Wow, that was a surprise…
So I guess it is safe to presume that Titan is “unlocked” for floating point CUDA calculations, and not for large and massive OpenGL mesh handling.
Btw: Great to see that Holomark shows the graphic card it uses at the top of the list.
Hi Andy.
The test only asks windows for information and presents what is returned.
So it probably just “finds” the onboard memory and not the theoretical shared amount.
Anyhow, what is the maximum amount of Vram that you have managed to use in Rhino?
An interesting app (widget) to run when Holomark2 is running is GPU Meter:
http://addgadgets.com/gpu_meter/
New motherboard and video card.
BDT - That’s an incredible score but I can’t read the JPG. Could you list the GPU, CPU and mother board please? I’m curious what’s giving you those blazing high numbers.
The last part of the test places the results in the lower left corner and then zooms to full screen, it wasn’t zooming. I restarted Rhino and ran the test again and here’s the results after zooming full screen.
Great to see the 970 perform that well!
Can you please answer this:
1- Did you adjust the driver settings, and if yes what?
2- Did you turn off AA in Rhino, if no then what settings did you use?
3- Have you done any other adjustments?
Thanks and enjoy!
It pleases me to see gaming cards hitting the peaks “You appear to have peaked off the chart and off the page dear” 
Ha! Awesome :)) You must have been sat there for a while while it cranked up 396 units!
My first score (way at the beginning of this thread) was 18,247. Since then I increased the RAM from 8 to 24. Also, I had the motherboard replaced which may have been a problem earlier on, not sure. Whether it was the increase in RAM, new motherboard or both, I get a substantially higher score now.
But nothing like that monster score of the GTX 970 / i7-4790 right above!
Holomark 2 v2,2,03
Total Score: 27488
Total Runtime: 127.41 sec
GPU scores: 14180
GPU_01 - 392.20 fps - Cube 4 tests
GPU_02 - 38.50 fps - UDT Shape
GPU_03 - 76.30 fps - Wireframe
GPU_04 - 37.50 fps - Shaded
GPU_05 - 29.00 fps - Rendered
GPU_06 - 28.30 fps - Block Rendered
GPU_07 - 16 units Nurbs @ 5 fps in Wireframe
GPU_08 - 8 units Nurbs @ 5 fps in Shaded
GPU_09 - 7 units Nurbs @ 5 fps in RenderSpeed
GPU_10 - 97.10 fps - RenderMesh Render
GPU_11 - 45.50 fps - RenderMesh RenderSpeed
GPU_12 - 133.30 fps - JoinedMesh Render
GPU_13 - 36.00 fps - JoinedMesh RenderSpeed
GPU_14 - 2 units mesh @ 15 fps in Shaded
GPU_15 - 6 units mesh @ 15 fps in Render
GPU_16 - 3 units mesh @ 15 fps in RenderSpeed
GPU_17 - 70.40 fps - mesh in Rendered Studio
GPU_18 - 9.20 fps - Nurbs in Rendered Studio
GPU_19 - 21.90 fps - Block Illustration
GPU_20 - 63.30 fps - 2D single
GPU_21 - 7.80 fps - 2D massive (20x)
CPU scores: 13308
CPU_01 - 11.24 sec - Booleans and Contours
CPU_02 - 1.50 sec - Twist and Taper (UDT)
CPU_03 - 4.71 sec - Meshing Mini
CPU_04 - 0.03 sec - Extract Render Mesh
CPU_05 - 0.05 sec - Join Render Mesh
CPU_06 - 14.74 sec - Reduce Mesh
CPU_07 - 1.72 sec - Calculating Technical display
CPU_08 - 3.58 sec - Making Silhouettes
To Be Filled By O.E.M.
To Be Filled By O.E.M.
Intel® HD Graphics 3000 - 2108.0 MB
DriverVersion: 9.17.10.2857
NVIDIA GeForce GTX 570 - 1280.0 MB
DriverVersion: 9.18.13.4465
Intel® Core™ i7-2600K CPU @ 3.40GHz
NumberOfCores: 4 NumberOfLogicalProcessors: 8
MaxClockSpeed: 3.4 GHz
TotalPhysicalMemory: 24.0 GB
Microsoft Windows 7 Professional
Rhino 5 sr 10 64 bit
The nvidia setting was for “performance” and Rhino was set to 2x antialiasing.
Yea, I was waiting a bit while it churned out 396 little minis!
Yeah, I must admit that I never expected any cards to perform THAT well in Rhino. The Quadro K5000 was the first to bust the charts and not too many users have those, so I figured it was better to release than wait half a year before I had time to fix it. But the 970 changes that game, many users will and can afford those. I should probably rewrite the test so it tests with a 10x10 and measure the speed for 5 seconds instead.
I’ll have to do some calculations to find the “perfect” score conversion rate, but that should be possible.
There is also a minor bug in SR10 that turns the illustration mode inside out, due to a change in how SR10 handles backface drawing of the polygons, and the holomark illustration mode accidently happens to push on all the odd buttons for this to happen, so I will rewrite this and have that fixed in the update as well.
Thanks to all of you for testing and sharing!
Holo,
Thanks for creating this test, it’s a great way to better understand our computer capabilities. This new 970 would not work in the original motherboard which was only 7 or 8 months old so I had to replace it with a board that would work with the 970.
I am surprised that the CPU doesn’t score much higher than the old computer which had an I7 3770, 3.4ghz, with 16gb ram. The old computer also had the Quadro 2000 which scored less than half of what the 970 can do now on the new computer.
Thanks again for all the work you’ve done on this benchmark and for sharing it with the Rhino community!
Phil
I totally agree, thanks to you for taking the time to pull together such an amazing tool for Rhino users, it helped me choose my next set up and gave me a decent understanding of what I should be looking for.
It also reminds me of the first car me and my mate use to bomb around in when we learnt to drive… Although that was white with a brown vinyl roof 
Thanks again!
Thanks Phil,
it makes it all worth while.
If you compare the pure GHz speeds of the 3.4GHz i7 3770 and your new 3.6GHz i7 4790 you will see that it’s only a 5% difference, and if you search them online you will find sites like the CpuBoss that does theoretical comparisons:
http://cpuboss.com/cpus/Intel-Core-i7-4790-vs-Intel-Core-i7-3770
And you’ll see that they too verifies a 5% difference bewteen the two.
Thanks man!
63813 points, but 66 fps in shaded mode - test from BDT. Other tests show 30fps by 32567 total score points (wattzie).
Just I looked through the results to see what is the current state of graphic power in simple shaded mode. For me this is the most interesting test part for getting an impression what can be expected for daily graphic use. I noted that the total score isn’t really informative, since GPU and CPU tests are matched. I ask me wouldn’t be better to write out different total values for graphic and modelling tests?
Hi Micha, I agree.
The GPU and CPU scores are separated already, and I was going to separate the 2D, nurbs and mesh scores, but decided to release it as it was.
I’ll see if I can update this for the upcoming SR2 as I already have two other things to fix.
Thanks for the reminder!
Ok, what has happened in the latest Rhino release? i turned of all plugins and now i have a 20k points drop :/…same graphics driver,…also, sli enabled and disabled made no difference in score,…
Jorgen,
Can Rhino use multiple GPU’s? Or multiple CPU’s?
Phil
BDT Manufacturing
505-515-6565
Holo,
Just downloaded this and ran it on my custom built hot rod, from going down the list I see that this machine of mine is pretty darn fast. Thank you very much for this testing download.
All my best … Danny
