[Bugfix] Fix GDN KKT precision loss on Hopper GPUs by aligning tl.dot operand layout with WGMMA (#42076)
Signed-off-by: kermit <ckeming@outlook.com>
This commit is contained in:
@@ -83,7 +83,7 @@ def chunk_scaled_dot_kkt_fwd_kernel(
|
|||||||
)
|
)
|
||||||
b_k = tl.load(p_k, boundary_check=(0, 1))
|
b_k = tl.load(p_k, boundary_check=(0, 1))
|
||||||
b_kb = b_k * b_beta[:, None]
|
b_kb = b_k * b_beta[:, None]
|
||||||
b_A += tl.dot(b_kb.to(b_k.dtype), tl.trans(b_k))
|
b_A += tl.dot(b_kb, tl.trans(b_k).to(b_kb.dtype))
|
||||||
|
|
||||||
if USE_G:
|
if USE_G:
|
||||||
p_g = tl.make_block_ptr(g + bos * H + i_h, (T,), (H,), (i_t * BT,), (BT,), (0,))
|
p_g = tl.make_block_ptr(g + bos * H + i_h, (T,), (H,), (i_t * BT,), (BT,), (0,))
|
||||||
|
|||||||
Reference in New Issue
Block a user