Expose kernel launch options as offload intrinsic args #150444
Conversation
|
The rustc-dev-guide subtree was changed. If this PR only touches the dev guide consider submitting a PR directly to rust-lang/rustc-dev-guide otherwise thank you for updating the dev guide with your changes. cc @BoxyUwU, @jieyouxu, @Kobzol, @tshepang Some changes occurred to the intrinsics. Make sure the CTFE / Miri interpreter |
rustbot
added
A-LLVM
rustbot
added
S-waiting-on-review
|
@kevinsala it would be nice to have an example in our docs where giving the same dimensions as runtime value (e.g. via command line) vs giving them at compile time (like in the example here) have some measurable perf difference. I assume it's hard to artificially come up with an example exactly for that, or do you know how to get one? |
|
@ZuseZ4 I'll try to find an example. |
Sa4dUs
force-pushed
the
offload-intrinsic2
branch
from
efcf026 to
330d170
Compare
|
As per discussion with offload devs, this should be u32 not i32, otherwise lgtm for now. Once we have a proper macro frontend for our intrinsic we can also consider changing it to |
Sa4dUs
force-pushed
the
offload-intrinsic2
branch
from
330d170 to
6e0d0da
Compare
This comment has been minimized.
This comment has been minimized.
Sa4dUs
force-pushed
the
offload-intrinsic2
branch
from
6e0d0da to
33d39a9
Compare
This comment has been minimized.
This comment has been minimized.
Sa4dUs
force-pushed
the
offload-intrinsic2
branch
from
33d39a9 to
58e2610
Compare
|
This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed. Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers. |
|
Thanks @oli-obk! As per LLVM side discussion, having the kernel launch dimensions at compile time isn't always beneficial. There was a recent case where LLVM actually generated worse code when it was given those dimensions at compile time rather than runtime. Compiler heuristics are fun. So we'll ignore that topic here, and I'll just post a note to our design overview in the dev guide. @bors r+ |
bors
added
S-waiting-on-bors
…uwer Rollup of 6 pull requests Successful merges: - #150425 (mapping an error from cmd.spawn() in npm::install) - #150444 (Expose kernel launch options as offload intrinsic args) - #150495 (Correct hexagon "unwinder_private_data_size") - #150578 (Fix a typo in the docs of AsMut for #149609) - #150581 (mir_build: Separate match lowering for string-equality and scalar-equality) - #150594 (Fix typo in the docs of `CString::from_vec_with_nul`) r? `@ghost` `@rustbot` modify labels: rollup
Rollup merge of #150444 - Sa4dUs:offload-intrinsic2, r=ZuseZ4 Expose kernel launch options as offload intrinsic args Allows modifying the workgroup and thread grid dimensions directly from the intrinsic call. ```rust core::intrinsics::offload(_kernel_1, [256, 1, 1], [32, 1, 1], (x,)) ``` r? `@ZuseZ4`
…uwer Rollup of 6 pull requests Successful merges: - rust-lang/rust#150425 (mapping an error from cmd.spawn() in npm::install) - rust-lang/rust#150444 (Expose kernel launch options as offload intrinsic args) - rust-lang/rust#150495 (Correct hexagon "unwinder_private_data_size") - rust-lang/rust#150578 (Fix a typo in the docs of AsMut for rust-lang/rust#149609) - rust-lang/rust#150581 (mir_build: Separate match lowering for string-equality and scalar-equality) - rust-lang/rust#150594 (Fix typo in the docs of `CString::from_vec_with_nul`) r? `@ghost` `@rustbot` modify labels: rollup
6 participants
Allows modifying the workgroup and thread grid dimensions directly from the intrinsic call.
r? @ZuseZ4