Has anyone had any luck getting ROCm working on the new AMD 300 series Framework 13?
I’ve got mine and generally like the experience. Though it doesn’t seem to be reconizing the GPU for machine learning purposes.
On Bazzite 42, though a slightly modified version to roll back the kernel to the previous version since Fedora 42’s default kernel doesn’t work with ROCm.
$ uname -r
6.13.7-108.bazzite.fc42.x86_64
If anyone’s had any luck getting it to recognize it I would love to hear.
$ rocm-clinfo
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.1 AMD-APP (3635.0)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_event_callback
Platform Name: AMD Accelerated Parallel Processing
Number of devices: 0
what does “rocminfo” give ?
I only have an FW16 with AMD 7840HS, so I cannot tell you “This should work for you”. It is more a few commands to see what part is not working.
James3:
rocminfo
Inside of my rocm distrobox container, where I have rocm actually installed: rocminfo
$ rocminfo
ROCk module is loaded
=====================
HSA System Attributes
=====================
Runtime Version: 1.15
Runtime Ext Version: 1.7
System Timestamp Freq.: 1000.000000MHz
Sig. Max Wait Duration: 18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model: LARGE
System Endianness: LITTLE
Mwaitx: DISABLED
XNACK enabled: NO
DMAbuf Support: YES
VMM Support: YES
==========
HSA Agents
==========
*******
Agent 1
*******
Name: AMD Ryzen AI 7 350 w/ Radeon 860M
Uuid: CPU-XX
Marketing Name: AMD Ryzen AI 7 350 w/ Radeon 860M
Vendor Name: CPU
Feature: None specified
Profile: FULL_PROFILE
Float Round Mode: NEAR
Max Queue Number: 0(0x0)
Queue Min Size: 0(0x0)
Queue Max Size: 0(0x0)
Queue Type: MULTI
Node: 0
Device Type: CPU
Cache Info:
L1: 49152(0xc000) KB
Chip ID: 0(0x0)
ASIC Revision: 0(0x0)
Cacheline Size: 64(0x40)
Max Clock Freq. (MHz): 5089
BDFID: 0
Internal Node ID: 0
Compute Unit: 16
SIMDs per CU: 0
Shader Engines: 0
Shader Arrs. per Eng.: 0
WatchPts on Addr. Ranges:1
Memory Properties:
Features: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: FINE GRAINED
Size: 65095832(0x3e14898) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 2
Segment: GLOBAL; FLAGS: EXTENDED FINE GRAINED
Size: 65095832(0x3e14898) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 3
Segment: GLOBAL; FLAGS: KERNARG, FINE GRAINED
Size: 65095832(0x3e14898) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 4
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 65095832(0x3e14898) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
ISA Info:
*******
Agent 2
*******
Name: gfx1152
Uuid: GPU-XX
Marketing Name: AMD Radeon Graphics
Vendor Name: AMD
Feature: KERNEL_DISPATCH
Profile: BASE_PROFILE
Float Round Mode: NEAR
Max Queue Number: 128(0x80)
Queue Min Size: 64(0x40)
Queue Max Size: 131072(0x20000)
Queue Type: MULTI
Node: 1
Device Type: GPU
Cache Info:
L1: 32(0x20) KB
L2: 1024(0x400) KB
Chip ID: 4372(0x1114)
ASIC Revision: 0(0x0)
Cacheline Size: 128(0x80)
Max Clock Freq. (MHz): 3000
BDFID: 49408
Internal Node ID: 1
Compute Unit: 8
SIMDs per CU: 2
Shader Engines: 1
Shader Arrs. per Eng.: 2
WatchPts on Addr. Ranges:4
Coherent Host Access: FALSE
Memory Properties: APU
Features: KERNEL_DISPATCH
Fast F16 Operation: TRUE
Wavefront Size: 32(0x20)
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Max Waves Per CU: 32(0x20)
Max Work-item Per CU: 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
Max fbarriers/Workgrp: 32
Packet Processor uCode:: 14
