修复CCE指令参数传输错误的问题-asc-devkit-AtomGit

Mimpl/c_api/instr_impl/npu_arch_3510/cube_datamove_impl/asc_copy_gm2l1_impl.h+1-1

	@@ -23,7 +23,7 @@ __aicore__ inline void asc_copy_gm2l1_impl(__cbuf__ void* dst, __gm__ void* src,
23	uint64_t src_stride, uint32_t dst_stride)	23	uint64_t src_stride, uint32_t dst_stride)
24	{	24	{
25	if ASC_IS_AIC {	25	if ASC_IS_AIC {
26	- copy_gm_to_cbuf_v2(dst, src, 0, n_burst, len_burst, pad_func_mode, src_stride, dst_stride);	26	+ copy_gm_to_cbuf_v2(dst, src, 0, n_burst, len_burst, pad_func_mode, 0, src_stride, dst_stride);
27	}	27	}
28	}	28	}
29		29

Mimpl/c_api/instr_impl/npu_arch_3510/cube_datamove_impl/asc_copy_l0c2l1_impl.h+40-24

	@@ -28,9 +28,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ half* dst, __cc__ float* sr
28	bool c0_pad_en)	28	bool c0_pad_en)
29	{	29	{
30	if ASC_IS_AIC {	30	if ASC_IS_AIC {
31	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	31	+ copy_matrix_cc_to_cbuf(
32	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	32	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
33	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	33	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		34	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		35	+ false, c0_pad_en, false, false, false);
34	}	36	}
35	}	37	}
36		38
	@@ -44,9 +46,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ int8_t* dst, __cc__ float*
44	bool c0_pad_en)	46	bool c0_pad_en)
45	{	47	{
46	if ASC_IS_AIC {	48	if ASC_IS_AIC {
47	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	49	+ copy_matrix_cc_to_cbuf(
48	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	50	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
49	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	51	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		52	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		53	+ false, c0_pad_en, false, false, false);
50	}	54	}
51	}	55	}
52		56
	@@ -60,9 +64,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ uint8_t* dst, __cc__ float*
60	bool c0_pad_en)	64	bool c0_pad_en)
61	{	65	{
62	if ASC_IS_AIC {	66	if ASC_IS_AIC {
63	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	67	+ copy_matrix_cc_to_cbuf(
64	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	68	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
65	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	69	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		70	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		71	+ false, c0_pad_en, false, false, false);
66	}	72	}
67	}	73	}
68		74
	@@ -75,9 +81,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ float* dst, __cc__ float* s
75	bool c0_pad_en)	81	bool c0_pad_en)
76	{	82	{
77	if ASC_IS_AIC {	83	if ASC_IS_AIC {
78	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	84	+ copy_matrix_cc_to_cbuf(
79	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	85	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
80	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	86	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		87	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		88	+ false, c0_pad_en, false, false, false);
81	}	89	}
82	}	90	}
83		91
	@@ -90,9 +98,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ half* dst, __cc__ int32_t*
90	bool c0_pad_en)	98	bool c0_pad_en)
91	{	99	{
92	if ASC_IS_AIC {	100	if ASC_IS_AIC {
93	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	101	+ copy_matrix_cc_to_cbuf(
94	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	102	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
95	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	103	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		104	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		105	+ false, c0_pad_en, false, false, false);
96	}	106	}
97	}	107	}
98		108
	@@ -105,9 +115,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ int8_t* dst, __cc__ int32_t
105	bool c0_pad_en)	115	bool c0_pad_en)
106	{	116	{
107	if ASC_IS_AIC {	117	if ASC_IS_AIC {
108	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	118	+ copy_matrix_cc_to_cbuf(
109	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	119	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
110	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	120	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		121	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		122	+ false, c0_pad_en, false, false, false);
111	}	123	}
112	}	124	}
113		125
	@@ -120,9 +132,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ uint8_t* dst, __cc__ int32_
120	bool c0_pad_en)	132	bool c0_pad_en)
121	{	133	{
122	if ASC_IS_AIC {	134	if ASC_IS_AIC {
123	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	135	+ copy_matrix_cc_to_cbuf(
124	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	136	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
125	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	137	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		138	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		139	+ false, c0_pad_en, false, false, false);
126	}	140	}
127	}	141	}
128		142
	@@ -135,9 +149,11 @@ __aicore__ inline void asc_copy_l0c2l1_impl(__cbuf__ int32_t* dst, __cc__ int32_
135	bool c0_pad_en)	149	bool c0_pad_en)
136	{	150	{
137	if ASC_IS_AIC {	151	if ASC_IS_AIC {
138	- copy_matrix_cc_to_cbuf(dst, src, 0, n_size, m_size, dst_stride, src_stride, clip_relu_pre, unit_flag_mode,	152	+ copy_matrix_cc_to_cbuf(
139	- static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en, static_cast<QuantMode_post>(quant_post),	153	+ dst, src, 0, n_size, m_size, dst_stride, src_stride, 0, clip_relu_pre, unit_flag_mode,
140	- relu_post, clip_relu_post, eltwise_op, eltwise_antq_en, c0_pad_en);	154	+ static_cast<QuantMode_t>(quant_pre), relu_pre, channel_split, nz2nd_en,
		155	+ static_cast<QuantMode_post>(quant_post), relu_post, clip_relu_post, false, eltwise_op, eltwise_antq_en,
		156	+ false, c0_pad_en, false, false, false);
141	}	157	}
142	}	158	}
143		159

