[SYCL][Doc] Release notes for Mar'25 release #18469
Conversation
There was a problem hiding this comment.
Some general comments: a new structure is used this time. We have less top-level categories, but more sub-categories which group items based on some feature/activity/area.
Motivation for that is because "Improvements & Bug Fixes -> Feature Name" is expected to be more interesting and useful for an end user, than a one huge list of Bug Fixes for everything.
We still have some kind of fallback section "all other bugfixes", but at least it is not that huge.
Any feedback on the new structure is appreciated
|
@intel/dpcpp-doc-reviewers, friendly ping. We need a review as soon as possible, thanks |
sycl/ReleaseNotes.md
Outdated
| @@ -349,8 +349,6 @@ SYCL runtime over low-level runtimes (such as Level Zero or OpenCL): | |||
| - Removed a busy-wait loop from the implementation of | |||
| `-fsycl-max-parallel-link-jobs` flag, making it consume less resources when | |||
| waiting. intel/llvm#17260 | |||
| - Made `-O0` to be the default optimization level when debug info is enabled | |||
| through `-g` flag. intel/llvm#16408 | |||
There was a problem hiding this comment.
@dm-vodopyanov, title of this PR states March. This change was reverted in April - will the revert be part of this release?
There was a problem hiding this comment.
@mdtoguchi yes, the revert will be present in Mar'25 release, as we currently align open-source releases with closed-source oneAPI releases. Upcoming oneAPI release have this revert, so upcoming open-source release will have cherry-pick of this revert.
Tagging @KornevNikita who is working on open-source releases, for awareness.
sycl/ReleaseNotes.md
Outdated
| - ada16682e8c3 [DevASAN] Only report warning if passing host ptr to kernel (#16654) | ||
| - seems like a potentially important bugfix | ||
| - ce4a320806b2 [DeivceASAN] Make ShadowMemory one instance per type (#16687) | ||
| - seems like some kind of bugfix | ||
| - ef4d66af3b74 [DeviceSAN] Fix kernel name addressspace (#16425) | ||
| - 1fba00d3be7d [DeviceASAN] Fix ASAN with kernel assert (#16256) | ||
| - 34aeabab551e [SYCL][DeviceASAN] Fix AcceeChain to a matrix for bfloat16 (#16323) | ||
| - 6f3b0e857d15 [DevASAN] Do allocation with USM pool to reduce memory overhead (#16280) | ||
| - a8c6e7715be2 [DeviceASAN] Re-use shadow if required size is not larger than last one (#16258) | ||
| - As the above, some optimization of memory usage by ASAN? | ||
| - 201725664cc5 [DeviceSanitizer] Fix device global type of KernelMetadata (#16357) |
There was a problem hiding this comment.
change them to the following:
- Only report warning if passing host ptr to kernel [DevASAN] Only report warning if passing host ptr to kernel #16654
- Make ShadowMemory one instance per type [DeivceASAN] Make ShadowMemory one instance per type #16687
- Fix kernel name addressspace [DeviceSAN] Fix kernel name addressspace #16425
- Fix ASAN with kernel assert [DeviceASAN] Fix ASAN with kernel assert #16256
- Fix AcceeChain to a matrix for bfloat16 intel/llvm/16323
- Reduce the frequency of shadow memory reallocation to reduce memory overhead and improve runtime performance [DevASAN] Do allocation with USM pool to reduce memory overhead #16280,[DeviceASAN] Re-use shadow if required size is not larger than last one #16258
- Fix device global type of KernelMetadata [DeviceSanitizer] Fix device global type of KernelMetadata #16357
There was a problem hiding this comment.
@cdai2 - can this be more descriptive and useful for customers? Using PR titles "as-is" is not enough.
BTW, if you see that something is not user-visible or not worth mentioning for customers, that can be also removed.
As an example, "Fix kernel name addressspace #16425" is not descriptive. I would say it should be something like this: "Fixed an error in SPIR-V module caused by incorrect casting from private/local/global address space to global address space. Such casting is not allowed. #16425"
There was a problem hiding this comment.
BTW, if you see that something is not user-visible or not worth mentioning for customers, that can be also removed.
Applied suggestion https://github.com/intel/llvm/pull/18469/files#r2099140914 from @yingcong-wu in 1079097
@intel/dpcpp-sanitizers-review @yingcong-wu @cdai2, please, review one more time
| - [`sycl_ext_oneapi_kernel_compiler`](https://github.com/intel/llvm/blob/b23d69e2c3fda1d69351137991897c96bf6a586d/sycl/doc/extensions/experimental/sycl_ext_oneapi_kernel_compiler.asciidoc) | ||
| extension specification was updated to accept `sycl` as source language, thus | ||
| providing functionality similar to | ||
| [NVRTC](https://docs.nvidia.com/cuda/nvrtc/). intel/llvm#11985, | ||
| intel/llvm#17446 | ||
| - Initial support for this feature was implemented. intel/llvm#16132, | ||
| intel/llvm#16222, intel/llvm#16132, intel/llvm#17640, intel/llvm#17356, | ||
| intel/llvm#16565, intel/llvm#17383, intel/llvm#17447, intel/llvm#17307, | ||
| intel/llvm#17373, intel/llvm#17331, intel/llvm#17329, intel/llvm#17266, | ||
| intel/llvm#17032, intel/llvm#16823, intel/llvm#16702, intel/llvm#16638, | ||
| intel/llvm#16316, intel/llvm#17359, intel/llvm#16485, intel/llvm#16821 | ||
| - Known issues and limitations are documented | ||
| [in the extension specification](https://github.com/intel/llvm/blob/b23d69e2c3fda1d69351137991897c96bf6a586d/sycl/doc/extensions/experimental/sycl_ext_oneapi_kernel_compiler.asciidoc#known-issues-and-limitations-when-the-language-is-sycl). intel/llvm#17307, | ||
| intel/llvm#17459 |
There was a problem hiding this comment.
