xinhui pan
c61fffa876
The test measures the bandwidth between GPUs. Currently we do not
care numa topology as some products really support across PCI-e root
complex p2p.
test result on two gfx900 system.
[ RUN ] KFDPerformanceTest.P2PBandWidthTest
[ ] Copy from node to node by [push, NONE]
[ ] [1 -> 0] 6.13477 - 6.12695 GB/s
[ ] [1 -> 2] 3.77734 - 3.76855 GB/s
[ ] [2 -> 0] 6.67676 - 6.6543 GB/s
[ ] [2 -> 1] 6.14453 - 6.12793 GB/s
[ ] Copy from node to node by [pull, NONE]
[ ] [1 -> 0] 6.10547 - 6.08105 GB/s
[ ] [1 -> 2] 9.65527 - 9.65039 GB/s
[ ] [2 -> 0] 6.49805 - 6.4873 GB/s
[ ] [2 -> 1] 8.95508 - 8.85254 GB/s
[ ] Full duplex copy from node to node by [push|pull, NONE]
[ ] [1 -> 0] 11.0986 - 11.0986 GB/s
[ ] [1 -> 2] 7.54297 - 7.54297 GB/s
[ ] [2 -> 0] 12.0264 - 11.9639 GB/s
[ ] [2 -> 1] 12.0469 - 12.0371 GB/s
[ ] Full duplex copy from node to node by [push, push]
[ ] [1 <-> 2] 11.7324 - 11.4541 GB/s
[ ] Full duplex copy from node to node by [pull, pull]
[ ] [1 <-> 2] 11.4824 - 11.0508 GB/s
[ ] Copy from node to multiple nodes by [push, NONE]
[ ] [1 -> [0...2]] 5.625 - 5.73633 GB/s
[ ] [2 -> [0...2]] 6.45801 - 6.4707 GB/s
[ ] Copy from multiple nodes to node by [push, NONE]
[ ] [[1...2] -> 0] 12.8379 - 12.2578 GB/s
Now we can get more timestamp info like below.
Copy from node to node by [push, NONE]
[1 -> 0]
[1 : 0] #-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-##-#-#-#-#-###############################
[1 : 1] ####################################################################################################
[1 -> 2]
[1 : 0] #--#-#-#-#-#--#-#-#-#-#--#-#-#-#-#--#-#-#-#-#-#--#-#-#-#-#--#-#-#-#-#--#-#-#-#-#-#--#-#-#-#-#--#-#-######################################
[1 : 1] ##################################################################################################-#
[2 -> 0]
[2 : 0] ##-###-##-###-###-##-###-##-###-###-##-###-###-##-###-###-##-###-##-###-###-##-###-###-##-###-###-#################
[2 : 1] ###############################################################################-#############-###-##
[2 -> 1]
[2 : 0] ##-##-##-##-##-###-##-##-##-##-##-###-##-##-##-##-###-##-##-##-##-##-###-##-##-##-##-###-##-##-##-####################
[2 : 1] ################################################################################-###-############-##
[snip]
Full duplex copy from node to node by [push, push]
[1 <-> 2]
[1 : 0] #-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-#-#-#-####################################
[1 : 1] ################-###################################################-############-####-#############
[2 : 2] #-##-##-##-#-##-##-##-##-#-##-##-##-##-#-##-##-##-##-#-##-##-##-##-#-##-##-##-##-##-#-##-##-##-##-#-##-##-##-##-##-#-##################
[2 : 3] #####-######-#####-######-#####-######-#####-######-#####-######-#####-######-#####-######-#####-######-#####-#####-##
Full duplex copy from node to node by [pull, pull]
[1 <-> 2]
[1 : 0] ######################################################################-##-#-###############-####-###
[1 : 1] #-#-#-##-#-#-##-#-#-##-#-#-##-#-#-##-#-#-##-#-#-##-#-#-#-##-#-#-##-#-#-##-#-#-##-#-#-##-#-#-##-#-#-############################
[2 : 2] ##-##-##-##-###-##-##-##-##-###-##-##-##-###-##-##-##-##-###-##-##-##-###-##-##-##-##-###-##-##-##-##-###-##-##-##-###-##-##-############
