Remove CommandQueue redirecting usages straight to HWCQ

tests/tt_metal/tt_metal/api/CMakeLists.txt

@@ -17,7 +17,6 @@ set(UNIT_TESTS_API_SRC
     ${CMAKE_CURRENT_SOURCE_DIR}/test_banked.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/test_bit_utils.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/test_buffer_region.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/test_CommandQueue.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/test_direct.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/test_dram_to_l1_multicast.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/test_dram.cpp

tests/tt_metal/tt_metal/dispatch/dispatch_buffer/test_EnqueueWriteBuffer_and_EnqueueReadBuffer.cpp

@@ -361,52 +361,6 @@ bool test_EnqueueWriteBuffer_and_EnqueueReadBuffer_multi_queue(
 namespace basic_tests {
 namespace dram_tests {
 
-TEST_F(CommandQueueBufferFixture, DISABLED_TestAsyncBufferRW) {
-    // Test Async Enqueue Read and Write + Get Addr + Buffer Allocation and Deallocation
-    auto& command_queue = this->device_->command_queue();
-    auto current_mode = CommandQueue::default_mode();
-    command_queue.set_mode(CommandQueue::CommandQueueMode::ASYNC);
-    Program program;
-    for (int j = 0; j < 10; j++) {
-        // Asynchronously initialize a buffer on device
-        uint32_t first_buf_value = j + 1;
-        uint32_t second_buf_value = j + 2;
-        uint32_t first_buf_size = 4096;
-        uint32_t second_buf_size = 2048;
-        // Asynchronously allocate buffer on device
-        std::shared_ptr<Buffer> buffer =
-            Buffer::create(this->device_, first_buf_size, first_buf_size, BufferType::DRAM);
-        std::shared_ptr<uint32_t> allocated_buffer_address = std::make_shared<uint32_t>();
-        EnqueueGetBufferAddr(this->device_->command_queue(), allocated_buffer_address.get(), buffer.get(), true);
-        // Ensure returned addr is correct
-        EXPECT_EQ((*allocated_buffer_address), buffer->address());
-
-        std::shared_ptr<std::vector<uint32_t>> vec =
-            std::make_shared<std::vector<uint32_t>>(first_buf_size / 4, first_buf_value);
-        std::vector<uint32_t> readback_vec = {};
-        // Write first vector to existing on device buffer.
-        EnqueueWriteBuffer(this->device_->command_queue(), buffer, vec, false);
-        // Reallocate the vector in the main thread after asynchronously pushing it (ensure that worker still has access
-        // to this data)
-        vec = std::make_shared<std::vector<uint32_t>>(second_buf_size / 4, second_buf_value);
-        // Simulate what tt-eager does: Share buffer ownership with program
-        AssignGlobalBufferToProgram(buffer, program);
-        // Reallocate buffer (this is safe, since the program also owns the existing buffer, which will not be
-        // deallocated)
-        buffer = Buffer::create(this->device_, second_buf_size, second_buf_size, BufferType::DRAM);
-        // Write second vector to second buffer
-        EnqueueWriteBuffer(this->device_->command_queue(), buffer, vec, false);
-        // Have main thread give up ownership immediately after writing
-        vec.reset();
-        // Read both buffer and ensure data is correct
-        EnqueueReadBuffer(this->device_->command_queue(), buffer, readback_vec, true);
-        for (int i = 0; i < readback_vec.size(); i++) {
-            EXPECT_EQ(readback_vec[i], second_buf_value);
-        }
-    }
-    command_queue.set_mode(current_mode);
-}
-
 TEST_F(CommandQueueSingleCardBufferFixture, WriteOneTileToDramBank0) {
     TestBufferConfig config = {.num_pages = 1, .page_size = 2048, .buftype = BufferType::DRAM};
     for (IDevice* device : devices_) {

tests/tt_metal/tt_metal/integration/test_flatten.cpp

@@ -299,17 +299,3 @@ TEST_F(DispatchFixture, TensixFlatten) {
         ASSERT_TRUE(test_flatten::flatten(this, this->devices_.at(id), num_tiles_r, num_tiles_c));
     }
 }
-
-TEST_F(CommandQueueProgramFixture, DISABLED_TensixTestAsyncFlattenStress) {
-    auto& command_queue = this->device_->command_queue();
-    auto current_mode = CommandQueue::default_mode();
-    command_queue.set_mode(CommandQueue::CommandQueueMode::ASYNC);
-    uint32_t num_tiles_r = 2;
-    uint32_t num_tiles_c = 2;
-    if (!this->IsSlowDispatch()) {
-        num_tiles_r = 1;
-        num_tiles_c = 1;
-    }
-    ASSERT_TRUE(test_flatten::flatten_stress(this->device_, num_tiles_r, num_tiles_c));
-    command_queue.set_mode(current_mode);
-}

tt_metal/api/tt-metalium/command_queue.hpp

@@ -22,20 +22,18 @@
 #include "worker_config_buffer.hpp"
 #include "program_impl.hpp"
 #include "trace_buffer.hpp"
+#include "hardware_command_queue.hpp"
 
 namespace tt::tt_metal {
 inline namespace v0 {
 
-class CommandQueue;
 class BufferRegion;
 class Event;
 class Trace;
 using RuntimeArgs = std::vector<std::variant<Buffer*, uint32_t>>;
 
