SYCL. Unify calculations for objectives reg:absoluteerror, reg:quantileerror, binary:hinge #10993
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
trivialfis
reviewed
Nov 14, 2024
Comment on lines
+48
to
+57
| template<typename Fn, typename TupleType, size_t ... I> | ||
| auto call(Fn&& fn, TupleType t, std::index_sequence<I ...>) { | ||
| return fn(std::get<I>(t) ...); | ||
| } | ||
|
|
||
| template<typename Fn, typename TupleType> | ||
| auto call(Fn&& fn, TupleType t) { | ||
| static constexpr auto size = std::tuple_size<TupleType>::value; | ||
| return call(fn, t, std::make_index_sequence<size>{}); | ||
| } |
There was a problem hiding this comment.
It is, but std::apply can not be executed in sycl-kernel.
src/objective/objective.cc
Outdated
| @@ -48,7 +48,7 @@ std::string ObjFunction::GetSyclImplementationName(const std::string& name) { | |||
| return name + sycl_postfix; | |||
| } else { | |||
| // Function hasn't specific sycl implementation | |||
| LOG(FATAL) << "`" << name << "` doesn't have sycl implementation yet\n"; | |||
| // LOG(FATAL) << "`" << name << "` doesn't have sycl implementation yet\n"; | |||
There was a problem hiding this comment.
Please remove the commented code.
src/common/linalg_op.h
Outdated
| @@ -49,6 +49,7 @@ void ElementWiseKernelHost(linalg::TensorView<T, D> t, std::int32_t n_threads, F | |||
| } | |||
|
|
|||
| #if !defined(XGBOOST_USE_CUDA) | |||
| #if !defined(XGBOOST_USE_SYCL) | |||
| @@ -45,7 +45,7 @@ namespace { | |||
| template <typename Fn> | |||
| PackedReduceResult Reduce(Context const* ctx, MetaInfo const& info, Fn&& loss) { | |||
| PackedReduceResult result; | |||
| auto labels = info.labels.View(ctx->Device()); | |||
| auto labels = info.labels.View(ctx->Device().IsSycl() ? DeviceOrd::CPU() : ctx->Device()); | |||
There was a problem hiding this comment.
I assume you want to use IsCPU instead of IsSycl?
There was a problem hiding this comment.
No. This function doesn't have sycl-specific implementation yet. So for the proper execution on cpu the data should be transferred back to host.
I plan to add sycl-implementation for it in the future. Didn't do it by single PR just to avoid unreadable PR size.
trivialfis
approved these changes
Nov 19, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi, I work on unification of sycl-plugin with the main part of the framework.
This PR unify calculations for some objective functions:
reg:absoluteerror,reg:quantileerrorandbinary:hingeThe procedure of unification occurs in the following way:
linalg::ElementWiseKernelthe sycl realisation is used.