derek/gem5
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

configs, gpu-compute: change default GPU reg allocator to dynamic

Browse Source 
The current default GPU register allocator is the "simple" policy,
which only allows 1 wavefront to run at a time on each CU.  This is
not very realistic and also means the tester (when not specifically
choosing the dynamic policy) is less rigorous in terms of validating
correctness.

To resolve this, this commit changes the default to the "dynamic"
register allocator, which runs as many waves per CU as there are
space in terms of registers and other resources -- thus it is more
realistic and does a better job of ensuring test coverage.

Change-Id: Ifca915130bb4f44da6a9ef896336138542b4e93e
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/57537
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
This commit is contained in:
Matt Sinclair
2022-03-13 19:37:53 -05:00
parent f64f05eff6
commit a8aa175c7f
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
configs/example/apu_se.py
View File

@@ -161,7 +161,7 @@ parser.add_argument('--fast-forward-pseudo-op', action='store_true',
' m5_switchcpu pseudo-ops will toggle back and forth')
parser.add_argument("--num-hw-queues", type=int, default=10,
help="number of hw queues in packet processor")
parser.add_argument("--reg-alloc-policy", type=str, default="simple",
parser.add_argument("--reg-alloc-policy", type=str, default="dynamic",
help="register allocation policy (simple/dynamic)")
parser.add_argument("--dgpu", action="store_true", default=False,