Huawei_Technology
/
mindspore-mindinsight
Not watched
1
0
Fork 0
Code
Releases 11 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
Browse Source

Changed Vulkan Vision build scripts to be based on releases of open-source repositories. Added link to CGO paper

tags/v1.2.0-rc1
dpankratz 4 years ago
parent
61fbee7450
commit
ea771381b6
4 changed files with 94 additions and 350 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +1
    -1
      README.md
  2. +5
    -11
      build/scripts/build_vulkan_vision_linux.sh
  3. +5
    -10
      build/scripts/build_vulkan_vision_windows.bat
  4. +83
    -328
      ecosystem_tools/VulkanVision/vv-patches/0001-layers-Added-auto-inst-layers.patch

+ 1
- 1
README.md View File

@@ -159,7 +159,7 @@ Check out how MindSpore Open Governance [works](https://gitee.com/mindspore/comm

## Vulkan Vision

Vulkan Vision(V-Vision) provides an unprecedented level of detail into the execution of Vulkan applications through dynamic instrumentation. V-Vision supports analyzing AI workloads implemented using the a compute pipeline as well as traditional raster and ray-tracing Vulkan applications. To use V-Vision please refer to the [build instructions](https://gitee.com/mindspore/mindinsight/blob/master/ecosystem_tools/VulkanVision/README.md).
Vulkan Vision(V-Vision) provides an unprecedented level of detail into the execution of Vulkan applications through dynamic instrumentation. V-Vision supports analyzing AI workloads implemented using the a compute pipeline as well as traditional raster and ray-tracing Vulkan applications. To use V-Vision please refer to the [build instructions](https://gitee.com/mindspore/mindinsight/blob/master/ecosystem_tools/VulkanVision/README.md). For more information, please refer to [the paper](https://webdocs.cs.ualberta.ca/~amaral/papers/PankratzCGO21) published at CGO 2021.

## Contributing



+ 5
- 11
build/scripts/build_vulkan_vision_linux.sh View File

@@ -21,40 +21,34 @@ cd "$(dirname "$0")"/../../ecosystem_tools/VulkanVision
if [[ ! -d "SPIRV-Tools" ]]
then
echo "Cloning SPIRV-Tools"
git clone https://github.com/KhronosGroup/SPIRV-Tools
git clone https://github.com/KhronosGroup/SPIRV-Tools --branch v2020.6
cp st-patches/*.patch SPIRV-Tools
cd SPIRV-Tools
git checkout 17ffa89097b26efeb323e6963220326b5ffb2baf
# These are the current stable changes and can be updated with new releases
git apply 0001-spirv-opt-Add-auto-inst-passes.patch
rm *.patch
git clone https://github.com/KhronosGroup/SPIRV-Headers.git external/spirv-headers --branch 1.5.4.raytracing.fixed
git clone https://github.com/google/effcee.git external/effcee --branch v2019.1
git clone https://github.com/google/re2.git external/re2 --branch 2020-11-01
cd ..
fi

if [[ ! -d "Vulkan-ValidationLayers" ]]
then
echo "Cloning Vulkan-ValidationLayers"
git clone https://github.com/KhronosGroup/Vulkan-ValidationLayers
git clone https://github.com/KhronosGroup/Vulkan-ValidationLayers --branch sdk-1.2.162
cp vv-patches/*.patch Vulkan-ValidationLayers
cd Vulkan-ValidationLayers
git checkout aa076dae88e282d7b6cada4f900b2fa7dac8ed08
# These are the current stable changes and can be updated with new releases
git apply 0001-layers-Added-auto-inst-layers.patch
rm *.patch
cd ..
fi


build_dir=$(pwd)

echo "Building SPIRV-Tools"
cd SPIRV-Tools
git clone https://github.com/KhronosGroup/SPIRV-Headers.git external/spirv-headers
cd external/spirv-headers
git checkout f027d53ded7e230e008d37c8b47ede7cd308e19d
cd ../..
git clone https://github.com/google/effcee.git external/effcee
git clone https://github.com/google/re2.git external/re2
mkdir build
cd build
mkdir install


+ 5
- 10
build/scripts/build_vulkan_vision_windows.bat View File

@@ -19,22 +19,23 @@ cd %~dp0..\..\ecosystem_tools\VulkanVision

IF NOT EXIST SPIRV-Tools (
echo Cloning SPIRV-Tools
git clone https://github.com/KhronosGroup/SPIRV-Tools
git clone https://github.com/KhronosGroup/SPIRV-Tools --branch v2020.6
copy st-patches\*.patch SPIRV-Tools
cd SPIRV-Tools
git checkout 17ffa89097b26efeb323e6963220326b5ffb2baf
REM These are the current stable changes and can be updated with new releases
git apply 0001-spirv-opt-Add-auto-inst-passes.patch
del *.patch
git clone https://github.com/KhronosGroup/SPIRV-Headers.git external\spirv-headers --branch 1.5.4.raytracing.fixed
git clone https://github.com/google/effcee.git external\effcee --branch v2019.1
git clone https://github.com/google/re2.git external\re2 --branch 2020-11-01
cd ..
)

IF NOT EXIST Vulkan-ValidationLayers (
echo Cloning Vulkan-ValidationLayers
git clone https://github.com/KhronosGroup/Vulkan-ValidationLayers
git clone https://github.com/KhronosGroup/Vulkan-ValidationLayers --branch sdk-1.2.162
copy vv-patches\*.patch Vulkan-ValidationLayers
cd Vulkan-ValidationLayers
git checkout aa076dae88e282d7b6cada4f900b2fa7dac8ed08
REM These are the current stable changes and can be updated with new releases
git apply 0001-layers-Added-auto-inst-layers.patch
del *.patch
@@ -45,12 +46,6 @@ set build_dir=%cd%

echo Building SPIRV-Tools
cd SPIRV-Tools
git clone https://github.com/KhronosGroup/SPIRV-Headers.git external\spirv-headers
cd external\spirv-headers
git checkout f027d53ded7e230e008d37c8b47ede7cd308e19d
cd ..\..
git clone https://github.com/google/effcee.git external\effcee
git clone https://github.com/google/re2.git external\re2
mkdir build
cd build
mkdir install


