Maxwell Render | Official public forum

Very interesting.

Thank you.

DELETED

thank you hrothgar....
but do you also I think these graphs are very huge in terms of pixel size...

edit: thanks for making them smaller, now they need antialiased text

Isn't there a mistake with the Dual Opteron score? I think it should say 2 threads, otherwise you're only using one processor......I think.

I'm doing the render on my dual 1.8 G5 and timing with a stopwatch...

I'll post results ASAP!

ok then here goes:

command line:
noseman$ /Applications/maxwell/mxcl -mxs:/maxwellfolder/skydome.mxs -d -o:/maxwellfolder/skydome01.tga -th:16 -res:640x480 -s:15

final image (for comparison):


time from terminal:
[2h 56m 18s] of [16h 39m 0s] , current sampling level : 15

time from stopwatch:
1 hour 31 minutes

maxwell.log:
**********************************
MAXWELL 1_1_30_0001_alpha_ ( Engine core: rs1b )
[ Physically and Unbiased Based Rendering ]


## Scene Information ##

44 meshes ( 37798 triangles )

## Render Information ##

*settings:

resolution: 640 x 480

*render time: 2 hours 56 minutes 19 seconds

normal quality mode
target sampling level: 15
sampling level reached: 15

MaxwellBench 1.0: 13.087601
******************************************

What I notice is that times from the command line are 200% the real time and MaxwellBench is 50%...

That's all for now!

very strange... is this a bug noseman?

NL says it's a dual G5 mac bug..

What type of chip does Intel make that is comparable to the Opteron 248?

As Spock would say; "Most curious!"

Maxer wrote:What type of chip does Intel make that is comparable to the Opteron 248?

Xeon 3.2Ghz or 3.4Ghz EMT HT = for maxwell much better performance/$ ratio. ....at least for now.

cy,
A

So a single Opteron 248 (2.2ghz) is better than a P4 2.6 ghz? That's pretty good

Are you planning to make an accurate dual opteron benchmark? I think it would be worth investing in a dual opteron system because when the dualcores come out all you have to do is upgrade the cpus.

PowerPC G5 - Dual 2.5 Ghz

2.5 Gig memory

OS 10.3.8

threads: 4

real world time: 1 hour 6 minutes 42 seconds

maxwell.log:
*****************************

MAXWELL 1_1_30_0001_alpha_ ( Engine core: rs1b )
[ Physically and Unbiased Based Rendering ]


## Scene Information ##

44 meshes ( 37798 triangles )

## Render Information ##

*settings:

resolution: 640 x 480

*render time: 2 hours 8 minutes 31 seconds

normal quality mode
target sampling level: 15
sampling level reached: 15

MaxwellBench 1.0: 17.818067

************************************

These result seem to confirm the problem stated above - Maxwell time and MaxwellBench don't reflect accurately what is actually happening.

In this case I specified only one thread in the command line - It appears from the Activity Monitor that Maxwell uses both processors even when one thread is specified - at least in the OS X version.


PowerPC G5 - Dual 2.5 Ghz

2.5 Gig memory

OS 10.3.8

threads: 1

real world time: 1 hour 34 minutes 20 seconds

maxwell.log:
*****************************

MAXWELL 1_1_30_0001_alpha_ ( Engine core: rs1b )
[ Physically and Unbiased Based Rendering ]


## Scene Information ##

44 meshes ( 37798 triangles )

## Render Information ##

*settings:

resolution: 640 x 480

*render time: 3 hours 1 minutes 19 seconds

normal quality mode
target sampling level: 15
sampling level reached: 15

MaxwellBench 1.0: 12.546945
************************************

The performance in Mac will improve in the next days, we are doing some fine-tuning for G5.