A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 5, 07:15 UTC
[ROCm] Add RDNA 3.5/4 device IDs (gfx1150, gfx1151, gfx1201) (#38455)
Signed-off-by: rdondeti <ravitez.dondeti@gmail.com> Signed-off-by: Ravitez Dondeti <ravitez.dondeti@gmail.com> Co-authored-by: Claude <noreply@anthropic.com>
Ravitez Dondeti committed
Apr 10, 2026 at 06:35 UTC
c1cc7344fb720547051bb44a2e102278ae9b7a6a
1 file changed
+6
-1
vllm/platforms/rocm.py
+6
-1
@@ -64,7 +64,12 @@ _ROCM_DEVICE_ID_NAME_MAP: dict[str, str] = {
64
"0x74a9": "AMD_Instinct_MI300X_HF",
65
"0x74bd": "AMD_Instinct_MI300X_HF",
66
"0x744c": "AMD_Radeon_RX7900XTX",
67
- "0x7551": "AMD_Radeon_R9700",
67
+ # RDNA 3.5 APUs (Strix Point / Strix Halo)
68
+ "0x150e": "AMD_Radeon_890M", # gfx1150, Strix Point
69
+ "0x1586": "AMD_Radeon_8060S", # gfx1151, Strix Halo
70
+ # RDNA 4 discrete (Navi 48)
71
+ "0x7550": "AMD_Radeon_RX9070XT", # gfx1201
72
+ "0x7551": "AMD_Radeon_R9700", # gfx1201
73
}
74
75