Computer Specs

[glorybound]

So your older 4 core Opteron is faster then the new i7 4 core?

[simmsimaging]

no - the list is progressively adding nodes to the first render, which was the dual-Xeon only. The Opteron would still be rendering if it was on it's own

b

[RErender]

So, how long does it take an i7 to render it by itself?

[simmsimaging]

Don't know - I have to port a license over for Max so I can test it on it's own and I"m working around the clock right now. I'll try and do it when I have some more time.

b

[rmejia]

The first i7920 that was added almost halved the time, so by itself it would probably be pretty close to the dual Xeon 5355 render time.

[TRGraphics]

Can you do a two frame animation of the scene and render with backburner? This would directly compare the two wouldn't it? Or at least come pretty close. One frame using the dual quad and the other frame using the one I7.

I'm itching to add myself.

[RErender]

Looks pretty good from the DR results indeed, I'm having a hard time not buying one don't really need one one now though...

I'm going to try and wait for the 8 core chips later this year, there is a 6 core xeon version available but still pretty pricey http://www.newegg.com/Product/Produc...82E16819117176

[simmsimaging]

Not sure why, but the 4 core i7 shows up as 8 cores in the task manager, and I get 8 buckets per machine. Something with the new hyperthreading?

Anyway, 8 "real" cores would be impressive indeed.

b

[simmsimaging]

Had a minute to start a new test with something that renders a bit longer

q600: 31:41 (rendered directly, this is my workstation)

2X 5355: 17:43 and 17:17 (rendered via BB, done twice to see BB added much variation)

i7: 16:17 and 16:20 (via BB, also done twice)

So...BB is pretty consistent, and the i7's are pretty good bang for the buck IMO. The eight-core 5355 machine was a lot more expensive when I bought it 1-1.5 years ago.

[simmsimaging]

Had a minute to start a new test with something that renders a bit longer

q600: 31:41 (rendered directly, this is my workstation)

2X 5355: 17:43 and 17:17 (rendered via BB, done twice to see BB added much variation)

i7: 16:17 and 16:20 (via BB, also done twice)

So...BB is pretty consistent, and the i7's are pretty good bang for the buck IMO. The eight-core 5355 machine was a lot more expensive when I bought it 1-1.5 years ago.

Updated testing:

EDIT:

Further tests: (all in minutes)

Xeon+Q6600+Opteron: 8:26 (direct) - This is my old 'farm'

All 4 i7's: 4:48 (via BB) - this is my new 'farm'

4 i7s +Xeon +Q6600+Opteron: 3:03 (rendered directly in this case, but I think with BB it would be about the same). - this is everything but the kitchen sink

The first of these (8:26 time) was all my machines up until yesterday so, for me, adding the 4 i7's has cut my render times in less than 1/2, or in 1/2 but also leaving me with my working machines available.

I could have gotten away with one less node (1 +3 configuration) and the times would not have been wildly different for the all-in renders, but using just my i7 farm would have been quite a bit slower. Having 4 allows me to offload the entire job to the "farm" and keep both my workstations moving along.

Cost of the 4 machines was around $7000 including an mid-range case. Another $600 for a small UPS for each one;the network hardware etc. that I needed (larger switch, KVM switch etc.). All in all only a few dollars more than I spent on two machines last go around.

I am happy.

[zeronove]

Something with the new hyperthreading?b

yes, it should be the HT!

how does it manage apps that aren't written to run multicore? 100% full load / 8 ? ..or 100% / 4 or what ever?

my old dual xeon with HT often showed me only 25% cpu usage because of that

[TRGraphics]

Had a minute to start a new test with something that renders a bit longer

q600: 31:41 (rendered directly, this is my workstation)

2X 5355: 17:43 and 17:17 (rendered via BB, done twice to see BB added much variation)

i7: 16:17 and 16:20 (via BB, also done twice)

So...BB is pretty consistent, and the i7's are pretty good bang for the buck IMO. The eight-core 5355 machine was a lot more expensive when I bought it 1-1.5 years ago.

Wow!!
I gotta get me one of those!!!!! Thanks for the tests.

[stevesideas]

Ive got my 2 i920 render machines in. They seem to be going well. Yet to test them against my 5355 dual xeon system. Will post once I have the time.

I was told by the guys I bought them from that they are 35% faster than the qx6700. But from the benchmarks shown in this thread they are showing to be faster than that.

[simmsimaging]

yes, it should be the HT!

how does it manage apps that aren't written to run multicore? 100% full load / 8 ? ..or 100% / 4 or what ever?

my old dual xeon with HT often showed me only 25% cpu usage because of that

Couldn't tell you: I have only installed Max on them for rendering - I won't be putting anything else on them to find out. These are strictly for rendering nodes.

[glorybound]

dual i7

So... I was told you can't get a dual i7. Is this correct?