-
Notifications
You must be signed in to change notification settings - Fork 116
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[UR][L0] Add several fixes to L0 adapter for test-usm #1105
Conversation
source/adapters/level_zero/usm.cpp
Outdated
return ReturnValue(Pool); | ||
} | ||
if (Pool->HostMemPool.get() == UMFPool) { | ||
return ReturnValue(Pool->HostMemPool.get()); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This should return ReturnValue(Pool);
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
thanks @kswiecicki! fixed.
source/adapters/level_zero/usm.cpp
Outdated
@@ -829,6 +884,10 @@ UR_APIEXPORT ur_result_t UR_APICALL urUSMPoolCreate( | |||
try { | |||
*Pool = reinterpret_cast<ur_usm_pool_handle_t>( | |||
new ur_usm_pool_handle_t_(Context, PoolDesc)); | |||
|
|||
std::shared_lock<ur_shared_mutex> ContextLock(Context->Mutex); | |||
Context->UsmPoolHandles.push_back(*Pool); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Context->UsmPoolHandles
should also be updated in urUSMPoolRelease
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
thanks @kswiecicki. missed it. fixed.
feda3bd
to
60ad384
Compare
There are some tests failing because we need to have UMF tracking. So we need to have this one merged #1071 |
1bdeaca
to
6caf12a
Compare
I have updated the target branch of this PR from the |
6caf12a
to
45b23e5
Compare
I'm merged #1071 now so will rebase this on top of the main branch. EDIT: I didn't notice the merge conflict, I'll let you resolve that as part of updating the branch @jandres742 |
I went ahead and rebased this on top of main. |
Codecov ReportAll modified and coverable lines are covered by tests ✅
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## main #1105 +/- ##
=======================================
Coverage 15.73% 15.74%
=======================================
Files 223 223
Lines 31466 31466
Branches 3556 3556
=======================================
+ Hits 4952 4953 +1
+ Misses 26463 26462 -1
Partials 51 51 ☔ View full report in Codecov by Sentry. |
I updated intel/llvm#11980 to point at the latest commit on this branch and its caused 121 failures. @nrspruit I think I've taken this one as far as I can. I'll need to hand that over to you. |
Thanks @kbenzie , I will be taking a look tomorrow to determine what is causing the failures. |
@@ -187,8 +187,15 @@ static ur_result_t USMDeviceAllocImpl(void **ResultPtr, | |||
ZeDesc.pNext = &RelaxedDesc; | |||
} | |||
|
|||
ZE2UR_CALL(zeMemAllocDevice, (Context->ZeContext, &ZeDesc, Size, Alignment, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks like this macro was needed for the l0 api tracing enabled with UR_L0_DEBUG
, that tracing it necessary for a lot of the sycl-e2e tests (they check for certain l0 api calls in the trace output). We should make these kinds of changes like this
ze_result_t ZeResult = ZE_CALL_NOCHECK(zeMemAllocDevice, (Context->ZeContext,
&ZeDesc, Size, Alignment, Device->ZeDevice, ResultPtr));
instead
Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
There is a discrepancy between Debug and Release build pass rates in the USM test suite on Level Zero. Make those tests optional until the fix is introduced.
Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
Signed-off-by: Spruit, Neil R <neil.r.spruit@intel.com>
Looks like the SYCL E2E failures have been resolved so I'll merge this. |
[UR][L0] Add several fixes to L0 adapter for test-usm
Corresponding UR chanages oneapi-src/unified-runtime#1105. --------- Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com> Co-authored-by: Kenneth Benzie (Benie) <k.benzie@codeplay.com>
[UR][L0] Add several fixes to L0 adapter for test-usm
Corresponding UR chanages oneapi-src/unified-runtime#1105. --------- Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com> Co-authored-by: Kenneth Benzie (Benie) <k.benzie@codeplay.com>
intel/llvm testing: intel/llvm#11980