Add a one workgroup argmax benchmark #49
Conversation
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
|
@kuhar @antiagainst Could you please review? Thanks |
There was a problem hiding this comment.
Looks good modulo copyright headers. For code that is primarily based on some existing benchmarks, we use dual copyright (the original author + the new one). You can see an example here: https://github.com/google/uVkCompute/blob/main/benchmarks/vmt/vmt_main.cc.
| #extension GL_KHR_shader_subgroup_arithmetic : enable | ||
| #extension GL_KHR_shader_subgroup_ballot : enable | ||
|
|
||
| layout(local_size_x = 16, local_size_y = 1, local_size_z = 1) in; |
There was a problem hiding this comment.
This is smaller than the native subgroup size for desktop GPU. How does this perform if we increase this to 32 or 64?
There was a problem hiding this comment.
I did not see any performance improvement
angelz913
force-pushed
the
argmax
branch
2 times, most recently
from
e0a8a84
to
6604dbd
Compare
This PR is based on #47. I opened a new one because the old one got stale.