[2 : 3] #-#-#-#-#-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#########-#############
Copy from node to multiple nodes by [push, NONE]
[1 -> [0...2]]
[1 : 0] #-#--#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-###############################
[1 : 1] ########################################################################################-###-###-###
[2 -> [0...2]]
[2 : 0] ##-##-##-###-##-###-##-##-###-##-###-##-##-###-##-###-##-###-##-##-###-##-###-##-##-###-##-###-##-##################
[2 : 1] -################################################################################################-##
Copy from multiple nodes to node by [push, NONE]
[[1...2] -> 0]
[1 : 0] #-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-##-#-#-#-#-#-#-#-#-##-#-#-#-###############################
[1 : 1] ################################################################################################-#-#
[2 : 2] ##-##-##-###-##-##-###-##-##-##-###-##-##-###-##-##-###-##-##-###-##-##-##-###-##-##-###-##-##-###-##-##################
[2 : 3] #########################-#########################-#########################-#########################
[ OK ] KFDPerformanceTest.P2PBandWidthTest (15982 ms)
Change-Id: Ia90044191d51650ccb220476d31fb317aa3ad6ce
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
[ROCm/ROCR-Runtime commit: e5a541eaf2]
133 wiersze
4.6 KiB
Plaintext
133 wiersze
4.6 KiB
Plaintext
# Permanent exclusions
|
|
# These tests are included for debugging, but are not executed in normal execution on any ASIC:
|
|
#
|
|
# Fragmentation test fails intermittently on different systems (see KFD-344). It
|
|
# stresses memory management in unusual ways by having lots of memory allocated but
|
|
# not mapped, which is not relevant to compute applications over ROCr.
|
|
#
|
|
# CU Masking Linear/Split are not working correctly due to how the HW distributes work over CUs.
|
|
# They are available for testing but are not currently expected to pass on CI/VI/AI.
|
|
PERMANENT_BLACKLIST_ALL_ASICS=\
|
|
"-KFDEventTest.MeasureInterruptConsumption:"\
|
|
"KFDLocalMemoryTest.Fragmentation:"\
|
|
"KFDQMTest.BasicCuMaskingLinear:"\
|
|
"RDMATest.GPUDirect"
|
|
|
|
# This is the temporary blacklist for all ASICs. This is to be used when a test is failing consistently
|
|
# on every ASIC (Kaveri, Carrizo, Hawaii, Tonga, Fiji, Polaris10, Polaris11 and Vega10 .
|
|
# TODO means that a JIRA ticket needs to be created for this issue, as no documentation regarding
|
|
# failures can be found
|
|
# NOTE: If you update this alphabetical listing, add the corresponding JIRA ticket for reference
|
|
#
|
|
# KFDGraphicsInterop.RegisterForeignDeviceMem (KFD-366)
|
|
# KFDPMTest.SuspendWith[ActiveProcess|IdleQueue|IdleQueueAfterWork] (KFD-326|327|328)
|
|
# KFDPNPTest.DisableAndCreateQueue (KFD-329)
|
|
# KFDQMTest.GPUDoorbellWrite fails intermittently (KFD-318)
|
|
# KFDQMTest.mGPUShareBO (KFD-334)
|
|
TEMPORARY_BLACKLIST_ALL_ASICS=\
|
|
"KFDGraphicsInterop.RegisterForeignDeviceMem:"\
|
|
"KFDPMTest.SuspendWithActiveProcess:"\
|
|
"KFDPMTest.SuspendWithIdleQueue:"\
|
|
"KFDPMTest.SuspendWithIdleQueueAfterWork:"\
|
|
"KFDPNPTest.DisableAndCreateQueue:"\
|
|
"KFDQMTest.GPUDoorbellWrite:"\
|
|
"KFDQMTest.mGPUShareBO"
|
|
|
|
BLACKLIST_ALL_ASICS=\
|
|
"$PERMANENT_BLACKLIST_ALL_ASICS:"\