Mtests/api/c_api/npu_arch_3510/cube_datamove/test_asc_copy_gm2l1.cpp+4-3

Mtests/api/c_api/stub/cce_stub.h+59-1

	@@ -66,6 +66,64 @@ using float8_e5m2_t = fp8_e5m2_t;
66	using float4_e1m2x2_t = fp4x2_e1m2_t;	66	using float4_e1m2x2_t = fp4x2_e1m2_t;
67	using float4_e2m1x2_t = fp4x2_e2m1_t;	67	using float4_e2m1x2_t = fp4x2_e2m1_t;
68		68
		69	+// ==========copy_matrix_cc_to_cbuf (25 params)===========
		70	+inline void copy_matrix_cc_to_cbuf(
		71	+ __cbuf__ half* dst_addr, __cc__ float* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		72	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		73	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		74	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		75	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		76	+{}
		77	+inline void copy_matrix_cc_to_cbuf(
		78	+ __cbuf__ int8_t* dst_addr, __cc__ float* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		79	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		80	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		81	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		82	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		83	+{}
		84	+inline void copy_matrix_cc_to_cbuf(
		85	+ __cbuf__ uint8_t* dst_addr, __cc__ float* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		86	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		87	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		88	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		89	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		90	+{}
		91	+inline void copy_matrix_cc_to_cbuf(
		92	+ __cbuf__ float* dst_addr, __cc__ float* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		93	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		94	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		95	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		96	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		97	+{}
		98	+inline void copy_matrix_cc_to_cbuf(
		99	+ __cbuf__ half* dst_addr, __cc__ int32_t* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		100	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		101	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		102	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		103	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		104	+{}
		105	+inline void copy_matrix_cc_to_cbuf(
		106	+ __cbuf__ int8_t* dst_addr, __cc__ int32_t* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		107	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		108	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		109	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		110	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		111	+{}
		112	+inline void copy_matrix_cc_to_cbuf(
		113	+ __cbuf__ uint8_t* dst_addr, __cc__ int32_t* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		114	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		115	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		116	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		117	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		118	+{}
		119	+inline void copy_matrix_cc_to_cbuf(
		120	+ __cbuf__ int32_t* dst_addr, __cc__ int32_t* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
		121	+ uint32_t loop_dst_stride, uint16_t loop_src_stride, uint8_t l2_cache_ctl, uint8_t clip_relu_pre,
		122	+ uint8_t unit_flag_ctl, uint64_t quant_pre, uint8_t relu_pre, bool split_en, bool NZ2ND_en, uint64_t quant_post,
		123	+ uint8_t relu_post, bool clip_relu_post, bool loop_enhance_en, uint8_t eltwise_op, bool eltwise_antq_en,
		124	+ bool loop_enhance_merge_en, bool C0_pad_en, bool wino_post_en, bool broadcast_en, bool NZ2DN_en)
		125	+{}
		126	+
69	// ==========copy_matrix_cc_to_cbuf_s4===========	127	// ==========copy_matrix_cc_to_cbuf_s4===========
70	inline void copy_matrix_cc_to_cbuf_s4(	128	inline void copy_matrix_cc_to_cbuf_s4(
71	__cbuf__ void* dst_addr, __cc__ float* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,	129	__cbuf__ void* dst_addr, __cc__ float* src_addr, uint8_t sid, uint16_t n_size, uint16_t m_size,
	@@ -501,7 +559,7 @@ inline void vbr(vector_f8e8m0& dst, fp8_e8m0_t value) {}
501		559
502	inline void copy_gm_to_cbuf_v2(	560	inline void copy_gm_to_cbuf_v2(
503	__cbuf__ void* dst, __gm__ void* src, uint8_t sid, uint32_t n_burst, uint32_t len_burst, uint8_t pad_func_mode,	561	__cbuf__ void* dst, __gm__ void* src, uint8_t sid, uint32_t n_burst, uint32_t len_burst, uint8_t pad_func_mode,
504	- uint64_t src_stride, uint32_t dst_stride)	562	+ uint8_t l2_cache_ctl, uint64_t src_stride, uint32_t dst_stride)
505	{}	563	{}
506	inline void img2colv2_cbuf_to_ca(	564	inline void img2colv2_cbuf_to_ca(
507	__ca__ int16_t* dst, __cbuf__ int16_t* src, uint16_t step_k, uint16_t step_m, uint16_t pos_k, uint16_t pos_m,	565	__ca__ int16_t* dst, __cbuf__ int16_t* src, uint16_t step_k, uint16_t step_m, uint16_t pos_k, uint16_t pos_m,

