diff options
Diffstat (limited to 'src/core/CL/cl_kernels/direct_convolution.cl')
-rw-r--r-- | src/core/CL/cl_kernels/direct_convolution.cl | 616 |
1 files changed, 144 insertions, 472 deletions
diff --git a/src/core/CL/cl_kernels/direct_convolution.cl b/src/core/CL/cl_kernels/direct_convolution.cl index 5d2a24e740..1de3737965 100644 --- a/src/core/CL/cl_kernels/direct_convolution.cl +++ b/src/core/CL/cl_kernels/direct_convolution.cl @@ -21,375 +21,12 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include "gemm_helpers.h" -#include "helpers_asymm.h" -#include "repeat.h" - -#if defined(IS_QUANTIZED) - -#if defined(ARM_COMPUTE_OPENCL_DOT8_ACC_ENABLED) && defined(cl_arm_integer_dot_product_accumulate_int8) -#define ARM_DOT(x, y, val) val = arm_dot_acc((x), (y), (val)); -#elif defined(ARM_COMPUTE_OPENCL_DOT8_ENABLED) && defined(cl_arm_integer_dot_product_int8) // defined(ARM_COMPUTE_OPENCL_DOT8_ENABLED) && defined(cl_arm_integer_dot_product_int8) -#define ARM_DOT(x, y, val) val += arm_dot((x), (y)); -#else // defined(ARM_COMPUTE_OPENCL_DOT8_ACC_ENABLED) && defined(cl_arm_integer_dot_product_accumulate_int8) -#define ARM_DOT(x, y, val) \ - ({ \ - val += (ACC_DATA_TYPE)x.s0 * (ACC_DATA_TYPE)y.s0; \ - val += (ACC_DATA_TYPE)x.s1 * (ACC_DATA_TYPE)y.s1; \ - val += (ACC_DATA_TYPE)x.s2 * (ACC_DATA_TYPE)y.s2; \ - val += (ACC_DATA_TYPE)x.s3 * (ACC_DATA_TYPE)y.s3; \ - }) -#endif // defined(ARM_COMPUTE_OPENCL_DOT8_ACC_ENABLED) && defined(cl_arm_integer_dot_product_accumulate_int8) - -#define ARM_DOT1(a, b, c) \ - ({ \ - ARM_DOT(((VEC_DATA_TYPE(SRC_DATA_TYPE, 4))(a, (VEC_DATA_TYPE(SRC_DATA_TYPE, 3))0)), ((VEC_DATA_TYPE(WEI_DATA_TYPE, 4))(b, (VEC_DATA_TYPE(WEI_DATA_TYPE, 3))0)), c); \ - }) -#define ARM_DOT2(a, b, c) \ - ({ \ - ARM_DOT(((VEC_DATA_TYPE(SRC_DATA_TYPE, 4))(a, (VEC_DATA_TYPE(SRC_DATA_TYPE, 2))0)), ((VEC_DATA_TYPE(WEI_DATA_TYPE, 4))(b, (VEC_DATA_TYPE(WEI_DATA_TYPE, 2))0)), c); \ - }) -#define ARM_DOT3(a, b, c) \ - ({ \ - ARM_DOT(((VEC_DATA_TYPE(SRC_DATA_TYPE, 4))(a, (SRC_DATA_TYPE)0)), ((VEC_DATA_TYPE(WEI_DATA_TYPE, 4))(b, (WEI_DATA_TYPE)0)), c); \ - }) -#define ARM_DOT4(a, b, c) \ - ({ \ - ARM_DOT(a, b, c); \ - }) -#define ARM_DOT8(a, b, c) \ - ({ \ - ARM_DOT4((a.lo), (b.lo), c); \ - ARM_DOT4((a.hi), (b.hi), c); \ - }) -#define ARM_DOT16(a, b, c) \ - ({ \ - ARM_DOT8((a.lo), (b.lo), c); \ - ARM_DOT8((a.hi), (b.hi), c); \ - }) - -#define ARM_OFFSET1(a, b, c) \ - ({ \ - c += (ACC_DATA_TYPE)a * (ACC_DATA_TYPE)b; \ - }) -#define ARM_OFFSET2(a, b, c) \ - ({ \ - c += (ACC_DATA_TYPE)a.s0 * (ACC_DATA_TYPE)b; \ - c += (ACC_DATA_TYPE)a.s1 * (ACC_DATA_TYPE)b; \ - }) -#define ARM_OFFSET3(a, b, c) \ - ({ \ - ARM_OFFSET2(a, b, c); \ - c += (ACC_DATA_TYPE)a.s2 * (ACC_DATA_TYPE)b; \ - }) -#define ARM_OFFSET4(a, b, c) \ - ({ \ - ARM_OFFSET3(a, b, c); \ - c += (ACC_DATA_TYPE)a.s3 * (ACC_DATA_TYPE)b; \ - }) -#define ARM_OFFSET8(a, b, c) \ - ({ \ - ARM_OFFSET4((a.lo), (b), c); \ - ARM_OFFSET4((a.hi), (b), c); \ - }) -#define ARM_OFFSET16(a, b, c) \ - ({ \ - ARM_OFFSET8((a.lo), (b), c); \ - ARM_OFFSET8((a.hi), (b), c); \ - }) - -#if N0 == 1 -#define ARM_OFFSET_K0XN0(k0, a, b, a_offset, b_offset, c) \ - ({ \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##0), (a_offset), (c)); \ - }) -#elif N0 == 2 // N) == 3 -#define ARM_OFFSET_K0XN0(k0, a, b, a_offset, b_offset, c) \ - ({ \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##0), (a_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##1), (a_offset), (c.s1)); \ - }) -#elif N0 == 3 // N0 == 3 -#define ARM_OFFSET_K0XN0(k0, a, b, a_offset, b_offset, c) \ - ({ \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##0), (a_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##1), (a_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##2), (a_offset), (c.s2)); \ - }) -#elif N0 == 4 // N0 == 4 -#define ARM_OFFSET_K0XN0(k0, a, b, a_offset, b_offset, c) \ - ({ \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##0), (a_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##1), (a_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##2), (a_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s3)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##3), (a_offset), (c.s3)); \ - }) -#elif N0 == 8 // N0 == 8 -#define ARM_OFFSET_K0XN0(k0, a, b, a_offset, b_offset, c) \ - ({ \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##0), (a_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##1), (a_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##2), (a_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s3)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##3), (a_offset), (c.s3)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s4)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##4), (a_offset), (c.s4)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s5)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##5), (a_offset), (c.s5)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s6)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##6), (a_offset), (c.s6)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s7)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##7), (a_offset), (c.s7)); \ - }) -#elif N0 == 16 // N0 == 16 -#define ARM_OFFSET_K0XN0(k0, a, b, a_offset, b_offset, c) \ - ({ \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##0), (a_offset), (c.s0)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##1), (a_offset), (c.s1)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##2), (a_offset), (c.s2)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s3)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##3), (a_offset), (c.s3)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s4)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##4), (a_offset), (c.s4)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s5)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##5), (a_offset), (c.s5)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s6)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##6), (a_offset), (c.s6)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s7)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##7), (a_offset), (c.s7)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s8)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##8), (a_offset), (c.s8)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.s9)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##9), (a_offset), (c.s9)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.sA)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##A), (a_offset), (c.sA)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.sB)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##B), (a_offset), (c.sB)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.sC)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##C), (a_offset), (c.sC)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.sD)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##D), (a_offset), (c.sD)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.sE)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##E), (a_offset), (c.sE)); \ - CONCAT(ARM_OFFSET, k0) \ - ((a), (b_offset), (c.sF)); \ - CONCAT(ARM_OFFSET, k0) \ - ((b##F), (a_offset), (c.sF)); \ - }) -#else // N0 not supported -#error "N0 value not supported" -#endif // N0 conditions -#else // defined(IS_QUANTIZED) - -#define ARM_DOT1(a, b, c) \ - ({ \ - c += (ACC_DATA_TYPE)a * (ACC_DATA_TYPE)b; \ - }) -#define ARM_DOT2(a, b, c) \ - ({ \ - c += (ACC_DATA_TYPE)a.s0 * (ACC_DATA_TYPE)b.s0; \ - c += (ACC_DATA_TYPE)a.s1 * (ACC_DATA_TYPE)b.s1; \ - }) -#define ARM_DOT3(a, b, c) \ - ({ \ - ARM_DOT2(a, b, c); \ - c += (ACC_DATA_TYPE)a.s2 * (ACC_DATA_TYPE)b.s2; \ - }) -#define ARM_DOT4(a, b, c) \ - ({ \ - ARM_DOT3(a, b, c); \ - c += (ACC_DATA_TYPE)a.s3 * (ACC_DATA_TYPE)b.s3; \ - }) -#define ARM_DOT8(a, b, c) \ - ({ \ - ARM_DOT4((a.lo), (b.lo), c); \ - ARM_DOT4((a.hi), (b.hi), c); \ - }) -#define ARM_DOT16(a, b, c) \ - ({ \ - ARM_DOT8((a.lo), (b.lo), c); \ - ARM_DOT8((a.hi), (b.hi), c); \ - }) -#endif // defined(IS_QUANTIZED) -#if N0 == 1 -#define ARM_DOT_K0XN0(k0, a, b, c) \ - ({ \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##0), (c)); \ - }) -#elif N0 == 2 // N) == 3 -#define ARM_DOT_K0XN0(k0, a, b, c) \ - ({ \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##0), (c.s0)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##1), (c.s1)); \ - }) -#elif N0 == 3 // N0 == 3 -#define ARM_DOT_K0XN0(k0, a, b, c) \ - ({ \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##0), (c.s0)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##1), (c.s1)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##2), (c.s2)); \ - }) -#elif N0 == 4 // N0 == 4 -#define ARM_DOT_K0XN0(k0, a, b, c) \ - ({ \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##0), (c.s0)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##1), (c.s1)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##2), (c.s2)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##3), (c.s3)); \ - }) -#elif N0 == 8 // N0 == 8 -#define ARM_DOT_K0XN0(k0, a, b, c) \ - ({ \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##0), (c.s0)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##1), (c.s1)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##2), (c.s2)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##3), (c.s3)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##4), (c.s4)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##5), (c.s5)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##6), (c.s6)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##7), (c.s7)); \ - }) -#elif N0 == 16 // N0 == 16 -#define ARM_DOT_K0XN0(k0, a, b, c) \ - ({ \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##0), (c.s0)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##1), (c.s1)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##2), (c.s2)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##3), (c.s3)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##4), (c.s4)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##5), (c.s5)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##6), (c.s6)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##7), (c.s7)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##8), (c.s8)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##9), (c.s9)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##A), (c.sA)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##B), (c.sB)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##C), (c.sC)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##D), (c.sD)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##E), (c.sE)); \ - CONCAT(ARM_DOT, k0) \ - ((a), (b##F), (c.sF)); \ - }) -#else // N0 not supported -#error "N0 value not supported" -#endif // N0 conditions +#include "helpers.h" +#include "helpers_asymm.h" +#include "tile_helpers.h" +//! @cond Doxygen_Suppress /** OpenCL kernel to compute the direct convolution. * * @note Data layout supported: NHWC @@ -403,6 +40,9 @@ * @note The spatial dimensions of the destination tensor must be passed at compile time using -DDST_WIDTH and -DDST_HEIGHT (e.g. -DDST_WIDTH=96, -DDST_HEIGHT=64) * @note The channels of the source tensor must be passed at compile time using -DSRC_CHANNELS (e.g. -DSRC_CHANNELS=64) * @note The channels of the destination tensor must be passed at compile time using -DDST_CHANNELS (e.g. -DDDST_CHANNELS=64) + * @note The tensor type ("BUFFER" or "IMAGE") of the source tensor must be passed at compile time using -DSRC_TENSOR_TYPE (e.g. -DSRC_TENSOR_TYPE=BUFFER) + * @note The tensor type ("BUFFER" or "IMAGE") of the weights tensor must be passed at compile time using -DWEI_TENSOR_TYPE (e.g. -DWEI_TENSOR_TYPE=BUFFER) + * @note The tensor type ("BUFFER" or "IMAGE") of the destination tensor must be passed at compile time using -DDST_TENSOR_TYPE (e.g. -DDST_TENSOR_TYPE=BUFFER) * @note The data type of the source tensor must be passed at compile time using -DSRC_DATA_TYPE (e.g. -DSRC_DATA_TYPE=float) * @note The data type of the weights tensor must be passed at compile time using -DWEI_DATA_TYPE (e.g. -DWEI_DATA_TYPE=float) * @note The data type of the destination tensor must be passed at compile time using -DDST_DATA_TYPE (e.g. -DDST_DATA_TYPE=float) @@ -410,12 +50,12 @@ * @note The number of M0 rows (width*height) to process must be passed at compile time using -DM0 (e.g. -DM0=2) * @note The number of N0 output channels to process must be passed at compile time using -DN0 (e.g. -DN0=2) * @note The number of K0 inner accumulations must be passed at compile time using -DK0 (e.g. -DK0=2) - * @note The size of the partial store block in y must be passed at compile time using -DPARTIAL_STORE_M0 (e.g. -DPARTIAL_STORE_M0=1) - * @note The size of the partial store block in x must be passed at compile time using -DPARTIAL_STORE_N0 (e.g. -DPARTIAL_STORE_N0=1) + * @note The size of the partial store block in x must be passed at compile time using -DPARTIAL_N0 (e.g. -DPARTIAL_N0=1) + * @note The zero value must be passed at compile time using -DZERO_VALUE (e.g. -DZERO_VALUE=0) * @note Only the following configurations of M0, N0 and K0 are currently supported: - * - M0 = 1 + * - M0 = 1, 2, 3, 4, 5, .... n * - N0 = 2, 3, 4, 8, 16 - * - K0 = 2, 3, 4, 8, 16 + * - K0 = 2, 3, 4, 8, 16 (only 4, 8 and 16 if WEI_TENSOR_TYPE=IMAGE) * *@note In case of QASYMM8/QASYMM8_SIGNED, the following extra information must be passed at compile time: * - -DIS_QUANTIZED @@ -426,13 +66,15 @@ * - The weights offset e.g. -DWEI_OFFSET=4 * - The quantized zero value e.g. -DZERO_VALUE=4 * - * @param[in] src_ptr Pointer to the source tensor. Supported data type: F16/F32 + * @param[in] src_ptr Pointer to the source tensor. Supported data type: F16/F32/QASYMM8 * @param[in] src_stride_x Stride of the source tensor in X dimension (in bytes) * @param[in] src_step_x src_stride_x * number of elements along X processed per workitem(in bytes) * @param[in] src_stride_y Stride of the source tensor in Y dimension (in bytes) * @param[in] src_step_y src_stride_y * number of elements along Y processed per workitem(in bytes) * @param[in] src_stride_z Stride of the source tensor in Z dimension (in bytes) * @param[in] src_step_z src_stride_z * number of elements along Z processed per workitem(in bytes) + * @param[in] src_stride_w Stride of the source tensor in W dimension (in bytes) + * @param[in] src_step_w src_stride_w * number of elements along W processed per workitem(in bytes) * @param[in] src_offset_first_element_in_bytes The offset of the first element in the source tensor * @param[out] dst_ptr Pointer to the destination tensor. Supported data type: same as @p src_ptr * @param[in] dst_stride_x Stride of the destination tensor in X dimension (in bytes) @@ -441,6 +83,8 @@ * @param[in] dst_step_y dst_stride_y * number of elements along Y processed per workitem(in bytes) * @param[in] dst_stride_z Stride of the destination tensor in Z dimension (in bytes) * @param[in] dst_step_z dst_stride_z * number of elements along Z processed per workitem(in bytes) + * @param[in] dst_stride_w Stride of the destination tensor in W dimension (in bytes) + * @param[in] dst_step_w dst_stride_w * number of elements along W processed per workitem(in bytes) * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination tensor * @param[in] wei_ptr Pointer to the weights tensor. Supported data type: same as @p src_ptr * @param[in] wei_stride_x Stride of the weights tensor in X dimension (in bytes) @@ -449,156 +93,184 @@ * @param[in] wei_step_y wei_stride_y * number of elements along Y processed per workitem(in bytes) * @param[in] wei_stride_z Stride of the weights tensor in Z dimension (in bytes) * @param[in] wei_step_z wei_stride_z * number of elements along Z processed per workitem(in bytes) + * @param[in] wei_stride_w Stride of the weights tensor in W dimension (in bytes) + * @param[in] wei_step_w wei_stride_w * number of elements along W processed per workitem(in bytes) * @param[in] wei_offset_first_element_in_bytes The offset of the first element in the bias matrix * @param[in] bia_ptr (Optional) Pointer to the bias tensor Supported data type: same as @p src_ptr (if F32/F16) or S32 (if QASYMM8/QASYMM8_SIGNED) * @param[in] bia_stride_x (Optional) Stride of the bias tensor in X dimension (in bytes) * @param[in] bia_step_x (Optional) bia_stride_x * number of elements along X processed per workitem(in bytes) * @param[in] bia_offset_first_element_in_bytes (Optional) The offset of the first element in the bias matrix - * @param[in] wei_stride_w Stride of the weights tensor in W dimension (in bytes) */ +//! @endcond __kernel void direct_convolution_nhwc( - TENSOR3D_DECLARATION(src), - TENSOR3D_DECLARATION(dst), - TENSOR3D_DECLARATION(wei), + TENSOR4D(src, SRC_TENSOR_TYPE), + TENSOR4D(dst, DST_TENSOR_TYPE), + TENSOR4D(wei, WEI_TENSOR_TYPE), #if defined(HAS_BIAS) - VECTOR_DECLARATION(bia), + VECTOR_DECLARATION(bia) #endif // defined(HAS_BIAS) - unsigned int wei_stride_w) +) { -#if M0 != 1 -#error "M0: Only supported 1" -#endif // M0 != 1 - - const int cout = max((int)(get_global_id(0) * N0 - (N0 - PARTIAL_STORE_N0) % N0), 0); // input channels - const int mout = get_global_id(1); // width x height - const int zout = get_global_id(2); // batch size index + // All the tensor dimensions are passed at compile time. + // In case of dynamic tensor support, the following dimensions should be passed as function argument. +#define _IWEI_WIDTH WEI_WIDTH +#define _IWEI_HEIGHT WEI_HEIGHT +#define _ISRC_WIDTH SRC_WIDTH +#define _ISRC_HEIGHT SRC_HEIGHT +#define _ISRC_CHANNELS SRC_CHANNELS +#define _IDST_WIDTH DST_WIDTH +#define _IDST_HEIGHT DST_HEIGHT +#define _IDST_CHANNELS DST_CHANNELS + + // If quantized, the output tile has to be quantized first before being stored to global memory +#if defined(IS_QUANTIZED) +#define _IOUTPUT_TILE cq +#else // defined(IS_QUANTIZED) +#define _IOUTPUT_TILE c +#endif // defined(IS_QUANTIZED) - REPEAT_VAR_INIT_TO_CONST(16, int, zero, 0); - REPEAT_VAR_INIT_TO_CONST(M0, int, xi, 0); - REPEAT_VAR_INIT_TO_CONST(M0, int, yi, 0); + const int cout = GET_SPATIAL_IDX(0, N0, PARTIAL_N0); // OFM + const int mout = GET_SPATIAL_IDX(1, M0, 0); // WIDTH x HEIGHT + const int bout = GET_SPATIAL_IDX(2, 1, 0); // BATCH SIZE IDX -#define LINEAR_2_COORDS(i) \ - xi##i = ((mout * M0 + i) % DST_WIDTH) * STRIDE_X; \ - yi##i = ((mout * M0 + i) / DST_WIDTH) * STRIDE_Y; \ - xi##i -= PAD_LEFT; \ - yi##i -= PAD_TOP; + // .v = access the whole vector (OpenCL vector) + // .s[x] = access the vector element at position x (scalar access) + TILE(int, M0, 1, xi) = { { 0 } }; + TILE(int, M0, 1, yi) = { { 0 } }; // Convert the linear index to coordinate - LINEAR_2_COORDS(0); - -#undef LINEAR_2_COORDS + LOOP_UNROLLING(int, i, 0, M0, 1) + { + xi[i].v = ((mout + i) % _IDST_WIDTH) * STRIDE_X; + yi[i].v = ((mout + i) / _IDST_WIDTH) * STRIDE_Y; + xi[i].v -= PAD_LEFT; + yi[i].v -= PAD_TOP; + } - uint src_offset = src_offset_first_element_in_bytes + zout * src_stride_y * (SRC_WIDTH * SRC_HEIGHT); - uint wei_offset = wei_offset_first_element_in_bytes + cout * wei_stride_w; + uint wei_x = 0; // Initialize the accumulators - REPEAT_VAR_INIT_TO_CONST(M0, VEC_DATA_TYPE(ACC_DATA_TYPE, N0), c, 0); + TILE(ACC_DATA_TYPE, M0, N0, c) = { { 0 } }; - for(int i = 0; i < (WEI_WIDTH * WEI_HEIGHT); ++i) + for(int i = 0; i < (_IWEI_WIDTH * _IWEI_HEIGHT); ++i) { - int xk = i % WEI_WIDTH; - int yk = i / WEI_WIDTH; - - REPEAT_VAR_INIT_TO_CONST(M0, int, mi_valid_row, 0); - REPEAT_VAR_INIT_TO_CONST(M0, int, mi_mask, 0); + uint src_x = 0; + int xk = i % _IWEI_WIDTH; + int yk = i / _IWEI_WIDTH; - // Calculate the input row to read from source tensor -#define MI_INIT(i) \ - mi_valid_row##i = max(min(xi##i + xk, SRC_WIDTH - 1), 0) + max(min(yi##i + yk, SRC_HEIGHT - 1), 0) * SRC_WIDTH; \ - mi_mask##i = (xi##i + xk) >= 0 && (xi##i + xk) < SRC_WIDTH && (yi##i + yk) >= 0 && (yi##i + yk) < SRC_HEIGHT; + TILE(int, M0, 1, src_indirect_y) = { { 0 } }; + TILE(int, M0, 1, src_indirect_mask) = { { 0 } }; - MI_INIT(0); - -#undef MI_INIT + // Calculate the source indirect Y and the source indirect mask + // Since the indirect Y is clamped when out-of-bound, the mask is used to + // force to zero the out-of-bound values + LOOP_UNROLLING(int, i, 0, M0, 1) + { + src_indirect_y[i].v = (CLAMP(xi[i].v + xk, 0, (int)_ISRC_WIDTH - 1) + CLAMP(yi[i].v + yk, 0, (int)_ISRC_HEIGHT - 1) * _ISRC_WIDTH); + src_indirect_y[i].v += bout * (int)_ISRC_WIDTH * (int)_ISRC_HEIGHT; + src_indirect_mask[i].v = ((xi[i].v + xk) >= 0 && (xi[i].v + xk) < (int)_ISRC_WIDTH && (yi[i].v + yk) >= 0 && (yi[i].v + yk) < (int)_ISRC_HEIGHT); + } int k = 0; - for(; k <= (SRC_CHANNELS - K0); k += K0) + for(; k <= (_ISRC_CHANNELS - K0); k += K0) { - // Load values from src tensor - LOAD_BLOCK_INDIRECT(M0, K0, SRC_DATA_TYPE, a, src_ptr, src_offset + k * sizeof(SRC_DATA_TYPE), src_stride_y, mi_valid_row, mi_mask); + TILE(SRC_DATA_TYPE, M0, K0, a); + TILE(WEI_DATA_TYPE, N0, K0, b); - // Load values from weights tensor - LOAD_BLOCK(N0, K0, WEI_DATA_TYPE, b, wei_ptr, wei_offset, wei_stride_w, zero); + // Load tile from the src tensor + T_LOAD_INDIRECT(SRC_DATA_TYPE, M0, K0, SRC_TENSOR_TYPE, src, src_x, src_stride_y, src_indirect_y, a); -#if defined(IS_QUANTIZED) -#define TENSOR_DOT(K0, i) \ - if(mi_mask##i != 0) \ - { \ - ARM_DOT_K0XN0(K0, a##i, b, c##i); \ - ARM_OFFSET_K0XN0(K0, a##i, b, SRC_OFFSET, WEI_OFFSET, c##i); \ - } \ - else \ - { \ - ARM_DOT_K0XN0(K0, ((VEC_DATA_TYPE(SRC_DATA_TYPE, K0))ZERO_VALUE), b, c##i); \ - ARM_OFFSET_K0XN0(K0, ((VEC_DATA_TYPE(SRC_DATA_TYPE, K0))ZERO_VALUE), b, SRC_OFFSET, WEI_OFFSET, c##i); \ - } -#else // defined(IS_QUANTIZED) -#define TENSOR_DOT(K0, i) \ - ARM_DOT_K0XN0(K0, a##i, b, c##i); -#endif // defined(IS_QUANTIZED) + // Load tile from the weights tensor + T_LOAD(WEI_DATA_TYPE, N0, K0, WEI_TENSOR_TYPE, wei, wei_x, cout, wei_stride_w, b); - TENSOR_DOT(K0, 0); + // Fill with zero the out-of-bound rows + T_ROWSET_MASK(SRC_DATA_TYPE, M0, K0, ZERO_VALUE, a, src_indirect_mask); - wei_offset += K0 * sizeof(WEI_DATA_TYPE); + // Compute the matrix multiplication between two tiles + T_MMUL(SRC_DATA_TYPE, WEI_DATA_TYPE, ACC_DATA_TYPE, M0, N0, K0, NT, T, a, b, c); + + // Apply the offset correction (correction usually needed for asymmetric quantized computation) + // The computation is not performed if both SRC_OFFSET and WEI_OFFSET are zero + T_OFFSET_CORRECTION(ACC_DATA_TYPE, M0, N0, K0, SRC_OFFSET, WEI_OFFSET, a, b, c); + + src_x += K0; + wei_x += K0; } -#if(SRC_CHANNELS % K0) != 0 + // We voluntarily use SRC_CHANNELS rather than _DSRC_CHANNELS + // This #if directive should be removed in case of dynamic tensor support +#if((SRC_CHANNELS % K0) != 0) // Left-over accumulations - for(; k < SRC_CHANNELS; ++k) + for(; k < _ISRC_CHANNELS; ++k) { - // Load values from src tensor - LOAD_BLOCK_INDIRECT(M0, 1, SRC_DATA_TYPE, a, src_ptr, src_offset + k * sizeof(SRC_DATA_TYPE), src_stride_y, mi_valid_row, mi_mask); + TILE(SRC_DATA_TYPE, M0, 1, a); + TILE(WEI_DATA_TYPE, N0, 1, b); - // Load values from weights tensor - LOAD_BLOCK(N0, 1, WEI_DATA_TYPE, b, wei_ptr, wei_offset, wei_stride_w, zero); + // Load tile from the src tensor + T_LOAD_INDIRECT(SRC_DATA_TYPE, M0, 1, SRC_TENSOR_TYPE, src, src_x, src_stride_y, src_indirect_y, a); - TENSOR_DOT(1, 0); + // Load tile from the weights tensor + T_LOAD(WEI_DATA_TYPE, N0, 1, WEI_TENSOR_TYPE, wei, wei_x, cout, wei_stride_w, b); -#undef TENSOR_DOT + // Fill with zero the out-of-bound rows + T_ROWSET_MASK(SRC_DATA_TYPE, M0, 1, ZERO_VALUE, a, src_indirect_mask); - wei_offset += sizeof(WEI_DATA_TYPE); - } -#endif // (SRC_CHANNELS % K0) != 0 + // Compute the matrix multiplication