ROCm/RELEASE.md

ROCm Core SDK 7.11.0 release notes

ROCm Core SDK 7.11.0 continues the technology preview release stream that began with ROCm 7.9.0, advancing the transition to the new TheRock build and release system. To learn more about TheRock, see ROCm Core SDK and TheRock Build System.

This release expands AMD GPU and APU support and adds more components to the ROCm Core SDK. Developers can expect a more consistent build experience and streamlined workflows that pave the way toward modular future ROCm releases planned for mid-2026.

(preview-stream-note)= :::{important} ROCm 7.11.0 follows the versioning discontinuity that began with 7.9.0 and remains separate from the 7.0 to 7.2 production releases. For the latest production stream release, see the ROCm documentation.

Maintaining parallel release streams -- preview and production -- gives users ample time to evaluate and adopt the new build system and dependency changes. The technology preview stream is planned to continue through mid‑2026, after which it will replace the current production stream. :::

Release highlights

ROCm Core SDK 7.11.0 with TheRock builds on the 7.10.0 release with several key enhancements:

Expanded AMD GPU support

The ROCm 7.11.0 preview adds support for RDNA 4 and other GPUs:

• Radeon AI PRO R9700

• Radeon AI PRO R9600D

• Radeon PRO W7900 Dual Slot

• Radeon PRO V710

• Radeon RX 9070 XT

• Radeon RX 9070 GRE

• Radeon RX 9070

• Radeon RX 9060 XT LP

• Radeon RX 9060 XT

• Radeon RX 9060

• Radeon RX 7700

For the full list of supported hardware, see Hardware support.

Expanded Linux distribution support

ROCm 7.11.0 adds support for the following Linux distributions on AMD Instinct and Radeon GPUs.

• Debian 13

• Red Hat Enterprise Linux (RHEL) 9.4, 8.10

• SUSE Linux Enterprise Server (SLES) 16.0

• Oracle Linux 10, 9, 8

• Rocky Linux 9

For the full list of supported Linux distributions, see Operating system support.

Added Linux package manager support

ROCm Core SDK packages are now available in .deb and .rpm formats and can be installed using your Linux distro's package manager. See the {doc}installation instructions </install/rocm> for details.

Added GPU virtualization support for Instinct GPUs

ROCm 7.11.0 supports GPU virtualization on AMD Instinct MI355X, MI350X, MI325X, and MI300X GPUs. For details, see GPU virtualization support.

Added debugging tools to the ROCm Core SDK

ROCm 7.11.0 adds the following debugging tools to the ROCm Core SDK on Linux:

• ROCdbgapi

• ROCm Debugger (ROCgdb)

• ROCr Debug Agent

For the full list of core components, see ROCm Core SDK components.

Compatibility notices

In terms of package compatibility, ROCm 7.11.0 diverges from the existing ROCm 7.0 production stream and future stable releases in that stream:

• Compute-focused: ROCm 7.11.0 enables support for primarily compute workloads. Future releases will support mixed workloads (compute and graphics).

  If you’re interested in testing AMD Radeon GPUs with preview support for graphics use cases with AMD ROCm 7.11.0, install Radeon Software for Linux version 25.35 from Linux Drivers for AMD Radeon and Radeon PRO Graphics.

  If you're interested in testing AMD Ryzen APUs with preview support for graphics use cases with AMD ROCm 7.11.0, use the inbox graphics drivers of Ubuntu 24.04.3.

• No upgrade path from existing production releases including ROCm 7.2.0 and earlier, as well as from upcoming stable releases. See the explanatory note.

• Not intended for production workloads: users running production environments should continue using the ROCm 7.0 stream. See the explanatory note.

• Not fully featured: this release is a stepping stone toward fully open software development.

• Limited hardware support: preview releases are only supported on some AMD Instinct GPUs, Radeon GPUs, and Ryzen APUs. See Supported hardware and operating systems.

