aboutsummaryrefslogtreecommitdiff
path: root/src/gpu/cl
diff options
context:
space:
mode:
authorPablo Marquez Tello <pablo.tello@arm.com>2023-07-12 14:29:58 +0100
committerPablo Marquez Tello <pablo.tello@arm.com>2023-07-13 15:43:57 +0000
commit205ba243309baaec4bccfc82229139978d1a354e (patch)
tree6f3f0cd00b969b77b89ce612d85a339c6644cc56 /src/gpu/cl
parenta359ee9ff349448a744140b352444171efc29899 (diff)
downloadComputeLibrary-205ba243309baaec4bccfc82229139978d1a354e.tar.gz
Added S64/U64 support for the input in CLCast
* Partially resolves MLCE-1089 Change-Id: Ie3d2fc2f755ae99cdb17b57cc90bb3f99a1843e0 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/9909 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/gpu/cl')
-rw-r--r--src/gpu/cl/kernels/ClCastKernel.cpp4
-rw-r--r--src/gpu/cl/kernels/ClCastKernel.h10
2 files changed, 8 insertions, 6 deletions
diff --git a/src/gpu/cl/kernels/ClCastKernel.cpp b/src/gpu/cl/kernels/ClCastKernel.cpp
index 6baa31e710..991867d1f3 100644
--- a/src/gpu/cl/kernels/ClCastKernel.cpp
+++ b/src/gpu/cl/kernels/ClCastKernel.cpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2022 Arm Limited.
+ * Copyright (c) 2016-2023 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -54,7 +54,7 @@ Status validate_arguments(const ITensorInfo *src, const ITensorInfo *dst, Conver
1,
DataType::U8, DataType::S8, DataType::QASYMM8, DataType::QASYMM8_SIGNED, DataType::QSYMM8_PER_CHANNEL, DataType::S16,
DataType::U16, DataType::U32, DataType::S32, DataType::F16,
- DataType::F32);
+ DataType::F32, DataType::S64, DataType::U64);
ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(dst,
1,
DataType::U8, DataType::S8, DataType::QASYMM8, DataType::S16,
diff --git a/src/gpu/cl/kernels/ClCastKernel.h b/src/gpu/cl/kernels/ClCastKernel.h
index 7fadfa73d0..a021b3c78c 100644
--- a/src/gpu/cl/kernels/ClCastKernel.h
+++ b/src/gpu/cl/kernels/ClCastKernel.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2022 Arm Limited.
+ * Copyright (c) 2016-2023 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -53,12 +53,14 @@ public:
* - U16 -> U8, S8, S16, U32, S32, F16, F32
* - S16 -> U8, S8, U16, U32, S32, F16, F32
* - U32 -> U8, S8, U16, S16, S32, F16, F32
+ * - S64 -> U8, S8, U16, S16, U32, S32, F16, F32
+ * - U64 -> U8, S8, U16, S16, U32, S32, F16, F32
* - S32 -> U8, S8, U16, S16, U32, F16, F32
- * - F16 -> U8, S8, U16, S16, U32, F32
- * - F32 -> U8, S8, U16, S16, U32, F16
+ * - F16 -> U8, S8, U16, S16, U32, S32, F32
+ * - F32 -> U8, S8, U16, S16, U32, S32, F16
*
* @param[in] compile_context The compile context to be used.
- * @param[in] src The source tensor to convert. Data types supported: U8/S8/QSYMM8_PER_CHANNEL/U16/S16/U32/S32/F16/F32.
+ * @param[in] src The source tensor to convert. Data types supported: U8/S8/QSYMM8_PER_CHANNEL/U16/S16/U32/S32/U64/S64/F16/F32.
* @param[out] dst The destination tensor. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
* @param[in] policy Conversion policy
*/