SDMA engine uCode:: 12
IOMMU Support:: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 32547916(0x1f0a44c) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:2048KB
Alloc Alignment: 4KB
Accessible by all: FALSE
Pool 2
Segment: GLOBAL; FLAGS: EXTENDED FINE GRAINED
Size: 32547916(0x1f0a44c) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:2048KB
Alloc Alignment: 4KB
Accessible by all: FALSE
Pool 3
Segment: GROUP
Size: 64(0x40) KB
Allocatable: FALSE
Alloc Granule: 0KB
Alloc Recommended Granule:0KB
Alloc Alignment: 0KB
Accessible by all: FALSE
ISA Info:
ISA 1
Name: amdgcn-amd-amdhsa--gfx1152
Machine Models: HSA_MACHINE_MODEL_LARGE
Profiles: HSA_PROFILE_BASE
Default Rounding Mode: NEAR
Default Rounding Mode: NEAR
Fast f16: TRUE
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
FBarrier Max Size: 32
ISA 2
Name: amdgcn-amd-amdhsa--gfx11-generic
Machine Models: HSA_MACHINE_MODEL_LARGE
Profiles: HSA_PROFILE_BASE
Default Rounding Mode: NEAR
Default Rounding Mode: NEAR
Fast f16: TRUE
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
FBarrier Max Size: 32
*******
Agent 3
*******
Name: aie2
Uuid: AIE-XX
Marketing Name: AIE-ML
Vendor Name: AMD
Feature: AGENT_DISPATCH
Profile: BASE_PROFILE
Float Round Mode: NEAR
Max Queue Number: 1(0x1)
Queue Min Size: 64(0x40)
Queue Max Size: 64(0x40)
Queue Type: SINGLE
Node: 0
Device Type: DSP
Cache Info:
L2: 1024(0x400) KB
Chip ID: 0(0x0)
ASIC Revision: 0(0x0)
Cacheline Size: 0(0x0)
Max Clock Freq. (MHz): 0
BDFID: 0
Internal Node ID: 0
Compute Unit: 0
SIMDs per CU: 0
Shader Engines: 0
Shader Arrs. per Eng.: 0
WatchPts on Addr. Ranges:0
Memory Properties:
Features: AGENT_DISPATCH
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: KERNARG, COARSE GRAINED
Size: 65095832(0x3e14898) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 2
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 65536(0x10000) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:0KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 3
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 65095832(0x3e14898) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
ISA Info:
*** Done ***
clinfo:
clinfo
Number of platforms: 1
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.1 AMD-APP (3649.0)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_event_callback
Platform Name: AMD Accelerated Parallel Processing
Number of devices: 1
Device Type: CL_DEVICE_TYPE_GPU
Vendor ID: 1002h
Board name: AMD Radeon Graphics
Device Topology: PCI[ B#193, D#0, F#0 ]
Max compute units: 4
Max work items dimensions: 3
Max work items[0]: 1024
Max work items[1]: 1024
Max work items[2]: 1024
Max work group size: 256
Preferred vector width char: 4
Preferred vector width short: 2
Preferred vector width int: 1
Preferred vector width long: 1
Preferred vector width float: 1
Preferred vector width double: 1
Native vector width char: 4
Native vector width short: 2
Native vector width int: 1
Native vector width long: 1
Native vector width float: 1
Native vector width double: 1
Max clock frequency: 3000Mhz
Address bits: 64
Max memory allocation: 28329706080
Image support: Yes
Max number of images read arguments: 128
Max number of images write arguments: 8
Max image 2D width: 16384
Max image 2D height: 16384
Max image 3D width: 16384
Max image 3D height: 16384
Max image 3D depth: 8192
Max samplers within kernel: 16
Max size of kernel argument: 1024
Alignment (bits) of base address: 2048
Minimum alignment (bytes) for any datatype: 128
