Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-

GPU programming

Name: Anonymous 2010-07-22 19:08

I posted this in /g/ a few minutes ago.

Is someone here familiar with GPU programming?
To make this clear in prior: I know Assembly better than C and I intend to use it.

How (from the CPU software in the main memory's point of view) does one initiate contact with the GPU? Hand it kernels and otherwise any communication between. Is the GPU already running in a loop from the very beginning and all we have to do is edit the video memory to let it "escape"?

On second thought: This question might apply just as much to any other CPU cores that might be running in parallel. Although I suspect that is way simpler.

Name: Anonymous 2010-07-22 19:32

I'll bite.

Choose CUDA or OpenCL. Download SDK. RTFM. It's not hard. They will tell you not to (but you can) use assembly in your kernels.

Name: Anonymous 2010-07-22 20:00

Thank you for replying.
I did read what I found about it but most of it was pre compilation C work and driver development dependent on underlying environments provided by OS'es. Guess I'll just have to dig deeper.

Name: Anonymous 2010-07-22 20:05

Oh if anyone wonders why;
I'm building a bootable system and I plan to exploit the GPU's superiority in mathematical operations as much as I can.

Name: Anonymous 2010-07-22 21:08

gpu's with 2MB of video memory? i've got one right now! it's got COLOR OUTPUT!

Name: Anonymous 2010-07-22 21:13

>>3,5
OS'es
gpu's
I hope you both get hit by a car.

Name: Anonymous 2010-07-22 21:36

>>4
The GPU doesn't have superiority with mathematical operations.  It has superiority in certain types of SIMD operations.  Learn to tell the difference.  Also, if you're not using CUDA or OpenGL, then what you're doing is writing your own GPU drivers.  That's an abyssal pit of suck from which you'll never recover.

Also, if the words "cache coherency" don't ring any bells (and be honest... you had to look it up, didn't you?) then you are not fit to write GPU drivers.

Name: Anonymous 2010-07-22 21:40

>>1
As for the GPU, you communicate with it through the PCI bus as always.  If you don't know how the PCI bus works, then you've got a long way to go.  If you don't know what a memory mapped register is, then just give up already.

As for other CPU cores, it depends on the CPU and the BIOS.  Some CPUs boot all cores at power on, sometimes the BIOS turns all but one of them off right away, and some CPUs leave all the extra cores in a power down state.

Name: Anonymous 2010-07-22 23:18

>>6
Did you mean, ``I hope you both get hit by a cudder''?

Name: Anonymous 2010-07-23 4:26

>>7
>SIMD
True. Guess I was being way to general.

>cache coherency
and lets be honest.. It IS scary stuff, yes. but I'm aware and I think I'll learn a lot from it.

>>6
>PCI bus
Well Duh.. , but no. I don't know how the PCI bus works. Memory mapped registers, I am familiar with (thank god). But that wasn't really that hard to learn.

>other CPU cores
Well that sucks... Still, way easier than gpu tho.

Thank you for your replies.

Don't change these.
Name: Email:
Entire Thread Thread List