Skip to content
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

#16134: add test cases for pre-allocated CreateBuffer / ttnn::event_query #16135

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

#16134: add test cases for pre-allocated CreateBuffer / ttnn::event_query #16135

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
6a091ac
add test case for pre-allocated CreateBuffer
ilkoo-lee Dec 24, 2024
6c891fb
add test case for ttnn::event_query
ilkoo-lee Dec 24, 2024
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
66 changes: 66 additions & 0 deletions tests/tt_metal/tt_metal/api/test_dram.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,6 +122,54 @@ bool dram_single_core(
fixture->ReadBuffer(device, output_dram_buffer, result_vec);
return result_vec == src_vec;
}

bool dram_single_core_pre_allocated(
DispatchFixture* fixture, tt_metal::Device* device, const DRAMConfig& cfg, std::vector<uint32_t> src_vec) {
// Create a program
tt_metal::Program program = CreateProgram();

tt_metal::InterleavedBufferConfig dram_config{
.device = device,
.size = cfg.dram_buffer_size,
.page_size = cfg.dram_buffer_size,
.buffer_type = tt_metal::BufferType::DRAM};

auto input_dram_buffer = tt_metal::CreateBuffer(dram_config);
uint32_t input_dram_buffer_addr = input_dram_buffer->address();
auto input_dram_pre_allocated_buffer = tt_metal::CreateBuffer(dram_config, input_dram_buffer_addr);
uint32_t input_dram_pre_allocated_buffer_addr = input_dram_pre_allocated_buffer->address();

EXPECT_EQ(input_dram_buffer_addr, input_dram_pre_allocated_buffer_addr);

auto output_dram_buffer = tt_metal::CreateBuffer(dram_config);
uint32_t output_dram_buffer_addr = output_dram_buffer->address();
auto output_dram_pre_allocated_buffer = tt_metal::CreateBuffer(dram_config, output_dram_buffer_addr);
uint32_t output_dram_pre_allocated_buffer_addr = output_dram_pre_allocated_buffer->address();

EXPECT_EQ(output_dram_buffer_addr, output_dram_pre_allocated_buffer_addr);

// Create the kernel
auto dram_kernel = tt_metal::CreateKernel(program, cfg.kernel_file, cfg.core_range, cfg.data_movement_cfg);
fixture->WriteBuffer(device, input_dram_pre_allocated_buffer, src_vec);

tt_metal::SetRuntimeArgs(
program,
dram_kernel,
cfg.core_range,
{cfg.l1_buffer_addr,
input_dram_pre_allocated_buffer_addr,
(std::uint32_t)0,
output_dram_pre_allocated_buffer_addr,
(std::uint32_t)0,
cfg.dram_buffer_size});

fixture->RunProgram(device, program);

std::vector<uint32_t> result_vec;
fixture->ReadBuffer(device, output_dram_pre_allocated_buffer, result_vec);

return result_vec == src_vec;
}
} // namespace unit_tests_common::dram::test_dram

TEST_F(DispatchFixture, TensixDRAMLoopbackSingleCore) {
Expand All @@ -142,6 +190,24 @@ TEST_F(DispatchFixture, TensixDRAMLoopbackSingleCore) {
}
}

TEST_F(DispatchFixture, TensixDRAMLoopbackSingleCorePreAllocated) {
uint32_t buffer_size = 2 * 1024 * 25;
std::vector<uint32_t> src_vec =
create_random_vector_of_bfloat16(buffer_size, 100, std::chrono::system_clock::now().time_since_epoch().count());
unit_tests_common::dram::test_dram::DRAMConfig dram_test_config = {
.core_range = {{0, 0}, {0, 0}},
.kernel_file = "tests/tt_metal/tt_metal/test_kernels/dataflow/dram_copy.cpp",
.dram_buffer_size = buffer_size,
.l1_buffer_addr = 400 * 1024,
.data_movement_cfg =
{.processor = tt_metal::DataMovementProcessor::RISCV_0, .noc = tt_metal::NOC::RISCV_0_default},
};
for (unsigned int id = 0; id < devices_.size(); id++) {
ASSERT_TRUE(unit_tests_common::dram::test_dram::dram_single_core_pre_allocated(
this, devices_.at(id), dram_test_config, src_vec));
}
}

TEST_F(DispatchFixture, TensixDRAMLoopbackSingleCoreDB) {
if (!this->IsSlowDispatch()) {
tt::log_info(tt::LogTest, "This test is only supported in slow dispatch mode");
Expand Down
2 changes: 2 additions & 0 deletions tests/ttnn/unit_tests/gtests/test_async_runtime.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -74,11 +74,13 @@ TEST_F(MultiCommandQueueSingleDeviceFixture, TestAsyncPreallocatedOutputs) {
ttnn::record_event(device_->command_queue(io_cq), write_event);
// Host stalls until write is completed, before sending workload
ttnn::event_synchronize(write_event);
EXPECT_EQ(ttnn::event_query(write_event), true);
// Dispatch workload. Preallocated output_tensor is populated by op/
ttnn::moreh_sum(input_tensor, /*dim*/ 3, false, output_tensor, std::nullopt, std::nullopt);
// Record completion of workload
ttnn::record_event(device_->command_queue(workload_dispatch_cq), workload_event);
ttnn::event_synchronize(workload_event);
EXPECT_EQ(ttnn::event_query(workload_event), true);
// Read output back, once workload is complete
ttnn::read_buffer(io_cq, output_tensor, {readback_data});
// Ensure that reference count book keeping is done correctly
Expand Down
Loading