Skip to content

[INTEL_HPU] fix addN 0x80 align issue #1873

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

Open
yanfeich wants to merge 1 commit into
base: develop
Choose a base branch
from
Open

[INTEL_HPU] fix addN 0x80 align issue #1873

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
15 changes: 14 additions & 1 deletion backends/intel_hpu/kernels/funcs.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@ limitations under the License. */
#pragma once

#include <cstdint>
#include <numeric>

#include "paddle/extension.h"
#include "paddle/phi/backends/device_ext.h"
Expand Down Expand Up @@ -348,7 +349,19 @@ class ConvertTensors {
for (auto d : dims) {
num_elements *= d;
}
int64_t addr_offset = num_elements * DataTypeSize(x.dtype());
int64_t element_bytes = num_elements * DataTypeSize(x.dtype());
auto addr_offset = element_bytes;

// Align addr_offset to the nearest multiple of 0x80 if it's not already
// aligned
if (element_bytes % 0x80 != 0) {
int64_t N = 1;
// 求 element_bytes 和 0x80 的最大公约数
int64_t gcd = std::gcd(element_bytes, 0x80);
N = 0x80 / gcd;
addr_offset = element_bytes * N;
num_list = num_list / N;
}

if (is_input) {
for (int64_t tensor_idx = 0; tensor_idx < num_list; tensor_idx++) {
Expand Down