huggingface / huggingface/candle
Flash Attention not working on CUDA 12.1
- Dominant language
- Rust
- Stars
- 21.1k
- Forks
- 1.8k
- Avg merge
- 16h 42m
- Merged PRs (30d)
- 25
Description
I'm trying to use Flash Attention on an environment with CUDA 12.1 but it fails to compile. Is it expected?
Reproducing:
1. Start a Docker container with CUDA version 12.1.1 `docker run -it --gpus all nvidia/cuda:12.1.1-devel-ubuntu20.04 /bin/bash`
2. Install dependencies (`curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y`)
3. Run an example with "flash-attn" feature `cargo run --example llama --release --features "cuda flash-attn" -- --prompt "hi"`
Here's the full error log:
```
Compiling candle-flash-attn v0.4.2 (/root/maritaca/testes/candle/candle-original/candle-flash-attn)
error: failed to run custom build command for `candle-flash-attn v0.4.2 (/root/maritaca/testes/candle/candle-original/candle-flash-attn)`
Caused by:
process didn't exit successfully: `/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-5a27d196c8d06b8f/build-script-build` (exit status: 101)
--- stdout
cargo:rerun-if-changed=build.rs
cargo:rerun-if-changed=kernels/flash_api.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim128_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim160_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim192_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim224_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim256_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim32_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim64_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim96_fp16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim128_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim160_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim192_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim224_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim256_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim32_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim64_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_hdim96_bf16_sm80.cu
cargo:rerun-if-changed=kernels/flash_fwd_kernel.h
cargo:rerun-if-changed=kernels/flash_fwd_launch_template.h
cargo:rerun-if-changed=kernels/flash.h
cargo:rerun-if-changed=kernels/philox.cuh
cargo:rerun-if-changed=kernels/softmax.h
cargo:rerun-if-changed=kernels/utils.h
cargo:rerun-if-changed=kernels/kernel_traits.h
cargo:rerun-if-changed=kernels/block_info.h
cargo:rerun-if-changed=kernels/static_switch.h
cargo:info=["/usr", "/usr/local/cuda", "/opt/cuda", "/usr/lib/cuda", "C:/Program Files/NVIDIA GPU Computing Toolkit", "C:/CUDA"]
cargo:rerun-if-env-changed=CUDA_COMPUTE_CAP
cargo:rustc-env=CUDA_COMPUTE_CAP=80
--- stderr
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _NVVM_BRANCH_=nvvm
#$ _CUDART_=cudart
#$ _SPACE_=
#$ _HERE_=/usr/local/cuda/bin
#$ _CUDART_=cudart
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ TOP=/usr/local/cuda/bin/..
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim192_fp16_sm80.cu" -o "/tmp/tmpxft_000005a8_00000000-5_flash_fwd_hdim192_fp16_sm80.cpp4.ii"
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_api.cu" -o "/tmp/tmpxft_00000599_00000000-5_flash_api.cpp4.ii"
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ _NVVM_BRANCH_=nvvm
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ _SPACE_=
#$ CUDAFE_FLAGS=
#$ _CUDART_=cudart
#$ PTXAS_FLAGS=
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _NVVM_BRANCH_=nvvm
#$ _TARGET_DIR_=
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_DIR_=targets/x86_64-linux
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim224_fp16_sm80.cu" -o "/tmp/tmpxft_0000059a_00000000-5_flash_fwd_hdim224_fp16_sm80.cpp4.ii"
#$ TOP=/usr/local/cuda/bin/..
#$ TOP=/usr/local/cuda/bin/..