• Software components: additional components are planned to be introduced in future preview releases as part of the ROCm Core SDK. Other libraries and tools not included in the future Core SDK will either be:

  • Released as standalone project-specific packages, or
  • Grouped into domain-specific toolkits.

(release-supported-hw)=

Hardware support

The following table lists supported AMD Instinct GPUs, Radeon GPUs, and Ryzen AI APUs. Each supported device is listed with its corresponding GPU microarchitecture and LLVM target.

::::{tab-set} :::{tab-item} Instinct :sync: instinct

AMD device series	Device	LLVM target	Architecture
Instinct MI350 Series	Instinct MI355X Instinct MI350X	gfx950	CDNA 4
Instinct MI300 Series	Instinct MI325X Instinct MI300X Instinct MI300A	gfx942	CDNA 3
Instinct MI200 Series	Instinct MI250X Instinct MI250 Instinct MI210	gfx90a	CDNA 2

:::

:::{tab-item} Radeon PRO :sync: radeon-pro

AMD device series	Device	LLVM target	Architecture
Radeon AI PRO R9000 Series	Radeon AI PRO R9700 Radeon AI PRO R9600D	gfx1201	RDNA 4
Radeon PRO W7000 Series	Radeon PRO W7900 Dual Slot Radeon PRO W7900 Radeon PRO W7800 48GB Radeon PRO W7800	gfx1100	RDNA 3
	Radeon PRO W7700	gfx1101
Radeon PRO V Series	Radeon PRO V710

:::

:::{tab-item} Radeon :sync: radeon

AMD device series	Device	LLVM target	Architecture
Radeon RX 9000 Series	Radeon RX 9070 XT Radeon RX 9070 GRE Radeon RX 9070	gfx1201	RDNA 4
	Radeon RX 9060 XT LP Radeon RX 9060 XT Radeon RX 9060	gfx1200
Radeon RX 7000 Series	Radeon RX 7900 XTX Radeon RX 7900 XT Radeon RX 7900 GRE	gfx1100	RDNA 3
	Radeon RX 7800 XT Radeon RX 7700 XT Radeon RX 7700	gfx1101

:::

:::{tab-item} Ryzen AI :sync: ryzen

AMD device series	Device	LLVM target	Architecture
Ryzen AI Max PRO 300 Series	Ryzen AI Max+ PRO 395 Ryzen AI Max PRO 390 Ryzen AI Max PRO 385 Ryzen AI Max PRO 380	gfx1151	RDNA 3.5
Ryzen AI Max 300 Series	Ryzen AI Max+ 395 Ryzen AI Max 390 Ryzen AI Max 385
Ryzen AI 300 Series	Ryzen AI 9 HX 375 Ryzen AI 9 HX 370 Ryzen AI 9 365	gfx1150

::: ::::

This preview release supports a limited number of GPU and APUs. Hardware
support will be expanded with future releases, following a six-week release
cadence.

(release-supported-os)=

Operating system support

ROCm supports the following Linux distribution and Microsoft Windows versions. If you're running ROCm on Linux, ensure your system is using a supported kernel version. Future preview releases will expand operating system support coverage.

:::{important} The following table is a general overview of supported OSes. Actual support might vary by GPU. Use the {doc}Compatibility matrix </compatibility/compatibility-matrix> to verify support for your specific setup before installation. :::

::::{tab-set} :::{tab-item} Instinct :sync: instinct

Linux distribution	Supported versions	Linux kernel version
Ubuntu	24.04.3	GA 6.8
	22.04.5	GA 5.15
Debian	13	6.12
Red Hat Enterprise Linux (RHEL)	10.1	6.12.0-124
	10.0	6.12.0-55
	9.7	5.14.0-611
	9.6	5.14.0-570
	9.4	5.14.0-427
	8.10	4.18.0-553
Oracle Linux	10.1	UEK 8.1
	9.6	UEK 8
	8.10	UEK 7
Rocky Linux	9	5.14.0-570
SUSE Linux Enterprise Server (SLES)	16.0	6.12
	15.7	6.4.0-150700.51

