I’ve been running current version on our lab computers. On Windows 8.1 (in contrary to Win 7) something strange is happening to CPU_02 UDT test. Instead of about 1.50 sec as before (with previous HM2 release) we are getting 25-45 sec, on i5, i7 and Xeon. Can someone with Win 8(.1) confirm that observation?
That part of test is not altered in any ways, so it should be something else. Is that with Rhino SR9 or SR10? And did you test it on win 8.1 the last time too?
this is my work unit, I will t
est my one at home to see if I have the same problem with the i7 3770 as others posted
SR 9
I don’t remember. There used to be Win 8 on those machines but I’m not sure when it changed to 8.1. I will try to find out.
I am not sure if it matters, just trying to help you limiting it down.
This from my build-your-own development box, courtesy of NewEgg…
Nice… you mcneel guys knows your stuff!
That is a steady performer if I ever saw one.
Hey, Dale, you can’t be better than me- I have an even newer New Egg and a fire-engine red video card… maybe it’s that ERROR on the right column. Hmmm. Tempting to uninstall Shoretel but I need it to run the phone.
With AA off in Rhino the score goes up to 58000 + and then at 2x I think is was lower than at 8x : 44000 and 47000, roughly, for those.
-Pascal
1 Like
With AA off it tests at 56712, with AA at “4” it tests at 53947…
Is this huge spike an error ? What is causing it?
Greg
Hi Greg, it is not an error, it just shows the true power of the K5000. And I didn’t predict the K5000 to be so fast with single meshes. So I probably didn’t balance the scores well enough.
@jeff just to be sure: Are there any reasons why the K5000 shouldn’t master a whopping 1250 frames per second on a single mesh?
(The test uses the build in TestMaxSpeed equivalent python tool)
@Holo Hey Jørgen, looks like it’s the GPU 14 that’s causing the spike in a bunch of cases above - “number of units mesh @ 15FPS in Shaded”. So maybe that needs to be looked at. The question is, is “Shaded” the standard shaded mode? Are mesh edges shown in the display?
–Mitch
Also interesting the difference that installation of a recent driver made with a GTX 780 - see below…
Edit: compared to my machine with a Quadro 400, the 780 is now doing better in a number of cases, although the performance of the Quadro is much less “spiky”. Somewhat disappointing, but the same machine tested only with ~26,000 in May (post #19 in this thread). I guess there were improvements in the nVidia drivers since, and looks like Jeff’s work on display management may have also paid off. Lastly, my laptop which only has a Quadro 2000M, scored almost as well as my workstation…
–Mitch
GTX 780 - 340.52 drivers
GTX 780 - 344.48 (latest) drivers
Quadro 4000 with 340.84 (latest) drivers (different machine)
And finally my laptop
Quadro 2000M 340.84 (latest) drivers
V-RAY CAUSING PROBLEM???
Here some other test.
I get much better result disabling Vray (2.0), that look like the cause for the cpu slowdown (?!?!)
I repeated this many times and look constant.
Anyone other have tried?
Without V-Ray: 
With V-Ray: 
Hi @skysurfer , What method did you use to disable Vray? I tried just switching the Renderer to Rhino Render, but I didn’t see any difference in my test results. Michael VS
@skysurfer Bingo! I can confirm that disabling VfR 2 solved CPU issue also here.
CPU_02 sank from 44.05 to 1.71 sec (issue I reported yesterday).
@mvyess , go to Options, Plug-ins and uncheck Enable for V-Ray for Rhino.
first run (Vray loaded) CPU_02 61.49 sec :
second run (Vray disabled) CPU_02 2.03sec
Hi, Yes! Disabling Vray fixed one of my bottlenecks in the test, the UDT time., but unfortunately the Technical View time on my PC is still ridiculously long at 131seconds. But this is progress… Michael VS
With VrayForRhino SP1 Enabled:
VrayForRhinoSp1 Disabled:
Interesting that GPU_13 test trippled in speed with Vray Disabled.
And UDT went from 37 seconds down to 1.6seconds…
@jeff - Perhaps you have some wise words for us on the reasoning behind this Vray effect?
also noticed in yours as in mine the GPU_01 fps were noticeably different…(yours 645 to 307) mine (769 to 211) all other tests seem to be have pretty similar results…
I use custom display modes for all the tests, so it should be correct… It shows mesh edges by default and is reinstalled from the ini file at each run, overwriting any modifications done to the display modes.
GTX 660, Xeon, GeForceTestEngine enabled (adds ~4k to GPU), latest driver 344.48 (adds ~4k to GPU):
1 Like