cancel
Showing results for 
Search instead for 
Did you mean: 

Archives Discussions

gat3way
Journeyman III

Suggest Feature you want in AMD APP

  • Minimize kernel launch latencies
  • Increase buffer size limits (so that we won't have to rely on experimental environment variables like GPU_INITIAL_HEAP_SIZE and GPU_MAX_HEAP_SIZE)
  • Some kind of integration of certain ADL features with OpenCL through clGetDeviceInfo() would be very useful and convinient as currently proper monitoring of device temperature/load/fanspeed is not very easy as there is no direct device mapping between ADL devices and OpenCL devices


And also certain OpenCL kernel optimizations like:

  • expose BFI_INT to IL and do the proper mapping with OCL's bitselect() as mrbpix suggested as currently, binary patching is the only way to get BFI_INT working
  • Improve long vectors support (ulongN/longN). There are certain problems with them like rotate() not working (and OpenCL documentation states it should be working with them). I don't except that those would be translated to BITALIGN_INT of course, I know that's not possible.
  • Fix cl_amd_printf support. Trying to put printf() in a branch (e.g if (smth) {printf("smth");} does not work and behaves rather erratic.
0 Likes
SiegeLord
Adept II

Suggest Feature you want in AMD APP

In order of importance:

- Reduce kernel launch times

- Remove X server requirement

- Make global atomics not force the complete path for every global memory access in the kernel

0 Likes
kiddoman
Journeyman III

Suggest Feature you want in AMD APP

Pretty good ideas!

It is useful for AMD fans!

0 Likes
Marix
Adept II

Suggest Feature you want in AMD APP

Originally posted by: SiegeLord

- Remove X server requirement



+1

0 Likes
Marix
Adept II

Suggest Feature you want in AMD APP

Another feature I would really like to have:

clGetDeviceInfo should return the GPU name if invoked with CL_DEVICE_NAME instead of the chipset name.

The chipset name makes it impossible for endusers to determine whether the application is using the device it's supposed to be using. In addition the current implementation is buggy anyhow, e.g. returning Cypress on the 5970, where I think it should be Hamlock. This makes it difficult for the developer to estimate the device performance (especially as the MAX_CLOCK also doesn't report correct values).

0 Likes
dukeleto
Adept I

Suggest Feature you want in AMD APP

Hello,

this is not quite directly software related, but might be useful to a number of people if it were made possible:

some kind of web-based service allowing the testing of opencl code on amd hardware.

As an example, I have easy access to reasonable nvidia cards, as they come on all of my laboratory's workstations. Having developed a simulation code in OpenCL, I would like to be able to test it on radeon/firestream cards, but am not prepared to go out and buy a card without having an idea of performance my code could reach.

Sorry if this is slightly off topic!

Regards,

Olivier

0 Likes
hduregger
Journeyman III

Suggest Feature you want in AMD APP

"Increase buffer size limits (so that we won't have to rely on experimental environment variables like GPU_INITIAL_HEAP_SIZE and GPU_MAX_HEAP_SIZE)"

+1

 

I think I already had highly varying performance values between application runs because of this.

 

edit:

oh yea,

  • continued and extended support for linux.
0 Likes
Marix
Adept II

Suggest Feature you want in AMD APP

Especially as it's already there for Windows: Zero-Copy on Linux! That should really speed up my Halo-Exchange, finally making my code work with more than one computer.

0 Likes
himanshu_gautam
Grandmaster

Suggest Feature you want in AMD APP

dukelato,

Thanks for your suggestion.

Here it is:http://forums.amd.com/forum/messageview.cfm?catid=390&threadid=149708&forumid=9

 

Have Fun

0 Likes
rahulgarg
Adept II

Suggest Feature you want in AMD APP

Adding my vote for the following:

cl_khr_fp64 on GPU

Zero-copy allocations for APUs on Linux

Removal of display connection requirement

0 Likes