| - [`sycl_ext_oneapi_kernel_compiler`](https://github.com/intel/llvm/blob/b23d69e2c3fda1d69351137991897c96bf6a586d/sycl/doc/extensions/experimental/sycl_ext_oneapi_kernel_compiler.asciidoc) | |
| extension specification was updated to accept `sycl` as source language, thus | |
| providing functionality similar to | |
| [NVRTC](https://docs.nvidia.com/cuda/nvrtc/). intel/llvm#11985, | |
| intel/llvm#17446 | |
| - Initial support for this feature was implemented. intel/llvm#16132, | |
| intel/llvm#16222, intel/llvm#16132, intel/llvm#17640, intel/llvm#17356, | |
| intel/llvm#16565, intel/llvm#17383, intel/llvm#17447, intel/llvm#17307, | |
| intel/llvm#17373, intel/llvm#17331, intel/llvm#17329, intel/llvm#17266, | |
| intel/llvm#17032, intel/llvm#16823, intel/llvm#16702, intel/llvm#16638, | |
| intel/llvm#16316, intel/llvm#17359, intel/llvm#16485, intel/llvm#16821 | |
| - Known issues and limitations are documented | |
| [in the extension specification](https://github.com/intel/llvm/blob/b23d69e2c3fda1d69351137991897c96bf6a586d/sycl/doc/extensions/experimental/sycl_ext_oneapi_kernel_compiler.asciidoc#known-issues-and-limitations-when-the-language-is-sycl). intel/llvm#17307, | |
| intel/llvm#17459 | |
| DPCPP gained support for runtime compilation of SYCL source code via the | |
| [`sycl_ext_oneapi_kernel_compiler`](https://github.com/intel/llvm/blob/b23d69e2c3fda1d69351137991897c96bf6a586d/sycl/doc/extensions/experimental/sycl_ext_oneapi_kernel_compiler.asciidoc) | |
| extension. SYCL applications and libraries can now define new kernels from a | |
| source string as a powerful new means for runtime specialization. This feature | |
| is built on top of a new in-memory compilation pipeline implemented in the | |
| SYCL-JIT library. | |
| [Known issues and limitations](https://github.com/intel/llvm/blob/b23d69e2c3fda1d69351137991897c96bf6a586d/sycl/doc/extensions/experimental/sycl_ext_oneapi_kernel_compiler.asciidoc#known-issues-and-limitations-when-the-language-is-sycl) | |
| are documented in the extension specification. See also the companion technical | |
| presentation, "Fast In-Memory Runtime Compilation of SYCL Code", at IWOCL 2025 | |
| ([slides](https://www.iwocl.org/wp-content/uploads/iwocl-2025-julian-oppermann-fast-in-memory-runtime.pdf), | |
| [recording](https://youtu.be/X9mS8xetZJY)). | |
| **Notable changes:** | |
| - Updated `sycl_ext_oneapi_kernel_compiler` extension specification to accept | |
| `sycl` as source language. intel/llvm#11985 | |
| - Aligned handling of normal and runtime-compiled kernels. intel/llvm#16316 | |
| - Allowed kernels to be queried by their (unmangled) source code name. | |
| intel/llvm#16485 intel/llvm#16821 intel/llvm#17032 | |
| - Implemented persistent cache for the frontend invocation. intel/llvm#16823 | |
| - Clarified handling of include search paths. intel/llvm#17307 | |
| - Made the JIT-based, in-memory pipeline the default compiler for the `sycl` | |
| source language. intel/llvm#17383 | |
| **Supplementary PRs** | |
| intel/llvm#16132, intel/llvm#16222, intel/llvm#16243, intel/llvm#16565, | |
| intel/llvm#16638, intel/llvm#16685, intel/llvm#16702, intel/llvm#17131, | |
| intel/llvm#17182, intel/llvm#17208, intel/llvm#17266, intel/llvm#17329, | |
| intel/llvm#17331, intel/llvm#17356, intel/llvm#17373, intel/llvm#17405, | |
| intel/llvm#17447, intel/llvm#17459, intel/llvm#17460 |
There was a problem hiding this comment.
Thanks Julian! I've already applied your feedback in accordance to overall style and level of details of release notes
|
@intel/dpcpp-doc-reviewers @steffenlarsen @tfzhu - tagging all code owners. We are currently waiting for a review from @intel/dpcpp-sanitizers-review, meanwhile, could you please review the entire patch to speed-up merging? |
|
@intel/dpcpp-doc-reviewers @steffenlarsen @tfzhu - this is ready for review |
No description provided.