Mtests/api/tensor_api/stub/cce_stub.h+1-1

	@@ -438,7 +438,7 @@ inline void vbr(vector_f8e4m3& dst, fp8_e4m3fn_t value) {}
438	inline void vbr(vector_f8e5m2& dst, fp8_e5m2_t value) {}	438	inline void vbr(vector_f8e5m2& dst, fp8_e5m2_t value) {}
439	inline void vbr(vector_f8e8m0& dst, fp8_e8m0_t value) {}	439	inline void vbr(vector_f8e8m0& dst, fp8_e8m0_t value) {}
440		440
441	-inline void copy_gm_to_cbuf_v2(__cbuf__ void* dst, __gm__ void* src, uint8_t sid, uint32_t n_burst, uint32_t len_burst, uint8_t pad_func_mode, uint64_t src_stride, uint32_t dst_stride) {}	441	+inline void copy_gm_to_cbuf_v2(__cbuf__ void* dst, __gm__ void* src, uint8_t sid, uint32_t n_burst, uint32_t len_burst, uint8_t pad_func_mode, uint8_t l2_cache_ctl, uint64_t src_stride, uint32_t dst_stride) {}
442	inline void img2colv2_cbuf_to_ca(__ca__ int16_t* dst, __cbuf__ int16_t* src, uint16_t step_k, uint16_t step_m, uint16_t pos_k, uint16_t pos_m, uint8_t stride_w, uint8_t stride_h, uint8_t w_k,	442	inline void img2colv2_cbuf_to_ca(__ca__ int16_t* dst, __cbuf__ int16_t* src, uint16_t step_k, uint16_t step_m, uint16_t pos_k, uint16_t pos_m, uint8_t stride_w, uint8_t stride_h, uint8_t w_k,
443	uint8_t h_k, uint8_t dilation_w, uint8_t dilation_h, bool filter_w, bool filter_h, bool transpose, bool fmatrix_ctrl, uint16_t size_channel) {}	443	uint8_t h_k, uint8_t dilation_w, uint8_t dilation_h, bool filter_w, bool filter_h, bool transpose, bool fmatrix_ctrl, uint16_t size_channel) {}
444	inline void img2colv2_cbuf_to_ca(__ca__ uint16_t* dst, __cbuf__ uint16_t* src, uint16_t step_k, uint16_t step_m, uint16_t pos_k, uint16_t pos_m, uint8_t stride_w, uint8_t stride_h, uint8_t w_k,	444	inline void img2colv2_cbuf_to_ca(__ca__ uint16_t* dst, __cbuf__ uint16_t* src, uint16_t step_k, uint16_t step_m, uint16_t pos_k, uint16_t pos_m, uint8_t stride_w, uint8_t stride_h, uint8_t w_k,

