[Tim Ellis]

http://www.maxwellrender.com/forum/view ... highlight=

Thread & speed comparison tests with conclusions and comments in the thread.

X64 is slower but can do more than x32. Use x32 for speed on simple scenes and x64 for high res & complex scenes.

Tim.

[b-kandor]

This is strange, I see you made two tests for 64bit (Maxwell also 64bit), one overclocked, one not overclocked. The one that wasn't overclocked was 1h 04m and the overclocked test was 1h 51m. Are you sure that's right?

Can it have something to do with the overclock, some step down feature in the motherboard that throttles back the CPUs if they get too hot?

Hi Mihai,

No, the seperator in the table is for two different systems (each overclocked) - look again at the chart under the cpu column (the q6600 is a quadcore and the e6400 is a dual core). Both systems are overclocked. Basically, for the quadcore system on winxp64 with 64bit maxwell the fastest time was achieved by setting the threads to 2 (37 mins). This is almost twice as slow as the 22 mins. I tested yesterday useing winxp32 and mr32bit. I'll split the chart into two so it's less confusing.

[b-kandor]

http://www.maxwellrender.com/forum/view ... highlight=

Thread & speed comparison tests with conclusions and comments in the thread.

X64 is slower but can do more than x32. Use x32 for speed on simple scenes and x64 for high res & complex scenes.

Tim.

mr64 on xp64 isn't just slower, it take 2 to 3 times longer to reach the same sl. It's a bug

[b-kandor]

I updated the table...... (previous page)

Also - my table shows that for DUAL core computers running winxp64 and 64bit MR1.5 - Setting the threads to 0 is a virtual death sentance!

Time to render benchwell to sl15 on my dual core conroe is 1 hr. 51 min. with threads set to 0. Set threads to 2 and the time is 320% faster or 35 mins.

[GM5]

Just to add to the confusion (I also posted these results in the benchwell thread).

With Adrian's benchmark set for SL 6 all test were run using XP 64 and M~R 1.5 on a Dual Quad Mac Pro. Didn't have time to run any more tests and had very strange results.

M~R 64 bit:

8 threads:

Test 1 = 1:28
Test 2 = 1:40
Test 3 = 1:36

4 threads:

Test 1 = 2:00

2 threads:

Test 1 = 1:14
Test 2 = 1:14

M~R 32 bit:

8 threads:

Test 1 = 1:25

The fastest render was XP 64 set for 2 threads on an 8 core machine.

[b-kandor]

Similiar to my result as well. My problem with this is how do you set coop? I have 2 dual cores, 1 quad and 2 singles. Setting 0 threads is not a viable option.