intel/compute-runtime
1
0
Fork 0
mirror of https://github.com/intel/compute-runtime.git synced 2025-12-24 21:18:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add internal argument to encode method

Browse Source 
Related-To: NEO-5244

Signed-off-by: Zbigniew Zdanowicz <zbigniew.zdanowicz@intel.com>
This commit is contained in:
Zbigniew Zdanowicz
2021-03-18 21:12:05 +00:00
committed by Compute-Runtime-Automation
parent 521cc56096
commit d6dde3df33
6 changed files with 53 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
shared/source/command_container/command_encoder.h
View File

@@ -38,7 +38,8 @@ struct EncodeDispatchKernel {
Device *device,
PreemptionMode preemptionMode,
bool &requiresUncachedMocs,
uint32_t &partitionCount);
uint32_t &partitionCount,
bool isInternal);
static void encodeAdditionalWalkerFields(const HardwareInfo &hwInfo, WALKER_TYPE &walkerCmd);
@@ -46,7 +47,7 @@ struct EncodeDispatchKernel {
static void *getInterfaceDescriptor(CommandContainer &container, uint32_t &iddOffset);
static size_t estimateEncodeDispatchKernelCmdsSize(Device *device, Vec3<size_t> groupStart, Vec3<size_t> groupCount);
static size_t estimateEncodeDispatchKernelCmdsSize(Device *device, Vec3<size_t> groupStart, Vec3<size_t> groupCount, bool isInternal);
static bool isRuntimeLocalIdsGenerationRequired(uint32_t activeChannels,
size_t *lws,

6
shared/source/command_container/command_encoder_bdw_plus.inl
View File

@@ -26,7 +26,7 @@ template <typename Family>
void EncodeDispatchKernel<Family>::encode(CommandContainer &container,
const void *pThreadGroupDimensions, bool isIndirect, bool isPredicate, DispatchKernelEncoderI *dispatchInterface,
uint64_t eventAddress, Device *device, PreemptionMode preemptionMode, bool &requiresUncachedMocs,
uint32_t &partitionCount) {
uint32_t &partitionCount, bool isInternal) {
using MEDIA_STATE_FLUSH = typename Family::MEDIA_STATE_FLUSH;
using MEDIA_INTERFACE_DESCRIPTOR_LOAD = typename Family::MEDIA_INTERFACE_DESCRIPTOR_LOAD;
@@ -49,7 +49,7 @@ void EncodeDispatchKernel<Family>::encode(CommandContainer &container,
if (!isIndirect) {
threadDimsVec = {threadDims[0], threadDims[1], threadDims[2]};
}
size_t estimatedSizeRequired = estimateEncodeDispatchKernelCmdsSize(device, threadStartVec, threadDimsVec);
size_t estimatedSizeRequired = estimateEncodeDispatchKernelCmdsSize(device, threadStartVec, threadDimsVec, isInternal);
if (container.getCommandStream()->getAvailableSpace() < estimatedSizeRequired) {
auto bbEnd = listCmdBufferStream->getSpaceForCmd<MI_BATCH_BUFFER_END>();
*bbEnd = Family::cmdInitBatchBufferEnd;
@@ -322,7 +322,7 @@ template <typename Family>
void EncodeDispatchKernel<Family>::appendAdditionalIDDFields(INTERFACE_DESCRIPTOR_DATA *pInterfaceDescriptor, const HardwareInfo &hwInfo, const uint32_t threadsPerThreadGroup, uint32_t slmTotalSize, SlmPolicy slmPolicy) {}
template <typename Family>
size_t EncodeDispatchKernel<Family>::estimateEncodeDispatchKernelCmdsSize(Device *device, Vec3<size_t> groupStart, Vec3<size_t> groupCount) {
size_t EncodeDispatchKernel<Family>::estimateEncodeDispatchKernelCmdsSize(Device *device, Vec3<size_t> groupStart, Vec3<size_t> groupCount, bool isInternal) {
using MEDIA_STATE_FLUSH = typename Family::MEDIA_STATE_FLUSH;
using MEDIA_INTERFACE_DESCRIPTOR_LOAD = typename Family::MEDIA_INTERFACE_DESCRIPTOR_LOAD;
using MI_BATCH_BUFFER_END = typename Family::MI_BATCH_BUFFER_END;