+ 83
- 328
ecosystem_tools/VulkanVision/vv-patches/0001-layers-Added-auto-inst-layers.patch View File

@@ -1,14 +1,13 @@
From 94f9a7c9ea98fa400d47b0a71efbea8792113b08 Mon Sep 17 00:00:00 2001
From 1bb2f039b0ba3f815c8a8fd6ac966a89f6ffe5a2 Mon Sep 17 00:00:00 2001
From: dpankratz <pankratz@ualberta.ca>
Date: Wed, 27 Jan 2021 09:17:26 -0700
Date: Thu, 18 Feb 2021 14:45:45 -0700
Subject: [PATCH] layers: Added auto-inst layers

---
CMakeLists.txt | 27 +-
CMakeLists.txt | 24 +-
build-android/jni/Android.mk | 7 +
build-android/known_good.json | 2 +-
docs/auto_instrument.md | 213 +++
layers/CMakeLists.txt | 25 +-
layers/CMakeLists.txt | 19 +
layers/auto_inst.cpp | 1205 +++++++++++++++++
layers/auto_inst.h | 465 +++++++
.../auto_inst_divergence_characterization.cpp | 157 +++
@@ -25,15 +24,14 @@ Subject: [PATCH] layers: Added auto-inst layers
layers/auto_inst_warp_entry_and_exit.h | 52 +
layers/debug_printf.cpp | 2 +-
layers/debug_printf.h | 1 +
layers/generated/chassis.cpp | 78 +-
layers/generated/chassis.h | 15 +-
layers/gpu_utils.h | 7 +
layers/gpu_validation.h | 4 +-
layers/generated/chassis.cpp | 53 +-
layers/generated/chassis.h | 13 +
layers/gpu_utils.h | 6 +
layers/gpu_validation.h | 1 +
layers/layer_options.cpp | 13 +
layers/layer_options.h | 19 +-
layers/layer_options.h | 14 +-
scripts/known_good.json | 17 +-
scripts/layer_chassis_generator.py | 72 +-
29 files changed, 3305 insertions(+), 37 deletions(-)
27 files changed, 3215 insertions(+), 8 deletions(-)
create mode 100644 docs/auto_instrument.md
create mode 100644 layers/auto_inst.cpp
create mode 100644 layers/auto_inst.h
@@ -51,7 +49,7 @@ Subject: [PATCH] layers: Added auto-inst layers
create mode 100644 layers/auto_inst_warp_entry_and_exit.h

