#4269: Supports specifying up to 2 hardware command queues on a singl…

…e device

tests/tt_metal/tt_metal/unit_tests_fast_dispatch/common/command_queue_fixture.hpp

@@ -34,34 +34,6 @@ class CommandQueueFixture : public ::testing::Test {
     }
 };
 
-class MultiCommandQueueFixture : public ::testing::Test {
-   protected:
-    void SetUp() override {
-        auto slow_dispatch = getenv("TT_METAL_SLOW_DISPATCH_MODE");
-        if (slow_dispatch) {
-            TT_THROW("This suite can only be run with fast dispatch or TT_METAL_SLOW_DISPATCH_MODE unset");
-            GTEST_SKIP();
-        }
-        arch_ = tt::get_arch_from_string(tt::test_utils::get_env_arch_name());
-
-        num_devices_ = tt::tt_metal::GetNumAvailableDevices();
-
-        for (unsigned int id = 0; id < num_devices_; id++) {
-            auto* device = tt::tt_metal::CreateDevice(id);
-            devices_.push_back(device);
-        }
-    }
-
-    void TearDown() override {
-        for (unsigned int id = 0; id < devices_.size(); id++) {
-            tt::tt_metal::CloseDevice(devices_.at(id));
-        }
-    }
-
-    std::vector<tt::tt_metal::Device*> devices_;
-    tt::ARCH arch_;
-    size_t num_devices_;
-};
 
 class CommandQueueMultiDeviceFixture : public ::testing::Test {
    protected:
@@ -73,6 +45,8 @@ class CommandQueueMultiDeviceFixture : public ::testing::Test {
         }
         arch_ = tt::get_arch_from_string(tt::test_utils::get_env_arch_name());
 
+        num_devices_ = tt::tt_metal::GetNumAvailableDevices();
+
         for (unsigned int id = 0; id < num_devices_; id++) {
             auto* device = tt::tt_metal::CreateDevice(id);
             devices_.push_back(device);

tt_metal/common/core_descriptor.hpp

@@ -138,6 +138,7 @@ inline const core_descriptor_t &get_core_descriptor_config(chip_id_t device_id,
         }
         producer_cores.push_back(coord);
     }
+    TT_ASSERT(producer_cores.size(), "Producer cores size must be positive");
 
     std::vector<RelativeCoreCoord> consumer_cores;
     for (const auto& core_node : desc_yaml["consumer_cores"]) {
@@ -150,6 +151,7 @@ inline const core_descriptor_t &get_core_descriptor_config(chip_id_t device_id,
         }
         consumer_cores.push_back(coord);
     }
+    TT_ASSERT(consumer_cores.size(), "Consumer cores size must be positive");
 
     config_by_num_cqs[num_hw_cqs] = core_descriptor_t{
         .compute_grid_size = compute_grid_size,

tt_metal/core_descriptors/grayskull_120_arch.yaml

@@ -51,5 +51,24 @@ E75:
     storage_cores: # Relative to grid of tensix cores
       [[11, 1], [11, 2], [11, 3], [11, 5], [11, 6], [11, 7]]
 
-    dispatch_cores: # Relative to grid of tensix cores
-      [[11, 0], [11, 4]]
+    producer_cores:
+      [[11, 0]]
+
+    consumer_cores:
+      [[11, 4]]
+  2:
+    l1_bank_size:
+      1048576
+
+    compute_with_storage_grid_range: # Logical only start and end [x, y]
+      start: [0, 0]
+      end: [10, 7]
+
+    storage_cores: # Relative to grid of tensix cores
+      [[11, 2], [11, 3], [11, 6], [11, 7]]
+
+    producer_cores:
+      [[11, 0], [11, 1]]
+
+    consumer_cores:
+      [[11, 4], [11, 5]]

tt_metal/detail/tt_metal.hpp

@@ -284,7 +284,7 @@ namespace tt::tt_metal{
             static std::mutex cq_creation_mutex;
             {
                 std::lock_guard<std::mutex> lock(cq_creation_mutex);
-                command_queues[device->id()] = std::make_unique<CommandQueue>(device, id);
+                command_queues[device->id()] = std::make_unique<CommandQueue>(device, 0);
             }
             return *(command_queues[id]);
         }

tt_metal/impl/dispatch/command_queue_interface.hpp

@@ -94,6 +94,8 @@ class SystemMemoryManager {
             tt::Cluster::instance().get_fast_pcie_static_tlb_write_callable(device_id)),
         worker_from_logical_callable(worker_from_logical) {
 
+        TT_ASSERT(cq_cores.size(), "cq_cores size must be positive");
+
         uint8_t num_hw_cqs = cq_cores.size();
         this->issue_byte_addrs.resize(num_hw_cqs);
         this->completion_byte_addrs.resize(num_hw_cqs);