

And the radiosity_box scene was around 5 minutes on the Mac and 6 minutes on the dualie.

Those times were under 8.0 - not 8.3 or 8.5.

And it's been a few weeks since I ran the tests. Both were done after fresh installs and XP was set up identically.

I'll try to remember to rerun them when I get a chance.

Point is the Mac Pro is wicked, wicked fast and - to me - the best bang for the buck hands down.

byte_fx

Captain Obvious
11-16-2006, 01:56 PM
True that!
I just tested on my Powermac G5 dual 2.3 with 4 GB ram.

render time 41.7 seconds!

Still my slower MacBook Pro is faster than my G5.

I guess Apple was right= the megahertz myth.
A slower PeeCee chip is faster than a more powerful G5.
Strange daze these are.
Um, what the heck are you talking about? :confused:

There really is no megahertz myth, because the idea that megahertz = speed is just too goddarneded idiotic to be taken seriously.

The Core 2 Duo chip is amazing, and the Core non-2 Duo is no slouch either. I've seen plenty of tests where the lowest-end Core 2 Duo (1.86GHz) outperforms the 3.73GHz Pentium D. Twice the clock frequency, worse performance. What does that tell you?

Apple certainly never claimed that the G5 was faster than the Core (2) Duo. Why do you think they switched to Intel in the first place?



And, as others have pointed out, you're running Lightwave in an emulation layer.





(you know, 3 layers of operating system vs XP's 2 layers)
No, I most certainly do NOT know. Please elaborate.

Chilton
11-16-2006, 02:29 PM
Hi everyone,


Asper request, screen grabs.

Just FYI, I asked sonofmickel to post these screens, and I am very glad he did. Performance is something I care very deeply about, and wanted to make sure I have screenshots to verify all benchmarks.

I've been using Chris' LightWave Benchmarks as well. What I'm really interested in right now is the performance of LW on WinXP, on Intel based Macs. If anyone has WinXP and LW9 on a MBP 2.1Ghz, I'd be very interested in any benchmarks from that.

Thank you,
-Chilton

John the Geek
11-16-2006, 05:38 PM
If anyone has WinXP and LW9 on a MBP 2.1Ghz, I'd be very interested in any benchmarks from that.

Raises hand. Will the Virus sample do? I'll dig it up as soon as I can get Win XP on this bad boy.

Let me know.

=)

Chilton
11-16-2006, 05:57 PM
Hi John,

Yes, Virus is perfect. That's the one I use the most here.

-Chilton

John the Geek
11-16-2006, 06:38 PM
Single frame is average 35 seconds. (34.6 to 35.1 over a few renders of frame 0.)

Screen cap to come if I can get it to work.

John the Geek
11-16-2006, 07:33 PM
MBP 2.16GHz with 2GB RAM.

It varies, so I posted two...

=)

dglidden
11-16-2006, 08:06 PM
Hi everyone,

Just FYI, I asked sonofmickel to post these screens, and I am very glad he did. Performance is something I care very deeply about, and wanted to make sure I have screenshots to verify all benchmarks.

I've been using Chris' LightWave Benchmarks as well. What I'm really interested in right now is the performance of LW on WinXP, on Intel based Macs. If anyone has WinXP and LW9 on a MBP 2.1Ghz, I'd be very interested in any benchmarks from that.

Thank you,
-Chilton

Oh sigh. That means I have to boot into windows again? Well if it makes the UB version better (or come out sooner... heheh).

BTW chilton, did you see my thread I started titled (I think) MacBook Pro Core 2 Duo? I posted a couple of other "benchmark" numbers from a different scene. This is a brand new MBP C2D 2.33Ghz. Is that an acceptable platform for testing? Also 9 or 9.2? Or are you specifically comparing against the 2.16Ghz MBP Core Duo?

Chilton
11-16-2006, 08:16 PM
Hi,

I'll take any benchmarks I can get screenshots for!

-Chilton

dglidden
11-16-2006, 09:53 PM
whew! it feels good to be back in OS X.

Here are a few renders with screencaps. Hope they are helpful.

MacBook Pro
Core 2 Duo 2.33Ghz
2GB RAM
LW9
XPSP2

Chilton
11-17-2006, 12:59 AM
Hi John,


MBP 2.16GHz with 2GB RAM.

It varies, so I posted two...

=)

Could you do that one more time, but with Adaptive Sampling turned off?

Thanks,
-Chilton

John the Geek
11-17-2006, 05:45 AM
Hi John,
Could you do that one more time, but with Adaptive Sampling turned off?

I think it is. It's unchecked in Render Globals, and the value (0.1) is greyed out so I can't change it.

Should I enable it, then set it at 0, and then re-disable it?

EDIT: Tried that, disabling it resets it to a greyed out 0.1.

avkills
11-17-2006, 06:36 AM
Can you guys do those renders with only 2 threads; supposedly LW9 runs faster if you set it to the number of cores or CPUs present.

-mark