@@ -764,7 +764,7 @@ HIP runtime has the following functional improvements which improves runtime per
#### Changed
* VX_RPP extension: Version 3.1.0 release.
* VX_RPP extension: Version 3.1.0 release.
* Update the parameters and kernel API of Blur, Fog, Jitter, LensCorrection, Rain, Pixelate, Vignette and ResizeCrop wrt tensor kernels replacing the legacy BatchPD API calls in VX_RPP extensions.
#### Known issues
@@ -1144,7 +1144,7 @@ Review the [README](https://github.com/ROCm/rocm_bandwidth_test/blob/amd-mainlin
#### Changed
* Completed migration from legacy [ROCProfiler](https://rocm.docs.amd.com/projects/rocprofiler/en/latest/) to [ROCprofiler-SDK](https://rocm.docs.amd.com/projects/rocprofiler-sdk/en/latest/).
* Reorganized the configuration files internally and improved [README/installation](https://github.com/ROCm/rdc/blob/amd-staging/README.md) instructions.
* Reorganized the configuration files internally and improved [README/installation](https://github.com/ROCm/rdc/blob/release/rocm-rel-7.0/README.md) instructions.
* Updated metrics and monitoring support for the latest AMD data center GPUs.
#### Optimized
@@ -1346,7 +1346,7 @@ The previous default accumulator types could lead to situations in which unexpec
@@ -42,7 +42,7 @@ The following are notable new features and improvements in ROCm 7.0.0. For chang
### Operating system, hardware, and virtualization support changes
ROCm 7.0.0 adds support for [AMD Instinct MI355X](https://www.amd.com/en/products/accelerators/instinct/mi350/mi355x.html) and [MI350X](https://www.amd.com/en/products/accelerators/instinct/mi350/mi350x.html). For details, see the full list of [Supported GPUs (Linux)](https://rocm.docs.amd.com/projects/install-on-linux-internal/en/latest/reference/system-requirements.html#supported-gpus).
ROCm 7.0.0 adds support for [AMD Instinct MI355X](https://www.amd.com/en/products/accelerators/instinct/mi350/mi355x.html) and [MI350X](https://www.amd.com/en/products/accelerators/instinct/mi350/mi350x.html). For details, see the full list of [Supported GPUs (Linux)](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/reference/system-requirements.html#supported-gpus).
ROCm 7.0.0 adds support for the following operating systems and kernel versions:
@@ -51,7 +51,7 @@ ROCm 7.0.0 adds support for the following operating systems and kernel versions:
ROCm 7.0.0 marks the end of support (EoS) for Ubuntu 24.04.2 (kernel: 6.8 [GA], 6.11 [HWE]) and SLES 15 SP6.
For more information about supported operating systems, see [Supported operating systems](https://rocm.docs.amd.com/projects/install-on-linux-internal/en/latest/reference/system-requirements.html#supported-operating-systems) and [install instructions](https://rocm.docs.amd.com/projects/install-on-linux-internal/en/latest/).
For more information about supported operating systems, see [Supported operating systems](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/reference/system-requirements.html#supported-operating-systems) and [install instructions](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/).
@@ -61,7 +61,7 @@ for more information about operating system and hardware compatibility.
ROCm 7.0.0 introduces support for KVM Passthrough for AMD Instinct MI350X and MI355X GPUs.
All KVM-based SR-IOV supported configurations require the GIM SR-IOV driver version 8.4.0.K. Refer to [GIM Release note](https://github.com/amd/MxGPU-Virtualization/releases) for more details. In addition, support for VMware ESXi 8 has been introduced for AMD Instinct MI300X GPUs. For more information, see [Virtualization Support](https://rocm.docs.amd.com/projects/install-on-linux-internal/en/latest/reference/system-requirements.html#virtualization-support).
