-
Notifications
You must be signed in to change notification settings - Fork 131
Conversation
// RUN: %CPU_RUN_PLACEHOLDER %t.out | ||
// RUN: %GPU_RUN_PLACEHOLDER %t.out | ||
|
||
// XFAIL: gpu |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
could you please clarify what is the reason for xfail on gpu? Is there lack of support?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, there is lack of support of these sizes on the GPU
// init_and_multiply<int8_t, int32_t, 4, 5, 16, 32>(); | ||
// init_and_multiply<int8_t, int32_t, 4, 6, 16, 32>(); | ||
// init_and_multiply<int8_t, int32_t, 4, 7, 16, 32>(); | ||
// init_and_multiply<int8_t, int32_t, 4, 8, 16, 32>(); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why do we have many lines commented out? What we need/wait to enable them?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I uncommented them. All these calls with different sizes need support to remove the xfail
for (int i = 0; i < MATRIX_M; i++) { | ||
for (int j = 0; j < MATRIX_N; j++) { | ||
if constexpr (std::is_same_v<Ta, bfloat16> && std::is_same_v<Tc, float>) { | ||
if ((fabs(C[i][j]) - fabs(D[i][j])) > BF16_EPSILON) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It should be: "(fabs(C[i][j] - D[i][j])) > BF16_EPSILON" otherwise the test will always pass when values in C are smaller or equal to those in D.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
done
was replaced by #1628 |
No description provided.