Skip to content

Commit

Permalink
Fix (2/2) [TOPI] conv2d schedule code (apache#3648)
Browse files Browse the repository at this point in the history
    * Fix the tile_rx and tile_ry issue.

    Note that this patch depends on pull request apache#9 in tvm-distro.
  • Loading branch information
mingwayzhang committed Aug 6, 2019
1 parent 3b287c4 commit fbd6229
Show file tree
Hide file tree
Showing 2 changed files with 3 additions and 3 deletions.
2 changes: 1 addition & 1 deletion python/tvm/autotvm/tophub.py
Original file line number Diff line number Diff line change
Expand Up @@ -39,7 +39,7 @@
'arm_cpu': "v0.04",
'llvm': "v0.03",

'cuda': "v0.04",
'cuda': "v0.05",
'rocm': "v0.02",
'opencl': "v0.02",
'mali': "v0.05",
Expand Down
4 changes: 2 additions & 2 deletions topi/python/topi/cuda/conv2d_direct.py
Original file line number Diff line number Diff line change
Expand Up @@ -90,8 +90,8 @@ def schedule_direct_cuda(cfg, s, conv):
n, f, y, x = s[OL].op.axis
rc, ry, rx = s[OL].op.reduce_axis
rco, rci = cfg['tile_rc'].apply(s, OL, rc)
ryo, ryi = cfg['tile_rx'].apply(s, OL, ry)
rxo, rxi = cfg['tile_ry'].apply(s, OL, rx)
ryo, ryi = cfg['tile_ry'].apply(s, OL, ry)
rxo, rxi = cfg['tile_rx'].apply(s, OL, rx)
s[OL].reorder(rco, ryo, rxo, rci, ryi, rxi, n, f, y, x)

s[AA].compute_at(s[OL], rxo)
Expand Down

0 comments on commit fbd6229

Please sign in to comment.