	@@ -91,13 +91,14 @@ TEST_CUBE_DATAMOVE_GM2L1(GM2L1, asc_copy_gm2l1, load_gm_to_cbuf_2dv2_s4, void);
91		91
92		92
93	__aicore__ inline void copy_gm_to_cbuf_v2_stub(__cbuf__ void* dst, __gm__ void* src, uint8_t sid, uint32_t n_burst, uint32_t len_burst, uint8_t pad_func_mode,	93	__aicore__ inline void copy_gm_to_cbuf_v2_stub(__cbuf__ void* dst, __gm__ void* src, uint8_t sid, uint32_t n_burst, uint32_t len_burst, uint8_t pad_func_mode,
94	- uint64_t src_stride, uint32_t dst_stride) {	94	+ uint8_t l2_cache_ctl, uint64_t src_stride, uint32_t dst_stride) {
95	EXPECT_EQ(dst, reinterpret_cast<__cbuf__ void *>(11));	95	EXPECT_EQ(dst, reinterpret_cast<__cbuf__ void *>(11));
96	EXPECT_EQ(src, reinterpret_cast<__gm__ void *>(22));	96	EXPECT_EQ(src, reinterpret_cast<__gm__ void *>(22));
97	EXPECT_EQ(sid, static_cast<uint8_t>(0));	97	EXPECT_EQ(sid, static_cast<uint8_t>(0));
98	EXPECT_EQ(n_burst, static_cast<uint32_t>(33));	98	EXPECT_EQ(n_burst, static_cast<uint32_t>(33));
99	EXPECT_EQ(len_burst, static_cast<uint32_t>(44));	99	EXPECT_EQ(len_burst, static_cast<uint32_t>(44));
100	EXPECT_EQ(pad_func_mode, static_cast<uint8_t>(55));	100	EXPECT_EQ(pad_func_mode, static_cast<uint8_t>(55));
		101	+ EXPECT_EQ(l2_cache_ctl, static_cast<uint8_t>(0));
101	EXPECT_EQ(src_stride, static_cast<uint64_t>(66));	102	EXPECT_EQ(src_stride, static_cast<uint64_t>(66));
102	EXPECT_EQ(dst_stride, static_cast<uint64_t>(77));	103	EXPECT_EQ(dst_stride, static_cast<uint64_t>(77));
103	}	104	}
	@@ -115,7 +116,7 @@ protected:
115	TEST_F(TEST_COPY_GM_TO_L1, TEST_COPY_GM_TO_L1)	116	TEST_F(TEST_COPY_GM_TO_L1, TEST_COPY_GM_TO_L1)
116	{	117	{
117	MOCKER_CPP(copy_gm_to_cbuf_v2, void(__cbuf__ void, __gm__ void, uint8_t, uint32_t, uint32_t, uint8_t,	118	MOCKER_CPP(copy_gm_to_cbuf_v2, void(__cbuf__ void, __gm__ void, uint8_t, uint32_t, uint32_t, uint8_t,
118	- uint64_t, uint32_t))	119	+ uint8_t, uint64_t, uint32_t))
119	.times(1)	120	.times(1)
120	.will(invoke(copy_gm_to_cbuf_v2_stub));	121	.will(invoke(copy_gm_to_cbuf_v2_stub));
121		122
	@@ -135,7 +136,7 @@ TEST_F(TEST_COPY_GM_TO_L1, TEST_COPY_GM_TO_L1)
135	TEST_F(TEST_COPY_GM_TO_L1, TEST_COPY_GM_TO_L1_SYNC)	136	TEST_F(TEST_COPY_GM_TO_L1, TEST_COPY_GM_TO_L1_SYNC)
136	{	137	{
137	MOCKER_CPP(copy_gm_to_cbuf_v2, void(__cbuf__ void, __gm__ void, uint8_t, uint32_t, uint32_t, uint8_t,	138	MOCKER_CPP(copy_gm_to_cbuf_v2, void(__cbuf__ void, __gm__ void, uint8_t, uint32_t, uint32_t, uint8_t,
138	- uint64_t, uint32_t))	139	+ uint8_t, uint64_t, uint32_t))
139	.times(1)	140	.times(1)
140	.will(invoke(copy_gm_to_cbuf_v2_stub));	141	.will(invoke(copy_gm_to_cbuf_v2_stub));
141		142