TheRock/rocm-libraries/projects/hipblaslt/tensilelite/src/hip/HipHardware.cpp

/*******************************************************************************
 *
 * MIT License
 *
 * Copyright (C) 2022-2026 Advanced Micro Devices, Inc. All rights reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 *
 *******************************************************************************/

#include <Tensile/AMDGPU.hpp>
#include <Tensile/AMDGPUPredicates.hpp>
#include <Tensile/hip/HipHardware.hpp>
#include <Tensile/hip/HipUtils.hpp>

#include <iostream>
#include <stdexcept>

namespace TensileLite
{
    namespace hip
    {
        HipAMDGPU::HipAMDGPU(hipDeviceProp_t const& prop,
                             int                    deviceId,
                             std::optional<int>     pciChipId)
            : AMDGPU(AMDGPU::toProcessor(prop.gcnArchName),
                     prop.multiProcessorCount,
                     std::string(prop.name),
                     pciChipId)
            , properties(prop)
            , deviceId(deviceId)
        {
            if(origami::hardware_t::is_hardware_supported(prop))
            {
                // Route analyticalHardware construction through Origami's device-id
                // entry point so selection uses the runtime-queried XCC count
                // (hipDeviceAttributeNumberOfXccs on HIP 7+) instead of the hardcoded
                // per-architecture default in get_default_num_xcds().
                //
                // Pass the caller-provided `prop` so any adjustments made upstream
                // (e.g. overriding multiProcessorCount with
                // hipDeviceAttributePhysicalMultiProcessorCount on multi-XCC
                // architectures) are preserved. The int-only overload re-queries
                // `hipGetDeviceProperties` and would discard those adjustments.
                //
                // Pass PCI chip ID when known (e.g. from hipDeviceAttributePciChipId upstream);
                // Origami does not query PCI chip ID itself—nullopt selects gfx950 id75a0.
                analyticalHardware = std::make_shared<origami::hardware_t>(
                    origami::hardware_t::get_hardware_for_device(deviceId, prop, pciChipId));
            }
        }

        std::string HipAMDGPU::archName() const
        {
            return properties.gcnArchName;
        }

        std::shared_ptr<Hardware> GetCurrentDevice()
        {
            int deviceId = 0;
            HIP_CHECK_EXC(hipGetDevice(&deviceId));
            return GetDevice(deviceId);
        }

        std::shared_ptr<Hardware> GetDevice(int deviceId)
        {
            hipDeviceProp_t prop;
            HIP_CHECK_EXC(hipGetDeviceProperties(&prop, deviceId));
#if HIP_VERSION >= 50220730
            int hip_version;
            HIP_CHECK_EXC(hipRuntimeGetVersion(&hip_version));
            if(hip_version >= 50220730)
            {
                HIP_CHECK_EXC(hipDeviceGetAttribute(&prop.multiProcessorCount,
                                                    hipDeviceAttributePhysicalMultiProcessorCount,
                                                    deviceId));
            }
#endif
            return GetDevice(prop, deviceId);
        }

        std::shared_ptr<Hardware> GetDevice(hipDeviceProp_t const& prop, int deviceId)
        {
            const auto processor = AMDGPU::toProcessor(prop.gcnArchName);
            if(!ChipIdRegistry::supportsChipIdPredicate(processor))
            {
                return std::make_shared<HipAMDGPU>(prop, deviceId);
            }

            int        pciChipId = 0;
            hipError_t chipIdResult
                = hipDeviceGetAttribute(&pciChipId, hipDeviceAttributePciChipId, deviceId);

            // Check hip runtime support for PCI Chip ID attribute
            if(chipIdResult == hipErrorInvalidValue)
                throw std::runtime_error(pciChipIdUnsupportedErrorMessage(prop));

            // For any other error, use standard error checking
            HIP_CHECK_EXC(chipIdResult);

            if(!ChipIdRegistry::isKnownChipId(pciChipId))
                logUnregisteredPciChipIdWarningOnce(prop, pciChipId);

            return std::make_shared<HipAMDGPU>(prop, deviceId, std::make_optional(pciChipId));
        }
    } // namespace hip
} // namespace TensileLite