Update PyTorch to 2.8.0 (#20358)

Signed-off-by: Huy Do <huydhn@gmail.com> Co-authored-by: Michael Goin <mgoin64@gmail.com>
2025-08-29 03:57:35 -07:00
parent 69f46359dd
commit 67c14906aa
12 changed files with 50 additions and 45 deletions
--- a/vllm/v1/attention/backends/flex_attention.py
+++ b/vllm/v1/attention/backends/flex_attention.py
@@ -789,6 +789,7 @@ def get_kernel_options(query, block_m, block_n,
            device_props = torch.cuda.get_device_properties()
            max_shared_memory = device_props.shared_memory_per_block_optin
            if max_shared_memory < 144 * 1024:
-                kernel_options["BLOCK_M"] = 32
-                kernel_options["BLOCK_N"] = 32
+                kernel_options["BLOCK_M"] = kernel_options["BLOCK_M"] // 2
+                kernel_options["BLOCK_N"] = kernel_options["BLOCK_N"] // 2
+
    return kernel_options