分类导航

程序问答发布时间：2022-06-01 发布网站：大佬教程 code.js-code.com

大佬教程收集整理的这篇文章主要介绍了从源故障构建 tensorflow 2.4：降低到 LLVM IR 失败，大佬教程大佬觉得挺不错的，现在分享给大家，也给大家做个参考。

如何解决从源故障构建 tensorflow 2.4：降低到 LLVM IR 失败？

开发过程中遇到从源故障构建 tensorflow 2.4：降低到 LLVM IR 失败的问题如何解决？下面主要结合日常开发的经验，给出你关于从源故障构建 tensorflow 2.4：降低到 LLVM IR 失败的解决方法建议，希望对你解决从源故障构建 tensorflow 2.4：降低到 LLVM IR 失败有所启发或帮助；

我正在使用 Clang 11.0 从源代码构建 Tensorflow 2.4。操作系统：Ubuntu 16.04。 CUDA：11.0

使用此类目标构建时会产生此类错误：

bazel build -c opt  --experimental_repo_remote_exec @org_tensorflow//tensorflow/core/kernels/mlir_generated:abs_f16_kernel

错误：

ERROR: /home/hzhao/.cache/bazel/_bazel_hzhao/dd20e335f94826116d98d72ed6038fbd/external/org_tensorflow/tensorflow/core/kernels/mlir_generated/BUILD:149:19: compile external/org_tensorflow/tensorflow/core/kernels/mlir_generated/abs_f16_kernel_cubin.7.5.bin Failed (Exit 1): linux-sandBox Failed: error execuTing command
  (cd /home/hzhao/.cache/bazel/_bazel_hzhao/dd20e335f94826116d98d72ed6038fbd/sandBox/linux-sandBox/5469/execroot/ && \
  exec env - \
    TMPDIR=/tmp \
  /home/hzhao/.cache/bazel/_bazel_hzhao/install/bc435791ccf8b3a8a0fe11b973c1ebb7/linux-sandBox -t 15 -w /home/hzhao/.cache/bazel/_bazel_hzhao/dd20e335f94826116d98d72ed6038fbd/sandBox/linux-sandBox/5469/execroot/ -w /tmp -w /dev/shm -D -- bazel-out/host/bin/external/org_tensorflow/tensorflow/compiler/mlir/tools/kernel_gen/tf_to_gpu_binary '--same_shape=0,1' '--unroll_factors=4' '--tile_sizes=256' '--arch=7.5' '--input=bazel-out/k8-opt/bin/external/org_tensorflow/tensorflow/core/kernels/mlir_generated/abs_f16.mlir' '--output=bazel-out/k8-opt/bin/external/org_tensorflow/tensorflow/core/kernels/mlir_generated/abs_f16_kernel_cubin.7.5.bin') linux-sandBox Failed: error execuTing command
  (cd /home/hzhao/.cache/bazel/_bazel_hzhao/dd20e335f94826116d98d72ed6038fbd/sandBox/linux-sandBox/5469/execroot/ && \
  exec env - \
    TMPDIR=/tmp \
  /home/hzhao/.cache/bazel/_bazel_hzhao/install/bc435791ccf8b3a8a0fe11b973c1ebb7/linux-sandBox -t 15 -w /home/hzhao/.cache/bazel/_bazel_hzhao/dd20e335f94826116d98d72ed6038fbd/sandBox/linux-sandBox/5469/execroot/ -w /tmp -w /dev/shm -D -- bazel-out/host/bin/external/org_tensorflow/tensorflow/compiler/mlir/tools/kernel_gen/tf_to_gpu_binary '--same_shape=0,1' '--unroll_factors=4' '--tile_sizes=256' '--arch=7.5' '--input=bazel-out/k8-opt/bin/external/org_tensorflow/tensorflow/core/kernels/mlir_generated/abs_f16.mlir' '--output=bazel-out/k8-opt/bin/external/org_tensorflow/tensorflow/core/kernels/mlir_generated/abs_f16_kernel_cubin.7.5.bin')
1615275075.984938474: src/main/tools/linux-sandBox-pID1.cc:208: writable: /home/hzhao/.cache/bazel/_bazel_hzhao/dd20e335f94826116d98d72ed6038fbd/sandBox/linux-sandBox/5469/execrootexternal/org_tensorflow/tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:194] disabling MliR crash reproducer,set env var `MliR_CRASH_reproduceR_DIRECTORY` to enable.
2021-03-08 23:31:16.021181: E external/org_tensorflow/tensorflow/compiler/mlir/tools/kernel_gen/tf_to_gpu_binary.cc:97] Internal: Lowering to LLVM IR Failed.
1615275076.027019459: src/main/tools/linux-sandBox-pID1.cc:410: wait returned pID=2,status=0x100
1615275076.027040792: src/main/tools/linux-sandBox-pID1.cc:428: child exited normally with code 1
1615275076.093379128: src/main/tools/linux-sandBox.cc:233: child exited normally with code 1
Target @org_tensorflow//tensorflow/core/kernels/mlir_generated:abs_f16_kernel Failed to build
Use --verbose_@R_870_4895@ to see the command lines of Failed build steps.
INFO: Elapsed time: 1.442s,Critical Path: 0.12s
INFO: 3 processes: 3 internal.
Failed: Build dID NOT complete successfully

abs_f16.mlir 的内容：

func @Abs(%arg0: tensor<?xf16>)
    -> tensor<?xf16> attributes {tf_entry,llvm.emit_c_interfacE} {
  %0 = "tf.Abs"(%arg0) : (tensor<?xf16>) -> tensor<?xf16>
  return %0 : tensor<?xf16>
}