rocm-systems/rocclr/runtime/device/pal/palconstbuf.cpp

//
// Copyright (c) 2010 Advanced Micro Devices, Inc. All rights reserved.
//

#include "device/pal/palconstbuf.hpp"
#include "device/pal/palvirtual.hpp"
#include "device/pal/paldevice.hpp"
#include "device/pal/palsettings.hpp"

namespace pal {

// ================================================================================================
ManagedBuffer::ManagedBuffer(VirtualGPU& gpu, uint32_t size)
    : gpu_(gpu),
      pool_(MaxNumberOfBuffers),
      activeBuffer_(0),
      size_(size),
      wrtOffset_(0),
      wrtAddress_(nullptr) {}

// ================================================================================================
void ManagedBuffer::release() {
  for (auto it : pool_) {
    if ((it.buf != nullptr) && (it.buf->data() != nullptr)) {
      it.buf->unmap(&gpu_);
    }
    delete it.buf;
  }
}

// ================================================================================================
bool ManagedBuffer::create(Resource::MemoryType type) {
  for (uint i = 0; i < pool_.size(); ++i) {
    pool_[i].buf = new Memory(const_cast<pal::Device&>(gpu_.dev()), size_);
    if (nullptr == pool_[i].buf || !pool_[i].buf->create(type)) {
      LogPrintfError("We couldn't create HW constant buffer, size(%d)!", size_);
      return false;
    }
    // Assign virtual gpu to the allocation. Buffer will be used only on a particular queue
    pool_[i].buf->memRef()->gpu_ = &gpu_;
    void* wrtAddress = pool_[i].buf->map(&gpu_);
    if (wrtAddress == nullptr) {
      LogPrintfError("We couldn't map HW constant buffer, size(%d)!", size_);
      return false;
    }
    // Make sure OCL touches every buffer in the queue to avoid delays on the first submit
    uint dummy = 0;
    static constexpr bool Wait = true;
    // Write 0 for the buffer paging by VidMM
    pool_[i].buf->writeRawData(gpu_, 0, sizeof(dummy), &dummy, Wait);
  }
  wrtAddress_ = pool_[activeBuffer_].buf->data();
  return true;
}

// ================================================================================================
address ManagedBuffer::reserve(uint32_t size, uint64_t* gpu_address) {
  // Align to the maximum data size available in OpenCL
  static constexpr uint32_t MemAlignment = sizeof(cl_double16);

  // Align reserve size on the vector's boundary
  uint32_t count = amd::alignUp(size, MemAlignment);

  // Save previous event
  pinGpuEvent();

  // Check if buffer has enough space for reservation
  if ((wrtOffset_ + count) > size_) {
    // Get the next buffer in the list
    ++activeBuffer_;
    activeBuffer_ %= MaxNumberOfBuffers;
    if (!gpu().dev().settings().disableSdma_) {
      // Make sure the buffer isn't busy
      gpu().waitForEvent(&pool_[activeBuffer_].events[SdmaEngine]);
    }
    gpu().waitForEvent(&pool_[activeBuffer_].events[MainEngine]);
    wrtAddress_ = pool_[activeBuffer_].buf->data();
    wrtOffset_ = 0;
  }

  *gpu_address = pool_[activeBuffer_].buf->vmAddress() + wrtOffset_;
  address cpu_address = wrtAddress_ + wrtOffset_;

  // Adjust the offset by the reserved size
  wrtOffset_ += count;

  return cpu_address;
}

// ================================================================================================
void ManagedBuffer::pinGpuEvent() {
  GpuEvent* event = activeMemory()->getGpuEvent(gpu());
  pool_[activeBuffer_].events[event->engineId_] = *event;
  activeMemory()->setBusy(gpu(), GpuEvent::InvalidID);
}

// ================================================================================================
ConstantBuffer::ConstantBuffer(ManagedBuffer& mbuf, uint32_t size)
    : mbuf_(mbuf), sys_mem_copy_(nullptr), size_(size) {}

// ================================================================================================
ConstantBuffer::~ConstantBuffer() { amd::AlignedMemory::deallocate(sys_mem_copy_); }

// ================================================================================================
bool ConstantBuffer::Create() {
  // Create sysmem copy for the constant buffer.
  sys_mem_copy_ = reinterpret_cast<address>(amd::AlignedMemory::allocate(size_, 256));
  if (sys_mem_copy_ == nullptr) {
    LogPrintfError("We couldn't allocate sysmem copy for constant buffer, size(%d)!", size_);
    return false;
  }
  memset(sys_mem_copy_, 0, size_);
  return true;
}

// ================================================================================================
uint64_t ConstantBuffer::UploadDataToHw(uint32_t size) const {
  uint64_t vm_address;
  address cpu_address = mbuf_.reserve(size, &vm_address);
  // Update memory with new CB data
  memcpy(cpu_address, sys_mem_copy_, size);
  return vm_address;
}

// ================================================================================================
uint64_t ConstantBuffer::UploadDataToHw(const void* sysmem, uint32_t size) const {
  uint64_t vm_address;
  address cpu_address = mbuf_.reserve(size, &vm_address);
  // Update memory with new CB data
  memcpy(cpu_address, sysmem, size);
  return vm_address;
}

// ================================================================================================
XferBuffer::XferBuffer(const Device& device, ManagedBuffer& mbuf, uint32_t size)
    : buffer_view_(device, size), mbuf_(mbuf), size_(size) {
  // Create a view for access
  Resource::ViewParams params = {};
  params.gpu_ = &mbuf_.gpu();
  params.offset_ = 0;
  params.size_ = size_;
  params.resource_ = mbuf_.activeMemory();
  bool result = buffer_view_.create(Resource::View, &params);
  assert(result && "View creaiton should never return an error!");
}

// ================================================================================================
Memory& XferBuffer::Acquire(uint32_t size) {
  uint64_t vm_address;
  // Reserve space in the managed buffer
  address cpu_address = mbuf_.reserve(size, &vm_address);
  // Update a view for access
  buffer_view_.updateView(mbuf_.activeMemory(), vm_address - mbuf_.vmAddress(), size);
  return buffer_view_;
}

}  // namespace pal