diff --git a/test/test_gpu/skip_tests_h100_triton_main.yaml b/test/test_gpu/skip_tests_h100_triton_main.yaml
index 4727832b..50b7aefe 100644
--- a/test/test_gpu/skip_tests_h100_triton_main.yaml
+++ b/test/test_gpu/skip_tests_h100_triton_main.yaml
@@ -7,17 +7,12 @@
 flash_attention:
   # thunderkittens cannot handle the default input shapes
   - tk
-  # FIXME: triton_tutorial_* kernels are broken
-  - triton_tutorial_flash_v2
-  - triton_tutorial_flash_v2_opt
-  - triton_tutorial_flash_v2_tma
+  # _ws kernels require Triton with warp specialization
   - triton_tutorial_flash_v2_ws
   - triton_tutorial_flash_v2_tma_ws
 fp8_attention:
   # fb-only kernel
   - colfax_fmha
-  # FIXME: triton_flash_v2 kernel is broken
-  - triton_flash_v2
 # fb-only kernels
 fp8_fused_quant_gemm_rowwise:
 fp8_gemm:
@@ -25,9 +20,9 @@ fp8_gemm:
   - triton_persistent_fp8_gemm
   - triton_tma_persistent_fp8_gemm
 gemm:
-  # out of shared memory
+  # FIXME: out of shared memory
   - triton_tma_persistent_matmul
-  # out of shared memory
+  # FIXME: out of shared memory
   - triton_tma_persistent_cached_matmul
   # internal only kernels
   - hstu_triton_matmul