Single precision floating point capability
Denorms: Yes
Quiet NaNs: Yes
Round to nearest even: Yes
Round to zero: Yes
Round to +ve and infinity: Yes
IEEE754-2008 fused multiply-add: Yes
Cache type: Read/Write
Cache line size: 128
Cache size: 32768
Global memory size: 33329065984
Constant buffer size: 28329706080
Max number of constant args: 8
Local memory type: Local
Local memory size: 65536
Max pipe arguments: 16
Max pipe active reservations: 16
Max pipe packet size: 2559902304
Max global variable size: 28329706080
Max global variable preferred total size: 33329065984
Max read/write image args: 64
Max on device events: 1024
Queue on device max size: 8388608
Max on device queues: 1
Queue on device preferred size: 262144
SVM capabilities:
Coarse grain buffer: Yes
Fine grain buffer: Yes
Fine grain system: No
Atomics: No
Preferred platform atomic alignment: 0
Preferred global atomic alignment: 0
Preferred local atomic alignment: 0
Kernel Preferred work group size multiple: 32
Error correction support: 0
Unified memory for Host and Device: 1
Profiling timer resolution: 1
Device endianess: Little
Available: Yes
Compiler available: Yes
Execution capabilities:
Execute OpenCL kernels: Yes
Execute native function: No
Queue on Host properties:
Out-of-Order: No
Profiling : Yes
Queue on Device properties:
Out-of-Order: Yes
Profiling : Yes
Platform ID: 0x7fdd7b43d050
Name: gfx1152
Vendor: Advanced Micro Devices, Inc.
Device OpenCL C version: OpenCL C 2.0
Driver version: 3649.0 (HSA1.1,LC)
Profile: FULL_PROFILE
Version: OpenCL 2.0
Extensions: cl_khr_fp64 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_fp16 cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_media_ops cl_amd_media_ops2 cl_khr_image2d_from_buffer cl_khr_subgroups cl_khr_depth_images cl_amd_copy_buffer_p2p cl_amd_assembly_program
Don’t have amdgpu_top installed
amdgpu_top
bash: amdgpu_top: command not found
Well, the rocminfo and clinfo both show it seeing your GPU. The “gfx1152”.
The amdgpu_top is available here:
Tool to display AMDGPU usage
I have a little test program that multiplies two matrix together here:
Examples of calling rocm from rust
I got it to work on a 50000 x 50000 matrix that takes up about 60GB RAM on my FW16 AMD 7840HS.
What exactly are you wishing to get going on the ROCM ?
I’m trying to run stable diffusion. Attempting to run that from within my distrobox container, where I ran clinfo et al from, gets me…
glibc version is 2.35
Check TCMalloc: libtcmalloc_minimal.so.4
libtcmalloc_minimal.so.4 is linked with libc.so,execute LD_PRELOAD=/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4
Python 3.10.12 (main, Feb 4 2025, 14:57:36) [GCC 11.4.0]
Version: f1.7.0-v1.10.1RC-latest-2183-g1480ec1d
Commit hash: 1480ec1d636aee32d913efd49ba6dc50978b300d
Traceback (most recent call last):
File "/home/nitrousoxide/Repos/stable-diffusion-webui-reForge/launch.py", line 51, in <module>
main()
File "/home/nitrousoxide/Repos/stable-diffusion-webui-reForge/launch.py", line 39, in main
prepare_environment()
File "/home/nitrousoxide/Repos/stable-diffusion-webui-reForge/modules/launch_utils.py", line 493, in prepare_environment
raise RuntimeError(
RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
Skipping the cuda test as the startup suggests…
RuntimeError: No HIP GPUs are available
So it does not appear to be able to see my GPU.
Hi,
What instructions are you trying to install from.
With the wget links updated to the latest like this.
