Commit Graph

27 Commits

Author SHA1 Message Date
Evgeny 5abb8e1a68 API tracing instrumentation
Change-Id: I257409b9fe299b009ded3e3a43287322d5f93a70
2020-05-14 11:03:09 -05:00
Payam 56509b0ffd adding support to disable lazy loading, HIP_ENABLE_LAZY_KERNEL_LOADING
Change-Id: Iafe0f396570ee2bccf642b537cbf9aad967e1370
2020-05-13 17:18:48 -04:00
Christophe Paquot a03fee04fe Make sure to remove the global var from amd::MemObjMap
We need this otherwise ROCr can give us a matching address
for another allocation and doing "insert" in ROCclr will not
update the map with the newest object. We would then end up
using stale objects (yikes)
SWDEV-234992

Change-Id: I3475adf9781a9309d64a024fae45181d7e5afb04
2020-05-13 00:26:19 -07:00
Christophe Paquot f100ae3679 Add lock to addFatBinary and removeFatBinary
In case hipModule(Un)Load is called from different thread as hipInit we need to grab the lock
as both are going to modify modules_
Also add some logging for __hipExtractCodeObjectFromFatBinary in case binary isn't found for GPU
SWDEV-236032

Change-Id: Icbd72b412502df80d5066cea42a4fbcd5b0b8a98
2020-05-12 16:04:08 -07:00
Saleel Kudchadker 7f548c5aa9 Merge "SWDEV-235495 Fix elapsed time calculation" into amd-master-next 2020-05-12 15:23:23 -04:00
Saleel Kudchadker 43986c6791 SWDEV-235495 Fix elapsed time calculation
This issue happens because we getLastQueuedCommand when recording
the event and do end_ - start_ so it takes the ticks for the
completion of the last command before event record. This may not
happen if one records a marker command for hipEventRecord

Change-Id: I1d6b06a5befb3b93f16b67692c59dca25c982e0f
2020-05-12 09:38:01 -07:00
Christophe Paquot 538f18f06c Merge "HPC : Intermittent hangs are observed while running Gromacs benchmarks" into amd-master-next 2020-05-12 11:21:47 -04:00
Karthik Jayaprakash 54d264c418 Merge "SWDEV-227602 - Fixing hipFuncGetAttribute parameters" into amd-master-next 2020-05-12 10:52:25 -04:00
Matt Arsenault 024764c337 Use find_package to find ROCclr
Maintain compatability with the old finding for now for the
convenience of commit order.

Change-Id: I99b236cbb3d61b00650e3da7fe5931d4c4b3fec6
2020-05-12 09:28:19 -04:00
kjayapra-amd 78e27488be SWDEV-227602 - Fixing hipFuncGetAttribute parameters
Change-Id: I46bd079372f453cabfaa2c709e2b30c69400dd33
2020-05-12 09:15:43 -04:00
Matt Arsenault ab285297e9 Use project relative include dir for generated header paths
Change-Id: I8c0834d68e6dac00b41a7059a978cbd2ebb933f9
2020-05-11 21:57:25 -04:00
Christophe Paquot 6b361bc1a0 HPC : Intermittent hangs are observed while running Gromacs benchmarks
SWDEV-235579
Move the lock before destroying the queue as there's a multithreaded race condition if the queue
is being destroy and right after we set queue_ to nullptr, another thread can call ihipWaitStreams
which will then call create on that same stream because queue is now nullptr.
Moving the lock on streamSet prevents this from happening because we would remove the stream from that
list and therefore ihipWait will not try to call asHostQueue which tries to create the queue if not created yet
since the stream won't be in the list anymore

Change-Id: I3108657ab403d39d4123e83294fcf1f0880e5563
2020-05-11 16:35:13 -07:00
Payam a477612a22 updating hipDeviceProps_t-< arch flags according to SWDEV-234277
Change-Id: I6238edf9a2df15a5dfe420dbb8723e10a0725f98
2020-05-11 16:43:13 -04:00
Matt Arsenault 16dbce3776 Start cleaning up rocclr include paths
Use target_include_directories instead of global includes.

Change-Id: I3adadc905843f152a548c71b2f12701476065207
2020-05-11 15:24:19 -04:00
Vlad Sytchenko fec51e85b0 Correct HIP_FUNC_ATTRIBUTE_NUM_REGS query
Change-Id: I526cc7871c690260df0fa8c1b3b4b15fbc5af219
2020-05-09 12:42:30 -04:00
Christophe Paquot 507f6bb0c3 Merge "Don't add a null command to waitList in hipMemcpy" into amd-master-next 2020-05-08 17:01:38 -04:00
Christophe Paquot 3ca2d1e6a9 Don't add a null command to waitList in hipMemcpy
SWDEV-235345

Change-Id: Ib1abd0ba6414d081891a9f5209df083c45734aee
2020-05-08 12:43:41 -07:00
Vlad Sytchenko 1b1c032e9f Correct HIP_FUNC_ATTRIBUTE_MAX_THREADS_PER_BLOCK query
We should be returning the max workgroup size calculated by the compiler.

Change-Id: If86590efbb9b291f470bdbe87e5df992e661c539
2020-05-08 14:36:47 -04:00
Vlad Sytchenko a373538d72 Fix confusion in hipFuncGetAttribute()
Cuda shared == OpenCL local

Cuda local == OpenCL private

Change-Id: I5a204945ecde35919b9e9def20bbb2662fffea2b
2020-05-08 14:36:36 -04:00
Matt Arsenault 6262ce482f Don't add llvm includes to rocclr build
There are no llvm includes here.

Change-Id: I0567c6e47e717580819f3854782ba08501bd069d
2020-05-08 12:24:49 -04:00
Sarbojit Sarkar 3612851809 Enabling hipGetDeviceFlags required in [SWDEV-229170]
Change-Id: I998d37e5847f9651345554bada86df6fce86d1eb
2020-05-08 01:37:23 -04:00
Aaron Enye Shi 1c1be71b63 Fix missed VDI names in hip-on-rocclr
Change-Id: I830feb37a043656136648e92a0c6f1eaae8402d7
2020-05-07 08:37:28 -04:00
Christophe Paquot 78df96024d Merge "SWDEV-232464 - Need to initialize image with ptr passed since they can pass image not of type __ClangOffloadBundler." into amd-master-next 2020-05-06 16:42:50 -04:00
kjayapra-amd 5e91bee221 SWDEV-232464 - Need to initialize image with ptr passed since they can pass image not of type __ClangOffloadBundler.
Change-Id: I2c50042220a0230bc445ed21728f114a229c53e1
2020-05-06 14:25:43 -04:00
Christophe Paquot f0b49d0560 Skip iHipWaitActiveStreams if stream is NonBlocking
SWDEV-234484

Change-Id: I279bdc8485d8218e0aaa89d094c08a84f002a608
2020-05-06 10:36:53 -07:00
Aakash Sudhanwa 6f86a74018 rocclr/CMakeLists.txt: Fixing unresolved merge conflict
Change-Id: I09b9a955b0f1ece8d278112a22bc8aef31b57258
Signed-off-by: Aakash Sudhanwa <Aakash.Sudhanwa@amd.com>
2020-05-06 07:24:25 -07:00
Payam c5f76c3de3 name change vdi to rocclr
Change-Id: I06d198bbb4a499e153b290b73a92afed3553b252
2020-05-06 09:14:30 -04:00