[gem5-dev] Change in gem5/gem5[develop]: tests, gpu-compute: test dynamic register policy in regressions

Matt Sinclair (Gerrit) via gem5-dev Tue, 02 Nov 2021 12:06:16 -0700

Matt Sinclair has submitted this change. (https://gem5-review.googlesource.com/c/public/gem5/+/52163 )

Change subject: tests, gpu-compute: test dynamic register policy inregressions

......................................................................


tests, gpu-compute: test dynamic register policy in regressions

The GPU models support a simple register allocation policy (1 WF/CU at a
time) and a dynamic register allocation policy (up to max WF/CU at a
time).  By default, the simple policy is used.  However, the dynamic
policy is much more realistic relative to real hardware and thus much
more important to ensure it works in the regressions.  This commit
updates the nightly and weekly regressions accordingly to run the
dynamic register allocation policy.

Change-Id: Id263d3d5e19e4ff47f0eb6d9b08cbafdf2177fb9
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/52163
Tested-by: kokoro <[email protected]>
Reviewed-by: Matt Sinclair <[email protected]>
Maintainer: Bobby R. Bruce <[email protected]>
---
M tests/weekly.sh
M tests/nightly.sh
2 files changed, 40 insertions(+), 8 deletions(-)

Approvals:
  Matt Sinclair: Looks good to me, approved
  Bobby R. Bruce: Looks good to me, approved
  kokoro: Regressions pass




diff --git a/tests/nightly.sh b/tests/nightly.sh
index b3708fd..41db369 100755
--- a/tests/nightly.sh
+++ b/tests/nightly.sh
@@ -83,7 +83,6 @@
         ./main.py run --length long -j${threads} -t${threads}

 # Run the GPU tests.
-
 # For the GPU tests we compile and run GCN3_X86 inside a gcn-gpu container.
 docker pull gcr.io/gem5-test/gcn-gpu:latest
 docker run --rm -u $UID:$GID --volume "${gem5_root}":"${gem5_root}" -w \
@@ -101,7 +100,7 @@
 # basic GPU functionality is working.
 docker run --rm -u $UID:$GID --volume "${gem5_root}":"${gem5_root}" -w \

"${gem5_root}" gcr.io/gem5-test/gcn-gpu:latest build/GCN3_X86/gem5.opt\

-    configs/example/apu_se.py -n3 -c square
+    configs/example/apu_se.py --reg-alloc-policy=dynamic -n3 -c square

 # get HeteroSync

wget -qNhttp://dist.gem5.org/dist/develop/test-progs/heterosync/gcn3/allSyncPrims-1kernel

@@ -112,8 +111,8 @@
 # atomics are tested.
 docker run --rm -u $UID:$GID --volume "${gem5_root}":"${gem5_root}" -w \

"${gem5_root}" gcr.io/gem5-test/gcn-gpu:latest build/GCN3_X86/gem5.opt\

-    configs/example/apu_se.py -n3  -c allSyncPrims-1kernel \
-    --options="sleepMutex 10 16 4"
+    configs/example/apu_se.py --reg-alloc-policy=dynamic -n3 -c \
+    allSyncPrims-1kernel --options="sleepMutex 10 16 4"

 # run HeteroSync LFBarr -- similar setup to sleepMutex above -- 16 WGs
 # accessing unique data and then joining a lock-free barrier, 10 Ld/St per
@@ -122,5 +121,5 @@
 # atomics are tested.
 docker run --rm -u $UID:$GID --volume "${gem5_root}":"${gem5_root}" -w \

"${gem5_root}" gcr.io/gem5-test/gcn-gpu:latest build/GCN3_X86/gem5.opt\

-    configs/example/apu_se.py -n3  -c allSyncPrims-1kernel \
-    --options="lfTreeBarrUniq 10 16 4"
+    configs/example/apu_se.py --reg-alloc-policy=dynamic -n3 -c \
+    allSyncPrims-1kernel --options="lfTreeBarrUniq 10 16 4"
diff --git a/tests/weekly.sh b/tests/weekly.sh
index 51376bd..172d955 100755
--- a/tests/weekly.sh
+++ b/tests/weekly.sh
@@ -95,7 +95,7 @@
 # stressing several GPU compute and memory components
 docker run --rm -u $UID:$GID --volume "${gem5_root}":"${gem5_root}" -w \
     "${gem5_root}" hacc-test-weekly build/GCN3_X86/gem5.opt \
-    configs/example/apu_se.py -n3 --mem-size=8GB \

+ configs/example/apu_se.py -n3 --mem-size=8GB--reg-alloc-policy=dynamic \--benchmark-root="${gem5_root}/gem5-resources/src/gpu/lulesh/bin" -clulesh


 # test DNNMark
@@ -137,6 +137,7 @@

"${gem5_root}/gem5-resources/src/gpu/DNNMark/cachefiles:/root/.cache/miopen/2.9.0"\

        -w "${gem5_root}/gem5-resources/src/gpu/DNNMark" hacc-test-weekly \

"${gem5_root}/build/GCN3_X86/gem5.opt" "${gem5_root}/configs/example/apu_se.py"-n3\