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui
cd stable-diffusion-webui
python -m venv venv
source venv/bin/activate
python -m pip install --upgrade pip wheel
wget https://repo.radeon.com/rocm/manylinux/rocm-rel-6.4/torch-2.6.0%2Brocm6.4.0.git2fb0ac2b-cp312-cp312-linux_x86_64.whl
wget https://repo.radeon.com/rocm/manylinux/rocm-rel-6.4/torchvision-0.21.0%2Brocm6.4.0.git4040d51f-cp312-cp312-linux_x86_64.whl
wget https://repo.radeon.com/rocm/manylinux/rocm-rel-6.4/pytorch_triton_rocm-3.2.0%2Brocm6.4.0.git6da9e660-cp312-cp312-linux_x86_64.whl
wget https://repo.radeon.com/rocm/manylinux/rocm-rel-6.4/torchaudio-2.6.0%2Brocm6.4.0.gitd8831425-cp312-cp312-linux_x86_64.whl
pip3 uninstall torch torchvision pytorch-triton-rocm
pip3 install "./torch-2.6.0+rocm6.4.0.git2fb0ac2b-cp312-cp312-linux_x86_64.whl" "./torchvision-0.21.0+rocm6.4.0.git4040d51f-cp312-cp312-linux_x86_64.whl" "./torchaudio-2.6.0+rocm6.4.0.gitd8831425-cp312-cp312-linux_x86_64.whl" "./pytorch_triton_rocm-3.2.0+rocm6.4.0.git6da9e660-cp312-cp312-linux_x86_64.whl"
python launch.py --precision full --no-half
But the last command fails, unable to compile something.
I’m using this distrobox assemble recipie to create the container, then running the webui.sh script.
[rocm]
image=ubuntu:22.04
init=false
additional_packages="build-essential libtcmalloc-minimal4 wget git python3 python3-venv libgl1 libglib2.0-0 libxkbcommon0 libsm6"
init_hooks="apt update -y;"
init_hooks="apt upgrade -y;"
init_hooks="cd ~/Downloads && wget https://repo.radeon.com/amdgpu-install/6.4/ubuntu/jammy/amdgpu-install_6.4.60400-1_all.deb && apt install -y ./amdgpu-install_6.4.60400-1_all.deb && rm amdgpu-install_6.4.60400-1_all.deb;"
init_hooks="amdgpu-install --usecase=rocm --no-dkms -y;"
init_hooks="mkdir -p ~/miniconda3 && wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh && bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3 && rm -rf ~/miniconda3/miniconda.sh;"
init_hooks="export HSA_OVERRIDE_GFX_VERSION=9.0.0;"
nvidia=false
pull=true
root=false
replace=true
start_now=true
After that I enter the distrobox container, clone the git repo and run the webui.sh.
It does successfully setup the application, but it just does not recognize the gpu at all.
I was able to get ROCm to “work” however it seems less optimal than it should considering the HX 370 should be more capable than the 7840U.
I tried force-host-alloction-APU as well to see if performance improved but sadly it didn’t, I attempted to allocate 8GB and 16GB of VRAM but testing Applio , it only ever uses 1.3GB and converting 18 seconds of audio takes around 123 seconds which I think is quite poor.
I’m not on Bazzite but I don’t think that makes a difference.
Just saw some news on kernel 6.14, no idea if the AMDXDNA accelerator driver will improve performance or not regarding AI tasks.
Operating System: Arch Linux
KDE Plasma Version: 6.3.4
KDE Frameworks Version: 6.13.0
Qt Version: 6.9.0
Kernel Version: 6.14.4-arch1-2 (64-bit)
Graphics Platform: Wayland
Processors: 24 × AMD Ryzen AI 9 HX 370 w/ Radeon 890M
Memory: 30.6 GiB of RAM
Graphics Processor: AMD Radeon Graphics
Manufacturer: Framework
Product Name: Laptop 13 (AMD Ryzen AI 300 Series)
System Version: A9
What steps did you take to get it to work, at least partially? I’m getting zero luck so far.
I’m on vanilla Arch so you’ll need to find the equivalent packages for Bazzite since I don’t know the package names.
