- benching GPU with 20000 triangles (max) by FPS (Hz) count
- checking AGPM working or not by using 20 up to a few 100/1000 triangles putting CPU load from very less % gpu load to > 80% gpu load
(use logging key in AGPM kext = 1 to show APGM change messages in the console)
Bildschirmfoto 2012-09-25 um 09.57.05.jpg 180.35K
114 downloads- be aware to put the slider very fast to the max - system may stall at least for low end GPUs!
- uses very less cpu time = FPS near independed from CPU speed. VRAM + Texture speed of GPU
Because bench uses vsync, FPS never get more than screen refesh (TFT 60 Hz, 75Hz,... others more).
Screenshoot displays great atMonitor tool displaying gpu % load (cpu load,....) in an small (extra) window. Added DL Link for that version (2.7.1) But avoid to update it to new version! For me newer atMonitor versions fail to get the gpu load % - which is used to check GPU load for AGPM.
(Where is the GPU info ? To make atMonitor compatible with MacOS 10.8 we have removed all GPU related features with V 2.8.0)
For me the older (2.71) version runs well with ML!! DL Link below.
To display the gpu load % running at Monitor: activate GPU in preferences / activity view / floanting window.
You dont need any gpu related (fakesmc) gpuplugin installed to get gpu load / gpu VRAM / gpu fps shown.
Only for the gpu temp you must have such an fakesmc plugin installed / same for CPU temp.
Results:
Gpu & load values for 20 triangles(min) + 2000 triangles + 5000 triangles(max) would be interesting.
9600 GT 512 MB:
20 tri : 3% gpu load
2000 : 57% gpu load
5000: 90% gpu load comment: above around 4800 tri the gpu is full loaded (>=91 %)
EDIT UPDATED 24.09.2012
- 20 .... 20000 triangles - wide range for AGPM checks
- compiled with 10.8 SDK
- added Link to DL older atMonitor(freeware) for gpu load % measurement
- added an finetuning (10-5000 triangles) OpenGL triangles AGPM version
Same as OpenGL triangles Bench Sep 2012 version but more finetunable slider for the gpu load .
OpenGL triangles AGPM usable for most gpus lowend to midrage.
Use other (20-20000 triangles) version if your (very fast / highend ) gpu will not get more than 75% gpu load with 5000 triangles.
atMonitor 2.7.1 (has gpu values support, for me runs well with ML):
http://download.atpu...onitor2.7.1.zip
Info atMonitor: http://www.atpurpose.com/atMonitor/
Happy benching / load you GPU / AGPM checks.
example AGPM log:
25.09.12 09:47:35,000 kernel[0]: AGPM: updateGPUHwPstate(1, 0): fHwPstate = 0 fFB = 0xffffff800b261000
25.09.12 09:47:35,000 kernel[0]: AGPM: updateGPUHwPstate(): state = 1. Calling fFB->setAggressiveness()...
25.09.12 09:47:35,000 kernel[0]: AGPM: GPU = display G-state set to 1 from 0, ControlID = 17. SW occupancy updated.
25.09.12 09:47:52,000 kernel[0]: AGPM: updateGPUHwPstate(2, 0): fHwPstate = 1 fFB = 0xffffff800b261000
25.09.12 09:47:52,000 kernel[0]: AGPM: updateGPUHwPstate(): state = 2. Calling fFB->setAggressiveness()...
25.09.12 09:47:52,000 kernel[0]: AGPM: GPU = display G-state set to 2 from 1, ControlID = 17. SW occupancy updated.
25.09.12 09:49:05,000 kernel[0]: AGPM: updateGPUHwPstate(3, 0): fHwPstate = 2 fFB = 0xffffff800b261000
25.09.12 09:49:05,000 kernel[0]: AGPM: updateGPUHwPstate(): state = 3. Calling fFB->setAggressiveness()...
25.09.12 09:49:05,000 kernel[0]: AGPM: GPU = display G-state set to 3 from 2, ControlID = 17. SW occupancy updated.



Sign In
Create Account













