From 8d709bc12f1a4d946630b92c5343e4dd2c57b831 Mon Sep 17 00:00:00 2001 From: "Baraldi, Giovanni" Date: Wed, 12 Feb 2025 01:12:41 +0100 Subject: [PATCH] SWDEV-513725: Update readme for gfx11+ power states (#193) * Update readme * Update README.md Co-authored-by: Madsen, Jonathan * Address review comments * Update README.md Co-authored-by: Madsen, Jonathan --------- Co-authored-by: Giovanni Baraldi Co-authored-by: Madsen, Jonathan [ROCm/rocprofiler-sdk commit: 831e46932083bdae8c89f7e98c318e384a8042d9] --- projects/rocprofiler-sdk/README.md | 30 +++++++++++++++++++++++------- 1 file changed, 23 insertions(+), 7 deletions(-) diff --git a/projects/rocprofiler-sdk/README.md b/projects/rocprofiler-sdk/README.md index 1b1df83a53..d7bae22d27 100644 --- a/projects/rocprofiler-sdk/README.md +++ b/projects/rocprofiler-sdk/README.md @@ -83,16 +83,32 @@ Please report in the Github Issues. - Using PC sampling on multi-threaded applications might fail with `HSA_STATUS_ERROR_EXCEPTION`.Furthermore, if three or more threads launch operations to the same agent, and if PC sampling is enabled, the `HSA_STATUS_ERROR_EXCEPTION` might appear. -- Navi3x requires a stable power state for counter collection. - Currently, this state needs to be set by the user. - To do so, set "power_dpm_force_performance_level" to be writeable for non-root users, then set performance level to profile_standard: - +- gfx11 and gfx12 requires a stable power state for counter collection. This includes Radeon 7000 GPUs. ```bash - sudo chmod 777 /sys/class/drm/card0/device/power_dpm_force_performance_level - echo profile_standard >> /sys/class/drm/card0/device/power_dpm_force_performance_level + # For device . Use 'rocm-smi' or 'amd-smi monitor' to see device number. + sudo amd-smi set -g -l stable_std + # After profiling, set power state back to 'auto' + sudo amd-smi set -g -l auto ``` - Recommended: "profile_standard" for counter collection and "auto" for all other profiling. Use rocm-smi to verify the current power state. For multiGPU systems (includes integrated graphics), replace "card0" by the desired card. + The gfx version can be found via `amd-smi static --asic -g ` in the `TARGET_GRAPHICS_VERSION` field: + + ```bash + $ amd-smi static -a -g 2 + GPU: 2 + ASIC: + MARKET_NAME: Navi 33 [Radeon Pro W7500] + VENDOR_ID: 0x1002 + VENDOR_NAME: Advanced Micro Devices Inc. [AMD/ATI] + SUBVENDOR_ID: 0x1002 + DEVICE_ID: 0x7489 + SUBSYSTEM_ID: 0x0e0d + REV_ID: 0x00 + ASIC_SERIAL: N/A + OAM_ID: N/A + NUM_COMPUTE_UNITS: 28 + TARGET_GRAPHICS_VERSION: gfx1102 + ``` > [!WARNING] > The latest mainline version of AQLprofile can be found at [https://repo.radeon.com/rocm/misc/aqlprofile/](https://repo.radeon.com/rocm/misc/aqlprofile/). However, it's important to note that updates to the public AQLProfile may not occur as frequently as updates to the rocprofiler-sdk. This discrepancy could lead to a potential mismatch between the AQLprofile binary and the rocprofiler-sdk source.