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 994a9ef0..da32e3b0 100644
index 994a9ef0..e72924fe 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -227,6 +227,8 @@ if(BUILD_TESTS OR BUILD_LAYERS)
@@ -63,13 +61,8 @@ index 994a9ef0..da32e3b0 100644
endif()
set(SPIRV_TOOLS_BINARY_ROOT "${SPIRV_TOOLS_INSTALL_DIR}/lib"
@@ -240,9 +242,12 @@ if(BUILD_TESTS OR BUILD_LAYERS)
set(SPIRV_TOOLS_OPT_DEBUG_SEARCH_PATH "${SPIRV_TOOLS_INSTALL_DIR}/lib")
@@ -242,7 +244,7 @@ if(BUILD_TESTS OR BUILD_LAYERS)
find_library(SPIRV_TOOLS_LIB NAMES SPIRV-Tools HINTS ${SPIRV_TOOLS_SEARCH_PATH})
+ if (NOT SPIRV_TOOLS_LIB)
+ message(FATAL_ERROR "Could not find SPIRV-Tools")
+ endif()
find_library(SPIRV_TOOLS_OPT_LIB NAMES SPIRV-Tools-opt HINTS ${SPIRV_TOOLS_OPT_SEARCH_PATH})
- if(WIN32)
@@ -77,7 +70,7 @@ index 994a9ef0..da32e3b0 100644
add_library(SPIRV-Tools-opt STATIC IMPORTED)
add_library(SPIRV-Tools STATIC IMPORTED)
@@ -269,6 +274,26 @@ if(BUILD_TESTS OR BUILD_LAYERS)
@@ -269,6 +271,26 @@ if(BUILD_TESTS OR BUILD_LAYERS)
set(SPIRV_TOOLS_INCLUDE_DIR "${spirv-tools_SOURCE_DIR}/include" CACHE PATH "Path to spirv tools headers")
endif()
@@ -105,10 +98,10 @@ index 994a9ef0..da32e3b0 100644
endif()
diff --git a/build-android/jni/Android.mk b/build-android/jni/Android.mk
index f0955dd5..75c49def 100644
index 871bdf30..e92895b9 100644
--- a/build-android/jni/Android.mk
+++ b/build-android/jni/Android.mk
@@ -45,6 +45,13 @@ LOCAL_SRC_FILES += $(SRC_DIR)/layers/generated/spirv_validation_helper.cpp
@@ -44,6 +44,13 @@ LOCAL_SRC_FILES += $(SRC_DIR)/layers/shader_validation.cpp
LOCAL_SRC_FILES += $(SRC_DIR)/layers/gpu_validation.cpp
LOCAL_SRC_FILES += $(SRC_DIR)/layers/gpu_utils.cpp
LOCAL_SRC_FILES += $(SRC_DIR)/layers/debug_printf.cpp
@@ -122,17 +115,6 @@ index f0955dd5..75c49def 100644
LOCAL_SRC_FILES += $(SRC_DIR)/layers/best_practices_utils.cpp
LOCAL_SRC_FILES += $(SRC_DIR)/layers/generated/best_practices.cpp
LOCAL_SRC_FILES += $(SRC_DIR)/layers/synchronization_validation.cpp
diff --git a/build-android/known_good.json b/build-android/known_good.json
index 1a77e5ae..d8635c37 100755
--- a/build-android/known_good.json
+++ b/build-android/known_good.json
@@ -1,5 +1,5 @@
{
- "repos" : [
+ "repos": [
{
"name" : "shaderc",
"url" : "https://github.com/google/shaderc.git",
diff --git a/docs/auto_instrument.md b/docs/auto_instrument.md
new file mode 100644
index 00000000..30d376ce
@@ -353,10 +335,10 @@ index 00000000..30d376ce
+
+* **warp_exits_vs_entires.csv** is generated per frame with the respective exits count/entry count for each pipeline invocation.
diff --git a/layers/CMakeLists.txt b/layers/CMakeLists.txt
index 851b98d5..93758bab 100644
index 7b1ba729..76b4b520 100644
--- a/layers/CMakeLists.txt
+++ b/layers/CMakeLists.txt
@@ -234,9 +234,27 @@ set(GPU_UTILITY_LIBRARY_FILES
@@ -232,6 +232,22 @@ set(GPU_UTILITY_LIBRARY_FILES
gpu_utils.cpp
gpu_utils.h)
@@ -364,61 +346,38 @@ index 851b98d5..93758bab 100644
+ auto_inst.cpp
+ auto_inst_divergence_characterization.cpp
+ auto_inst_dyn_shader_trace.cpp
+ auto_inst_dyn_trace_ray_trace.cpp
+ auto_inst_execution_trace.cpp
+ auto_inst_simt_efficiency.cpp
+ auto_inst_warp_entry_and_exit.cpp
+ auto_inst_dyn_trace_ray_trace.cpp
+ auto_inst_execution_trace.cpp
+ auto_inst_simt_efficiency.cpp
+ auto_inst_warp_entry_and_exit.cpp
+ auto_inst.h
+ auto_inst_divergence_characterization.h
+ auto_inst_dyn_shader_trace.h
+ auto_inst_dyn_trace_ray_trace.h
+ auto_inst_execution_trace.h
+ auto_inst_simt_efficiency.h
+ auto_inst_warp_entry_and_exit.h
+ )
+ auto_inst_dyn_trace_ray_trace.h
+ auto_inst_execution_trace.h
+ auto_inst_simt_efficiency.h
+ auto_inst_warp_entry_and_exit.h)
+
set(SYNC_VALIDATION_LIBRARY_FILES
synchronization_validation.cpp
- synchronization_validation.h)
+ synchronization_validation.h
+ )
# Validation Layer performance instrumentation support using Optick.
# https://optick.dev/ https://github.com/bombomby/optick
@@ -268,6 +286,7 @@ if(INSTRUMENT_OPTICK)
endif()
endif()
+
if(BUILD_LAYERS)
AddVkLayer(khronos_validation "${KHRONOS_LAYER_COMPILE_DEFINITIONS}"
${CHASSIS_LIBRARY_FILES}
@@ -279,6 +298,7 @@ if(BUILD_LAYERS)
synchronization_validation.h)
@@ -247,13 +263,16 @@ if(BUILD_LAYERS)
${GPU_UTILITY_LIBRARY_FILES}
${GPU_ASSISTED_LIBRARY_FILES}
${DEBUG_PRINTF_LIBRARY_FILES}
+ ${AUTO_INST_LIBRARY_FILES}
${SYNC_VALIDATION_LIBRARY_FILES}
${OPTICK_SOURCE_FILES})
@@ -290,14 +310,17 @@ if(BUILD_LAYERS)
target_link_libraries(VkLayer_khronos_validation PRIVATE "$<$<CONFIG:Release>:-DEBUG:FULL>")
endif()
${SYNC_VALIDATION_LIBRARY_FILES})
+
# Khronos validation additional dependencies
target_include_directories(VkLayer_khronos_validation PRIVATE ${GLSLANG_SPIRV_INCLUDE_DIR})
target_include_directories(VkLayer_khronos_validation PRIVATE ${SPIRV_TOOLS_INCLUDE_DIR})
+ target_include_directories(VkLayer_khronos_validation PRIVATE ${SPIRV_CROSS_INCLUDE_DIR})
target_include_directories(VkLayer_khronos_validation PRIVATE ${SPIRV_HEADERS_INCLUDE_DIR})
if(INSTRUMENT_OPTICK)
target_include_directories(VkLayer_khronos_validation PRIVATE ${OPTICK_SOURCE_DIR})
endif()
+ target_include_directories(VkLayer_khronos_validation PRIVATE ${SPIRV_CROSS_INCLUDE_DIR})
target_link_libraries(VkLayer_khronos_validation PRIVATE ${SPIRV_TOOLS_LIBRARIES})
