fix lint

Author: airMeng

python/sgl_kernel/flash_attn.py

@@ -190,7 +190,12 @@ def flash_attn_with_kvcache(
         max_seqlen_k = cache_seqlens.max().item()
         assert cache_seqlens.size(0) + 1 == cu_seqlens_q.size(0)
         max_page_size_per_seq = page_table.shape(1)
-        num_pages_per_seq = torch.arange(0, cache_seqlens.size(0) * max_page_size_per_seq, max_page_size_per_seq, device=cache_seqlens.device).to(torch.int32)
+        num_pages_per_seq = torch.arange(
+            0,
+            cache_seqlens.size(0) * max_page_size_per_seq,
+            max_page_size_per_seq,
+            device=cache_seqlens.device,
+        ).to(torch.int32)
         cu_seqlens_k = torch.concat(
             (
                 torch.zeros(1, dtype=torch.int32, device=cache_seqlens.device),

tests/kernel-dyn-perf.txt

@@ -0,0 +1,17 @@
+Loop 5 @ level 1: 3 total, 0 fill, 0 spill, 0 subroutine calls
+Loop 1 @ level 1: 1522 total, 0 fill, 0 spill, 0 subroutine calls
+Loop 4 @ level 2: 49 total, 0 fill, 0 spill, 0 subroutine calls
+Loop 3 @ level 2: 11 total, 0 fill, 0 spill, 0 subroutine calls
+Loop 2 @ level 2: 11 total, 0 fill, 0 spill, 0 subroutine calls
+Kernel name: kernel
+ # BBs : 69
+ # Asm Insts: 2565
+ # RA Iters = 1
+ Spill Size = 0
+ # Spills: 0
+ # Fills : 0
+Total dyn inst: 22680
+Fill dyn inst: 0
+Spill dyn inst: 0
+Percent Fill/Total dyn inst: 0.000000
+LL1(#2): {S-0, F-0}, LL2(#3): {S-0, F-0},