+       --reg-alloc-policy=dynamic \

--benchmark-root="${gem5_root}/gem5-resources/src/gpu/DNNMark/build/benchmarks/test_fwd_softmax"\

        -c dnnmark_test_fwd_softmax \

--options="-config${gem5_root}/gem5-resources/src/gpu/DNNMark/config_example/softmax_config.dnnmark\

@@ -146,6 +147,7 @@

"${gem5_root}/gem5-resources/src/gpu/DNNMark/cachefiles:/root/.cache/miopen/2.9.0"\

        -w "${gem5_root}/gem5-resources/src/gpu/DNNMark" hacc-test-weekly \

"${gem5_root}/build/GCN3_X86/gem5.opt" "${gem5_root}/configs/example/apu_se.py"-n3\

+       --reg-alloc-policy=dynamic \

--benchmark-root="${gem5_root}/gem5-resources/src/gpu/DNNMark/build/benchmarks/test_fwd_pool"\

        -c dnnmark_test_fwd_pool \

--options="-config${gem5_root}/gem5-resources/src/gpu/DNNMark/config_example/pool_config.dnnmark\

@@ -155,6 +157,7 @@

"${gem5_root}/gem5-resources/src/gpu/DNNMark/cachefiles:/root/.cache/miopen/2.9.0"\

        -w "${gem5_root}/gem5-resources/src/gpu/DNNMark" hacc-test-weekly \

"${gem5_root}/build/GCN3_X86/gem5.opt" "${gem5_root}/configs/example/apu_se.py"-n3\

+       --reg-alloc-policy=dynamic \

--benchmark-root="${gem5_root}/gem5-resources/src/gpu/DNNMark/build/benchmarks/test_bwd_bn"\

        -c dnnmark_test_bwd_bn \

--options="-config${gem5_root}/gem5-resources/src/gpu/DNNMark/config_example/bn_config.dnnmark\

@@ -170,7 +173,7 @@
 # the GPU memory system
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
-       ${gem5_root}/configs/example/apu_se.py -n3 \

+ ${gem5_root}/configs/example/apu_se.py -n3--reg-alloc-policy=dynamic \--benchmark-root=${gem5_root}/gem5-resources/src/gpu/halo-finder/src/hip \

        -c ForceTreeTest --options="0.5 0.1 64 0.1 1 N 12 rcb"

@@ -191,6 +194,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \
        --benchmark-root=gem5-resources/src/gpu/pannotia/bc/bin -c bc.gem5 \
        --options="1k_128k.gr"

@@ -204,6 +208,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/color/bin \

        -c color_max.gem5 --options="1k_128k.gr 0"

@@ -217,6 +222,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/color/bin \

        -c color_maxmin.gem5 --options="1k_128k.gr 0"

@@ -230,6 +236,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/fw/bin \

        -c fw_hip.gem5 --options="1k_128k.gr"

@@ -243,6 +250,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/mis/bin \

        -c mis_hip.gem5 --options="1k_128k.gr 0"

@@ -258,6 +266,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/pagerank/bin \

        -c pagerank.gem5 --options="coAuthorsDBLP.graph 1"

@@ -271,6 +280,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/pagerank/bin \

        -c pagerank_spmv.gem5 --options="coAuthorsDBLP.graph 1"

@@ -284,6 +294,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/sssp/bin \

        -c sssp.gem5 --options="1k_128k.gr 0"

@@ -297,6 +308,7 @@
 docker run --rm -v ${gem5_root}:${gem5_root} -w ${gem5_root} -u $UID:$GID \
        hacc-test-weekly ${gem5_root}/build/GCN3_X86/gem5.opt \
        ${gem5_root}/configs/example/apu_se.py -n3 --mem-size=8GB \
+       --reg-alloc-policy=dynamic \

--benchmark-root=${gem5_root}/gem5-resources/src/gpu/pannotia/sssp/bin \

        -c sssp_ell.gem5 --options="1k_128k.gr 0"


--
To view, visit https://gem5-review.googlesource.com/c/public/gem5/+/52163

To unsubscribe, or for help writing mail filters, visithttps://gem5-review.googlesource.com/settings


Gerrit-Project: public/gem5
Gerrit-Branch: develop
Gerrit-Change-Id: Id263d3d5e19e4ff47f0eb6d9b08cbafdf2177fb9
Gerrit-Change-Number: 52163
Gerrit-PatchSet: 5
Gerrit-Owner: Matt Sinclair <[email protected]>
Gerrit-Reviewer: Bobby R. Bruce <[email protected]>
Gerrit-Reviewer: Jason Lowe-Power <[email protected]>
Gerrit-Reviewer: Matt Sinclair <[email protected]>
Gerrit-Reviewer: Matthew Poremba <[email protected]>
Gerrit-Reviewer: kokoro <[email protected]>
Gerrit-CC: Kyle Roarty <[email protected]>
Gerrit-MessageType: merged

_______________________________________________
gem5-dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s

[gem5-dev] Change in gem5/gem5[develop]: tests, gpu-compute: test dynamic register policy in regressions

Reply via email to