+ target_link_libraries(VkLayer_khronos_validation PRIVATE ${SPIRV_CROSS_LIBRARIES})
# The output file needs Unix "/" separators or Windows "\" separators On top of that, Windows separators actually need to be doubled
# because the json format uses backslash escapes
# Force generation of the PDB file for Release builds.
# Note that CMake reduces optimization levels for RelWithDebInfo builds.
diff --git a/layers/auto_inst.cpp b/layers/auto_inst.cpp
new file mode 100644
index 00000000..8f3669d3
@@ -2846,7 +2805,7 @@ index 00000000..769b0cc4
+};
diff --git a/layers/auto_inst_execution_trace.cpp b/layers/auto_inst_execution_trace.cpp
new file mode 100644
index 00000000..ec4c876b
index 00000000..6f27d804
--- /dev/null
+++ b/layers/auto_inst_execution_trace.cpp
@@ -0,0 +1,174 @@
@@ -2993,7 +2952,7 @@ index 00000000..ec4c876b
+ // Output annotated shaders
+ std::unordered_map<uint32_t, std::string> annotations;
+ for (const auto& entry : inst_id2bb_inst_ids_) {
+ auto instrumentation_id = *entry.second.begin();
+ auto instrumentation_id = entry.first;
+ auto visits = inst_id2dyn_execution_count[instrumentation_id];
+ if (visits > 0) {
+ auto simt_efficiency = inst_id2active_and_total_thread_counts[entry.first].SimtEfficiency();
@@ -3350,7 +3309,7 @@ index 00000000..17adfdfb
+ };
+};
diff --git a/layers/debug_printf.cpp b/layers/debug_printf.cpp
index f04f4852..417341b5 100644
index e9181812..a8ba17df 100644
--- a/layers/debug_printf.cpp
+++ b/layers/debug_printf.cpp
@@ -80,7 +80,7 @@ void DebugPrintf::PostCallRecordCreateDevice(VkPhysicalDevice physicalDevice, co
@@ -3375,10 +3334,10 @@ index 915d5a6d..deb85031 100644
enum vartype { varsigned, varunsigned, varfloat };
diff --git a/layers/generated/chassis.cpp b/layers/generated/chassis.cpp
index 75a4ebfa..6548ddc9 100644
index 9a4799dd..6ecc487a 100644
--- a/layers/generated/chassis.cpp
+++ b/layers/generated/chassis.cpp
@@ -50,6 +50,12 @@ bool wrap_handles = true;
@@ -49,6 +49,12 @@ bool wrap_handles = true;
#include "gpu_validation.h"
#include "object_lifetime_validation.h"
#include "debug_printf.h"
@@ -3391,7 +3350,7 @@ index 75a4ebfa..6548ddc9 100644
#include "stateless_validation.h"
#include "synchronization_validation.h"
#include "thread_safety.h"
@@ -306,6 +312,24 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
@@ -299,6 +305,24 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
auto sync_validation_obj = new SyncValidator;
sync_validation_obj->RegisterValidationObject(local_enables[sync_validation], api_version, report_data, local_object_dispatch);
@@ -3409,31 +3368,18 @@ index 75a4ebfa..6548ddc9 100644
+
+ auto auto_inst_warp_entry_and_exit_obj = new AutoInstWarpEntryAndExit;
+ auto_inst_warp_entry_and_exit_obj->RegisterValidationObject(local_enables[auto_inst_warp_entry_and_exit], api_version, report_data, local_object_dispatch);
+
+
+ auto auto_inst_dyn_shader_trace_obj = new AutoInstDynShaderTrace;
+ auto_inst_dyn_shader_trace_obj->RegisterValidationObject(local_enables[auto_inst_dyn_shader_trace], api_version, report_data, local_object_dispatch);
+
// If handle wrapping is disabled via the ValidationFeatures extension, override build flag
if (local_disables[handle_wrapping]) {
wrap_handles = false;
@@ -338,7 +362,7 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
framework->report_data = report_data;
framework->api_version = api_version;
framework->instance_extensions.InitFromInstanceCreateInfo(specified_version, pCreateInfo);
-
+
OutputLayerStatusInfo(framework);
thread_checker_obj->FinalizeInstanceValidationObject(framework);
@@ -348,9 +372,15 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
core_checks_obj->instance = *pInstance;
core_checks_obj->instance_state = core_checks_obj;
best_practices_obj->FinalizeInstanceValidationObject(framework);
- gpu_assisted_obj->FinalizeInstanceValidationObject(framework);
+ gpu_assisted_obj->FinalizeInstanceValidationObject(framework);
@@ -344,6 +368,12 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
gpu_assisted_obj->FinalizeInstanceValidationObject(framework);
debug_printf_obj->FinalizeInstanceValidationObject(framework);
sync_validation_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_simt_efficiency_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_simt_efficiency_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_execution_trace_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_dyn_trace_ray_trace_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_divergence_characterization_obj->FinalizeInstanceValidationObject(framework);
@@ -3442,90 +3388,59 @@ index 75a4ebfa..6548ddc9 100644
for (auto intercept : framework->object_dispatch) {
auto lock = intercept->write_lock();
@@ -360,8 +390,9 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
// Delete unused validation objects to avoid memory leak.
@@ -354,7 +384,8 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
std::vector<ValidationObject*> local_objs = {
thread_checker_obj, object_tracker_obj, parameter_validation_obj,
- core_checks_obj, best_practices_obj, gpu_assisted_obj, debug_printf_obj,
core_checks_obj, best_practices_obj, gpu_assisted_obj, debug_printf_obj,
- sync_validation_obj,
+ core_checks_obj, best_practices_obj, gpu_assisted_obj, debug_printf_obj, sync_validation_obj,
