David Yat Sin 38e832a682 Reserve scratch on first queue allocation ...

Some workloads running on multi-GPU create 1 process per GPU. So each
process creates a GPU agent on every GPU, but will only create queues on
one GPU. This would cause un-necessary scratch reservation.

Change-Id: I50a216f0bcc0b5f707f3943147390b0ecec1ac22

2023-05-15 17:10:57 -04:00

..

cmake_modules

Move Images code to hsa-runtime folder

2020-04-30 19:35:57 -05:00

hsa-ext-finalize

Prepare for hsakmt build system cleanup

2018-06-05 16:01:03 -04:00

hsa-ext-image

Build ROCr core and image libraries as one shared object

2020-06-19 22:33:36 -04:00

hsa-runtime

Reserve scratch on first queue allocation

2023-05-15 17:10:57 -04:00

hsa-runtime-tools

Adding RUNPATH to find libhsakmt.so for Centos and SLES

2020-02-14 17:43:29 -05:00

packages

packaging: set arch, field separators, vendor info

2020-08-21 11:33:05 -04:00