|
|
"$TEMPORARY_BLACKLIST_ALL_ASICS"
|
|
|
|
# SDMA-based tests (KFDIPCTest.BasicTest, KFDQM.*Sdma*, KFDMemoryTest.MMBench) are all
|
|
# disabled on non-Hawaii due to SDMA instability - SWDEV-101666
|
|
SDMA_BLACKLIST=\
|
|
"KFDIPCTest.*:"\
|
|
"KFDMemoryTest.MMBench:"\
|
|
"KFDQMTest.*Sdma*:"\
|
|
"KFDQMTest.P2PTest:"\
|
|
"KFDPerformanceTest.P2PBandWidthTest"
|
|
|
|
# Anything involving CP queue creation is failing on Kaveri. Separate them here for convenience (KFD-336)
|
|
KV_QUEUE_BLACKLIST=\
|
|
"KFDExceptionTest.InvalidBadAddress:"\
|
|
"KFDExceptionTest.InvalidWriteAddress:"\
|
|
"KFDExceptionTest.InvalidPPR*:"\
|
|
"KFDLocalMemoryTest.*:"\
|
|
"KFDEventTest.Signal*Event*:"\
|
|
"KFDQMTest.CreateQueueStressSingleThreaded:"\
|
|
"KFDQMTest.*CpQueue*:"\
|
|
"KFDQMTest.*Dispatch*:"\
|
|
"KFDQMTest.Atomics:"\
|
|
"KFDQMTest.GPUDoorbellWrite"
|
|
|
|
# KFDQMTest.CpuWriteCoherence fails. 0 dwordsAvailable (KFD-338)
|
|
# KFDMemoryTest.MemoryRegister fails on SDMA queue creation (KFD-337)
|
|
KV_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"$SDMA_BLACKLIST:"\
|
|
"$KV_QUEUE_BLACKLIST:"\
|
|
"KFDMemoryTest.MemoryRegister:"\
|
|
"KFDQMTest.CpuWriteCoherence"
|
|
|
|
# KFDLocalMemoryTest.BasicTest is failing intermittently (KFD-368)
|
|
# KFDMemoryTest.BigBufferStressTest was failing intermittently on 4.9 and hangs when executed twice (KFD-312)
|
|
# KFDQMTest.GPUDoorbellWrite fails on Hawaii. Could be HW-related (KFD-342)
|
|
HI_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"KFDLocalMemoryTest.BasicTest:"\
|
|
"KFDMemoryTest.BigBufferStressTest:"\
|
|
"KFDQMTest.GPUDoorbellWrite"
|
|
|
|
CZ_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"$SDMA_BLACKLIST:"\
|
|
"KFDExceptionTest.InvalidWriteAddress"
|
|
|
|
# KFDPerfCountersTest.*Trace fail (KFD-339)
|
|
# KFDMemoryTest.QueryPointerInfo/MemoryRegister* (KFD-341)
|
|
# The remaining tests listed here fail on map memory to GPU with a VA conflict (KFD-340)
|
|
TONGA_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"$SDMA_BLACKLIST:"\
|
|
"KFDCWSRTest.BasicTest:"\
|
|
"KFDEventTest.*:"\
|
|
"KFDLocalMemoryTest.*:"\
|
|
"KFDMemoryTest.FlatScratchAccess:"\
|
|
"KFDMemoryTest.MapMemoryToGPU:"\
|
|
"KFDMemoryTest.MemoryRegister*:"\
|
|
"KFDMemoryTest.QueryPointerInfo:"\
|
|
"KFDPerfCountersTest.*:"\
|
|
"KFDDBGTest.BasicAddressWatch:"\
|
|
"KFDQMTest.*"
|
|
|
|
FIJI_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"$SDMA_BLACKLIST"
|
|
|
|
ELLESMERE_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"$SDMA_BLACKLIST"
|
|
|
|
BAFFIN_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"$SDMA_BLACKLIST"
|
|
|
|
# KFDDBG.BasicAddressWatch is not supported on GFXv9 and later. FIXME: it should disable itself
|
|
VEGA10_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"KFDDBGTest.BasicAddressWatch"
|
|
|
|
VEGA20_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"KFDDBGTest.BasicAddressWatch:"\
|
|
"KFDExceptionTest.InvalidBadAddress:"\
|
|
"KFDExceptionTest.InvalidWriteAddress:"\
|
|
"KFDQMTest.CreateQueueStressSingleThreaded:"\
|
|
"KFDQMTest.GPUDoorbellWrite:"\
|
|
"KFDMemoryTest.MemoryRegister"
|
|
|
|
RAVEN_TESTS_BLACKLIST=\
|
|
"$BLACKLIST_ALL_ASICS:"\
|
|
"KFDDBGTest.BasicAddressWatch:"\
|
|
"KFDQMTest.QueueLatency:"\
|
|
"KFDQMTest.SdmaEventInterrupt"
|