local-llm-rn 1.0.0

package/cpp/ggml/src/ggml-hexagon/htp/softmax-ops.c

@@ -0,0 +1,421 @@
+#pragma clang diagnostic ignored "-Wunused-variable"
+#pragma clang diagnostic ignored "-Wunused-function"
+#pragma clang diagnostic ignored "-Wunused-but-set-variable"
+
+#include <HAP_farf.h>
+#include <HAP_perf.h>
+
+#include <math.h>
+#include <string.h>
+
+#include "hex-dma.h"
+#include "hvx-utils.h"
+#include "hex-fastdiv.h"
+
+#define GGML_COMMON_DECL_C
+#include "ggml-common.h"
+#include "htp-ctx.h"
+#include "htp-msg.h"
+#include "htp-ops.h"
+
+#define htp_softmax_preamble3                              \
+    const uint32_t ne00 = src0->ne[0];                     \
+    const uint32_t ne01 = src0->ne[1];                     \
+    const uint32_t ne02 = src0->ne[2];                     \
+    const uint32_t ne03 = src0->ne[3];                     \
+                                                           \
+    const uint32_t nb00 = src0->nb[0];                     \
+    const uint32_t nb01 = src0->nb[1];                     \
+    const uint32_t nb02 = src0->nb[2];                     \
+    const uint32_t nb03 = src0->nb[3];                     \
+                                                           \
+    const uint32_t ne10 = (src1->ne[0]) ? src1->ne[0] : 1; \
+    const uint32_t ne11 = (src1->ne[0]) ? src1->ne[1] : 1; \
+    const uint32_t ne12 = (src1->ne[0]) ? src1->ne[2] : 1; \
+    const uint32_t ne13 = (src1->ne[0]) ? src1->ne[3] : 1; \
+                                                           \
+    const uint32_t nb10 = (src1->ne[0]) ? src1->nb[0] : 1; \
+    const uint32_t nb11 = (src1->ne[0]) ? src1->nb[1] : 1; \
+    const uint32_t nb12 = (src1->ne[0]) ? src1->nb[2] : 1; \
+    const uint32_t nb13 = (src1->ne[0]) ? src1->nb[3] : 1; \
+                                                           \
+    const uint32_t ne0 = dst->ne[0];                       \
+    const uint32_t ne1 = dst->ne[1];                       \
+    const uint32_t ne2 = dst->ne[2];                       \
+    const uint32_t ne3 = dst->ne[3];                       \
+                                                           \
+    const uint32_t nb0 = dst->nb[0];                       \
+    const uint32_t nb1 = dst->nb[1];                       \
+    const uint32_t nb2 = dst->nb[2];                       \
+    const uint32_t nb3 = dst->nb[3];
+
+struct htp_softmax_context {
+    bool     use_f16;
+    bool     use_src1;
+    uint32_t n_head;
+    uint32_t n_head_log2;
+
+    float scale;
+    float max_bias;
+    float m0;
+    float m1;
+
+    uint32_t src0_nrows_per_thread;
+    struct fastdiv_values fastdiv_ne01;
+    struct fastdiv_values fastdiv_ne02;
+    struct fastdiv_values fastdiv_ne12; // For mask broadcasting
+    struct fastdiv_values fastdiv_ne13; // For mask broadcasting
+    size_t spad_stride;
+
+    struct htp_ops_context * octx;
+};
+
+static void init_softmax_ctx(struct htp_softmax_context * smctx, struct htp_ops_context * octx) {
+    const struct htp_tensor * src0 = &octx->src0;
+    const struct htp_tensor * src1 = &octx->src1;
+
+    memset(smctx, 0, sizeof(struct htp_softmax_context));
+
