ROCm Core SDK 7.10.0 release notes#
2025-12-11
13 min read time
ROCm Core SDK 7.10.0 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 coverage 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 2026.
Important
ROCm 7.10.0 follows the versioning discontinuity that began with 7.9.0 and remains separate from the 7.0 and 7.1 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#
This preview of the ROCm Core SDK with TheRock introduces several improvements following the previous 7.9.0 release, including expanded hardware support, operating system coverage, and additional ROCm Core SDK components.
Expanded AMD hardware support#
ROCm 7.10.0 builds on ROCm 7.9.0, adding new support for the following AMD Instinct GPUs and Ryzen AI APUs:
Instinct MI250X
Instinct MI250
Instinct MI210
Radeon PRO W7900D
Radeon PRO W7900
Radeon PRO W7800 48GB
Radeon PRO W7800
Radeon PRO W7700
Radeon RX 7900 XTX
Radeon RX 7900 XT
Radeon RX 7900 GRE
Radeon RX 7800 XT
Radeon RX 7700 XT
Ryzen AI 9 HX 375
Ryzen AI 9 HX 370
Ryzen AI 9 365
For the full list of supported GPUs and APUs, see Supported hardware and operating systems.
Expanded Linux distribution support on Instinct GPUs#
ROCm 7.10.0 builds on ROCm 7.9.0, adding new support for the following Linux distributions on AMD Instinct MI350 Series, MI300 Series, and MI200 Series GPUs.
Red Hat Enterprise Linux (RHEL) 10.1, 10.0, 9.7, 9.6, and 8.10
SUSE Linux Enterprise Server (SLES) 15.7
For the full list of supported operating system versions, see Supported hardware and operating systems.
Expanded ROCm Core SDK components#
ROCm 7.10.0 adds the following tools and libraries to the ROCm Core SDK:
System utilities, profiling, and debugging tools: ROCm Compute Profiler and SPIRV-LLVM-Translator
Math and compute libraries: hipSPARSELt, Composable Kernel, and rocWMMA
Compatibility notices#
In terms of package compatibility, ROCm 7.10.0 diverges from the existing ROCm 7.0 stream and upcoming stable releases in that stream:
Compute-focused: ROCm 7.10.0 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 7.10.0, install Radeon Software for Linux version 25.30.1 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 7.10.0, use the inbox graphics drivers of Ubuntu 24.04.3.
No upgrade path from existing production releases including ROCm 7.1.1 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.
Packaging formats: RPM and Debian packages are not available in this release. Instead, Python wheels and tarballs are provided. See the ROCm 7.10.0 installation instructions.
Software components: some components of the ROCm Core SDK are not yet available in this release. 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.
Looking ahead#
Subsequent technology preview releases will follow a 6-week cadence, filling gaps and introducing new ROCm expansions. AMD continues to maintain traditional ROCm releases in parallel with the preview stream until mid-2026.
Supported hardware and operating systems#
ROCm 7.10.0 adds support for Instinct MI200 Series GPUs and Ryzen AI Series APUs. The following table lists supported AMD Instinct GPUs, Radeon GPUs, and Ryzen AI APUs. Each supported device is listed with its corresponding GPU architecture, LLVM target, and supported operating systems.
Note
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.
