vrr/kvcache-ai-ktransformers
2
0
Fork 0
mirror of https://github.com/kvcache-ai/ktransformers.git synced 2026-04-26 10:50:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

Revert "kt-kernel: enable CPUInfer stream bridge for ROCm (#1918)" (#1925)
Some checks failed
Book-CI / test (push) Has been cancelled
Details
Book-CI / test-1 (push) Has been cancelled
Details
Book-CI / test-2 (push) Has been cancelled
Details
Deploy / deploy (macos-latest) (push) Has been cancelled
Details
Deploy / deploy (ubuntu-latest) (push) Has been cancelled
Details
Deploy / deploy (windows-latest) (push) Has been cancelled
Details

Browse source 
This reverts commit 1dd0a78899.
This commit is contained in:
ErvinXie 2026-04-09 18:43:03 +08:00 • committed by GitHub
parent 1dd0a78899
commit 279c920a69
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 3 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

11
kt-kernel/cpu_backend/cpuinfer.h
View file

@ -83,13 +83,11 @@ class CPUInfer {
}
#ifndef KTRANSFORMERS_CPU_ONLY
void submit_with_cuda_stream(intptr_t user_cuda_stream, std::pair<intptr_t, intptr_t> params) {
#if defined(KTRANSFORMERS_USE_CUDA) || defined(KTRANSFORMERS_USE_ROCM)
#if defined(KTRANSFORMERS_USE_CUDA)
void (*func)(void*) = (void (*)(void*))params.first;
void* args = (void*)params.second;
*((CPUInfer**)args) = this;
cudaLaunchHostFunc((cudaStream_t)user_cuda_stream, (cudaHostFn_t)func, args);
#else
submit(params);
#endif
}
#endif
@ -102,7 +100,6 @@ class CPUInfer {
static void sync_(void* sync_args) {
SyncArgs* args = (SyncArgs*)sync_args;
args->cpuinfer->task_queue_->sync(args->allow_n_pending);
delete args;
}
void sync(size_t allow_n_pending = 0) {
@ -111,11 +108,9 @@ class CPUInfer {
}
#ifndef KTRANSFORMERS_CPU_ONLY
void sync_with_cuda_stream(intptr_t user_cuda_stream, size_t allow_n_pending = 0) {
#if defined(KTRANSFORMERS_USE_CUDA) || defined(KTRANSFORMERS_USE_ROCM)
#if defined(KTRANSFORMERS_USE_CUDA)
SyncArgs* args = new SyncArgs{this, allow_n_pending};
cudaLaunchHostFunc((cudaStream_t)user_cuda_stream, (cudaHostFn_t)&sync_, (void*)args);
#else
sync(allow_n_pending);
#endif
}
#endif
@ -124,4 +119,4 @@ class CPUInfer {
TaskQueue* task_queue_;
};
#endif
#endif