[SYCL][ESIMD] Update ESIMD tests and add raw send support. #2482
Conversation
kychendev
force-pushed
the
private/kchen-esimd-patch1
branch
from
597f92e to
ced5805
Compare
|
@kbobrovs @rbegam @DenisBakhvalov Could you please review? |
| @@ -643,6 +643,198 @@ media_block_store(AccessorTy acc, unsigned x, unsigned y, simd<T, m * n> vals) { | |||
| SYCL_EXTERNAL void slm_init(uint32_t size) {} | |||
|
|
|||
| #endif | |||
|
|
|||
| template <typename AccessorTy> | |||
| } | ||
|
|
||
| /// \brief Raw sends load. | ||
| /// |
There was a problem hiding this comment.
Please provide some comments what is "Raw send".
Does it model "send" instruction from https://01.org/sites/default/files/documentation/intel-gfx-prm-osrc-icllp-vol02a-commandreference-instructions_2.pdf? Maybe it makes sense to add a link then?
There was a problem hiding this comment.
Yes, this API is used to implement the send messages. Added link to the open source document.
| @@ -1838,7 +1838,7 @@ template <typename T0, typename T1, int SZ> struct esimd_apply_reduced_min { | |||
|
|
|||
| template <typename T0, typename T1, int SZ, | |||
| template <typename RT, typename T, int N> class OpType> | |||
| T1 esimd_reduce_single(simd<T1, SZ> v) { | |||
| T0 esimd_reduce_single(simd<T1, SZ> v) { | |||
There was a problem hiding this comment.
This change does not seem to be related to raw_send.
There was a problem hiding this comment.
Right, as you can see from the change list, this patch includes other local changes including ESIMD tests update and a reduction test fix.
There was a problem hiding this comment.
NIT: it's good to have proper and meaningful names for template types instead of indexed ones.
| /// | ||
| /// @execSize the execution size, which must be a compile time constant. | ||
| /// | ||
| /// @sfid the shared function ID, which must be a compile time constant. |
There was a problem hiding this comment.
the API would be hard to use w/o some more info on what exDesc, msgDesc and sfid can be.
There was a problem hiding this comment.
These correspond to the different send message encoding fields and are described in the open source reference document (added link in the comment). This API is intended for expert programmers who will need to read the reference manual in order to program the desired send message function.
| /// a compile time constant (optional - default to 0). | ||
| /// | ||
| /// @mask the predicate to specify enabled channels (optional - default to on). | ||
| /// |
There was a problem hiding this comment.
Please add info what this operation returns
| /// \brief Raw sends load. | ||
| /// | ||
| /// @msgDst the destination operand of send message. | ||
| /// |
There was a problem hiding this comment.
it is not clear how this parameter is to be used - it is passed by value, yet it is a destination.
There was a problem hiding this comment.
It's the old value of the destination operand - updated comment to clarify this.
kychendev
force-pushed
the
private/kchen-esimd-patch1
branch
2 times, most recently
from
9d01503 to
2e742b0
Compare
kychendev
force-pushed
the
private/kchen-esimd-patch1
branch
from
2e742b0 to
64968c1
Compare
|
@kychendev, I took the liberty to commit few minor comments fixes to approve and speed up the process. Feel free to roll back if needed. |
|
@kbobrovs, LGTM. However I see 4 checks failing now, which were passing before. Could you help double check if it's test env issue, or what needs to be changed? |
|
@kychendev, sorry for typo - restarted the checks. |
|
@intel/llvm-reviewers-runtime , ping. Please review/approve |
…_wrapper * upstream/sycl: (1021 commits) [SYCL] Enable async_work_group_copy for scalar and vector bool types (intel#2582) [SYCL] Fix element type in handler::copy (intel#2590) [NFC][SYCL] Remove unnecessary if condition (intel#2585) [SYCL][NFC] Fix SYCL lit test execution on a system w/o GPU (intel#2584) [SYCL] Add error handling for non-uniform work group size case (intel#2569) [SYCL][ESIMD] Preserve undef initializer for globals in ESIMDLowerVecArg pass (intel#2555) [SYCL] Make Level-Zero events visible on the host (intel#2576) [Driver][SYCL][NFC] Add help information for -Wno-sycl-strict (intel#2570) [SYCL] Relax test to work in Win32 environment. (intel#2580) [SYCL] Emit suppressed warnings from SYCL headers (intel#2575) [SYCL][NFC] Cover more classes with ABI tests (intel#2577) [SYCL][ESIMD] Update ESIMD tests and add raw send support. (intel#2482) [SYCL] Make ESIMD on-device tests require linux,gpu,opencl. (intel#2560) [SYCL] Release commands with no dependencies after they're enqueued (intel#2492) [SYCL] Add multi-device and multi-platform support for SYCL_DEVICE_ALLOWLIST (intel#2483) [SYCL] Try to enqueue host command depencies (intel#2561) [SYCL][ESIMD][NFC] Align namespace name with the spec guidelines (intel#2573) [SYCL][NFC] Add class layout ABI tests for memory objects (intel#2559) [SYCL] Change adress space for global variables (intel#2534) [NFC][SYCL] Fix comment. (intel#2541) ...
…2482) This addresses p.2 and p.3 of #2460 Original commit: KhronosGroup/SPIRV-LLVM-Translator@4b4508438a6dfe5
…2482) This addresses p.2 and p.3 of #2460 Original commit: KhronosGroup/SPIRV-LLVM-Translator@4b4508438a6dfe5
Get rid of extra-confounding test helper code
Get rid of extra-confounding test helper code
Update ESIMD tests and add raw send support.