 }  // namespace v0
 
-class HWCommandQueue;
-
 // Only contains the types of commands which are enqueued onto the device
 enum class EnqueueCommandType {
     ENQUEUE_READ_BUFFER,
@@ -228,98 +226,6 @@ struct CommandInterface {
     tt::stl::Span<const SubDeviceId> sub_device_ids;
 };
 
-inline namespace v0 {
-
-class CommandQueue {
-    friend class Trace;
-
-public:
-    enum class CommandQueueMode {
-        PASSTHROUGH = 0,
-        ASYNC = 1,
-        TRACE = 2,
-    };
-    enum class CommandQueueState {
-        IDLE = 0,
-        RUNNING = 1,
-        TERMINATE = 2,
-    };
-
-    CommandQueue() = delete;
-
-    CommandQueue(IDevice* device, uint32_t id, CommandQueueMode mode = CommandQueue::default_mode());
-    ~CommandQueue();
-
-    // Trace queue constructor
-    CommandQueue(Trace& trace);
-
-    // Getters for private members
-    IDevice* device() const { return this->device_ptr; }
-    uint32_t id() const { return this->cq_id; }
-
-    // Blocking method to wait for all commands to drain from the queue
-    // Optional if used in passthrough mode (async_mode = false)
-    void wait_until_empty();
-
-    // Schedule a command to be run on the device
-    // Blocking if in passthrough mode. Non-blocking if in async mode
-    void run_command(CommandInterface&& command);
-
-    // API for setting/getting the mode of the command queue
-    // TODO: disallow changing the mode of the queue. This is error prone, because changing mode requires
-    // accordingly updating the higher-level abstractions.
-    void set_mode(const CommandQueueMode& mode);
-    CommandQueueMode get_mode() const { return this->mode; }
-
-    // Reference to the underlying hardware command queue, non-const because side-effects are allowed
-    HWCommandQueue& hw_command_queue();
-
-    // The empty state of the worker queue
-    bool empty() const { return this->worker_queue.empty(); }
-
-    // Dump methods for name and pending commands in the queue
-    void dump();
-    std::string name();
-
-    static CommandQueueMode default_mode() {
-        // Envvar is used for bringup and debug only. Will be removed in the future and should not be relied on in
-        // production.
-        static int value =
-            parse_env<int>("TT_METAL_CQ_ASYNC_MODE", /*default_value=*/static_cast<int>(CommandQueueMode::PASSTHROUGH));
-        return static_cast<CommandQueue::CommandQueueMode>(value);
-    }
-    // Determine if any CQ is using Async mode
-    static bool async_mode_set() { return num_async_cqs > 0; }
-
-   private:
-    // Initialize Command Queue Mode based on the env-var. This will be default, unless the user excplictly sets the
-    // mode using set_mode.
-    CommandQueueMode mode;
-    CommandQueueState worker_state;
-    std::unique_ptr<std::thread> worker_thread;
-    WorkerQueue worker_queue;
-    uint32_t cq_id = 0;
-    IDevice* device_ptr = nullptr;
-    Trace* trace_ptr = nullptr;
-
-    void start_worker();
-    void stop_worker();
-    void run_worker();
-    void run_command_impl(const CommandInterface& command);
-
-    bool async_mode() { return this->mode == CommandQueueMode::ASYNC; }
-    bool trace_mode() { return this->mode == CommandQueueMode::TRACE; }
-    bool passthrough_mode() { return this->mode == CommandQueueMode::PASSTHROUGH; }
-
-    std::atomic<std::size_t> worker_thread_id = -1;
-    std::atomic<std::size_t> parent_thread_id = -1;
-    // Track the number of CQs using async vs pt mode
-    inline static uint32_t num_async_cqs = 0;
-    inline static uint32_t num_passthrough_cqs = 0;
-};
-
-}  // namespace v0
-
 // Primitives used to place host only operations on the SW Command Queue.
 // These are used in functions exposed through tt_metal.hpp or host_api.hpp
 void EnqueueGetBufferAddr(CommandQueue& cq, uint32_t* dst_buf_addr, const Buffer* buffer, bool blocking);
@@ -337,5 +243,4 @@ void EnqueueAddBufferToProgram(
 
 }  // namespace tt::tt_metal
 
-std::ostream& operator<<(std::ostream& os, tt::tt_metal::EnqueueCommandType const& type);
-std::ostream& operator<<(std::ostream& os, tt::tt_metal::CommandQueue::CommandQueueMode const& type);
+std::ostream& operator<<(std::ostream& os, const tt::tt_metal::EnqueueCommandType& type);

tt_metal/api/tt-metalium/device.hpp

@@ -39,13 +39,12 @@ inline namespace v0 {
 
 class Buffer;
 class Program;
-class CommandQueue;
 class SubDevice;
 
 }  // namespace v0
 
 class JitBuildEnv;
-class HWCommandQueue;
+class CommandQueue;
 class TraceBuffer;
 
 inline namespace v0 {
@@ -175,7 +174,6 @@ class IDevice {
     virtual const JitBuildStateSubset build_kernel_states(uint32_t programmable_core, uint32_t processor_class) const = 0;
 
     virtual SystemMemoryManager& sysmem_manager() = 0;
-    virtual HWCommandQueue& hw_command_queue(size_t cq_id = 0) = 0;
     virtual CommandQueue& command_queue(size_t cq_id = 0) = 0;
 
     // Metal trace device capture mode