[monochrome]

I have a Ninja Scythe cooler. I fixed the Orthos problem...
I've been experimenting with different air routing inside my LianLi case and so far the best I've managed in Maxwell is upper 60's.

[GM5]

I just tried M~W 1.5.1 (32bit) on my Mac Pro.

Using XP64 and set to 0 threads:

Benchmark: 865.18

Time: 12m24s

-Greg

[numerobis]

nice speedgain GM5! i'll redo the test with the new release too.
i just took a look at the list and saw that there is a new fastest machine:

pinkSky IBM PS3 Cell 3.00 4 1.5 (64bit) 10m 12s 956

[numerobis]

i just rerun the test with 1.5.1 and 1.5

for me 1.5.1 is ~3% slower than 1.5

Q6600@3GHz
WinXPx64
Maxwell 32bit
threads 4
priority normal


1.5.1 _ 20m 30s _ bench 521.71

1.5.0 _ 19m 51s _ bench 539.70


(last 1.5 test: 20m 03s - bench 533.54)

[sandykoufax]

IBM PS3 Cell ?

What is it?

Who is pinksky?

[GM5]

Play Station 3 processor - I guess it is supposed to be funny...

[GM5]

i just rerun the test with 1.5.1 and 1.5

for me 1.5.1 is ~3% slower than 1.5

Q6600@3GHz
WinXPx64
Maxwell 32bit
threads 4
priority normal


1.5.1 _ 20m 30s _ bench 521.71

1.5.0 _ 19m 51s _ bench 539.70


(last 1.5 test: 20m 03s - bench 533.54)

Strange... mine went from 13m 30s with 1.5.0 to 12m24s with 1.5.1 - it could be a fluke that it rendered faster this time (maybe it rendered slower than it should have the last time?). Did you try 0 threads? In the 1.5.1 announcement it states "Rendering on 64 bits systems might not be able to make use of all the cores" so the bugs aren't worked out yet.

-Greg

[thxraph]

IBM PS3 Cell ?

What is it?

Who is pinksky?

what what? i know there is some memory extender over there & you need to flash the firmware, but hell, maxwell isn't cell ready? or is it x86_64 emulated? some info plz

Raph

[b-kandor]

I thought linux ran on ps3's?

[3DBob]

Hi all, after looking at what seemed to be slowing down my machine I found it was first the read/writes to hard disk first - apparently a U320 Fujitsu 15K drive is not fast enough. So I created a ram drive and made maxwell load/render there.

Next up - I found that writing the preview image to the screen slowed things down a fair bit. So I turned off the preview - Still writes out the exact same image at the end of the render.

My question is, is the render time still valid. It was:

12m17s

and the Maxwell benchmark

816.4

3DBob

[martgreg]

IBM PS3 Cell ?

What is it?

Who is pinksky?

what what? i know there is some memory extender over there & you need to flash the firmware, but hell, maxwell isn't cell ready? or is it x86_64 emulated? some info plz

Raph

Well with the name "PINK SKY" i think het are being silly , if not i want one

[Tim Ellis]

PS3 is running Gentoo. If you can install any Linux application through that os, then Maxwell will work.

I have PS2 with Linux which I ran Blender on, PS3 & Linux is on my list and if this Benchwell entry can be confirmed, then I'll be getting one sooner. If it's possible to combine the RSX & the Cell, then that bm will be huge.

Tim.

[Bubbaloo]

I don't believe it. I have a sneaking suspicion that it's that def4d guy that's still upset. The name gives it away as someone who is disgruntled...

I want some proof! (It has nothing to do with him knocking me off the top...)

[thxraph]

PS3 is running Gentoo. If you can install any Linux application through that os, then Maxwell will work.

I have PS2 with Linux which I ran Blender on, PS3 & Linux is on my list and if this Benchwell entry can be confirmed, then I'll be getting one sooner. If it's possible to combine the RSX & the Cell, then that bm will be huge.

Tim.

yes, you won't be able to combine RSX & the Cell, but only use the Cell, and that's enough ^^, BUT, as MR isn't compiled for the Cell, but for x86 & x86_64, so you need to emulate it, with wine... and i think this rendertime & benchark is possible, but would like to see a proof, otherwise i'll have todo it by myself, cause that is the cheapest ultra powerfull rendernode i've seen, the cell is a monster, and all the cores are in the same chip, it is far faster than have 2x4cores for exemple.

Raph

[Tim Ellis]

...........and i think this rendertime & benchark is possible, but would like to see a proof, otherwise i'll have todo it by myself, cause that is the cheapest ultra powerfull rendernode i've seen, the cell is a monster, and all the cores are in the same chip, it is far faster than have 2x4cores for exemple.

Raph

Me too, cheapest & fastest rendernode system so far.

Tim.