|
AMD device series |
Device |
LLVM target |
Architecture |
Supported OS |
|---|---|---|---|---|
|
Instinct MI350 Series |
Instinct MI355X Instinct MI350X |
gfx950 |
CDNA 4 |
Ubuntu 24.04.3 |
|
Instinct MI300 Series |
Instinct MI325X Instinct MI300X Instinct MI300A |
gfx942 |
CDNA 3 |
|
|
Instinct MI200 Series |
Instinct MI250X Instinct MI250 Instinct MI210 |
gfx90a |
CDNA 2 |
|
AMD device series |
Device |
LLVM target |
Architecture |
Supported OS |
|---|---|---|---|---|
|
Radeon PRO W7000 Series |
Radeon PRO W7900D Radeon PRO W7900 Radeon PRO W7800 48GB Radeon PRO W7800 |
gfx1100 |
RDNA 3 |
Ubuntu 24.04.3 |
|
Radeon PRO W7700 |
gfx1101 |
|
AMD device series |
Device |
LLVM target |
Architecture |
Supported OS |
|---|---|---|---|---|
|
Radeon RX 7000 Series |
Radeon RX 7900 XTX Radeon RX 7900 XT Radeon RX 7900 GRE |
gfx1100 |
RDNA 3 |
Ubuntu 24.04.3 |
|
Radeon RX 7800 XT Radeon RX 7700 XT |
gfx1101 |
|
AMD device series |
Device |
LLVM target |
Architecture |
Supported OS |
|---|---|---|---|---|
|
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 |
Ubuntu 24.04.3 |
|
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 |
Note
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.
Supported kernel driver and firmware bundles#
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 drivers and ROCm user space components, your server or infrastructure provider publishes the GPU and baseboard firmware by bundling AMD firmware releases through Platform Level Data Model (PLDM) bundles – which include the Integrated Firmware Image (IFWI).
Note
GPU virtualization is not supported in ROCm 7.10.0.
|
AMD device |
Firmware |
Linux driver |
|---|---|---|
|
Instinct MI355X |
PLDM bundle 01.25.15.04 PLDM bundle 01.25.13.09 |
|
|
Instinct MI350X |
||
|
Instinct MI325X |
PLDM bundle 01.25.04.02 PLDM bundle 01.25.03.03 |
|
|
Instinct MI300X |
PLDM bundle 01.25.05.00 (or later) PLDM bundle 01.25.03.12 |
|
|
Instinct MI300A |
BKC 26 BKC 25 |
|
|
Instinct MI250X |
IFWI 47 (or later) |
|
|
Instinct MI250 |
Maintenance update 5 with IFWI 75 (or later) |
|
|
Instinct MI210 |
|
AMD device |
Linux driver |
Windows driver |
|---|---|---|
|
Radeon PRO W7900D |
AMD Software: Adrenalin Edition |
|
|
Radeon PRO W7900 |
||
|
Radeon PRO W7800 48GB |
||
|
Radeon PRO W7800 |
||
|
Radeon PRO W7700 |
|
AMD device |
Linux driver |
Windows driver |
|---|---|---|
|
Radeon RX 7900 XTX |
AMD Software: Adrenalin Edition |
|
|
Radeon RX 7900 XT |
||
|
Radeon RX 7900 GRE |
||
|
Radeon RX 7800 XT |
||
|
Radeon RX 7700 XT |
|
AMD device |
Linux driver |
Windows driver |
|---|---|---|
|
Ryzen AI Max+ PRO 395 |
Inbox kernel driver |
AMD Software: Adrenalin Edition |
|
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 |
Deep learning frameworks#
ROCm 7.10.0 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 |
ROCm Core SDK components#
The following table lists core components included in the ROCm 7.10.0 release. Expect future releases in this stream to expand the list of components.
|
Component group |
Component name |
Supported operating systems |
|---|---|---|
|
Runtime and compilers |
HIP |
Linux and Windows |
| HIPIFY | ||
| LLVM | ||
| ROCr Runtime |
Linux |
|
| SPIRV-LLVM-Translator | ||
|
Control and monitoring tools |
AMD SMI |
Linux |
| rocminfo | ||
|
hipinfo |
Windows |
|
|
Profiling and debugging tools |
ROCm Compute Profiler (rocprofiler-compute) |
Linux |
| ROCprofiler-SDK |
Linux (Instinct GPUs only) |
|
|
Math and compute libraries |
rocBLAS |
Linux and Windows |
| hipBLAS | ||
| hipBLASLt | ||
| rocFFT | ||
| hipFFT | ||
| rocRAND | ||
| hipRAND | ||
| rocSOLVER | ||
| hipSOLVER | ||
| rocSPARSE | ||
| hipSPARSE | ||
| rocPRIM | ||
| rocThrust | ||
| hipCUB | ||
| rocWMMA | ||
| hipSPARSELt |
Linux |
|
|
Composable Kernel (partial, limited support) |
||
| MIOpen |
Linux (Instinct GPUs only) |
|
|
Communication libraries |
RCCL |
Linux |
|
Support libraries |
ROCm CMake |
Linux and Windows |
Known issues#
The following sections describe some known issues identified in ROCm 7.10.0.