:::

:::{tab-item} Radeon PRO :sync: radeon-pro

Operating system	Supported versions	Linux kernel version
Ubuntu	24.04.3	GA 6.8
	22.04.5	GA 5.15
Red Hat Enterprise Linux (RHEL)	10.1	6.12.0-124
	9.7	5.14.0-611
Windows	11 25H2	—

:::

:::{tab-item} Radeon :sync: radeon

Operating system	Supported versions	Linux kernel version
Ubuntu	24.04.3	GA 6.8
	22.04.5	GA 5.15
Red Hat Enterprise Linux (RHEL)	10.1	6.12.0-124
	9.7	5.14.0-611
Windows	11 25H2	—

:::

:::{tab-item} Ryzen AI :sync: ryzen

Operating system	Supported versions	Linux kernel version
Ubuntu	24.04.3	HWE 6.14
Windows	11 25H2	—

::: ::::

(release-supported-fw)=

Kernel driver and firmware bundle support

ROCm requires a coordinated stack of compatible firmware, driver, and user space components. Maintaining version alignment between these layers ensures correct GPU operation and performance, especially for AMD data center products. While AMD publishes the AMD GPU driver and ROCm user space components, your server OEM or infrastructure provider distributes the firmware packages. AMD supplies those firmware images (PLDM bundles), which the OEM integrates and distributes.

::::{tab-set} :::{tab-item} Instinct :sync: instinct

AMD device	Firmware	Linux driver
Instinct MI355X	PLDM bundle 01.25.16.03	AMD GPU Driver (amdgpu) 31.10.0 30.20.1 30.20.0 30.10.2 30.10.1 30.10.0
Instinct MI350X
Instinct MI325X	PLDM bundle 01.25.04.02
Instinct MI300X	PLDM bundle 01.25.03.12
Instinct MI300A	BKC 26.1
Instinct MI250X	IFWI 75 (or later)
Instinct MI250	Maintenance update 5 with IFWI 75 (or later)
Instinct MI210

:::

:::{tab-item} Radeon PRO :sync: radeon-pro

AMD device	Linux driver	Windows driver
Radeon AI PRO R9700	AMD GPU Driver (amdgpu) 31.10.0 30.20.1 30.20.0 30.10.2 30.10.1 30.10.0	AMD Software: Adrenalin Edition 26.1.1
Radeon AI PRO R9600D		—
Radeon PRO W7900 Dual Slot		AMD Software: Adrenalin Edition 26.1.1
Radeon PRO W7900
Radeon PRO W7800 48GB
Radeon PRO W7800
Radeon PRO W7700
Radeon PRO V710		—

:::

:::{tab-item} Radeon RX :sync: radeon

AMD device	Linux driver	Windows driver
Radeon RX 9070 XT	AMD GPU Driver (amdgpu) 31.10.0 30.20.1 30.20.0 30.10.2 30.10.1 30.10.0	—
Radeon RX 9070 GRE
Radeon RX 9070
Radeon RX 9060 XT LP
Radeon RX 9060 XT
Radeon RX 9060
Radeon RX 7900 XTX		AMD Software: Adrenalin Edition 26.1.1
Radeon RX 7900 XT
Radeon RX 7900 GRE
Radeon RX 7800 XT
Radeon RX 7700 XT

:::

:::{tab-item} Ryzen AI :sync: ryzen

AMD device	Linux driver	Windows driver
Ryzen AI Max+ PRO 395	Inbox kernel driver in Ubuntu 24.04.3	AMD Software: Adrenalin Edition 26.1.1
Ryzen AI Max PRO 390
Ryzen AI Max PRO 385
Ryzen AI Max PRO 380
Ryzen AI Max+ 395
Ryzen AI Max 390
Ryzen AI Max 385
Ryzen AI 9 HX 375
Ryzen AI 9 HX 370
Ryzen AI 9 365

::: ::::

(release-virtualization-support)=

GPU virtualization support

AMD Instinct data center GPUs support virtualization in the following configurations. Supported SR-IOV configurations require the AMD GPU Virtualization Driver (GIM) 8.7.0K -- see the AMD Instinct Virtualization Driver documentation to learn more.

