Hello List

Hello List,
I just made my first OpenCL program (for Snow Leopard).
It is not optimized yet!
I suspect I am losing most of my performance on the ATI 4850 device from reading back to host memory.
It is sort of difficult finding good examples for things like Shared CL/GL textures.
I was wondering if anyone might be willing to run it and post the performance results of their machines onto this board (reply to this post).
If you don’t mind post your CPU and GPU scores.
I am just curious what relative performances are between the devices are.
If anyone gets any crashes I will try to address it ASAP.

you can find it here:
http://www.tweakoz.com/portfolio/index.php?page=OpenCL

Thanks,
mtm

I should mention, I am interested in the following values (for each device):

FPS:
MPPS:
MSPS:
CalcSize:
GlobalSize:
LocalSize:
DeviceName:
MaxFrq (Max core frequency)
NumCU (Number of Compute Units)
MaxWIS (Max Work Items)
MaxWGS (Max Work Groups)

Thanks,

mtm

One more thing,
you can resize the window to vary the fill.
Please post results at at the biggest size you can (to amortize the readback)!

Thanks,

mtm

Mac Pro Quad 2.66 Ghz (First Generation with PCI-Express 1.0)

FPS : 20
MPPS : 30
MSPS : 780
CalcSize : 1664,896
GlobalSize : 52,28
LocalSize : 13,7
DeviceName : GeForce 8800 GT
MaxFrq : 1500 MHz
NumCU : 112
MaxWIS : 512,512,64
MaxWGS : 512