+ auto_inst_simt_efficiency_obj, auto_inst_execution_trace_obj, auto_inst_dyn_trace_ray_trace_obj,
+ sync_validation_obj, auto_inst_simt_efficiency_obj, auto_inst_execution_trace_obj, auto_inst_dyn_trace_ray_trace_obj,
+ auto_inst_divergence_characterization_obj, auto_inst_warp_entry_and_exit_obj, auto_inst_dyn_shader_trace_obj
};
for (auto obj : local_objs) {
if (std::find(local_object_dispatch.begin(), local_object_dispatch.end(), obj) == local_object_dispatch.end()) {
@@ -493,22 +524,27 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateDevice(VkPhysicalDevice gpu, const VkDevice
auto debug_printf_obj = new DebugPrintf;
debug_printf_obj->InitDeviceValidationObject(enables[debug_printf], instance_interceptor, device_interceptor);
-
+
@@ -490,6 +521,24 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateDevice(VkPhysicalDevice gpu, const VkDevice
auto sync_validation_obj = new SyncValidator;
sync_validation_obj->InitDeviceValidationObject(enables[sync_validation], instance_interceptor, device_interceptor);
- // Delete unused validation objects to avoid memory leak.
- std::vector<ValidationObject *> local_objs = {
- thread_safety_obj, stateless_validation_obj, object_tracker_obj,
- core_checks_obj, best_practices_obj, gpu_assisted_obj, debug_printf_obj,
- sync_validation_obj,
- };
- for (auto obj : local_objs) {
- if (std::find(device_interceptor->object_dispatch.begin(), device_interceptor->object_dispatch.end(), obj) ==
- device_interceptor->object_dispatch.end()) {
- delete obj;
- }
- }
+ auto auto_inst_simt_efficiency_obj = new AutoInstSimtEfficiency;
+ auto_inst_simt_efficiency_obj->InitDeviceValidationObject(enables[auto_inst_simt_efficiency], instance_interceptor, device_interceptor);
+
+ auto auto_inst_execution_trace_obj = new AutoInstExecutionTrace;
+ auto_inst_execution_trace_obj->InitDeviceValidationObject(enables[auto_inst_execution_trace], instance_interceptor, device_interceptor);
+
+
+ auto auto_inst_dyn_trace_ray_trace_obj = new AutoInstDynTraceRayTrace;
+ auto_inst_dyn_trace_ray_trace_obj->InitDeviceValidationObject(enables[auto_inst_dyn_trace_ray_trace], instance_interceptor, device_interceptor);
+
+
+ auto auto_inst_divergence_characterization_obj = new AutoInstDivergenceCharacterization;
+ auto_inst_divergence_characterization_obj->InitDeviceValidationObject(enables[auto_inst_divergence_characterization], instance_interceptor, device_interceptor);
+
+
+ auto auto_inst_warp_entry_and_exit_obj = new AutoInstWarpEntryAndExit;
+ auto_inst_warp_entry_and_exit_obj->InitDeviceValidationObject(enables[auto_inst_warp_entry_and_exit], instance_interceptor, device_interceptor);
+
+
+ auto auto_inst_dyn_shader_trace_obj = new AutoInstDynShaderTrace;
+ auto_inst_dyn_shader_trace_obj->InitDeviceValidationObject(enables[auto_inst_dyn_shader_trace], instance_interceptor, device_interceptor);
for (auto intercept : instance_interceptor->object_dispatch) {
auto lock = intercept->write_lock();
@@ -574,7 +610,8 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateGraphicsPipelines(
}
+
// Delete unused validation objects to avoid memory leak.
std::vector<ValidationObject *> local_objs = {
thread_safety_obj, stateless_validation_obj, object_tracker_obj,
@@ -568,6 +617,7 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateGraphicsPipelines(
auto usepCreateInfos = (!cgpl_state[LayerObjectTypeGpuAssisted].pCreateInfos) ? pCreateInfos : cgpl_state[LayerObjectTypeGpuAssisted].pCreateInfos;
- if (cgpl_state[LayerObjectTypeDebugPrintf].pCreateInfos) usepCreateInfos = cgpl_state[LayerObjectTypeDebugPrintf].pCreateInfos;
+ if (cgpl_state[LayerObjectTypeDebugPrintf].pCreateInfos) usepCreateInfos = cgpl_state[LayerObjectTypeDebugPrintf].pCreateInfos;
if (cgpl_state[LayerObjectTypeDebugPrintf].pCreateInfos) usepCreateInfos = cgpl_state[LayerObjectTypeDebugPrintf].pCreateInfos;
+ else if (cgpl_state[LayerObjectTypeAutoInst].pCreateInfos) usepCreateInfos = cgpl_state[LayerObjectTypeAutoInst].pCreateInfos;
VkResult result = DispatchCreateGraphicsPipelines(device, pipelineCache, createInfoCount, usepCreateInfos, pAllocator, pPipelines);
@@ -610,7 +647,8 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateComputePipelines(
}
@@ -604,6 +654,7 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateComputePipelines(
auto usepCreateInfos = (!ccpl_state[LayerObjectTypeGpuAssisted].pCreateInfos) ? pCreateInfos : ccpl_state[LayerObjectTypeGpuAssisted].pCreateInfos;
- if (ccpl_state[LayerObjectTypeDebugPrintf].pCreateInfos) usepCreateInfos = ccpl_state[LayerObjectTypeDebugPrintf].pCreateInfos;
+ if (ccpl_state[LayerObjectTypeDebugPrintf].pCreateInfos) usepCreateInfos = ccpl_state[LayerObjectTypeDebugPrintf].pCreateInfos;
if (ccpl_state[LayerObjectTypeDebugPrintf].pCreateInfos) usepCreateInfos = ccpl_state[LayerObjectTypeDebugPrintf].pCreateInfos;
+ else if (ccpl_state[LayerObjectTypeAutoInst].pCreateInfos) usepCreateInfos = ccpl_state[LayerObjectTypeAutoInst].pCreateInfos;
VkResult result = DispatchCreateComputePipelines(device, pipelineCache, createInfoCount, usepCreateInfos, pAllocator, pPipelines);
@@ -654,6 +692,8 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateRayTracingPipelinesNV(
pPipelines, result, &(crtpl_state[intercept->container_type]));
}
return result;
+
+
}