All KVM-based SR-IOV supported configurations require the GIM SR-IOV driver version 8.4.0.K. Refer to [GIM Release note](https://github.com/amd/MxGPU-Virtualization/releases) for more details. In addition, support for VMware ESXi 8 has been introduced for AMD Instinct MI300X GPUs. For more information, see [Virtualization Support](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/reference/system-requirements.html#virtualization-support).
### Deep learning and AI framework updates
@@ -117,14 +117,14 @@ ROCm 7.0.0 enables support for Triton 3.3.0.
AMD ROCm has officially added support for the following Deep learning and AI frameworks:
* Ray is a unified framework for scaling AI and Python applications from your laptop to a full cluster, without changing your code. Ray consists of a core distributed runtime and a set of AI libraries for simplifying machine learning computations. It is currently supported on ROCm 6.4.1. For more information, see [Ray compatibility](https://advanced-micro-devices-rocm-internal--500.com.readthedocs.build/en/500/compatibility/ml-compatibility/ray-compatibility.html).
* Ray is a unified framework for scaling AI and Python applications from your laptop to a full cluster, without changing your code. Ray consists of a core distributed runtime and a set of AI libraries for simplifying machine learning computations. It is currently supported on ROCm 6.4.1. For more information, see [Ray compatibility](https://rocm.docs.amd.com/en/latest/compatibility/ml-compatibility/ray-compatibility.html).
* llama.cpp is an open-source framework for Large Language Model (LLM) inference that runs on both central processing units (CPUs) and graphics processing units (GPUs). It is written in plain C/C++, providing a simple, dependency-free setup. It is currently supported on ROCm 6.4.0. For more information, see [llama.cpp compatibility](https://advanced-micro-devices-rocm-internal--500.com.readthedocs.build/en/500/compatibility/ml-compatibility/llama-cpp-compatibility.html).
* llama.cpp is an open-source framework for Large Language Model (LLM) inference that runs on both central processing units (CPUs) and graphics processing units (GPUs). It is written in plain C/C++, providing a simple, dependency-free setup. It is currently supported on ROCm 6.4.0. For more information, see [llama.cpp compatibility](https://rocm.docs.amd.com/en/latest/compatibility/ml-compatibility/llama-cpp-compatibility.html).
### AMD GPU Driver/ROCm packaging separation
The AMD GPU Driver (amdgpu) is now distributed separately from the ROCm software stack and is stored under in its own location ``/amdgpu/`` in the package repository at [repo.radeon.com](https://repo.radeon.com/amdgpu/). The first release is designated as AMD GPU Driver (amdgpu) version 30.10. See the [User and kernel-space support matrix](https://rocm.docs.amd.com/projects/install-on-linux-internal/en/latest/reference/user-kernel-space-compat-matrix.html) for more information.
The AMD GPU Driver (amdgpu) is now distributed separately from the ROCm software stack and is stored under in its own location ``/amdgpu/`` in the package repository at [repo.radeon.com](https://repo.radeon.com/amdgpu/). The first release is designated as AMD GPU Driver (amdgpu) version 30.10. See the [User and kernel-space support matrix](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/reference/user-kernel-space-compat-matrix.html) for more information.
[AMD SMI](https://github.com/ROCm/amdsmi) continues to stay with the ROCm software stack under the ROCm organization repository.
@@ -139,7 +139,7 @@ Other ROCm libraries are also in the process of migration along with ROCm tools
### HIP API compatibility improvements
To improve code portability between AMD ROCm and other programming models, HIP API has been updated in ROCm 7.0.0 to simplify cross-platform programming. These changes are incompatible with prior ROCm releases and might require recompiling existing HIP applications for use with ROCm 7.0.0. For more information, see the [HIP API 7.0.0 changes](https://rocm.docs.amd.com/projects/HIP/en/docs-develop/hip-7-changes.html) and the [HIP changelog](#hip-7-0-0) below.
