Update for 10.5-GA (#4185)

* Updated submodule onnx to 10.5

Signed-off-by: Michal Guzek <[email protected]>

* Updated files to 10.5

Signed-off-by: Michal Guzek <[email protected]>

* Restore the previous .gitmodules

Signed-off-by: Michal Guzek <[email protected]>

* .gitmodules new line

Signed-off-by: Michal Guzek <[email protected]>

* Update public docker files

Signed-off-by: Michal Guzek <[email protected]>

* Update 10.5 oss deleted files

Signed-off-by: Michal Guzek <[email protected]>

* Update deprecated APIs

Signed-off-by: Michal Guzek <[email protected]>

---------

Signed-off-by: Michal Guzek <[email protected]>

Author: moraxu

CHANGELOG.md

@@ -1,6 +1,30 @@
 # TensorRT OSS Release Changelog
 
-## 10.4.0 GA - 2024-09-11
+
+## 10.5.0 GA - 2024-09-30
+Key Features and Updates:
+
+- Demo changes
+    - Added [Flux.1-dev](demo/Diffusion) pipeline
+- Sample changes
+    - None
+- Plugin changes
+    - Migrated `IPluginV2`-descendent versions of `bertQKVToContextPlugin` (1, 2, 3) to newer versions (4, 5, 6 respectively) which implement `IPluginV3`.
+    - Note:
+        - The newer versions preserve the attributes and I/O of the corresponding older plugin version
+        - The older plugin versions are deprecated and will be removed in a future release
+- Quickstart guide
+    - None
+- Parser changes
+    - Added support for real-valued `STFT` operations
+    - Improved error handling in `IParser`
+
+Known issues:
+
+- Demos:
+    - TensorRT engine might not be build successfully when using `--fp8` flag on H100 GPUs.
+
+## 10.4.0 GA - 2024-09-18
 Key Features and Updates:
 
 - Demo changes

CMakeLists.txt

@@ -158,7 +158,11 @@ endif()
 
 find_library_create_target(nvinfer ${nvinfer_lib_name} SHARED ${TRT_LIB_DIR})
 
-find_library(CUDART_LIB cudart_static HINTS ${CUDA_TOOLKIT_ROOT_DIR} PATH_SUFFIXES lib lib/x64 lib64)
+if (DEFINED USE_CUGFX)
+    find_library(CUDART_LIB cugfx_dll HINTS ${CUDA_TOOLKIT_ROOT_DIR} PATH_SUFFIXES lib lib/x64 lib64)
+else()
+    find_library(CUDART_LIB cudart_static HINTS ${CUDA_TOOLKIT_ROOT_DIR} PATH_SUFFIXES lib lib/x64 lib64)
+endif()
 
 if (NOT MSVC)
     find_library(RT_LIB rt)
@@ -174,7 +178,6 @@ if (DEFINED GPU_ARCHS)
   separate_arguments(GPU_ARCHS)
 else()
   list(APPEND GPU_ARCHS
-      70
       75
     )
 
@@ -202,7 +205,7 @@ endif()
 set(BERT_GENCODES)
 # Generate SASS for each architecture
 foreach(arch ${GPU_ARCHS})
-    if (${arch} GREATER_EQUAL 70)
+    if (${arch} GREATER_EQUAL 75)
         set(BERT_GENCODES "${BERT_GENCODES} -gencode arch=compute_${arch},code=sm_${arch}")
     endif()
     set(GENCODES "${GENCODES} -gencode arch=compute_${arch},code=sm_${arch}")
@@ -211,7 +214,7 @@ endforeach()
 # Generate PTX for the last architecture in the list.
 list(GET GPU_ARCHS -1 LATEST_SM)
 set(GENCODES "${GENCODES} -gencode arch=compute_${LATEST_SM},code=compute_${LATEST_SM}")
-if (${LATEST_SM} GREATER_EQUAL 70)
+if (${LATEST_SM} GREATER_EQUAL 75)
     set(BERT_GENCODES "${BERT_GENCODES} -gencode arch=compute_${LATEST_SM},code=compute_${LATEST_SM}")
 endif()
 

README.md

@@ -26,7 +26,7 @@ You can skip the **Build** section to enjoy TensorRT with Python.
 To build the TensorRT-OSS components, you will first need the following software packages.
 
 **TensorRT GA build**
-* TensorRT v10.4.0.26
+* TensorRT v10.5.0.18
   * Available from direct download links listed below
 
 **System Packages**
@@ -73,25 +73,25 @@ To build the TensorRT-OSS components, you will first need the following software
     If using the TensorRT OSS build container, TensorRT libraries are preinstalled under `/usr/lib/x86_64-linux-gnu` and you may skip this step.
 
     Else download and extract the TensorRT GA build from [NVIDIA Developer Zone](https://developer.nvidia.com) with the direct links below:
-      - [TensorRT 10.4.0.26 for CUDA 11.8, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.4.0/tars/TensorRT-10.4.0.26.Linux.x86_64-gnu.cuda-11.8.tar.gz)
-      - [TensorRT 10.4.0.26 for CUDA 12.6, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.4.0/tars/TensorRT-10.4.0.26.Linux.x86_64-gnu.cuda-12.6.tar.gz)
-      - [TensorRT 10.4.0.26 for CUDA 11.8, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.4.0/zip/TensorRT-10.4.0.26.Windows.win10.cuda-11.8.zip)
-      - [TensorRT 10.4.0.26 for CUDA 12.6, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.4.0/zip/TensorRT-10.4.0.26.Windows.win10.cuda-12.6.zip)
+      - [TensorRT 10.5.0.18 for CUDA 11.8, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.5.0/tars/TensorRT-10.5.0.18.Linux.x86_64-gnu.cuda-11.8.tar.gz)
+      - [TensorRT 10.5.0.18 for CUDA 12.6, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.5.0/tars/TensorRT-10.5.0.18.Linux.x86_64-gnu.cuda-12.6.tar.gz)
+      - [TensorRT 10.5.0.18 for CUDA 11.8, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.5.0/zip/TensorRT-10.5.0.18.Windows.win10.cuda-11.8.zip)
+      - [TensorRT 10.5.0.18 for CUDA 12.6, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.5.0/zip/TensorRT-10.5.0.18.Windows.win10.cuda-12.6.zip)
 
 
     **Example: Ubuntu 20.04 on x86-64 with cuda-12.6**
 
     ```bash
     cd ~/Downloads
-    tar -xvzf TensorRT-10.4.0.26.Linux.x86_64-gnu.cuda-12.6.tar.gz
-    export TRT_LIBPATH=`pwd`/TensorRT-10.4.0.26
+    tar -xvzf TensorRT-10.5.0.18.Linux.x86_64-gnu.cuda-12.6.tar.gz
+    export TRT_LIBPATH=`pwd`/TensorRT-10.5.0.18
     ```
 
     **Example: Windows on x86-64 with cuda-12.6**
 
     ```powershell
-    Expand-Archive -Path TensorRT-10.4.0.26.Windows.win10.cuda-12.6.zip
-    $env:TRT_LIBPATH="$pwd\TensorRT-10.4.0.26\lib"
+    Expand-Archive -Path TensorRT-10.5.0.18.Windows.win10.cuda-12.6.zip
+    $env:TRT_LIBPATH="$pwd\TensorRT-10.5.0.18\lib"
     ```
 
 ## Setting Up The Build Environment

VERSION

@@ -1 +1 @@
-10.4.0.26
+10.5.0.18