VKAPI_ATTR VkResult VKAPI_CALL CreateRayTracingPipelinesKHR(
diff --git a/layers/generated/chassis.h b/layers/generated/chassis.h
index 7f67fbe3..269aa727 100644
index 23649ec6..5b432a9e 100644
--- a/layers/generated/chassis.h
+++ b/layers/generated/chassis.h
@@ -52,6 +52,12 @@
@@ -3549,27 +3464,24 @@ index 7f67fbe3..269aa727 100644
LayerObjectTypeCommandCounter, // Command Counter validation object, child of corechecks
LayerObjectTypeSyncValidation, // Instance or device synchronization validation layer object
LayerObjectTypeMaxEnum, // Max enum count
@@ -2897,8 +2904,14 @@ typedef enum EnableFlags {
gpu_validation_reserve_binding_slot,
best_practices,
@@ -2899,6 +2906,12 @@ typedef enum EnableFlags {
vendor_specific_arm,
- debug_printf,
+ debug_printf,
debug_printf,
sync_validation,
+ auto_inst_simt_efficiency,
+ auto_inst_simt_efficiency,
+ auto_inst_execution_trace,
+ auto_inst_dyn_trace_ray_trace,
+ auto_inst_divergence_characterization,
+ auto_inst_warp_entry_and_exit,
+ auto_inst_dyn_shader_trace,
+ auto_inst_dyn_shader_trace,
// Insert new enables above this line
kMaxEnableFlags,
} EnableFlags;
diff --git a/layers/gpu_utils.h b/layers/gpu_utils.h
index 01197b94..e04ee285 100644
index 01197b94..0e31f00e 100644
--- a/layers/gpu_utils.h
+++ b/layers/gpu_utils.h
@@ -314,14 +314,21 @@ void UtilPostCallRecordPipelineCreations(const uint32_t count, const CreateInfo
@@ -314,11 +314,17 @@ void UtilPostCallRecordPipelineCreations(const uint32_t count, const CreateInfo
VkShaderModule shader_module = VK_NULL_HANDLE;
if (bind_point == VK_PIPELINE_BIND_POINT_GRAPHICS) {
shader_module = pipeline_state->graphicsPipelineCI.pStages[stage].module;
@@ -3587,12 +3499,8 @@ index 01197b94..e04ee285 100644
} else {
assert(false);
}
+
object_ptr->shader_map[shader_state->gpu_validation_shader_id].shader_module = shader_module;
object_ptr->shader_map[shader_state->gpu_validation_shader_id].pgm = std::move(code);
}
diff --git a/layers/gpu_validation.h b/layers/gpu_validation.h
index 706d3fb7..b48b84ab 100644
index 291cfb85..2ff6e0fc 100644
--- a/layers/gpu_validation.h
+++ b/layers/gpu_validation.h
@@ -54,6 +54,7 @@ struct GpuAssistedShaderTracker {
@@ -3603,18 +3511,8 @@ index 706d3fb7..b48b84ab 100644
};
struct GpuAssistedAccelerationStructureBuildValidationBufferInfo {
@@ -149,7 +150,8 @@ class GpuAssisted : public ValidationStateTracker {
VkPipelineStageFlags srcStageMask, VkPipelineStageFlags dstStageMask,
uint32_t memoryBarrierCount, const VkMemoryBarrier* pMemoryBarriers,
uint32_t bufferMemoryBarrierCount, const VkBufferMemoryBarrier* pBufferMemoryBarriers,
- uint32_t imageMemoryBarrierCount, const VkImageMemoryBarrier* pImageMemoryBarriers) const override;
+ uint32_t imageMemoryBarrierCount,
+ const VkImageMemoryBarrier* pImageMemoryBarriers) const override;
void PreCallRecordCreateBuffer(VkDevice device, const VkBufferCreateInfo* pCreateInfo, const VkAllocationCallbacks* pAllocator,
VkBuffer* pBuffer, void* cb_state_data) override;
void CreateAccelerationStructureBuildValidationState(GpuAssisted* device_GpuAssisted);
diff --git a/layers/layer_options.cpp b/layers/layer_options.cpp
index 3c6f5dfe..0b6ec389 100644
index 7a791087..022a9466 100644
--- a/layers/layer_options.cpp
+++ b/layers/layer_options.cpp
@@ -92,6 +92,19 @@ void SetValidationEnable(CHECK_ENABLED &enable_data, const ValidationCheckEnable
@@ -3638,71 +3536,44 @@ index 3c6f5dfe..0b6ec389 100644
case VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_EXT:
enable_data[gpu_validation] = true;
diff --git a/layers/layer_options.h b/layers/layer_options.h
index 861b9abe..d73768cd 100644
index 861b9abe..d1b5ea68 100644
--- a/layers/layer_options.h
+++ b/layers/layer_options.h
@@ -45,11 +45,16 @@ static const std::unordered_map<std::string, VkValidationFeatureDisableEXT> VkVa
static const std::unordered_map<std::string, VkValidationFeatureEnableEXT> VkValFeatureEnableLookup = {
{"VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_EXT", VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_EXT},
- {"VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_RESERVE_BINDING_SLOT_EXT",
- VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_RESERVE_BINDING_SLOT_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_RESERVE_BINDING_SLOT_EXT", VK_VALIDATION_FEATURE_ENABLE_GPU_ASSISTED_RESERVE_BINDING_SLOT_EXT},
@@ -50,6 +50,12 @@ static const std::unordered_map<std::string, VkValidationFeatureEnableEXT> VkVal
{"VK_VALIDATION_FEATURE_ENABLE_BEST_PRACTICES_EXT", VK_VALIDATION_FEATURE_ENABLE_BEST_PRACTICES_EXT},
{"VK_VALIDATION_FEATURE_ENABLE_DEBUG_PRINTF_EXT", VK_VALIDATION_FEATURE_ENABLE_DEBUG_PRINTF_EXT},
{"VK_VALIDATION_FEATURE_ENABLE_SYNCHRONIZATION_VALIDATION_EXT", VK_VALIDATION_FEATURE_ENABLE_SYNCHRONIZATION_VALIDATION_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_EXECUTION_TRACE_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_EXECUTION_TRACE_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_TRACE_RAY_TRACE_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_TRACE_RAY_TRACE_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DIVERGENCE_CHARACTERIZATION_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DIVERGENCE_CHARACTERIZATION_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_WARP_ENTRY_AND_EXIT_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_WARP_ENTRY_AND_EXIT_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_SHADER_TRACE_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_SHADER_TRACE_EXT},
