Add -gsplit-dwarf flag to reduce debug section size and fix ckProfiler link errors (#2611)

Resolves R_X86_64_32 relocation out of range errors in grouped conv2d instances
by splitting debug information into separate .dwo files.

Add explicit cast to avoid signed/unsigned comparison warning.
This commit is contained in:
Jinchao Xu
2025-08-05 02:26:08 +08:00
committed by GitHub
parent 8655ba989c
commit 15eb493152
3 changed files with 11 additions and 6 deletions

View File

@@ -1,5 +1,5 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
@@ -14,7 +14,7 @@ CK_TILE_HOST void
reference_softmax(const HostTensor<InputType>& x, HostTensor<OutputType>& y, index_t dim = -1)
{
index_t rank = x.get_num_of_dimension();
assert(rank == y.get_num_of_dimension());
assert(static_cast<std::size_t>(rank) == y.get_num_of_dimension());
assert(dim == -1 || dim < rank);
index_t target_dim = dim == -1 ? (rank - 1) : dim;

View File

@@ -1,5 +1,5 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
@@ -38,8 +38,8 @@ CK_TILE_HOST void reference_topk(const HostTensor<DataType>& x,
{
// rank must be the same
index_t rank = x.get_num_of_dimension();
assert(rank == y_values.get_num_of_dimension());
assert(rank == y_indices.get_num_of_dimension());
assert(static_cast<std::size_t>(rank) == y_values.get_num_of_dimension());
assert(static_cast<size_t>(rank) == y_indices.get_num_of_dimension());
assert(dim == -1 || dim < rank);
index_t topk_dim = dim == -1 ? (rank - 1) : dim;
@@ -47,7 +47,8 @@ CK_TILE_HOST void reference_topk(const HostTensor<DataType>& x,
auto x_len = x.get_lengths();
assert(k <= topk_src_len);
assert(k == y_values.get_length(topk_dim) && k == y_indices.get_length(topk_dim));
assert(static_cast<size_t>(k) == y_values.get_length(topk_dim) &&
static_cast<size_t>(k) == y_indices.get_length(topk_dim));
index_t n_parallel = x.get_element_size() / topk_src_len;