To improve code portability between AMD ROCm and other programming models, HIP API has been updated in ROCm 7.0.0 to simplify cross-platform programming. These changes are incompatible with prior ROCm releases and might require recompiling existing HIP applications for use with ROCm 7.0.0. For more information, see the [HIP API 7.0.0 changes](https://rocm.docs.amd.com/projects/HIP/en/docs-7.0.0/hip-7-changes.html) and the [HIP changelog](#hip-7-0-0) below.
* Added support for an in-memory virtual file system (VFS) for storing temporary files generated during intermediate compilation steps. This is designed to improve performance by reducing on-disk file I/O. Currently, VFS is supported only for the device library link step, with plans for expanded support in future releases.
* Added a new target-specific builtin ``__builtin_amdgcn_processor_is`` for late or deferred queries of the current target processor.
* Added a new target-specific builtin ``__builtin_amdgcn_is_invocable``, enabling fine-grained, per-builtin feature availability.
* The compiler driver now uses parallel code generation by default when compiling using full LTO (including when using the `-fgpu-rdc` option) for HIP. This divides the optimized LLVM IR module into roughly equal partitions before instruction selection and lowering, which can help improve build times.
@@ -247,12 +247,12 @@ See the [ROCgdb changelog](#rocgdb-16-3) for more details.
ROCm Compute Profiler includes the following key changes:
* Interactive command line with a Textual User Interface (TUI) has been added to analyze mode. For more details, see [TUI analysis](https://rocm.docs.amd.com/projects/rocprofiler-compute/en/amd-staging/how-to/analyze/tui.html).
* Interactive command line with a Textual User Interface (TUI) has been added to analyze mode. For more details, see [TUI analysis](https://rocm.docs.amd.com/projects/rocprofiler-compute/en/docs-7.0.0/how-to/analyze/tui.html).
* Support added for advanced data types: `FP4` and `FP6`
* Support for AMD Instinct MI355X and MI350X with addition of performance counters: CPC, SPI, SQ, TA/TD/TCP, and TCC.
* Roofline enhancement added for AMD Instinct MI350 Series.
* Improved support for Selective Kernel profiling.
* Program Counter (PC) sampling (Software-based) feature has been enabled for AMD Instinct MI200, MI300X, MI350X, and MI355X GPUs. This feature helps in GPU profiling to understand code execution patterns and hotspots during GPU kernel execution. For more details, see [Using PC sampling in ROCm Compute Profiler](https://rocm.docs.amd.com/projects/rocprofiler-compute/en/amd-staging/how-to/pc_sampling.html).
* Program Counter (PC) sampling (Software-based) feature has been enabled for AMD Instinct MI200, MI300X, MI350X, and MI355X GPUs. This feature helps in GPU profiling to understand code execution patterns and hotspots during GPU kernel execution. For more details, see [Using PC sampling in ROCm Compute Profiler](https://rocm.docs.amd.com/projects/rocprofiler-compute/en/docs-7.0.0/how-to/pc_sampling.html).
* Program Counter (PC) sampling (Hardware-based, Stochastic) feature has been enabled for AMD Instinct MI300X, MI350, and MI355X GPUs.
* Docker files has been added to package the application and dependencies into a single portable and executable standalone binary file.
@@ -356,7 +356,7 @@ ROCm documentation continues to be updated to provide clearer and more comprehen
In addition, the [AI agent with MCPs using vLLM and PydanticAI](https://rocm.docs.amd.com/projects/ai-developer-hub/en/latest/notebooks/inference/build_airbnb_agent_mcp.html) tutorial has been updated. For more information about the changes, see [Changelog for the AI Developer Hub](https://rocm.docs.amd.com/projects/ai-developer-hub/en/latest/changelog.html).
* Documentation for [rocCV](https://advanced-micro-devices-roccv--28.com.readthedocs.build/en/28/), an efficient GPU-accelerated library for image pre- and post-processing, has been added. rocCV is in an early access state, and using it on production workloads is not recommended.
