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 15, 01:00 UTC
Remove torch_xla.tpu.version() from pallas.py. (#21065)
Signed-off-by: Qiliang Cui <derrhein@gmail.com>
QiliangCui committed
Jul 16, 2025 at 17:25 UTC
72ad2735823e23b4e1cc79b7c73c3a5f3c093ab0
1 file changed
-4
vllm/v1/attention/backends/pallas.py
-4
@@ -167,10 +167,6 @@ class PallasAttentionBackendImpl(AttentionImpl):
167
"are not implemented for "
168
"PallasAttentionBackendImpl")
169
170
- tpu_version = torch_xla.tpu.version()
171
- if tpu_version < 4:
172
- raise NotImplementedError("TPU version must be 4 or higher.")
173
-
170
def forward(
171
self,
172
layer: AttentionLayer,