merge cuda and rocm files

[njzjz]

Merge source/lib/src/cuda and source/lib/src/rocm into source/lib/src/gpu.

• Define macros gpuGetLastError, gpuDeviceSynchronize, gpuMemcpy, gpuMemcpyDeviceToHost, gpuMemcpyHostToDevice, and gpuMemset to make them available for both CUDA and ROCm.
• Use <<< >>> syntax for both CUDA and ROCm. Per ROCm/HIP@cf78d85, it has been supported in HIP since 2018.
• Fix several int const numbers that should be double or float.
• For tabulate:
  • Fix WARP_SIZE for ROCm. Per [ROCm] define C10_WARP_SIZE to warpSize HIP constant pytorch/pytorch#64302, WARP_SIZE can be 32 or 64, so it should not be hardcoded to 64.
  • Add GpuShuffleSync. Per [HIP][device] 4 __shfl_sync functions are missing ROCm/HIP#1491, __shfl_sync is not supported by HIP.
  • After merging the code, Accelerate model compression #1274 should also work for ROCm.
  • Use the same ii for speedup tabulate cuda kernel by reducing shm using #830 and fix ROCm tabulate_fusion_se_a_grad_fifth_order_polynomial #2357. Although both of them work, ii has different meanings in these two PRs, but now it should be the same.
    • However, ii in tabulate_fusion_se_a_fifth_order_polynomial (rocm) added by Enable model compression for se_atten #2532 is wrong. After merging the codes, it should be corrected.
  • Optimization in speedup tabulate cuda kernel by reducing shm using #830 was not applied to ROCm.
  • __syncwarp is not supported by ROCm.
• After merging the code, synchronize in the beginning of all CUDA functions #2661 will be applied to ROCm. Although TF ROCm stream is still blocking (https://github.com/tensorflow/tensorflow/blob/9d1262082e761cd85d6726bcbdfdef331d6d72c6/tensorflow/compiler/xla/stream_executor/rocm/rocm_driver.cc#L566), we don't know whether it will change to non-blocking.
• There are several other differences between CUDA and ROCm.

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: +0.42% 🎉

Comparison is base (544875e) 75.24% compared to head (06c3eb6) 75.66%.

Additional details and impacted files

@@            Coverage Diff             @@
##            devel    #2844      +/-   ##
==========================================
+ Coverage   75.24%   75.66%   +0.42%     
==========================================
  Files         242      243       +1     
  Lines       24457    24789     +332     
  Branches     1580     1613      +33     
==========================================
+ Hits        18403    18757     +354     
+ Misses       5123     5070      -53     
- Partials      931      962      +31     

Files Changed	Coverage Δ
source/lib/include/gpu_cuda.h	70.00% <ø> (ø)

... and 15 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.