* Documentation for [rocCV](https://rocm.docs.amd.com/projects/rocCV/en/latest/index.html), an efficient GPU-accelerated library for image pre- and post-processing, has been added. rocCV is in an early access state, and using it on production workloads is not recommended.
* ROCm Math libraries support a wide range of data types, enabling optimized performance across various precision requirements. The following Math libraries are now updated with new precision content. For more information, click the Math library’s link:
@@ -530,7 +530,7 @@ For more information on these features, see [AMD SMI changelog](https://github.c
#### KVM SR-IOV virtualization changes dependent on open source AMD GPU Virtualization Driver (GIM)
KVM SR-IOV support for all Instinct GPUs require the open source AMD GPU Virtualization Driver (GIM) 8.4.0.K. For detailed support information, see [virtualization support](https://rocm.docs.amd.com/projects/install-on-linux-internal/en/latest/reference/system-requirements.html#virtualization-support) and [GIM Release Note](https://github.com/amd/MxGPU-Virtualization/releases).
KVM SR-IOV support for all Instinct GPUs require the open source AMD GPU Virtualization Driver (GIM) 8.4.0.K. For detailed support information, see [virtualization support](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/reference/system-requirements.html#virtualization-support) and [GIM Release Note](https://github.com/amd/MxGPU-Virtualization/releases).
#### GPU partitioning support for AMD Instinct MI355X and MI350X GPUs
@@ -562,47 +562,47 @@ Click {fab}`github` to go to the component's source code on GitHub.
<tr>
<th rowspan="9">Libraries</th>
<th rowspan="9">Machine learning and computer vision</th>
@@ -1629,7 +1629,7 @@ HIP runtime has the following functional improvements which improves runtime per
#### Changed
* VX_RPP extension: Version 3.1.0 release.
* VX_RPP extension: Version 3.1.0 release.
* Update the parameters and kernel API of Blur, Fog, Jitter, LensCorrection, Rain, Pixelate, Vignette and ResizeCrop wrt tensor kernels replacing the legacy BatchPD API calls in VX_RPP extensions.
#### Known issues
@@ -2009,7 +2009,7 @@ Review the [README](https://github.com/ROCm/rocm_bandwidth_test/blob/amd-mainlin
#### Changed
* Completed migration from legacy [ROCProfiler](https://rocm.docs.amd.com/projects/rocprofiler/en/latest/) to [ROCprofiler-SDK](https://rocm.docs.amd.com/projects/rocprofiler-sdk/en/latest/).
* Reorganized the configuration files internally and improved [README/installation](https://github.com/ROCm/rdc/blob/amd-staging/README.md) instructions.
* Reorganized the configuration files internally and improved [README/installation](https://github.com/ROCm/rdc/blob/release/rocm-rel-7.0/README.md) instructions.
* Updated metrics and monitoring support for the latest AMD data center GPUs.
#### Optimized
@@ -2211,7 +2211,7 @@ The previous default accumulator types could lead to situations in which unexpec
* `warpSize` is only available as a non-`constexpr` variable. Where required,
the wavefront size should be queried via the `warpSize` variable in device code,
or via `hipGetDeviceProperties` in host code. Neither of these will result in a compile-time constant. For more information, see [warpSize](https://rocm.docs.amd.com/projects/HIP/en/docs-6.4.3/how-to/hip_cpp_language_extensions.html#warpsize).
or via `hipGetDeviceProperties` in host code. Neither of these will result in a compile-time constant. For more information, see [warpSize](https://rocm.docs.amd.com/projects/HIP/en/docs-7.0.0/how-to/hip_cpp_language_extensions.html#warpsize).
* For cases where compile-time evaluation of the wavefront size cannot be avoided,
uses of `__AMDGCN_WAVEFRONT_SIZE`, `__AMDGCN_WAVEFRONT_SIZE__`, or `warpSize`
can be replaced with a user-defined macro or `constexpr` variable with the wavefront
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
