crashes when Strix Point iGPU is available

[weslowski]

I have an AMD Ryzen 395 AI box which has a Radeon 8060S iGPU.
I installed the rocm-smi package which shows these information (apparently AMD itself did not adapt their software to this hardware):

========================= ROCm System Management Interface =========================
=================================== Concise Info ===================================
Exception caught: map::at
ERROR: GPU[0]	: sclk clock is unsupported
====================================================================================
GPU[0]		: mclk current clock frequency not found
====================================================================================
GPU[0]		: get_power_cap, Not supported on the given system
GPU  Temp (DieEdge)  AvgPwr   SCLK  MCLK  Fan  Perf  PwrCap       VRAM%  GPU%  
0    41.0c           13.034W  None  None  0%   auto  Unsupported   29%   1%    
====================================================================================
=============================== End of ROCm SMI Log ================================

btop crashes with that debug log:

2026/05/04 (16:21:49) | ===> btop++ v.1.3.0
2026/05/04 (16:21:49) | INFO: Logger set to DEBUG
2026/05/04 (16:21:49) | DEBUG: Using locale en_US.UTF-8
2026/05/04 (16:21:49) | INFO: Running on /dev/pts/2
2026/05/04 (16:21:49) | INFO: Failed to load libnvidia-ml.so, NVIDIA GPUs will not be detected: libnvidia-ml.so.1: cannot open shared object file: No such file or directory
2026/05/04 (16:21:49) | WARNING: ROCm SMI: Failed to get maximum GPU power draw, defaulting to 225W
2026/05/04 (16:21:49) | WARNING: ROCm SMI: Failed to get maximum GPU temperature, defaulting to 110°C
2026/05/04 (16:21:49) | WARNING: ROCm SMI: Failed to get VRAM utilization
2026/05/04 (16:21:49) | WARNING: ROCm SMI: Failed to get VRAM clock speed: 
2026/05/04 (16:21:49) | WARNING: ROCm SMI: Failed to get PCIe throughput

As far as I can see: the lookout for Nvidia will not be an issue but it cannot collect the AMD GPU values while the first 2 and last 2 are not available. As the user can partition the installed RAM for system memory and VRAM himself Btop seems to collect the amout of VRAM/used VRAM from the OS. Before I set up my machine again to have a more recent OS base to match the modern hardware, every now and then Btop was able to launch and it was showing a correct temperature and VRAM usage (on which I was really looking while using my LLMs) and it was showing always 95% GPU usage - apparently wrong. However, it stalled after a few seconds to up to a Minute silently.
Before I installed the AMD drivers Btop run as perfect as always.

I forgot: the OS error output
Segementation fault (core dumped)

thanks for your awesome work!

[shreechaturvedi]

Upgrade to 1.4.7. v1.3.0 predates ROCm 6 support, which landed in 1.3.2.

git clone https://github.com/aristocratos/btop && cd btop
make && sudo make install

Your log has no "Failed to get GPU clock speed" line, so rsmi_dev_gpu_clk_freq_get succeeded. 1.3.0 hands it a ROCm 5 rsmi_frequencies_t (uint32_t frequency[32]), and ROCm 7 writes the v6 layout into it: a leading bool, then uint64_t frequency[33]. Stack buffer overrun. The next two probes still log, then the collect function returns and you get the segfault.

1.3.2 added rsmi_version_get and picks the matching struct.