Lines Matching refs:ohi
280 oho, ohi = cfg.define_split("tile_oh", oh, num_outputs=2)
285 [n, oho, owo, oco, kh, kw, ic, ohi, owi, oci],
288 [n, oho, owo, oco, kh, kw, ic, ohi, owi, oci],
289 [n, oho, owo, oco, ohi, kh, kw, ic, owi, oci],
290 [n, oho, owo, oco, ohi, kh, kw, owi, ic, oci],
291 [n, oho, owo, ohi, oco, kh, kw, owi, ic, oci],
296 cfg.define_annotate("ann_spatial", [ohi, owi, oci], policy="try_unroll_vec")
315 lambda n, oho, owo, kh, kw, ic, ohi, owi: data_pad[n][
316 (oho * OHI + ohi) * HSTR + kh * dilation_h
324 lambda n, oho, owo, ohi, owi, ic: data_pad[n][oho * OHI * HSTR + ohi][
346 lambda n, oho, owo, oco, ohi, owi, oci: te.sum(
347 data_vec[n, oho, owo, kh, kw, ohi, owi, ic].astype(out_dtype)
356 lambda n, oho, owo, oco, ohi, owi, oci: te.sum(
357 data_vec[n, oho, owo, ohi * HSTR + kh, owi * WSTR + kw, ic].astype(out_dtype)
393 oho, ohi = cfg["tile_oh"].apply(s, output, oh)
395 s[output].reorder(n, oho, owo, oco, ohi, owi, oci)
397 s, output, [ohi, owi, oci], axis_lens=[OHI, OWI, OCI], max_unroll=16, cfg=cfg
407 n, oho, owo, oco, ohi, owi, oci = s[conv].op.axis
409 cfg["reorder_conv"].apply(s, conv, [n, oho, owo, oco, kh, kw, ohi, owi, ic, oci])
419 s, conv, [ohi, owi, oci], axis_lens=[OHI, OWI, OCI], max_unroll=16, cfg=cfg
436 n, oho, owo, kh, kw, ic, ohi, owi = s[data_vec].op.axis
438 s[data_vec].unroll(ohi)
440 n, oho, owo, ohi, owi, ic = s[data_vec].op.axis