PyTorch float64 test failures#
The PyTorch unit test test_reduction_fns_name_softmax_float64 might fail when
using the float64 data type. This issue will be fixed in a future release.
PyTorch binary ufunc test failures#
Multiple tests in test_binary_ufuncs (for example, pow_cuda_complex64 and
related slicing tests) might fail when using TheRock package.
ComfyUI hangs on Windows when using older drivers#
When running ComfyUI-based AI workflows on Windows using the TheRock packages and Adrenalin driver 25.11.1, Timeout Detection and Recovery (TDR) events may occur. As a workaround, use the Adrenalin driver 25.20.01.17 as noted in Supported kernel driver and firmware bundles for stable ComfyUI model use.
Wan2.2 14B video generation hangs on Windows#
Running Wan2.2 14B video generation workflows in ComfyUI on Windows might trigger Timeout Detection and Recovery (TDR) events and unusable output. This issue will be fixed in a future release. As a workaround, avoid large-scale video generation workloads on Windows for now.
Long-duration inference workloads hang on Windows#
Long-duration inference workloads (for example, Stable Diffusion or Llama models) might cause hangs or system unresponsiveness, requiring a reboot. This issue will be fixed in a future release. As a workaround, avoid long-duration inference workloads on Windows for now.
MIOpen incorrect results#
On Ryzen AI Max 300 Series APUs, MIOpen might produce incorrect results due to solvers that don’t support non-packed tensors. This issue will be fixed in a future release.
MIOpen smoke test failures#
MIOpen unit tests Smoke/CPU_UnitTestImplicitGemmCKUtil_NONE.TestParsing/0 and
TestParsing/5 might fail on TheRock builds. This issue will be fixed in
a future release.
ROCr unit test failures#
rocrtstFunc.Concurrent_Init_Shutdown_Test test hangs intermittently and can
cause timeouts or segmentation faults under specific system configurations (for
example, MI325X or Ryzen AI Max 300 Series APUs on Ubuntu 22.04).
Build failures in HIP samples and rocrtst tests#
Some HIP samples and rocrtst tests fail to build due to differences between
TheRock and legacy ROCm build environments. As a workaround, set the
environment variable before building: export CXX=hipcc.
HIP-based, LDS, and SPIR-V test failures#
Tests related to HIP-based, LDS, and SPIR-V fail with segmentation faults. This issue will be fixed in a future release.
Intermittent failures in KFD tests when using older drivers#
KFD tests might intermittently fail when using ROCm 7.10.0 user-space components with older kernel drivers (for example, ROCm 6.4 drivers) due to feature mismatches. As a workaround, upgrade to the latest drivers as noted in Supported kernel driver and firmware bundles for stability.
PyTorch DDP with RCCL failures#
Distributed training across multiple nodes on Instinct MI325X clusters using PyTorch DistributedDataParallel (DDP) with RCCL might fail with collective operation timeout errors or mismatched parameter shapes. This issue will be fixed in a future release.
HIP runtime error on PyTorch DDP with RCCL#
PyTorch unit test DistributedDataParallelTest.test_failure_recovery fails
with a HIP runtime error invalid device ordinal when running RCCL-based
distributed tests on TheRock builds. This issue will be fixed in a future release.
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 coverage
GPU virtualization support