Split bwd into more .cu files to speed up compilation

csrc/flash_attn/src/flash_fwd_split_hdim128_bf16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim128_fp16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim128_fp16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim160_bf16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim160_bf16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim160_fp16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim160_fp16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim192_bf16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim192_bf16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim192_fp16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim192_fp16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim256_bf16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim256_bf16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim256_fp16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim256_fp16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim32_bf16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim32_bf16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim32_fp16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim32_fp16_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"
 

csrc/flash_attn/src/flash_fwd_split_hdim64_bf16_causal_sm80.cu

-// Copyright (c) 2023, Tri Dao.
+// Copyright (c) 2024, Tri Dao.
 // Splitting the different head dimensions to different files to speed up compilation.
 // This file is auto-generated. See "generate_kernels.py"