[GitHub] [tvm] tkonolige commented on a change in pull request #10310: [ARM_CPU] Conv2d int8 intrinsic for cortex-A72

[GitBox <gi...@apache.org>]

tkonolige commented on a change in pull request #10310:
URL: https://github.com/apache/tvm/pull/10310#discussion_r810370223



##########
File path: python/tvm/topi/arm_cpu/conv2d_int8.py
##########
@@ -39,10 +40,10 @@ def _get_default_config(cfg, data, kernel, strides, padding, dilation, out_dtype
     wkl = _get_conv2d_workload(data, kernel, strides, padding, dilation, out_dtype)
     is_kernel_1x1 = wkl.kernel_h == 1 and wkl.kernel_w == 1
     if is_kernel_1x1:
-        conv2d_generic.fallback_schedule_cpu_1x1_int8(cfg, wkl, int32_lanes=2, num_int8_elements=4)
+        conv2d_generic.fallback_schedule_cpu_1x1_int8(cfg, wkl, int32_lanes=4, num_int8_elements=4)

Review comment:
       Yes, it was the wrong value all along. If you look down at `schedule_conv2d_NCHWc_int8` (https://github.com/apache/tvm/blob/2c0a7c2a7e1df8203c14e4f0326a99aeaceb967a/python/tvm/topi/arm_cpu/conv2d_int8.py#L110-L116), `int32_lanes` was set to 4 already.




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org