+ {"VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_SHADER_TRACE_EXT", VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_SHADER_TRACE_EXT},
};
static const std::unordered_map<std::string, VkValidationFeatureEnable> VkValFeatureEnableLookup2 = {
@@ -93,7 +98,13 @@ static const std::vector<std::string> EnableFlagNameHelper = {
@@ -93,7 +99,13 @@ static const std::vector<std::string> EnableFlagNameHelper = {
"VK_VALIDATION_FEATURE_ENABLE_BEST_PRACTICES_EXT", // best_practices,
"VALIDATION_CHECK_ENABLE_VENDOR_SPECIFIC_ARM", // vendor_specific_arm,
"VK_VALIDATION_FEATURE_ENABLE_DEBUG_PRINTF_EXT", // debug_printf,
- "VK_VALIDATION_FEATURE_ENABLE_SYNCHRONIZATION_VALIDATION" // sync_validation,
+ "VK_VALIDATION_FEATURE_ENABLE_SYNCHRONIZATION_VALIDATION", // sync_validation,
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT", // auto_inst_simt_efficiency
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT", // auto_inst_simt_efficiency
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_EXECUTION_TRACE_EXT", // auto_inst_execution_trace
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_TRACE_RAY_TRACE_EXT", // auto_inst_dyn_trace_ray_trace
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DIVERGENCE_CHARACTERIZATION_EXT", // auto_inst_divergence_characterization
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_WARP_ENTRY_AND_EXIT_EXT", // auto_inst_warp_entry_and_exit
+ "VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_SHADER_TRACE_EXT", // auto_inst_dyn_shader_trace
};
-
+
void ProcessConfigAndEnvSettings(ConfigAndEnvSettings *settings_data);
diff --git a/scripts/known_good.json b/scripts/known_good.json
index 9a53e452..b4fef368 100755
index 9a53e452..3a27076e 100755
--- a/scripts/known_good.json
+++ b/scripts/known_good.json
@@ -1,6 +1,7 @@
{
- "repos" : [
+ "repos": [
{
+
"name" : "glslang",
"url" : "https://github.com/KhronosGroup/glslang.git",
"sub_dir" : "glslang",
@@ -10,7 +11,7 @@
"prebuild" : [
"python update_glslang_sources.py"
],
- "cmake_options" : [
+ "cmake_options": [
"-DUSE_CCACHE=ON"
]
},
@@ -28,12 +29,22 @@
"sub_dir": "SPIRV-Headers",
@@ -29,11 +29,20 @@
"build_dir": "SPIRV-Headers/build",
"install_dir": "SPIRV-Headers/build/install",
+
"commit": "f027d53ded7e230e008d37c8b47ede7cd308e19d"
+ },
+ {
@@ -3710,137 +3581,21 @@ index 9a53e452..b4fef368 100755
+ "url": "https://github.com/KhronosGroup/SPIRV-Cross.git",
+ "sub_dir": "spirv-cross",
+ "build_dir": "spirv-cross/build",
+ "install_dir": "spirv-cross/build/install",
+ "install_dir": "spirv-cross/build/install",
+ "commit": "e50f7d1ce8e162d0c826e84168cfa234e4de4ec9"
}
],
"install_names" : {
"glslang" : "GLSLANG_INSTALL_DIR",
"Vulkan-Headers" : "VULKAN_HEADERS_INSTALL_DIR",
- "install_names" : {
- "glslang" : "GLSLANG_INSTALL_DIR",
- "Vulkan-Headers" : "VULKAN_HEADERS_INSTALL_DIR",
- "SPIRV-Headers" : "SPIRV_HEADERS_INSTALL_DIR"
+ "install_names": {
+ "glslang": "GLSLANG_INSTALL_DIR",
+ "Vulkan-Headers": "VULKAN_HEADERS_INSTALL_DIR",
+ "SPIRV-Headers": "SPIRV_HEADERS_INSTALL_DIR",
+ "SPIRV-Cross" : "SPIRV_CROSS_INSTALL_DIR"
+ "SPIRV-Cross": "SPIRV_CROSS_INSTALL_DIR"
}
}
diff --git a/scripts/layer_chassis_generator.py b/scripts/layer_chassis_generator.py
index 8c4a4c4d..f53055c7 100644
--- a/scripts/layer_chassis_generator.py
+++ b/scripts/layer_chassis_generator.py
@@ -241,6 +241,13 @@ class LayerChassisOutputGenerator(OutputGenerator):
#include "vk_safe_struct.h"
#include "vk_typemap_helper.h"
+// Define here as a placeholder during development.
+#define VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_SIMT_EFFICIENCY_EXT (VkValidationFeatureEnableEXT)5
+#define VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_EXECUTION_TRACE_EXT (VkValidationFeatureEnableEXT)7
+#define VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_TRACE_RAY_TRACE_EXT (VkValidationFeatureEnableEXT)8
+#define VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DIVERGENCE_CHARACTERIZATION_EXT (VkValidationFeatureEnableEXT)9
+#define VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_WARP_ENTRY_AND_EXIT_EXT (VkValidationFeatureEnableEXT)10
+#define VK_VALIDATION_FEATURE_ENABLE_AUTO_INST_DYN_SHADER_TRACE_EXT (VkValidationFeatureEnableEXT)11
extern std::atomic<uint64_t> global_unique_id;
@@ -343,6 +350,12 @@ typedef enum EnableFlags {
vendor_specific_arm,
debug_printf,
sync_validation,
+ auto_inst_simt_efficiency,
+ auto_inst_execution_trace,
+ auto_inst_dyn_trace_ray_trace,
+ auto_inst_divergence_characterization,
+ auto_inst_warp_entry_and_exit,
+ auto_inst_dyn_shader_trace,
// Insert new enables above this line
kMaxEnableFlags,
} EnableFlags;
@@ -685,6 +698,12 @@ bool wrap_handles = true;
#include "gpu_validation.h"
#include "object_lifetime_validation.h"
#include "debug_printf.h"
+#include "auto_inst_dyn_shader_trace.h"
+#include "auto_inst_dyn_trace_ray_trace.h"
+#include "auto_inst_execution_trace.h"
+#include "auto_inst_simt_efficiency.h"
+#include "auto_inst_divergence_characterization.h"
+#include "auto_inst_warp_entry_and_exit.h"
#include "stateless_validation.h"
#include "synchronization_validation.h"