I installed rocm-hip-sdk and python-conda then created a 3.10 python environment and installed gcc, then modified the install script to download ROCm for pytorch instead of CUDA, waited for it to finish installing and it runs (but not well).
Replace the normal CUDA pytorch download with this one:
python -m pip install torch==2.7.0 torchvision torchaudio==2.7.0 --upgrade --index-url https://download.pytorch.org/whl/rocm6.3
Add this before running the python script to force ROCm to “accept” incompatible GPUs like the iGPU we have:HSA_OVERRIDE_GFX_VERSION=11.0.2
When it runs, it does utilize 100% of the iGPU however I’m unable to get it to use more VRAM which results in slower speeds.
This was the steps I used when I tried to run some AI stuff on the 7840U but gave up because it was worse than the HX 370.
That’s interesting. The same command doesn’t show the APU/GPU agent at all in WSL, even though I was following the same AMD documentation for WSL that I’ve successfully used with the 7900 XTX
I’ve used ComfyUI with PyTorch running on AMD’s ROCm AI framework on my desktop with an AMD 7900XTX dedicated GPU successfully, and I was curious to see how a laptop APU designed for AI workloads would compare. Sadly, I can’t get PyTorch to work with the Framework Laptop 13 AMD Ryzen AI 9 HX 370 with Radeon 890M with 96 GB of system memory.
It turns out, AMD AMD ROCm does not support the Radeon 890M. In fact, when support was requested, AMD pointed users to third-party patches ! So, if you were…
Looks like even if ROCm worked correctly, the memory bandwidth is a huge bottleneck when trying to run LLMs.
gfx1150 support (890M) not working · Issue #40 · likelovewant/ollama-for-amd
Interestingly, inference performance stays the same, about 7.1-7.8 tok/s for deepseek-r1:14b q4 on this hardware (AMD HX 370, LPDDR5X 7500), when I run all of these variants:
Stock Ollama CPU, num_threads=20 (it’s a 12-core + SMT CPU)
llama.cpp with Vulkan
llama.cpp with AVX512
ollama-for-amd with ROCm
So, most likely it’s a memory bandwidth limitation. The CPU is free to do other things if GPU is used, of course.
For what it’s worth, I got ROCm working with Pytorch today on my new Framework 13 (AMD Ryzen AI 9 HX 370 w/ Radeon 890M).
I’m running Fedora 42:
Linux fedora 6.14.5-300.fc42.x86_64 #1 SMP PREEMPT_DYNAMIC Fri May 2 14:16:46 UTC 2025 x86_64 GNU/Linux
I think all I did to get it installed was:
sudo dnf update
sudo dnf install rocminfo
sudo usermod -a -G render,video $LOGNAME
Here’s the output from rocminfo:
ROCk module is loaded
=====================
HSA System Attributes
=====================
Runtime Version: 1.1
Runtime Ext Version: 1.6
System Timestamp Freq.: 1000.000000MHz
Sig. Max Wait Duration: 18446744073709551615 (0xFFFFFFFFFFFFFFFF) (timestamp count)
Machine Model: LARGE
System Endianness: LITTLE
Mwaitx: DISABLED
DMAbuf Support: YES
==========
HSA Agents
==========
*******
Agent 1
*******
Name: AMD Ryzen AI 9 HX 370 w/ Radeon 890M
Uuid: CPU-XX
Marketing Name: AMD Ryzen AI 9 HX 370 w/ Radeon 890M
Vendor Name: CPU
Feature: None specified
Profile: FULL_PROFILE
Float Round Mode: NEAR
Max Queue Number: 0(0x0)
Queue Min Size: 0(0x0)
Queue Max Size: 0(0x0)
Queue Type: MULTI
Node: 0
Device Type: CPU
Cache Info:
L1: 49152(0xc000) KB