AMD GPU	Hypervisor	Virtualization technology	Virtualization driver	Host OS	Guest OS
Instinct MI355X	KVM	Passthrough	—	Ubuntu 24.04	Ubuntu 24.04
		SR-IOV	GIM 8.7.0K
Instinct MI350X	KVM	Passthrough	—
		SR-IOV	GIM 8.7.0K
Instinct MI325X	KVM	SR-IOV	GIM 8.7.0K	Ubuntu 22.04	Ubuntu 22.04
Instinct MI300X	KVM	Passthrough	—
		SR-IOV	GIM 8.7.0K

Deep learning framework support

ROCm 7.11.0 provides optimized support for popular deep learning frameworks. The following table lists supported frameworks, their compatible operating systems, and validated framework versions.

Framework	Supported OS	Supported framework versions
PyTorch	Linux	2.9.1, 2.8.0, 2.7.1
	Windows	2.9.1

(release-components)=

ROCm Core SDK components

The following table lists tools and libraries included in the ROCm 7.11.0 release. Expect future releases to expand the list of components.

:::{important} The following table is a general overview of ROCm Core SDK components. Actual support for these libraries and tools might vary by GPU and OS. Use the {doc}Compatibility matrix </compatibility/compatibility-matrix> to verify support for your specific setup. :::

Component group	Component name	Support
Math and compute libraries	Composable Kernel	Linux, Windows
	hipBLAS
	hipBLASLt
	hipCUB
	hipFFT
	hipRAND
	hipSOLVER
	hipSPARSE
	MIOpen
	rocBLAS
	rocFFT
	rocRAND
	rocSOLVER
	rocSPARSE
	rocPRIM
	rocThrust
	rocWMMA
	hipSPARSELt	Linux only (Instinct MI350, M300 Series, Ryzen APUs)
Communication libraries	RCCL	Linux only
Support libraries	ROCm CMake	Linux, Windows
Runtimes and compilers	HIP	Linux, Windows
	HIPIFY
	LLVM
	SPIRV-LLVM-Translator
	ROCr Runtime	Linux only
Profiling and debugging tools	ROCm Compute Profiler (rocprofiler-compute)	Linux only (Instinct)
	ROCm Systems Profiler (rocprofiler-systems)
	ROCprofiler-SDK	Linux only (Instinct, Radeon PRO, Radeon)
	ROCdbgapi
	ROCm Debugger (ROCgdb)
	ROCr Debug Agent
Control and monitoring tools	AMD SMI	Linux only (Instinct, Radeon PRO, Radeon)
	rocminfo	Linux only
System validation tools	ROCm Bandwidth Test (RBT)
	TransferBench

Known issues

The following are known issues identified in ROCm 7.11.0.

DistilBERT model performance regression on Instinct MI350 Series

The DistilBERT base model experiences reduced GPU kernel performance on Instinct MI350 Series GPUs (gfx950). This issue is under investigation.

ROCgdb GPU core dump limitation

ROCgdb currently has a limitation that preventing proper GPU core dump generation. This blocks effective root‑cause analysis of GPU faults. ROCm/rocm-systems PR #2851 introduces a fix to the ROCr runtime's core dump generation and will be included in a future release.

Clang illegal instruction error on Radeon GPUs

Using Clang with the -O0 optimization level on certain supported Radeon PRO and Radeon GPUs might trigger an illegal instructions detected error. This failure typically occurs in code paths using ockl_wfred_* functions, which handle wavefront operations and synchronization. Projects like llama.cpp are known to be affected. As a workaround, use -Og optimization level instead of -O0 for debug builds:

• Linux (CMake)

  DCMAKE_CXX_FLAGS_DEBUG="-Og -g"
  

• Windows (CMake)

  -DCMAKE_CXX_FLAGS_DEBUG="-Og -g -Xclang -gcodeview -D_DEBUG -D_DLL -D_MT -Xclang --dependent-lib=msvcrtd"
  

PyTorch model training validation issues

The following models failed validation on PyTorch for ROCm 7.11.0 due to compilation errors and other issues: Llama 3.1 8B, Llama 3.1 70B, and DeepSpeed Megatron-LM GPT2.

