PDA

View Full Version : When one million equals 4 million


Bobo
10-04-2008, 08:54 AM
I assume the thread started by Jigu was closed because of the language in the title, not because of the questions he asked.

So I am going to comment here, if you moderators don't mind.

While I haven't looked at the rest of the files and applications and admit that Max does not scale well on multi-core systems yet (except when rendering), I opened the file entitled "One_Million_Particles.max" which was part of the package Jon Peddie Research and Avid / Softimage produced.

I really have to wonder whether they believe nobody is going to look at that file.

Imagine my surprise when I opened the scene to find a BIRTH RATE of 1 million particles, which as anyone who can multiply knows is equal to 1 million particles every second. At 24 fps, this results in 4,166,667 particles at the end of frame 100. Multiplication skills and not even necessary, as the Birth Operator actually reports the total particle count in a field below the birth rate...

Now I can understand that somebody not familiar with Max could make an honest mistake to create over 4 million particles using 1 million birth rate, then name the file "one_million_particles.max" and use it in a benchmark. But whoever made the scene actually took care to go to the System Settings and set the upper limit of the system to exactly FOUR MILLION particles, not one.

Quote from the white paper:
"One Million Particles: In the one million particles test, the animation builds as one million particles are generated. Once again, we see that Maya is better able to handle the overhead imposed by Vista but XSI is the fastest in XP. Throughout these tests, itís clear that Max is challenged by particles."

The corresponding graph shows Max being about 3 times as slow creating 4 times more particles "by accident".

I am all for quantifying how fast XSI is, but let's stay fair and objective if possible.

As Jigu mentioned, any single-CPU machine can generate that many particles on a single frame in around 2 seconds. Krakatoa renders around 1.5 million per second. With two lights, it would take 5 seconds. Btw, half of the time in the benchmark is spent drawing points in the viewports. With reduced Display>Visible settings, it really flies, so the generating of particles is not an issue.

erilaz
10-04-2008, 09:08 AM
It was closed mainly due to it starting with a rather aggressive "this is better than that" road, which ends up in fruitless arguing.

We're okay with objective discussion on the topic, but if it starts getting messy then we'll have to close it again.

jigu
10-04-2008, 09:15 AM
I apologize for bad title of thread. I didn't have intention to create Application war thread. I felt wrong in comparison. As most of max users know that pflow uses only one core so felt to correct misleading information of max going on.

Thanks for closing that thread, I wanted to apologize for the bad title of the thread(or it was a bit offensive language) but thread were closed before that. I wanted to make information fair for max not misleading.

jigu
10-04-2008, 09:23 AM
Thanks Bobo for you comments and correcting it.

Also isn't that in that file max has to render 2 million of polygones of tetra shape after 1million of particles calculated in 2 secs?

erilaz
10-04-2008, 12:41 PM
I apologize for bad title of thread. I didn't have intention to create Application war thread. I felt wrong in comparison. As most of max users know that pflow uses only one core so felt to correct misleading information of max going on.

Thanks for closing that thread, I wanted to apologize for the bad title of the thread(or it was a bit offensive language) but thread were closed before that. I wanted to make information fair for max not misleading.

No worries jigu. We just need to be aware of the community at large. :)

JohnnyRandom
10-04-2008, 05:48 PM
Well not having opened the xsi or maya files either, when adjusting the particle count to a rate of 24000 to get a generation of 1 million particles (my completely unorthodox method of other applications running, not a fresh boot, testing on a laptop, and typing this reply) has generated a time of 31 seconds which seems to fall fairly close to XSI time, considering I am not running an 8 core with 16 gig of ram:)

I really wonder what is messed up in Maya scene file? Anybody got Maya to try it?

I am all for better performance in ANY application, seems like a very biased comparision to me too, just read the white paper, it is not very concise.

PiotrekM
10-09-2008, 12:00 AM
The benchmark is OK.

Both MAX and XSI generate 1 million particles per second.

On frame 20:

MAX hitting 0,3 fps, XSI 4,1 fps

MAX ~2-6 sec lag when trying to rotate/pan/maximize viewport, XSI ~17fps

pflow is slow and outdated. like max.


(im max user)


(dual quad core xeon, 8800 gts, vista x64, max 2009, xsi 7)

spacefrog
11-08-2008, 03:31 PM
Ha - just found this thread !

I too had some very strange findings in those scenes ..
these tests are HEAVILY biased against Max

Look at this thread in Max's main forum for what i found till now
http://forums.cgsociety.org/showthread.php?f=6&t=693497

EDIT:

i opened a new thread with results from a rerun i did with corrected Max scenes
Exact test-descriptions in a test PDF etc.

http://forums.cgsociety.org/showthread.php?f=6&t=693979

CGTalk Moderation
11-08-2008, 03:31 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.