OS: Win 7 x64
Blender: 2.71.6
Build: Buildbot
Revision: 526ae63
CPU: i7-3930K 3.6GHZ
Tiles: 16x16
Optimization: -
Time: 012:29.66
GPU: Titan Black
Tiles: - 256
Optimization: -
Time: - 5:10.36
2 x Titan Black
Time: - 2:55.12
OS: Windows 7 Ultimate x64
Blender: 2.71.6
Build: Buildbot
Revision: 526ae63
CPU: Intel i7-920 @2.67GHz
Tiles: 256x256
Optimisation: -
Time: -
GPU: 660Ti OC 3GB + 750Ti 2GB
Tiles: 256x256
Optimisation: -
Time: 08:57.00
GPU: 660Ti OC 3GB + 750Ti 2GB
Tiles: 64x64
Optimisation: -
Time: 09:49.27
OS: Opensuse 13.1 64Bit
Blender: 2.71.6
Build: Own
Revision: cebf728
CPU: i5 3570K Stock
Tiles: 16x16
Optimization: -
Time: 20:03.98
GPU: GTX 760 4 GB
Tiles: 512x432
Optimization: -
Time: 12:55.07
so the list is up to date 
congratz blokebious 
blazraidr please mention your cpu model and clock speed!
@nudelZ, What do you mean regarding your GT 630 is not supported? Is not really supported (error, bad result), or you say that because it takes a long time?
My old GT 430 1GB (nvidia 331.38 in Linux) can perform the render with 2.71.6 from buildbot. But with an estimated time of hours, insane. Reducing the size a lot seem perform the render ok.
By the way, I updated my previous post with version 2.71.6. Not very important because only use the CPU, but to be in tune with others.
YAFU which tile size works or you? i’ve tried a couple of times… - with “multires 3” always out of memory… (transparent render)
Do you have this error when starting the render?
With the default settings in .blend file, I have not any kind of error (mem 472.21M peak 482.21M). The render starts and goes fast on the white part, it takes very long time in the part of the skin, but it seems that it does well. Then I set the size to 15% just so it does not take hours, and be able to check if the end result was good.
This at 15% size has taken 14:24.31 min!
it renders but transparent
buildbot build
8x8 tiles
-> CUDA error: Out of memory in cuLaunchKernel(cuPathTrace, xblocks , yblocks, 1, xthreads, ythreads, 1, 0, 0, args, 0)
I’ve updated my post 
OS: Win 7 x64
Blender: 2.71.6
Build: Buildbot
Revision: 6891f1c
CPU: i7 4770K @4.0GHz
Tiles: 16x16
Optimization: -
Time: 13:10.44
GPU: GTX 780 Ti
Tiles: 256x256
Optimization: -
Time: 05:15.07
OS: OS X Mavericks 10.9.4 - Nvidia Web Driver 334.01.02f02 - CUDA 6.5
Blender: 2.71.6
Build: Buildbot official
Revision: 526ae63
CPU: i7-3770K stock
Tiles: 16x16
Optimization: -
Time: 00:15:38.30
GPU: 3x 780 Strix OC GHz Edition
Tiles: 256 x 288
Optimization: -
Time: 00:02:38:31
EDIT: Updated CPU Bench, LOL, 
you were right, 18 secs faster without spatial splits.
thank you enilnacs and congratz for the first place
i’ll add your gpu result but not (yet) your cpu result - only software optimizations are allowed (compiling)
please render again with spatial splits off
btw: spatial splits is not better - it’s for sure slower then non spatial split render
My try
OS: Windows 7 64 bit
Blender: 2.71.6
Build: Buildbot
Revision: 35bc266
CPU: Athlon II x4 630 2.8 GHz
Tiles: 16x16
Optimization: lowered resolution (40%) - 379x432 px
Time: 12:09.72
GPU: 750ti 2GB
Tiles: 95x96
Optimization: -
Time: 19:31.13
In background was AIMPP, firefox (GPU). No problems with rendering at all.
Added some time for CPU but with lowered version… No time for such high rendering times
Sorry, something wrong under typing 
tjindy: I cant find any information about your gpu - maybe its a notebook card? 675M/MX?
also your cpu clock speed is missing
OS: CentOS 6.4 64-bit
Blender: 2.71.6
Build: Buildbot
Revision: 526ae63
GPU: 8 x GTX 690 (over GPUBox Artist :))
Tiles: 128x108
Optimization: -
Time: 01:15.38
Hehe ;), well i could switch at work also together with GPUBox 
(15x 780 6GB Strix over 10 GbE, imagine that speed!), but what is the speed of your PC system itself, not a networked cluster, that is the real benchmark question. Gosh i could even get the old 580 out at work over GPUBox ethernet. LOL 
I mean, we could go so far as to get even into a GPU cloud from Amazon over fast GbE,etc. with 100 GPUs… but is that a “System” benchmark ? I really doubt it. So i didn’t use GPUBox cause I think it distorts the benchmark, especially regarding PCIe GPUs only.
Really would want to see some strict rules for benchmarking Blender workstations (Host system only, no networking, no cloud, system latency rendering, no rack server hardware, etc…). Benchmarks are for system hardware reference (buying,etc.), NOT for the fastest times.
If you are using GPUBox at your work, launch the benchmark there and post your result. On 15x780 and 10GbE you should easily go under 1 minute and regain your 1st position When it comes to Blender, GPU virtualization (especially GTX-class cards) is something absolutely new so believe me that lots of people would find this information much more valuable than benchmark result of another Titan or 780.
Besides, using GPUBox is not very different to installing another cards to your PCIe slots, so we should also forbid to post any other multi-GPU result. But this is a discussion for a new thread, let’s keep this one clean
OS: Win 7 Pro 64
Blender: 2.71.6
Build: Buildbot
Revision: 4776d80
CPU: i7-2670QM @ 2.2 GHz
Tiles: 16 x 16
Optimization: -
Time: 28:35:82
GPU: Quadro 1000M
Tiles: 380 x 216
Optimization: -
Time: 1:16:03.88
Where are you getting the “.6” build? I can only find the “.5” at the bulidbot site.
OS: Win 7 Home Premium 64
Blender: 2.71.6
Build: Official
Revision: 9460fe3
CPU: AMD 1090t @ 3.2 GHz
Tiles: 16 x 16
Optimization: -
Time: 0:26:43.14
GPU: 3x 750 Ti 2GB
Tiles: 253 x 288
Optimization: -
Time: 0:07:04.09
Tiles: 253 x 72
Optimization: -
Time: 0:06:13.18
OS: Linux Mint 17
Blender: 2.71.6
Build: Official
Revision: 4e732b9
CPU: AMD 1090t @ 3.2 GHz
Tiles: 16 x 16
Optimization: -
Time: 0:22:35.35
GPU: 3x 750 Ti 2GB
Tiles: 253 x 288
Optimization: -
Time: 0:07:43.90
Tiles: 253 x 72
Optimization: -
Time:0:07:04.37