diff options
| author | Joseph Huber <jhuber6@vols.utk.edu> | 2023-05-01 12:10:04 -0500 |
|---|---|---|
| committer | Joseph Huber <jhuber6@vols.utk.edu> | 2023-05-05 10:12:19 -0500 |
| commit | aea866c12cb428eb5fe062ffa910a63daff62b01 (patch) | |
| tree | 9bf301a8c3c7f66949f2a36399ba2a5580b7b195 /libc/startup | |
| parent | a1be6f0290effb26ff5e0d7a43ab20931f1353a2 (diff) | |
| download | llvm-aea866c12cb428eb5fe062ffa910a63daff62b01.zip llvm-aea866c12cb428eb5fe062ffa910a63daff62b01.tar.gz llvm-aea866c12cb428eb5fe062ffa910a63daff62b01.tar.bz2 | |
[libc] Support concurrent RPC port access on the GPU
Previously we used a single port to implement the RPC. This was
sufficient for single threaded tests but can potentially cause deadlocks
when using multiple threads. The reason for this is that GPUs make no
forward progress guarantees. Therefore one group of threads waiting on
another group of threads can spin forever because there is no guarantee
that the other threads will continue executing. The typical workaround
for this is to allocate enough memory that a sufficiently large number
of work groups can make progress. As long as this number is somewhat
close to the amount of total concurrency we can obtain reliable
execution around a shared resource.
This patch enables using multiple ports by widening the arrays to a
predetermined size and indexes into them. Empty ports are currently
obtained via a trivial linker scan. This should be imporoved in the
future for performance reasons. Portions of D148191 were applied to
achieve parallel support.
Depends on D149581
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D149598
Diffstat (limited to 'libc/startup')
| -rw-r--r-- | libc/startup/gpu/amdgpu/start.cpp | 5 | ||||
| -rw-r--r-- | libc/startup/gpu/nvptx/start.cpp | 5 |
2 files changed, 6 insertions, 4 deletions
diff --git a/libc/startup/gpu/amdgpu/start.cpp b/libc/startup/gpu/amdgpu/start.cpp index d1dfc7b..84adb3b 100644 --- a/libc/startup/gpu/amdgpu/start.cpp +++ b/libc/startup/gpu/amdgpu/start.cpp @@ -15,7 +15,7 @@ extern "C" int main(int argc, char **argv, char **envp); namespace __llvm_libc { -static cpp::Atomic<uint32_t> lock = 0; +static cpp::Atomic<uint32_t> lock[rpc::default_port_count] = {0}; extern "C" uintptr_t __init_array_start[]; extern "C" uintptr_t __init_array_end[]; @@ -43,7 +43,8 @@ extern "C" [[gnu::visibility("protected"), clang::amdgpu_kernel]] void _begin(int argc, char **argv, char **env, void *in, void *out, void *buffer) { // We need to set up the RPC client first in case any of the constructors // require it. - __llvm_libc::rpc::client.reset(__llvm_libc::gpu::get_lane_size(), + __llvm_libc::rpc::client.reset(__llvm_libc::rpc::default_port_count, + __llvm_libc::gpu::get_lane_size(), &__llvm_libc::lock, in, out, buffer); // We want the fini array callbacks to be run after other atexit diff --git a/libc/startup/gpu/nvptx/start.cpp b/libc/startup/gpu/nvptx/start.cpp index 83453ae..1d366dc 100644 --- a/libc/startup/gpu/nvptx/start.cpp +++ b/libc/startup/gpu/nvptx/start.cpp @@ -15,7 +15,7 @@ extern "C" int main(int argc, char **argv, char **envp); namespace __llvm_libc { -static cpp::Atomic<uint32_t> lock = 0; +static cpp::Atomic<uint32_t> lock[rpc::default_port_count] = {0}; extern "C" { // Nvidia's 'nvlink' linker does not provide these symbols. We instead need @@ -47,7 +47,8 @@ extern "C" [[gnu::visibility("protected"), clang::nvptx_kernel]] void _begin(int argc, char **argv, char **env, void *in, void *out, void *buffer) { // We need to set up the RPC client first in case any of the constructors // require it. - __llvm_libc::rpc::client.reset(__llvm_libc::gpu::get_lane_size(), + __llvm_libc::rpc::client.reset(__llvm_libc::rpc::default_port_count, + __llvm_libc::gpu::get_lane_size(), &__llvm_libc::lock, in, out, buffer); // We want the fini array callbacks to be run after other atexit |