ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-29 19:28:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

changes to profiler for streamk

Browse Source
This commit is contained in:
Astha Rai
2025-06-17 23:41:56 +00:00
parent 6e4ddf54ef
commit fcf50b211d
2 changed files with 103 additions and 147 deletions
Download Patch File Download Diff File Expand all files Collapse all files

222
profiler/include/profiler/profile_gemm_universal_streamk_impl.hpp Executable file → Normal file
View File

@@ -44,11 +44,10 @@ bool profile_gemm_universal_streamk_impl(int do_verification,
int StrideA,
int StrideB,
int StrideC,
int Streamk_sel,
int Grid_size,
int n_warmup,
int n_iter,
uint64_t rotating = 0)
uint64_t rotating = 0,
uint32_t NumSKBlocks = 0xffffffff)
{
bool pass = true;
@@ -152,144 +151,105 @@ bool profile_gemm_universal_streamk_impl(int do_verification,
}
std::string best_op_name;
float best_ave_time = 0;
float best_tflops = 0;
float best_gb_per_sec = 0;
float best_grid_size = 0;
float best_streamk_sel = 0;
float best_ave_time = 0;
float best_tflops = 0;
float best_gb_per_sec = 0;
// profile device GEMM instances
for(auto& op_ptr : op_ptrs)
{
std::vector<int> grid_size_list = {38, 76, 114, 152, 190, 228, 266, 304, 342, 380};
std::vector<int> streamk_sel_list = {
0, 1, 2, 3, 4}; // 0: Data Parallel (DP) mode (Stream-K OFF), 1: 1-tile Stream-K+ DP,
// 2:2-tile Stream-K + DP
auto argument_ptr =
op_ptr->MakeArgumentPointer(static_cast<ADataType*>(a_device_buf.GetDeviceBuffer()),
static_cast<BDataType*>(b_device_buf.GetDeviceBuffer()),
static_cast<CDataType*>(c_device_buf.GetDeviceBuffer()),
M,
N,
K,
StrideA,
StrideB,
StrideC,
a_element_op,
b_element_op,
c_element_op,
NumSKBlocks); // NumSKBlocks parameter
if(Grid_size == -1)
auto invoker_ptr = op_ptr->MakeInvokerPointer();
if(op_ptr->IsSupportedArgument(argument_ptr.get()))
{
grid_size_list = {Grid_size};
}
if(Streamk_sel != -1)
{
streamk_sel_list = {Streamk_sel};
}
for(std::size_t j = 0; j < streamk_sel_list.size(); j++)
{
for(std::size_t i = 0; i < grid_size_list.size(); i++)
// re-init C to zero before profiling next kernel
c_device_buf.SetZero();
invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, false, 0, n_warmup, n_iter});
if(do_verification)
{
auto grid_size_curr = grid_size_list[i];
index_t streamk_sel_curr = streamk_sel_list[j];
printf("streamk_sel_curr=%0d\n", streamk_sel_curr);
auto argument_ptr = op_ptr->MakeArgumentPointer(
static_cast<ADataType*>(a_device_buf.GetDeviceBuffer()),
static_cast<BDataType*>(b_device_buf.GetDeviceBuffer()),
static_cast<CDataType*>(c_device_buf.GetDeviceBuffer()),
M,
N,
K,
StrideA,
StrideB,
StrideC,
streamk_sel_curr,
grid_size_curr,
a_element_op,
b_element_op,
c_element_op);
c_device_buf.FromDevice(c_m_n_device_result.mData.data());
auto invoker_ptr = op_ptr->MakeInvokerPointer();
// Always compare against CPU reference results computed earlier
pass = pass & ck::utils::check_err(c_m_n_device_result, c_m_n_host_result);
if(op_ptr->IsSupportedArgument(argument_ptr.get()))
if(do_log)
{
// re-init C to zero before profiling next kernel
c_device_buf.SetZero();
invoker_ptr->Run(argument_ptr.get(),
StreamConfig{nullptr, false, 0, n_warmup, n_iter});
if(do_verification)
{
c_device_buf.FromDevice(c_m_n_device_result.mData.data());
// Always compare against CPU reference results computed earlier
pass = pass & ck::utils::check_err(c_m_n_device_result, c_m_n_host_result);
if(do_log)
{
LogRangeAsType<float>(std::cout << "a : ", a_m_k.mData, ",")
<< std::endl;
LogRangeAsType<float>(std::cout << "b: ", b_k_n.mData, ",")
<< std::endl;
LogRangeAsType<float>(
std::cout << "c_host : ", c_m_n_host_result.mData, ",")
<< std::endl;
LogRangeAsType<float>(
