All posts relating to Maxwell Render 1.x
By garbage75
#190407
Does anyone can do te speed test under OSX now that the universal binary version have been released??

Naturally using a MAC PRO.... :)
User avatar
By glassbathroom
#190408
garbage75 wrote:Does anyone can do te speed test under OSX now that the universal binary version have been released??

Naturally using a MAC PRO.... :)
I'm on it. It's running now. Obviously it is Release 1.1, so it is not comparable with the tabled results, but I have an XP (Bootcamp) score that we can compare it with.
By garbage75
#190409
thanks a lot... :) :)
post your results as soon as possible....

good work Walter!!
User avatar
By glassbathroom
#190422
Firstly I am thrilled to bits to see the new Universal version for the Mac. I really appreciate the work that the Next Limit team have done to get this out to us. Nearly all of the little bugs that showed up as differences between the XP and OSX version have been sorted out. Thanks guys.

Now the results of my test are slightly disappointing for OSX fans.

The History

My Mac Pro scored very well running Version 1.0 on XP (with Boot Camp)

Dual Processor Apple Mac Pro (4 cores)
Xeon 2.66 GHz 4GB RAM

Version 1.0 - Windows XP Pro in Boot Camp
Benchmark= 91.31
Time= 46 minutes 58 seconds

Now we know that Version 1.1 takes longer to reach the same SL, but is better quality for that SL level. The test with XP installed on Boot Camp and Version 1.1 resulted in the following:-

Version 1.1 - Windows XP Pro in Boot Camp
Benchmark= 39.68
Time= 1 Hour 47 minutes 58 seconds

The UB Version

Bit slower than the XP version.

Version 1.1UB - OSX
Benchmark= 32.13
Time= 2 Hrs 13 minutes 10 seconds

Now it may be that the SL levels are again not comparable, but if not then the OSX version is around 20% slower than the XP version.

Still really thrilled with it though. It is a huge speed jump from my old Pentium 4!
By garbage75
#190426
but the results of the Version 1.0 of speed test, running on a MAC PRO under win XP are very good...

so in there any table that we can compare this results with other machine like opteron or other PC??? (with results of Version 1.0)??

where can I find this table?

I'd like to know the position of this MAC PRO....
User avatar
By glassbathroom
#190427
garbage75 wrote:but the results of the Version 1.0 of speed test, running on a MAC PRO under win XP are very good...

so in there any table that we can compare this results with other machine like opteron or other PC??? (with results of Version 1.0)??

where can I find this table?

I'd like to know the position of this MAC PRO....
The table is on the first page of this thread. It's a bit out of data and the Mac Pro is not on it. It was the best result (briefly) but now has been beaten by others since. You will have to scroll back a couple of pages.
By garbage75
#190430
but we can say that DUAL XEON DUAL CORE on a MAC PRO or on a PC are the best machine by now running this test with Version 1.0??
User avatar
By glassbathroom
#190433
garbage75 wrote:but we can say that DUAL XEON DUAL CORE on a MAC PRO or on a PC are the best machine by now running this test with Version 1.0??
I am not sure what you are getting at. The current fastest speed for Version 1.0 is jrandom's Dell as follows. It beats the Mac Pro by 4 minutes. Probably because its a 3.0GHz. My Mac Pro is 2.66GHz

Dell 690
Dual Xeon Woodcrest 5160 (3.0 ghz)
8 gb ram (i'm sure maxwell can only use 4 as 32-bit process)
XP64
time=00:44:33
Benchmark=96.04
By garbage75
#190434
Yes OK.... I agree with this.. probably cause it is a 3 GHZ but the difference is minimal... 4 minute are not hours!!!! :) :)

Maybe if we done the test in a MAC PRO 3 GHZ the results will be better... do you agree with me??
User avatar
By Stephen
#190542
Dual 5140 (2.33) 2gig Ram / Quad core
Version 1.1 - Server 2003 R2 64bit
Benchmark = 35.93
Time = 1h59m26s

Not too bad. I will OC in a week or two.
#190548
chrisvconley wrote:Hey there,

For what it is worth, I ran the test scene on our 4 machine x 2 processors per machine x 2 core per processors ( 16 cores) 3Ghz Woodcrests and set the time to 30 minutes. So if you ran the test in version1.0 you should finish at sl12 in 9.25 minutes!!!!!!!

We're running maxwll 1.1 so i am not sure how the scene for 1.0 works.

It achieved SL 12.05 in the 30 minutes.

Maybe daros can run it on his 70 x processor farm...

Chris
Hi Chris, I'm confused? Have you run the test on this system in version1.0? I would like to see the difference. I ran the test on a single conroe chip 6400 with 2 cores) and achieved sl12 in 1:14min.

Thanks,

Kandor
Last edited by b-kandor on Sat Oct 21, 2006 3:45 am, edited 1 time in total.
By chrisvconley
#190549
Hey kandor,

No, I ran it with mxcl 1.1 which is why i do not know how it really compares.

With 1.1 and eight 3GHz woodcrests, I achieved 12.x in 30 minutes...

Chris
User avatar
By b-kandor
#190550
Hi Chris, you beat me to my edit, but in my above post I mentioned that according to the math (very approx.) you should achieve sl12 in 9.25minutes with that setup!
By chrisvconley
#190551
so do you think that means that 1.0 is not at all comparable to 1.1? probably not, eh?

Maybe we can get a new scene for 1.1 and begin a new benchmark list. especially now that UB for OSX is out!

Chris
User avatar
By b-kandor
#190553
I think it's discussed somewhere in this long thread - but there is simply no way that 16 cores at 3ghz take 30 minutes to do something that 2 cores (at 3.2ghz) can do in 74 minutes!
  • 1
  • 8
  • 9
  • 10
  • 11
  • 12

> .\maxwell.exe -benchwell -nowait -priority:[…]

render engines and Maxwell

You could be right about AI, but actually I prefe[…]