Apex fails to build using TheRock

Apex encounters crashes, missing module errors, and segmentation faults related to the HIP runtime during testing with the TheRock build system. This will be fixed in a future release.

PyTorch unit tests freeze on Windows

The test_cublas_config_nondeterministic_alert_cuda and test_graph_error PyTorch tests fail and hang indefinitely on Windows. This issue will be fixed in a future release.

hipRTC rocWMMA unknown type name compilation errors

Any hipRTC program using the rocwmma.hpp header will fail to compile and produce a number of "unknown type name" errors. This issue will be fixed in a future release. As a workaround, add the following code before including the rocwmma.hpp header:

typedef signed char int8_t;
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef unsigned int uint32_t;

hipBLASLt TF32 matmul INF propagation on Instinct MI350 Series

On Instinct MI350 Series GPUs (gfx950), hipblasLtMatmul with computeType = HIPBLAS_COMPUTE_32F_FAST_TF32 returns NaN instead of propagating INF values from input tensors. As a workaround, sanitize or clamp tensors before TF32 matmul operations if acceptable for the workload. This issue will be fixed in a future release.

rocPRIM unit test fails on Windows on Radeon RX 9060 GPUs

The rocprim.device_adjacent_find unit test on Windows on Radeon RX 9060 XT LP, 9060 XT, and 9060 GPUs might hang intermittently. This issue will be fixed in a future release.

MIOpen unit test header include failures

MIOpen unit tests fail to find the rocrand_xorwow.h header during runtime compilation of certain kernels. This occurs because ROCm can be installed in various locations and the include path is not automatically resolved. There's minimal impact to users. This issue will be fixed in a future release.

CRIU checkpoint fails on Instinct MI300X with Debian 13

ROCm Bandwidth Test (RBT) test cases for Checkpoint/Restore In Userspace (CRIU) fail on Instinct MI300X GPUs running Debian 13. Debian 13 users might experience CRIU failures. This issue will be fixed in a future release.

Multi-node RCCL test segmentation faults

Multi-node RCCL tests are experiencing intermittent segmentation faults on Instinct GPUs with ROCm 7.11.0. This issue will be fixed in a future release.

hipify-clang errors with CUDA 12.x

Users running hipify-clang with CUDA 12.x or later will encounter errors when processing CUDA source files:

error: must pass in an explicit nvptx64 gpu architecture to 'ptxas'
error: must pass in an explicit nvptx64 gpu architecture to 'nvlink'

As a workaround, explicitly pass the cuda-gpu-arch flag:

hipify-clang --cuda-gpu-arch=sm_70...

This issue will be fixed in a future release.

HIP Graph memory leak tests report invalid reads

The memLeak_Phase2 HIP Graph test suite fails with hipGraphInstantiate_leak and hipGraph_AddMemAlloc tests reporting invalid memory reads. Impact is minimal as the issue only occurs during shutdown after the workload is complete. This issue will be fixed in a future release.

ROCm examples HIP Graph tutorial code build fails

The HIP Graph tutorial code fails to build on Linux due to a missing -fPIC compiler flag. To resolve this issue, enable position-independent code in the main CMakeLists.txt by adding set(CMAKE_POSITION_INDEPENDENT_CODE ON) as in the fix in ROCm/rocm-examples PR #408.

Upcoming changes

Future preview releases will expand support for:

• Additional ROCm Core SDK components

• Domain-specific expansion toolkits (data science, life science, finance, simulation, and other HPC domains)

• Extended AMD hardware support