std::cout << "c_device: ", c_m_n_device_result.mData, ",")
<< std::endl;
}
}
std::string op_name = op_ptr->GetTypeString();
float ave_time = invoker_ptr->Run(argument_ptr.get(),
StreamConfig{nullptr,
time_kernel,
0,
n_warmup,
n_iter,
rotating_count > 1,
rotating_count});
std::size_t flop = std::size_t(2) * M * N * K;
std::size_t num_btype = sizeof(ADataType) * M * K + sizeof(BDataType) * K * N +
sizeof(CDataType) * M * N;
float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
float gb_per_sec = num_btype / 1.E6 / ave_time;
std::cout << "Perf: " << std::setw(10) << ave_time << " ms, " << tflops
<< " TFlops, " << gb_per_sec << " GB/s, " << op_name << ", Grid_size "
<< grid_size_curr << ", streamk selection strategy"
<< streamk_sel_curr << std::endl;
#if defined CK_ENABLE_FP8
// set softer tolerances for fp8
if constexpr(is_same_v<ADataType, f8_t> || is_same_v<BDataType, f8_t> ||
is_same_v<CDataType, f8_t>)
{
std::string msg = "Error: Incorrect results!";
double rtol = 1e-1;
double atol = 1e-1;
pass = pass & ck::utils::check_err(
c_m_n_device_result, c_m_n_host_result, msg, rtol, atol);
}
else
{
#endif
pass = pass & ck::utils::check_err(c_m_n_device_result, c_m_n_host_result);
#if defined CK_ENABLE_FP8
}
#endif
if(tflops > best_tflops)
{
best_op_name = op_name;
best_tflops = tflops;
best_ave_time = ave_time;
best_gb_per_sec = gb_per_sec;
best_grid_size = grid_size_curr;
best_streamk_sel = streamk_sel_curr;
}
}
else
{
std::cout << op_ptr->GetTypeString() << " does not support this problem"
<< std::endl;
LogRangeAsType<float>(std::cout << "a : ", a_m_k.mData, ",") << std::endl;
LogRangeAsType<float>(std::cout << "b: ", b_k_n.mData, ",") << std::endl;
LogRangeAsType<float>(std::cout << "c_host : ", c_m_n_host_result.mData, ",")
<< std::endl;
LogRangeAsType<float>(std::cout << "c_device: ", c_m_n_device_result.mData, ",")
<< std::endl;
}
}
std::string op_name = op_ptr->GetTypeString();
float ave_time = invoker_ptr->Run(
argument_ptr.get(),
StreamConfig{
nullptr, time_kernel, 0, n_warmup, n_iter, rotating_count > 1, rotating_count});
std::size_t flop = std::size_t(2) * M * N * K;
std::size_t num_btype =
sizeof(ADataType) * M * K + sizeof(BDataType) * K * N + sizeof(CDataType) * M * N;
float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
float gb_per_sec = num_btype / 1.E6 / ave_time;
std::cout << "Perf: " << std::setw(10) << ave_time << " ms, " << tflops << " TFlops, "
<< gb_per_sec << " GB/s, " << op_name << std::endl;
#if defined CK_ENABLE_FP8
// set softer tolerances for fp8
if constexpr(is_same_v<ADataType, f8_t> || is_same_v<BDataType, f8_t> ||
is_same_v<CDataType, f8_t>)
{
std::string msg = "Error: Incorrect results!";
double rtol = 1e-1;
double atol = 1e-1;
pass = pass & ck::utils::check_err(
c_m_n_device_result, c_m_n_host_result, msg, rtol, atol);
}
else
{
#endif
pass = pass & ck::utils::check_err(c_m_n_device_result, c_m_n_host_result);
#if defined CK_ENABLE_FP8
}
#endif
if(tflops > best_tflops)
{
best_op_name = op_name;
best_tflops = tflops;
best_ave_time = ave_time;
best_gb_per_sec = gb_per_sec;
}
}
else
{
std::cout << op_ptr->GetTypeString() << " does not support this problem" << std::endl;
}
}
@@ -329,9 +289,7 @@ bool profile_gemm_universal_streamk_impl(int do_verification,
}
std::cout << " M = " << M << " N = " << N << " K = " << K << " StrideA = " << StrideA
<< " StrideB = " << StrideB << " StrideC = " << StrideC
<< " Grid_size = " << best_grid_size
<< " Stream-K selection strategy = " << best_streamk_sel << " : " << best_ave_time
<< " StrideB = " << StrideB << " StrideC = " << StrideC << "Time : " << best_ave_time
<< " ms, " << best_tflops << " TFlops, " << best_gb_per_sec << " GB/s, "
<< best_op_name << std::endl;

