AI/rocm-systems
2
0
Bifurcation 0
Code Tickets Demandes d'ajout Actions Paquets Projets Publications Wiki Activité

[SWDEV-518325/SWDEV-518320/SWDEV-443309] Changelog addition

Parcourir la source 
Change-Id: I29567229f0e27d307ac3df935b5a5fab8ca43409
Signed-off-by: Charis Poag <Charis.Poag@amd.com>


[ROCm/amdsmi commit: 19a4775d32]
Cette révision appartient à :
Charis Poag
2025-04-13 20:22:17 -05:00
révisé par Arif, Maisam
Parent 8d4a4d7b14
révision 1413ae1431
1 fichiers modifiés avec 93 ajouts et 1 suppressions
Télécharger le Fichier Patch Télécharger le Fichier des Différences Développer tous les fichiers Réduire tous les fichiers
projects/amdsmi/CHANGELOG.md
+93 -1
Voir le fichier
@@ -203,7 +203,97 @@ This change was needed for BM/Guest setups due to other related partition output
### Resolved issues
- **Fixed `amd-smi static --partition` for guest systems with MIx ASICs being unable to run**
- **Fixed partition enumeration - `amd-smi list -e`, `amdsmi_get_gpu_enumeration_info()`'s `amdsmi_enumeration_info_t` `drm_card` and `drm_render` fields**
Previously, partitions incorrectly reflected the primary node (1st GPU) and showed the DRM Render Minor as renderD128. Partition nodes mirrored renderD128's information, which was incorrect. See the "<i>Previous Outputs in CPX</i>" example below.
Device enumeration was updated to correctly map DRM Render Minor paths. See the "<i>Corrected Outputs in CPX</i>" example below.
These changes impact what information is readable/writable for the partition nodes.
<b><i>Example: Previous Outputs in CPX</b></i>
```shell
$ amd-smi list -e
GPU: 0
BDF: 0000:0c:00.0
UUID: <Redacted>
KFD_ID: 18421
NODE_ID: 2
PARTITION_ID: 0
RENDER: renderD128
CARD: card0
HSA_ID: 2
HIP_ID: 0
HIP_UUID: <Redacted>
GPU: 1
BDF: 0000:0c:00.1
UUID: <Redacted>
KFD_ID: 48116
NODE_ID: 3
PARTITION_ID: 1
RENDER: N/A
CARD: N/A
HSA_ID: 3
HIP_ID: 1
HIP_UUID: GPU-<Redacted>
...
```
```shell
$ amd-smi monitor
GPU POWER GPU_T MEM_T GFX_CLK GFX% MEM% ENC% DEC% VRAM_USAGE
0 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
1 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
2 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
3 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
4 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
5 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
6 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
7 201 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
8 210 W 46 °C 42 °C 2104 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
...
```
<b><i>Example: Corrected outputs in CPX</i></b>
```shell
$ amd-smi list -e
GPU: 0
BDF: 0000:0c:00.0
UUID: <Redacted>
KFD_ID: 18421
NODE_ID: 2
PARTITION_ID: 0
RENDER: renderD128
CARD: card0
HSA_ID: 2
HIP_ID: 0
HIP_UUID: GPU-<Redacted>
GPU: 1
BDF: 0000:0c:00.1
UUID: <Redacted>
KFD_ID: 48116
NODE_ID: 3
PARTITION_ID: 1
RENDER: renderD129
CARD: card1
HSA_ID: 3
HIP_ID: 1
HIP_UUID: GPU-<Redacted>
...
```
```shell
$ amd-smi monitor
GPU POWER GPU_T MEM_T GFX_CLK GFX% MEM% ENC% DEC% VRAM_USAGE
0 202 W 46 °C 42 °C 2107 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
1 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
2 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
3 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
4 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
5 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
6 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
7 N/A N/A N/A N/A N/A N/A N/A N/A 0.5/ 24.0 GB
8 210 W 46 °C 42 °C 2104 MHz 0 % 0 % N/A 0 % 0.3/192.0 GB
...
```
### Upcoming changes
@@ -633,6 +723,8 @@ Updated structure `amdsmi_vram_info_t`:
### Resolved issues
- **Fixed `amd-smi static --partition` for guest systems with MIx ASICs being unable to run**
- **Fixed `amdsmi_get_gpu_asic_info` and `amd-smi static --asic` not displaying graphics version properly for MI2x, MI1x or Navi 3x ASICs.**
Before on MI100: