ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-12 01:10:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert "feature:tf32:add initial conv3d fwd kernel support (#2763)" (#2848)

Browse Source 
This reverts commit c51102144f.
This commit is contained in:
Illia Silin
2025-09-15 08:27:04 -07:00
committed by GitHub
parent c51102144f
commit 03b59f8c76
44 changed files with 175 additions and 1085 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/CMakeLists.txt
View File

@@ -7,7 +7,6 @@ set(GROUPED_CONV3D_FWD
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_instance.cpp
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f16_instance.cpp
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_instance.cpp
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instance.cpp
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_16x16_instance.cpp
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f16_16x16_instance.cpp
xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_16x16_instance.cpp
@@ -35,7 +34,7 @@ set(GROUPED_CONV3D_FWD
xdl/mem/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_mem_intra_instance.cpp
xdl/mem/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f16_mem_intra_instance.cpp
xdl/mem/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_mem_intra_instance.cpp
xdl/comp/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_comp_instance.cpp
xdl/comp/device_grouped_conv3d_fwd_xdl_ngcdhw_gkczyx_ngkdhw_f32_comp_instance.cpp
xdl/comp/device_grouped_conv3d_fwd_xdl_ngcdhw_gkczyx_ngkdhw_bf16_comp_2x_instance.cpp

56
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd/xdl/device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instance.cpp
View File

@@ -1,56 +0,0 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
#include "ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp"
#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
void add_device_grouped_conv3d_fwd_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances(
std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleABD<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
F32,
F32,
Empty_Tuple,
F32,
PassThrough,
PassThrough,
PassThrough,
TF32,
TF32>>>& instances)
{
add_device_operation_instances(
instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
ConvFwdDefault>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
ConvFwd1x1P0>{});
add_device_operation_instances(
instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Empty_Tuple,
NDHWGK,
ConvFwd1x1S1P0>{});
}
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck

59
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_bias_bnorm_clamp/CMakeLists.txt
View File

@@ -2,7 +2,7 @@
set(GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP)
include(ShardInstantiation)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_instances
@@ -11,7 +11,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_instances
@@ -20,7 +20,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_instances
@@ -29,16 +29,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances
TEMPLATE_FILE xdl/device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instance.in
NUM_SHARDS 16
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_16x16_instances
@@ -47,7 +38,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_16x16_instances
@@ -56,7 +47,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_16x16_instances
@@ -67,7 +58,7 @@ generate_sharded_instantiations(
)
# large tensor
# NDHWGC, GKZYXC, NDHWGK
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_large_tensor_ndhwgc_gkzyxc_ndhwgk_bf16_instances
@@ -76,7 +67,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/large_tensor
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_large_tensor_ndhwgc_gkzyxc_ndhwgk_f16_instances
@@ -85,7 +76,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/large_tensor
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_large_tensor_ndhwgc_gkzyxc_ndhwgk_f32_instances
@@ -96,7 +87,7 @@ generate_sharded_instantiations(
)
# merged groups
# NDHWGC, GKZYXC, NDHWGK
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_merged_groups_ndhwgc_gkzyxc_ndhwgk_bf16_instances
@@ -105,7 +96,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/merged_groups
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_merged_groups_ndhwgc_gkzyxc_ndhwgk_f16_instances
@@ -114,7 +105,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/merged_groups
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_merged_groups_ndhwgc_gkzyxc_ndhwgk_f32_instances
@@ -125,7 +116,7 @@ generate_sharded_instantiations(
)
#mem
# NDHWGC, GKZYXC, NDHWGK
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_mem_intra_instances
@@ -134,7 +125,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/mem
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_mem_intra_instances
@@ -143,7 +134,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/mem
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_mem_intra_instances
@@ -153,7 +144,7 @@ generate_sharded_instantiations(
OUTPUT_DIR ${GENERATED_DIR}/xdl/mem
)
# NDHWGC, GKZYXC, NDHWGK
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_mem_inter_instances
@@ -162,7 +153,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/mem
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_mem_inter_instances
@@ -171,7 +162,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/mem
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_mem_inter_instances
@@ -182,7 +173,7 @@ generate_sharded_instantiations(
)
#comp
# NDHWGC, GKZYXC, NDHWGK
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_comp_instances
@@ -191,7 +182,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/comp
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_comp_instances
@@ -200,7 +191,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/comp
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_comp_instances
@@ -209,7 +200,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/comp
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_comp_2x_instances
@@ -218,7 +209,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/comp
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_comp_2x_instances
@@ -227,7 +218,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/comp
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_bf16_comp_part2_instances
@@ -236,7 +227,7 @@ generate_sharded_instantiations(
SRC_LIST GROUPED_conv3d_FWD_BIAS_BNORM_CLAMP
OUTPUT_DIR ${GENERATED_DIR}/xdl/comp
)
set(GENERATED_DIR ${CMAKE_CURRENT_BINARY_DIR}/generated)
generate_sharded_instantiations(
INSTANCES_NAME device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f16_comp_part2_instances

81
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_bias_bnorm_clamp/xdl/device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instance.in
View File

@@ -1,81 +0,0 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
#include "ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp"
#include "ck/utility/filter_tuple.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
using device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances =
std::vector<std::unique_ptr<
DeviceGroupedConvFwdMultipleABD<3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK, NDHWGK, NDHWGK, NDHWGK, NDHWGK>,
NDHWGK,
F32,
F32,
Tuple<F32, F32, F32, F32, F32>,
F32,
PassThrough,
PassThrough,
BiasNormalizeInInferClamp,
TF32,
TF32>>>;
// Compilation parameters for in[n, hi, wi, g, c] * wei[g, k, y, x, c] = out[n, ho, wo, g, k]
template <int Shards, int ShardIndex>
void add_device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances_shard(
device_grouped_conv3d_fwd_bias_bn_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances& instances)
{
add_device_operation_instances(
instances,
ck::util::filter_tuple_by_modulo_t<device_grouped_conv_fwd_xdl_f32_tf32_instances<
3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK, NDHWGK, NDHWGK, NDHWGK, NDHWGK>,
NDHWGK,
ConvFwdDefault,
Tuple<F32, F32, F32, F32, F32>,
BiasNormalizeInInferClamp>,
Shards,
ShardIndex>{});
add_device_operation_instances(
instances,
ck::util::filter_tuple_by_modulo_t<device_grouped_conv_fwd_xdl_f32_tf32_instances<
3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK, NDHWGK, NDHWGK, NDHWGK, NDHWGK>,
NDHWGK,
ConvFwd1x1P0,
Tuple<F32, F32, F32, F32, F32>,
BiasNormalizeInInferClamp>,
Shards,
ShardIndex>{});
add_device_operation_instances(
instances,
ck::util::filter_tuple_by_modulo_t<device_grouped_conv_fwd_xdl_f32_tf32_instances<
3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK, NDHWGK, NDHWGK, NDHWGK, NDHWGK>,
NDHWGK,
ConvFwd1x1S1P0,
Tuple<F32, F32, F32, F32, F32>,
BiasNormalizeInInferClamp>,
Shards,
ShardIndex>{});
}
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck

4
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_bias_clamp/CMakeLists.txt
View File

@@ -23,8 +23,6 @@ set(GROUPED_CONV3D_FWD
xdl/mem/device_grouped_conv3d_fwd_bias_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_mem_inter_instance.cpp
xdl/mem/device_grouped_conv3d_fwd_bias_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_mem_intra_instance.cpp
xdl/comp/device_grouped_conv3d_fwd_bias_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_comp_instance.cpp
xdl/device_grouped_conv3d_fwd_bias_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_tf32_instance.cpp
)
)
add_instance_library(device_grouped_conv3d_fwd_bias_clamp_instance ${GROUPED_CONV3D_FWD})

60
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_bias_clamp/xdl/device_grouped_conv3d_fwd_bias_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_tf32_instance.cpp
View File

@@ -1,60 +0,0 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
#include "ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp"
#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
void add_device_grouped_conv3d_fwd_bias_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances(
std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleABD<3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK>,
NDHWGK,
F32,
F32,
Tuple<F32>,
F32,
PassThrough,
PassThrough,
AddClamp,
TF32,
TF32>>>& instances)
{
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK>,
NDHWGK,
ConvFwdDefault,
Tuple<F32>,
AddClamp>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK>,
NDHWGK,
ConvFwd1x1P0,
Tuple<F32>,
AddClamp>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Tuple<NDHWGK>,
NDHWGK,
ConvFwd1x1S1P0,
Tuple<F32>,
AddClamp>{});
}
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck

4
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_clamp/CMakeLists.txt
View File

@@ -23,8 +23,6 @@ set(GROUPED_CONV3D_FWD
xdl/mem/device_grouped_conv3d_fwd_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_mem_inter_instance.cpp
xdl/mem/device_grouped_conv3d_fwd_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_mem_intra_instance.cpp
xdl/comp/device_grouped_conv3d_fwd_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_comp_instance.cpp
xdl/device_grouped_conv3d_fwd_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_tf32_instance.cpp
)
)
add_instance_library(device_grouped_conv3d_fwd_clamp_instance ${GROUPED_CONV3D_FWD})

60
library/src/tensor_operation_instance/gpu/grouped_conv3d_fwd_clamp/xdl/device_grouped_conv3d_fwd_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_fp32_tf32_instance.cpp
View File

@@ -1,60 +0,0 @@
// SPDX-License-Identifier: MIT
// Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
#include "ck/library/tensor_operation_instance/gpu/grouped_conv_fwd/device_grouped_conv_fwd_xdl_instance.hpp"
#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
namespace ck {
namespace tensor_operation {
namespace device {
namespace instance {
void add_device_grouped_conv3d_fwd_clamp_xdl_ndhwgc_gkzyxc_ndhwgk_f32_tf32_instances(
std::vector<std::unique_ptr<DeviceGroupedConvFwdMultipleABD<3,
NDHWGC,
GKZYXC,
Tuple<>,
NDHWGK,
F32,
F32,
Tuple<>,
F32,
PassThrough,
PassThrough,
Clamp,
TF32,
TF32>>>& instances)
{
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Tuple<>,
NDHWGK,
ConvFwdDefault,
Tuple<>,
Clamp>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Tuple<>,
NDHWGK,
ConvFwd1x1P0,
Tuple<>,
Clamp>{});
add_device_operation_instances(instances,
device_grouped_conv_fwd_xdl_f32_tf32_instances<3,
NDHWGC,
GKZYXC,
Tuple<>,
NDHWGK,
ConvFwd1x1S1P0,
Tuple<>,
Clamp>{});
}
} // namespace instance
} // namespace device
} // namespace tensor_operation
} // namespace ck

2
library/src/utility/host_tensor.cpp
View File

@@ -53,7 +53,7 @@ std::ostream& operator<<(std::ostream& os, const HostTensorDescriptor& desc)
os << "strides {";
LogRange(os, desc.GetStrides(), ", ");
os << "} ";
os << "}";
return os;
}