Chip ID: 0(0x0)
ASIC Revision: 0(0x0)
Cacheline Size: 64(0x40)
Max Clock Freq. (MHz): 2000
BDFID: 0
Internal Node ID: 0
Compute Unit: 24
SIMDs per CU: 0
Shader Engines: 0
Shader Arrs. per Eng.: 0
WatchPts on Addr. Ranges:1
Memory Properties:
Features: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: FINE GRAINED
Size: 98125092(0x5d94524) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 2
Segment: GLOBAL; FLAGS: EXTENDED FINE GRAINED
Size: 98125092(0x5d94524) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 3
Segment: GLOBAL; FLAGS: KERNARG, FINE GRAINED
Size: 98125092(0x5d94524) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
Pool 4
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 98125092(0x5d94524) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:4KB
Alloc Alignment: 4KB
Accessible by all: TRUE
ISA Info:
*******
Agent 2
*******
Name: gfx1150
Uuid: GPU-XX
Marketing Name: AMD Radeon Graphics
Vendor Name: AMD
Feature: KERNEL_DISPATCH
Profile: BASE_PROFILE
Float Round Mode: NEAR
Max Queue Number: 128(0x80)
Queue Min Size: 64(0x40)
Queue Max Size: 131072(0x20000)
Queue Type: MULTI
Node: 1
Device Type: GPU
Cache Info:
L1: 32(0x20) KB
L2: 2048(0x800) KB
Chip ID: 5390(0x150e)
ASIC Revision: 4(0x4)
Cacheline Size: 128(0x80)
Max Clock Freq. (MHz): 2900
BDFID: 49408
Internal Node ID: 1
Compute Unit: 16
SIMDs per CU: 2
Shader Engines: 1
Shader Arrs. per Eng.: 2
WatchPts on Addr. Ranges:4
Coherent Host Access: FALSE
Memory Properties: APU
Features: KERNEL_DISPATCH
Fast F16 Operation: TRUE
Wavefront Size: 32(0x20)
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Max Waves Per CU: 32(0x20)
Max Work-item Per CU: 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
Max fbarriers/Workgrp: 32
Packet Processor uCode:: 29
SDMA engine uCode:: 11
IOMMU Support:: None
Pool Info:
Pool 1
Segment: GLOBAL; FLAGS: COARSE GRAINED
Size: 49062544(0x2eca290) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:2048KB
Alloc Alignment: 4KB
Accessible by all: FALSE
Pool 2
Segment: GLOBAL; FLAGS: EXTENDED FINE GRAINED
Size: 49062544(0x2eca290) KB
Allocatable: TRUE
Alloc Granule: 4KB
Alloc Recommended Granule:2048KB
Alloc Alignment: 4KB
Accessible by all: FALSE
Pool 3
Segment: GROUP
Size: 64(0x40) KB
Allocatable: FALSE
Alloc Granule: 0KB
Alloc Recommended Granule:0KB
Alloc Alignment: 0KB
Accessible by all: FALSE
ISA Info:
ISA 1
Name: amdgcn-amd-amdhsa--gfx1150
Machine Models: HSA_MACHINE_MODEL_LARGE
Profiles: HSA_PROFILE_BASE
Default Rounding Mode: NEAR
Default Rounding Mode: NEAR
Fast f16: TRUE
Workgroup Max Size: 1024(0x400)
Workgroup Max Size per Dimension:
x 1024(0x400)
y 1024(0x400)
z 1024(0x400)
Grid Max Size: 4294967295(0xffffffff)
Grid Max Size per Dimension:
x 4294967295(0xffffffff)
y 4294967295(0xffffffff)
z 4294967295(0xffffffff)
FBarrier Max Size: 32
*** Done ***
I setup a virtualenv for Python3 and installed pytorch with rockm support:
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm6.3
And it appears to work:
$ python
Python 3.13.3 (main, Apr 22 2025, 00:00:00) [GCC 15.0.1 20250418 (Red Hat 15.0.1-0)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
>>> print(f"Using device: {device}")
Using device: cuda
>>> quit()
I haven’t used it in anger yet, but so far so good!
Adrian.
1 Like