Skip to content

Pure FP8 (W8A8) GEMM support (draft) #306

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
wants to merge 2 commits into
base: sycl-develop
Choose a base branch
from
Open

Pure FP8 (W8A8) GEMM support (draft) #306

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
74dc9d3
a method to avoid data shuffle
jiyang1011 Apr 2, 2025
f9cf3ee
All major combination functionality
jiyang1011 Apr 15, 2025
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
6 changes: 6 additions & 0 deletions examples/sycl/pvc/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,6 +97,12 @@ cutlass_example_add_executable(
TEST_BATCHES
)

cutlass_example_add_executable(
pvc_gemm_fp8
pvc_gemm_fp8.cpp
TEST_COMMAND_OPTIONS
TEST_BATCHES
)
cutlass_example_add_executable(
pvc_gemm_group
pvc_gemm_group.cpp
Expand Down
21 changes: 21 additions & 0 deletions examples/sycl/pvc/common.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,6 +61,27 @@ bool initialize_block(
return true;
}

template<class LowElement, class HiElement>
void intialize_block(cutlass::DeviceAllocation<LowElement>& block_device, cutlass::DeviceAllocation<HiElement>& block_device_ref,
uint64_t seed, int M, int N) {
static_assert(cute::sizeof_bits_v<HiElement> > 8);
std::ranlux24_base rng(std::random_device{}());
rng.seed(seed);

using Limits = cutlass::platform::numeric_limits<LowElement>;
std::uniform_int_distribution<> dist(Limits::lowest(), Limits::max());

auto block_host = std::vector<LowElement>(block_device.size());
auto block_host_ref = std::vector<HiElement>(block_device_ref.size());
for (int i = 0; i < block_host.size(); i++) {
block_host[i] = static_cast<LowElement>(dist(rng));
block_host_ref[i]= static_cast<HiElement>(block_host[i]);
}

block_device.copy_from_host(block_host.data());
block_device_ref.copy_from_host(block_host_ref.data());
}

template <typename T1, typename T2>
void initialize_mixed_dtype_block(cutlass::DeviceAllocation<T1>& block_device,
cutlass::DeviceAllocation<T2>& block_device_dq,
Expand Down
Loading