#include "thread_safety.h"
@@ -941,6 +960,24 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
auto sync_validation_obj = new SyncValidator;
sync_validation_obj->RegisterValidationObject(local_enables[sync_validation], api_version, report_data, local_object_dispatch);
+ auto auto_inst_simt_efficiency_obj = new AutoInstSimtEfficiency;
+ auto_inst_simt_efficiency_obj->RegisterValidationObject(local_enables[auto_inst_simt_efficiency], api_version, report_data, local_object_dispatch);
+
+ auto auto_inst_execution_trace_obj = new AutoInstExecutionTrace;
+ auto_inst_execution_trace_obj->RegisterValidationObject(local_enables[auto_inst_execution_trace], api_version, report_data, local_object_dispatch);
+
+ auto auto_inst_dyn_trace_ray_trace_obj = new AutoInstDynTraceRayTrace;
+ auto_inst_dyn_trace_ray_trace_obj->RegisterValidationObject(local_enables[auto_inst_dyn_trace_ray_trace], api_version, report_data, local_object_dispatch);
+
+ auto auto_inst_divergence_characterization_obj = new AutoInstDivergenceCharacterization;
+ auto_inst_divergence_characterization_obj->RegisterValidationObject(local_enables[auto_inst_divergence_characterization], api_version, report_data, local_object_dispatch);
+
+ auto auto_inst_warp_entry_and_exit_obj = new AutoInstWarpEntryAndExit;
+ auto_inst_warp_entry_and_exit_obj->RegisterValidationObject(local_enables[auto_inst_warp_entry_and_exit], api_version, report_data, local_object_dispatch);
+
+ auto auto_inst_dyn_shader_trace_obj = new AutoInstDynShaderTrace;
+ auto_inst_dyn_shader_trace_obj->RegisterValidationObject(local_enables[auto_inst_dyn_shader_trace], api_version, report_data, local_object_dispatch);
+
// If handle wrapping is disabled via the ValidationFeatures extension, override build flag
if (local_disables[handle_wrapping]) {
wrap_handles = false;
@@ -986,7 +1023,13 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateInstance(const VkInstanceCreateInfo *pCreat
gpu_assisted_obj->FinalizeInstanceValidationObject(framework);
debug_printf_obj->FinalizeInstanceValidationObject(framework);
sync_validation_obj->FinalizeInstanceValidationObject(framework);
-
+ auto_inst_simt_efficiency_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_execution_trace_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_dyn_trace_ray_trace_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_divergence_characterization_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_warp_entry_and_exit_obj->FinalizeInstanceValidationObject(framework);
+ auto_inst_dyn_shader_trace_obj->FinalizeInstanceValidationObject(framework);
+
for (auto intercept : framework->object_dispatch) {
auto lock = intercept->write_lock();
intercept->PostCallRecordCreateInstance(pCreateInfo, pAllocator, pInstance, result);
@@ -1132,11 +1175,30 @@ VKAPI_ATTR VkResult VKAPI_CALL CreateDevice(VkPhysicalDevice gpu, const VkDevice
auto sync_validation_obj = new SyncValidator;
sync_validation_obj->InitDeviceValidationObject(enables[sync_validation], instance_interceptor, device_interceptor);
+ auto auto_inst_simt_efficiency_obj = new AutoInstSimtEfficiency;
+ auto_inst_simt_efficiency_obj->InitDeviceValidationObject(enables[auto_inst_simt_efficiency], instance_interceptor, device_interceptor);
+
+ auto auto_inst_execution_trace_obj = new AutoInstExecutionTrace;
+ auto_inst_execution_trace_obj->InitDeviceValidationObject(enables[auto_inst_execution_trace], instance_interceptor, device_interceptor);
+
+ auto auto_inst_dyn_trace_ray_trace_obj = new AutoInstDynTraceRayTrace;
+ auto_inst_dyn_trace_ray_trace_obj->InitDeviceValidationObject(enables[auto_inst_dyn_trace_ray_trace], instance_interceptor, device_interceptor);
+
+ auto auto_inst_divergence_characterization_obj = new AutoInstDivergenceCharacterization;
+ auto_inst_divergence_characterization_obj->InitDeviceValidationObject(enables[auto_inst_divergence_characterization], instance_interceptor, device_interceptor);
+
+ auto auto_inst_warp_entry_and_exit_obj = new AutoInstWarpEntryAndExit;
+ auto_inst_warp_entry_and_exit_obj->InitDeviceValidationObject(enables[auto_inst_warp_entry_and_exit], instance_interceptor, device_interceptor);
+
+ auto auto_inst_dyn_shader_trace_obj = new AutoInstDynShaderTrace;
+ auto_inst_dyn_shader_trace_obj->InitDeviceValidationObject(enables[auto_inst_dyn_shader_trace], instance_interceptor, device_interceptor);
+
// Delete unused validation objects to avoid memory leak.
- std::vector<ValidationObject *> local_objs = {
- thread_safety_obj, stateless_validation_obj, object_tracker_obj,
- core_checks_obj, best_practices_obj, gpu_assisted_obj, debug_printf_obj,
- sync_validation_obj,
+ std::vector<ValidationObject*> local_objs = {
+ thread_checker_obj, object_tracker_obj, parameter_validation_obj,
+ core_checks_obj, best_practices_obj, gpu_assisted_obj, debug_printf_obj, sync_validation_obj,
+ auto_inst_simt_efficiency_obj, auto_inst_execution_trace_obj, auto_inst_dyn_trace_ray_trace_obj,
+ auto_inst_divergence_characterization_obj, auto_inst_warp_entry_and_exit_obj, auto_inst_dyn_shader_trace_obj
};
for (auto obj : local_objs) {
if (std::find(device_interceptor->object_dispatch.begin(), device_interceptor->object_dispatch.end(), obj) ==
--
2.29.2.windows.2


Write Preview
Loading…
Cancel
Save