#$ _NVVM_BRANCH_=nvvm
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ _SPACE_=
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ _CUDART_=cudart
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ _HERE_=/usr/local/cuda/bin
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ _THERE_=/usr/local/cuda/bin
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ _TARGET_SIZE_=
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ _TARGET_DIR_=
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ _NVVM_BRANCH_=nvvm
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ _SPACE_=
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ _CUDART_=cudart
#$ _TARGET_DIR_=targets/x86_64-linux
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ _TARGET_DIR_=targets/x86_64-linux
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim96_bf16_sm80.cu" -o "/tmp/tmpxft_000005a9_00000000-5_flash_fwd_hdim96_bf16_sm80.cpp4.ii"
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim64_fp16_sm80.cu" -o "/tmp/tmpxft_0000059e_00000000-5_flash_fwd_hdim64_fp16_sm80.cpp4.ii"
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim224_bf16_sm80.cu" -o "/tmp/tmpxft_0000059b_00000000-5_flash_fwd_hdim224_bf16_sm80.cpp4.ii"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim128_fp16_sm80.cu" -o "/tmp/tmpxft_0000059c_00000000-5_flash_fwd_hdim128_fp16_sm80.cpp4.ii"
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ _NVVM_BRANCH_=nvvm
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _HERE_=/usr/local/cuda/bin
#$ TOP=/usr/local/cuda/bin/..
#$ _TARGET_DIR_=
#$ _THERE_=/usr/local/cuda/bin
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ _NVVM_BRANCH_=nvvm
#$ _TARGET_DIR_=targets/x86_64-linux
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ _SPACE_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ _NVVM_BRANCH_=nvvm
#$ CUDAFE_FLAGS=
#$ _CUDART_=cudart
#$ PTXAS_FLAGS=
#$ _HERE_=/usr/local/cuda/bin
#$ _SPACE_=
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _CUDART_=cudart
#$ _TARGET_DIR_=
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim32_fp16_sm80.cu" -o "/tmp/tmpxft_000005a3_00000000-5_flash_fwd_hdim32_fp16_sm80.cpp4.ii"
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ CUDAFE_FLAGS=
#$ CUDAFE_FLAGS=
#$ TOP=/usr/local/cuda/bin/..
#$ PTXAS_FLAGS=
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ PTXAS_FLAGS=
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim256_bf16_sm80.cu" -o "/tmp/tmpxft_000005a7_00000000-5_flash_fwd_hdim256_bf16_sm80.cpp4.ii"
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim128_bf16_sm80.cu" -o "/tmp/tmpxft_000005a1_00000000-5_flash_fwd_hdim128_bf16_sm80.cpp4.ii"
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim256_fp16_sm80.cu" -o "/tmp/tmpxft_000005a5_00000000-5_flash_fwd_hdim256_fp16_sm80.cpp4.ii"
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim160_bf16_sm80.cu" -o "/tmp/tmpxft_0000059f_00000000-5_flash_fwd_hdim160_bf16_sm80.cpp4.ii"
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ _NVVM_BRANCH_=nvvm
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ _TARGET_SIZE_=
#$ CUDAFE_FLAGS=
#$ _TARGET_DIR_=
#$ PTXAS_FLAGS=
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _TARGET_DIR_=targets/x86_64-linux
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim160_fp16_sm80.cu" -o "/tmp/tmpxft_000005a0_00000000-5_flash_fwd_hdim160_fp16_sm80.cpp4.ii"
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ TOP=/usr/local/cuda/bin/..
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ CUDAFE_FLAGS=
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PTXAS_FLAGS=
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim96_fp16_sm80.cu" -o "/tmp/tmpxft_0000059d_00000000-5_flash_fwd_hdim96_fp16_sm80.cpp4.ii"
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim64_bf16_sm80.cu" -o "/tmp/tmpxft_000005a6_00000000-5_flash_fwd_hdim64_bf16_sm80.cpp4.ii"
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim192_bf16_sm80.cu" -o "/tmp/tmpxft_000005a2_00000000-5_flash_fwd_hdim192_bf16_sm80.cpp4.ii"
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/root/maritaca/testes/candle/candle-original/target/release/deps:/root/maritaca/testes/candle/candle-original/target/release:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib:/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/root/.cargo/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ gcc -std=c++17 -D__CUDA_ARCH_LIST__=800 -E -x c++ -D__CUDACC__ -D__NVCC__ -D__CUDACC_EXTENDED_LAMBDA__ -D__CUDACC_RELAXED_CONSTEXPR__ -O3 -I"cutlass/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -U "__CUDA_NO_HALF_OPERATORS__" -U "__CUDA_NO_HALF_CONVERSIONS__" -U "__CUDA_NO_HALF2_OPERATORS__" -U "__CUDA_NO_BFLOAT16_CONVERSIONS__" -D__CUDACC_VER_MAJOR__=12 -D__CUDACC_VER_MINOR__=1 -D__CUDACC_VER_BUILD__=105 -D__CUDA_API_VER_MAJOR__=12 -D__CUDA_API_VER_MINOR__=1 -DCUDA_API_PER_THREAD_DEFAULT_STREAM=1 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -include "cuda_runtime.h" -m64 "kernels/flash_fwd_hdim32_bf16_sm80.cu" -o "/tmp/tmpxft_000005a4_00000000-5_flash_fwd_hdim32_bf16_sm80.cpp4.ii"
