Skip to content

topk: add an 8192-page tier and make the overflow check actionable #4

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.

Sign up for GitHub

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

Open
GeoffreyWang1117 wants to merge 1 commit into
base: v1
Choose a base branch
from
Open

topk: add an 8192-page tier and make the overflow check actionable #4

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 19 additions & 1 deletion csrc/topk.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -196,8 +196,26 @@ const int64_t max_num_pages
reserved_bos,
reserved_eos
);
} else if (max_num_pages <= 8192){
TopKOutput_BF16_Kernel<1024, 8><<<nblks, 1024, 0, stream>>>(
reinterpret_cast<__nv_bfloat16*>(x.data_ptr<at::BFloat16>()),
dense_kv_indptr.data_ptr<int>(),
sparse_kv_indptr.data_ptr<int>(),
dense_kv_indices.data_ptr<int>(),
sparse_kv_indices.data_ptr<int>(),
topk_val,
reserved_bos,
reserved_eos
);
} else {
TORCH_CHECK(false);
// The next tier up (<1024, 16>) would need ~66 KB of *static* shared
// memory for the CUB block-sort scratch, above the 48 KB per-block
// limit ptxas enforces, so 8192 is the highest tier this dispatch
// shape can reach without switching to dynamic shared memory.
TORCH_CHECK(false, "topk_output: max_num_pages (", max_num_pages,
") exceeds the maximum supported (8192). Reduce the "
"context length / raise page_size so that "
"ceil(max_seq_len / page_size) <= 8192.");
}

}