Skip to content
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

[rls-v3.8-pc] Fixed int4 weights alignment check in matmul #2945

Merged
merged 2 commits into from
Mar 27, 2025
Merged

[rls-v3.8-pc] Fixed int4 weights alignment check in matmul #2945

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
adb9d66
benchdnn: matmul: adjust int4 invalid cases
dzarukin Mar 18, 2025
44b0e82
common: matmul: adjust check for int4 tensors w.r.t. strides
dzarukin Mar 18, 2025
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
48 changes: 33 additions & 15 deletions src/common/matmul.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -355,21 +355,39 @@ status_t matmul_desc_init(matmul_desc_t *matmul_desc,
? utils::get_dims_mask(dst_desc->dims, op_d.bias_desc.dims, ndims)
: 0;

// TODO: requirement is for innermost dim to be multiple of 2 for
// the memory to be byte aligned.

// s4/u4/f4 weights requires n to be multiple of 2 to be byte aligned
VCHECK_MATMUL(IMPLICATION(utils::one_of(weights_desc->data_type,
data_type::s4, data_type::u4,
data_type::f4_e2m1, data_type::f4_e3m0),
weights_desc->dims[n_idx] % 2 == 0),
VERBOSE_BAD_DIM, "weights", n_idx);
// s4/u4/f4 src requires k to be multiple of 2 to be byte aligned
VCHECK_MATMUL(IMPLICATION(utils::one_of(src_desc->data_type, data_type::s4,
data_type::u4, data_type::f4_e2m1,
data_type::f4_e3m0),
src_desc->dims[k_idx_src] % 2 == 0),
VERBOSE_BAD_DIM, "src", n_idx);
using namespace data_type;
if (weights_desc->format_kind == format_kind::blocked
&& utils::one_of(
weights_desc->data_type, s4, u4, f4_e2m1, f4_e3m0)) {
const auto &wei_strides = weights_desc->format_desc.blocking.strides;

int n_unit_strides = 0;
for (int d = 0; d < ndims; d++) {
if (wei_strides[d] == 1) {
n_unit_strides++;
VCHECK_MATMUL(
n_unit_strides <= 1, VERBOSE_BAD_DIM, "weights", d);
}
VCHECK_MATMUL(
IMPLICATION(wei_strides[d] > 1, wei_strides[d] % 2 == 0),
VERBOSE_BAD_DIM, "weights", d);
}
}
if (src_desc->format_kind == format_kind::blocked
&& utils::one_of(src_desc->data_type, s4, u4, f4_e2m1, f4_e3m0)) {
const auto &src_strides = src_desc->format_desc.blocking.strides;

int n_unit_strides = 0;
for (int d = 0; d < ndims; d++) {
if (src_strides[d] == 1) {
n_unit_strides++;
VCHECK_MATMUL(n_unit_strides <= 1, VERBOSE_BAD_DIM, "src", d);
}
VCHECK_MATMUL(
IMPLICATION(src_strides[d] > 1, src_strides[d] % 2 == 0),
VERBOSE_BAD_DIM, "src", d);
}
}
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems like we should factor out the check, but fine for this backport:

bool valid_subbyte_strides(const memory_desc &md, int *bad_dim) { ... }

/* ... */

VCHECK_MATMUL(valid_subbyte_strides(weights_desc, &bad_dim), VERBOSE_BAD_DIM, "weights", bad_dim);
VCHECK_MATMUL(valid_subbyte_strides(src_desc, &bad_dim), VERBOSE_BAD_DIM, "src", bad_dim);


// check if other dims match.
for (int d = 0; d < ndims - 2; ++d) {
Expand Down
42 changes: 34 additions & 8 deletions tests/benchdnn/matmul/matmul.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -676,15 +676,41 @@ void skip_invalid_prb(const prb_t *prb, res_t *res) {
}
}

// Check int4 weights byte alignment if format is specified.
if ((prb->wei_dt() == dnnl_s4 || prb->wei_dt() == dnnl_u4)
&& (prb->n % 2)) {
BENCHDNN_PRINT(2,
"[INVALID][%s:%d]: Int4 Weights decompression requires OC "
"('%d') to be even.\n",
__FILE__, __LINE__, (int)prb->n);
res->state = SKIPPED;
res->reason = skip_reason::invalid_case;
return;
&& (!prb->strides[WEI].empty()
|| (prb->wtag != tag::any && prb->wtag != tag::undef))) {
const auto &weights_rt_dims = get_runtime_dims(
prb->weights_dims(), prb->weights_runtime_dim_mask());
const auto wei_md
= dnn_mem_t::init_md(prb->ndims, weights_rt_dims.data(),
prb->wei_dt(), prb->wtag, prb->strides[STRIDES_WEI]);

const auto wei_strides = query_md_strides(wei_md);
int n_unit_strides = 0;
for (int d = 0; d < query_md_ndims(wei_md); d++) {
if (wei_strides[d] == 1) {
n_unit_strides++;
if (n_unit_strides > 1) {
BENCHDNN_PRINT(2,
"[INVALID][%s:%d]: Int4 Weights decompression "
"requires byte alignment for the tensor.\n",
__FILE__, __LINE__);
res->state = SKIPPED;
res->reason = skip_reason::invalid_case;
return;
}
}
if (wei_strides[d] > 1 && (wei_strides[d] % 2)) {
BENCHDNN_PRINT(2,
"[INVALID][%s:%d]: Int4 Weights decompression requires "
"byte alignment for the tensor.\n",
__FILE__, __LINE__);
res->state = SKIPPED;
res->reason = skip_reason::invalid_case;
return;
}
}
}

auto src_rt_mask = prb->src_runtime_dim_mask();
Expand Down
Loading