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim128_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim128_bf16_sm80-f1ff254233809e96.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim128_bf16_sm80.cu"
# stdout
# stderr
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim32_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim64_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_api.cu:1:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim32_bf16_sm80-aca7d8fdce93ef53.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim32_bf16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim64_fp16_sm80-a93563dad84e2972.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim64_fp16_sm80.cu"
# stdout
# stderr
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_api-59d12f2bec85f63.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_api.cu"
# stdout
# stderr
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim224_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim128_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim96_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
# --error 0x1 --
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim192_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim224_fp16_sm80-54d101fd022eab36.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim224_fp16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim160_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim128_fp16_sm80-759fdfecd1f0ed1c.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim128_fp16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim96_bf16_sm80-f51ba409eb93ce41.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim96_bf16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim256_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim32_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim192_fp16_sm80-3981fe996a7e8814.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim192_fp16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim160_bf16_sm80-b8e226bc00ecbaf1.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim160_bf16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim32_fp16_sm80-3a7585e74a278dc3.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim32_fp16_sm80.cu"
# stdout
# stderr
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim192_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim256_fp16_sm80-6bbb415157454ca9.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim256_fp16_sm80.cu"
# stdout
# stderr
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim96_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim224_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim256_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim192_bf16_sm80-f7453c8601d43b17.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim192_bf16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim64_bf16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim96_fp16_sm80-791226771e2c8c97.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim96_fp16_sm80.cu"
# stdout
# stderr
In file included from kernels/flash_fwd_launch_template.h:9,
from kernels/flash_fwd_hdim160_fp16_sm80.cu:5:
kernels/flash_fwd_kernel.h:7:10: fatal error: cute/algorithm/copy.hpp: No such file or directory
7 | #include
| ^~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
# --error 0x1 --
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim224_bf16_sm80-9b2b93dbac21043c.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim224_bf16_sm80.cu"
# stdout
# stderr
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim256_bf16_sm80-21dd0f7dd998e506.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim256_bf16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim64_bf16_sm80-eaa7ce7f57eb7351.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim64_bf16_sm80.cu"
# stdout
# stderr
# --error 0x1 --
thread '' panicked at /root/.cargo/registry/src/index.crates.io-6f17d22bba15001f/bindgen_cuda-0.1.4/src/lib.rs:262:21:
nvcc error while executing compiling: "nvcc" "--gpu-architecture=sm_80" "-c" "-o" "/root/maritaca/testes/candle/candle-original/target/release/build/candle-flash-attn-e3705818c6c58200/out/flash_fwd_hdim160_fp16_sm80-17db6cdd19f7f98b.o" "--default-stream" "per-thread" "-std=c++17" "-O3" "-U__CUDA_NO_HALF_OPERATORS__" "-U__CUDA_NO_HALF_CONVERSIONS__" "-U__CUDA_NO_HALF2_OPERATORS__" "-U__CUDA_NO_BFLOAT16_CONVERSIONS__" "-Icutlass/include" "--expt-relaxed-constexpr" "--expt-extended-lambda" "--use_fast_math" "--verbose" "kernels/flash_fwd_hdim160_fp16_sm80.cu"
# stdout
# stderr
```
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with candle-flash-attn/build.rs and the CUDA kernels listed in the build output, then reproduce the failure using the CUDA 12.1.1 Docker command and the llama example with the flash-attn feature. Done means the example compiles and runs successfully in that environment, with the CUDA build failure addressed.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- docker, rust
- Domain
- build-system, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100