28
profiler/src/profile_gemm_universal_streamk.cpp Executable file → Normal file
View File

@@ -34,7 +34,7 @@ enum struct GemmDataType
int profile_gemm_universal_streamk(int argc, char* argv[])
{
if(argc != 16 && argc != 19)
if(argc != 14 && argc != 18)
{
printf("arg1: tensor operation (" OP_NAME ": " OP_DESC ")\n");
printf("arg2: data type (0: fp32; 1: fp16; 2: bf16; 3: int8; 4: f8@f16; 5: f16@f8; 6: f16, "
@@ -48,12 +48,11 @@ int profile_gemm_universal_streamk(int argc, char* argv[])
printf("arg6: print tensor value (0: no; 1: yes)\n");
printf("arg7: time kernel (0=no, 1=yes)\n");
printf("arg8 to 13: M, N, K, StrideA, StrideB, StrideC\n");
printf("arg14: Stream-k select strategy 0: all DP, 1: 1-tile SK, 2: 2-tile SK\n");
printf("arg15: Grid-size, -1 for max persistent kernel occupancy\n");
printf("optional:\n");
printf("arg16: number of warm-up cycles (default 1)\n");
printf("arg17: number of iterations (default 10)\n");
printf("arg18: memory for rotating buffer (default 0, size in MB)\n");
printf("arg14: number of warm-up cycles (default 1)\n");
printf("arg15: number of iterations (default 10)\n");
printf("arg16: memory for rotating buffer (default 0, size in MB)\n");
printf("arg17: NumSKBlocks\n");
exit(1);
}
@@ -86,18 +85,18 @@ int profile_gemm_universal_streamk(int argc, char* argv[])
const int K = std::stoi(argv[10]);
const int StrideC = std::stoi(argv[13]);
const int Streamk_sel = std::stoi(argv[14]);
const int Grid_size = std::stoi(argv[15]);
const int StrideC = std::stoi(argv[13]);
const uint32_t NumSKBlocks =
argc >= 18 ? static_cast<uint32_t>(std::stoul(std::string(argv[17]))) : 0xffffffff;
int n_warmup = 20;
int n_iter = 50;
uint64_t rotating = 0;
if(argc == 19)
if(argc == 17)
{
n_warmup = std::stoi(argv[16]);
n_iter = std::stoi(argv[17]);
rotating = std::stoull(argv[18]) * 1024 * 1024;
n_warmup = std::stoi(argv[14]);
n_iter = std::stoi(argv[15]);
rotating = std::stoull(argv[16]) * 1024 * 1024;
}
using F32 = float;
@@ -151,8 +150,7 @@ int profile_gemm_universal_streamk(int argc, char* argv[])
(StrideA < 0) ? DefaultStrideA : StrideA,
(StrideB < 0) ? DefaultStrideB : StrideB,
(StrideC < 0) ? DefaultStrideC : StrideC,
Streamk_sel,
Grid_size,
NumSKBlocks,
n_warmup,
n_iter,
rotating);