between two tiles + T_MMUL(SRC_DATA_TYPE, WEI_DATA_TYPE, ACC_DATA_TYPE, M0, N0, 1, NT, T, a, b, c); - c0 += (SRC_CHANNELS * SRC_OFFSET * WEI_OFFSET); - } + // Apply the offset correction (operation usually needed for asymmetric quantized computation) + // The computation is not performed if both SRC_OFFSET and WEI_OFFSET are zero + T_OFFSET_CORRECTION(ACC_DATA_TYPE, M0, N0, 1, SRC_OFFSET, WEI_OFFSET, a, b, c); - __global uchar *dst_addr = dst_ptr + dst_offset_first_element_in_bytes + (cout * sizeof(DST_DATA_TYPE)) + (mout * M0 * dst_stride_y); + ++src_x; + ++wei_x; + } +#endif // ((SRC_CHANNELS % K0) != 0) + } - // Batched direct convolution - dst_addr += zout * dst_stride_y * (DST_WIDTH * DST_HEIGHT); + // Offset correction required for the quantized asymmetric computation + // The computation is not performed if both SRC_OFFSET and WEI_OFFSET are zero + T_ADD_CONSTANT(ACC_DATA_TYPE, M0, N0, c, (_IWEI_WIDTH * _IWEI_HEIGHT * _ISRC_CHANNELS * SRC_OFFSET * WEI_OFFSET), c); #if defined(HAS_BIAS) - __global uchar *bias_addr = bia_ptr + bia_offset_first_element_in_bytes + (cout * sizeof(BIA_DATA_TYPE)); + TILE(BIA_DATA_TYPE, 1, N0, bias0); - LOAD_BLOCK(1, N0, BIA_DATA_TYPE, bias, bias_addr, 0, zero0, zero); + T_LOAD(BIA_DATA_TYPE, 1, N0, BUFFER, bia, cout, 0, 0, bias0); // c = c + bias[broadcasted] - ADD_BLOCK_BROADCAST(M0, c, bias0); + T_ADD_BROADCAST_X(ACC_DATA_TYPE, M0, N0, c, bias0, c); + #endif // HAS_BIAS -#if defined(IS_QUANTIZED) + TILE(uint, M0, 1, dst_indirect_y); - REPEAT_VAR_INIT_TO_CONST(M0, VEC_DATA_TYPE(DST_DATA_TYPE, N0), cq, 0); + // Calculate the destination indirect Y + LOOP_UNROLLING(int, i, 0, M0, 1) + { + dst_indirect_y[i].v = (uint)min(mout + i, (int)(_IDST_WIDTH * _IDST_HEIGHT) - 1); + dst_indirect_y[i].v += bout * (int)(_IDST_WIDTH * _IDST_HEIGHT); + } -#if DST_SHIFT < 0 -#define QUANTIZE(i) \ - c##i = ASYMM_MULT_BY_QUANT_MULTIPLIER_GREATER_THAN_ONE(c##i, DST_MULTIPLIER, DST_SHIFT, N0); \ - c##i = c##i + DST_OFFSET; \ - cq##i = CONVERT_SAT(c##i, VEC_DATA_TYPE(DST_DATA_TYPE, N0)); -#else // OUTPUT_SHIFT < 0 -#define QUANTIZE(i) \ - c##i = ASYMM_MULT_BY_QUANT_MULTIPLIER_LESS_THAN_ONE(c##i, DST_MULTIPLIER, DST_SHIFT, N0); \ - c##i = c##i + DST_OFFSET; \ - cq##i = CONVERT_SAT(c##i, VEC_DATA_TYPE(DST_DATA_TYPE, N0)); -#endif // OUTPUT_SHIFT < 0 + bool x_cond = PARTIAL_N0 != 0 && get_global_id(0) == 0; - QUANTIZE(0); +#if defined(IS_QUANTIZED) -#undef QUANTIZE + TILE(DST_DATA_TYPE, M0, N0, cq); - STORE_VECTOR_SELECT(cq, DST_DATA_TYPE, dst_addr, N0, PARTIAL_STORE_N0, PARTIAL_STORE_N0 != 0 && get_global_id(0) == 0); -#else // defined(IS_QUANTIZED) - STORE_VECTOR_SELECT(c, DST_DATA_TYPE, dst_addr, N0, PARTIAL_STORE_N0, PARTIAL_STORE_N0 != 0 && get_global_id(0) == 0); + // Quantize the tile + T_QUANTIZE8_ASYMMETRIC(ACC_DATA_TYPE, DST_DATA_TYPE, M0, N0, DST_OFFSET, DST_SHIFT, DST_MULTIPLIER, c, cq); #endif // defined(IS_QUANTIZED) + + // _IOUTPUT_TILE: c = fp32/fp16, cq=qasymm8 + // Store the tile in reverse order so the invalid values are overwritten with the valid ones + T_STORE_INDIRECT_WIDTH_SELECT(DST_DATA_TYPE, M0, N0, PARTIAL_N0, DST_TENSOR_TYPE, dst, cout, dst_stride_y, x_cond, _IOUTPUT_TILE, dst_indirect_y); + +#undef _IWEI_WIDTH +#undef _IWEI_HEIGHT +#undef _ISRC_WIDTH +#undef _ISRC_HEIGHT +#undef _ISRC_CHANNELS +#undef _IDST_WIDTH +#undef _IDST_HEIGHT +#undef _IDST_CHANNELS }
\ No newline at end of file |