Skip to content

fix undefined values for tail elements in act quant kernels #4186

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
iamzainhuda merged 1 commit into from
Mar 27, 2026
+6 −4
Merged

fix undefined values for tail elements in act quant kernels #4186

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 6 additions & 4 deletions torchao/prototype/blockwise_fp8_training/kernels.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -318,7 +318,7 @@ def triton_fp8_blockwise_act_quant_lhs_kernel(
k_offs = pid_k * BLOCK_SIZE + tl.arange(0, BLOCK_SIZE)
x_offs = m_offs[:, None] * x_stride_dim_0 + k_offs[None, :] * x_stride_dim_1
x_mask = (m_offs[:, None] < M) & (k_offs[None, :] < K)
x = tl.load(x_ptr + x_offs, mask=x_mask)
x = tl.load(x_ptr + x_offs, mask=x_mask, other=0.0)

# Scales for (1 x block_size) groups, shape will be (NUM_GROUPS, 1)
amax = tl.clamp(tl.max(tl.abs(x), axis=1), min=EPS, max=float("inf")).to(tl.float64)
Expand All @@ -333,7 +333,8 @@ def triton_fp8_blockwise_act_quant_lhs_kernel(

# Write reciprocal scales
scale_offs = m_offs[:, None] * s_stride_dim_0 + pid_k * s_stride_dim_1
tl.store(s_ptr + scale_offs, tl.div_rn(1.0, scale))
scale_mask = m_offs[:, None] < M
tl.store(s_ptr + scale_offs, tl.div_rn(1.0, scale), mask=scale_mask)


@triton_op("torchao::triton_fp8_blockwise_act_quant_lhs", mutates_args={})
Expand Down Expand Up @@ -412,7 +413,7 @@ def triton_fp8_blockwise_act_quant_rhs_kernel(
k_offs = pid_k * NUM_GROUPS + tl.arange(0, NUM_GROUPS)
x_offs = m_offs[:, None] * x_stride_dim_0 + k_offs[None, :] * x_stride_dim_1
x_mask = (m_offs[:, None] < M) & (k_offs[None, :] < K)
x = tl.load(x_ptr + x_offs, mask=x_mask)
x = tl.load(x_ptr + x_offs, mask=x_mask, other=0.0)

# Column-wise scales for RHS operand, shape (1, block_size)
amax = tl.clamp(tl.max(tl.abs(x), axis=0), min=EPS, max=float("inf")).to(tl.float64)
Expand All @@ -427,7 +428,8 @@ def triton_fp8_blockwise_act_quant_rhs_kernel(

# Write scales
scale_offs = pid_m * s_stride_dim_0 + k_offs[None, :] * s_stride_dim_1
tl.store(s_ptr + scale_offs, tl.div_rn(1.0, scale))
scale_mask = k_offs[None, :] < K
tl.store(s_ptr + scale_offs, tl.div_rn(1.0, scale), mask=scale_mask)


@triton_op("torchao::triton_fp8_blockwise_act_quant_rhs", mutates_args={})
Expand Down
Loading