+    memcpy(&smctx->scale, (float *) octx->op_params, sizeof(float));
+    memcpy(&smctx->max_bias, (float *) octx->op_params + 1, sizeof(float));
+
+    smctx->n_head      = src0->ne[2];
+    smctx->n_head_log2 = 1u << (uint32_t) floor(log2(smctx->n_head));
+
+    smctx->m0 = powf(2.0f, -(smctx->max_bias) / smctx->n_head_log2);
+    smctx->m1 = powf(2.0f, -(smctx->max_bias / 2.0f) / smctx->n_head_log2);
+
+    smctx->use_src1 = (src1->ne[0] != 0);
+    smctx->use_f16  = (src1->ne[0] != 0) && (src1->type == HTP_TYPE_F16);
+
+    smctx->octx = octx;
+
+    // Initialize fastdiv values
+    const uint32_t ne01 = src0->ne[1];
+    const uint32_t ne02 = src0->ne[2];
+
+    if (ne01 > 0) smctx->fastdiv_ne01 = init_fastdiv_values(ne01);
+    if (ne02 > 0) smctx->fastdiv_ne02 = init_fastdiv_values(ne02);
+
+    const uint32_t ne12 = (src1->ne[0]) ? src1->ne[2] : 1;
+    const uint32_t ne13 = (src1->ne[0]) ? src1->ne[3] : 1;
+
+    if (ne12 > 0) smctx->fastdiv_ne12 = init_fastdiv_values(ne12);
+    if (ne13 > 0) smctx->fastdiv_ne13 = init_fastdiv_values(ne13);
+}
+
+static void hvx_fast_softmax_prep_f32(const uint8_t * restrict src,
+                                      uint8_t * restrict dst,
+                                      const int num_elems,
+                                      float     scale,
+                                      const uint8_t * restrict mask,
+                                      float slope) {
+    const uint8_t * restrict src_curr  = src;
+    uint8_t * restrict dst_curr        = dst;
+    const uint8_t * restrict mask_curr = mask;
+
+    HVX_Vector scale_vec = hvx_vec_splat_f32(scale);
+    HVX_Vector slope_vec = hvx_vec_splat_f32(slope);
+
+    int step_of_1 = num_elems >> 5;
+
+    #pragma unroll(4)
+    for (int i = 0; i < step_of_1; i++) {
+        HVX_Vector v1 = *(HVX_Vector *) src_curr;
+
+        HVX_Vector v3 = *(HVX_Vector *) mask_curr;
+
+        HVX_Vector v2 = Q6_Vqf32_vmpy_VsfVsf(v1, scale_vec);
+
+        HVX_Vector v4 = Q6_Vqf32_vmpy_VsfVsf(v3, slope_vec);
+
+        HVX_Vector v5 = Q6_Vqf32_vadd_Vqf32Vqf32(v2, v4);
+
+        *(HVX_Vector *) dst_curr = Q6_Vsf_equals_Vqf32(v5);
+
+        src_curr += VLEN;
+        dst_curr += VLEN;
+        mask_curr += VLEN;
+    }
+}
+
+static void hvx_fast_softmax_f32(const uint8_t * restrict src,
+                                 uint8_t * restrict dst,
+                                 uint8_t * restrict pad,
+                                 const int num_elems) {
+    const HVX_Vector * restrict v_src = (HVX_Vector *) src;
+    HVX_Vector * restrict v_pad       = (HVX_Vector *) pad;
+    HVX_Vector * restrict v_dst       = (HVX_Vector *) dst;
+
+    HVX_Vector sum_vec = Q6_V_vsplat_R(0x00000000);
+    HVX_Vector max_vec = hvx_vec_splat_f32(((const float *) src)[0]);
+    HVX_Vector zero_v  = Q6_V_vzero();
+    HVX_Vector one_v   = hvx_vec_splat_f32(1.0);
+
+    int step_of_1 = num_elems >> 5;
+
+    #pragma unroll(4)
+    for (int i = 0; i < step_of_1; i++) {
+        HVX_Vector v1 = v_src[i];
+        max_vec       = Q6_Vsf_vmax_VsfVsf(max_vec, v1);
+    }
+
+    max_vec = hvx_vec_reduce_max_f32(max_vec); // replicated over all lanes
+
+    #pragma unroll(4)
+    for (int i = 0; i < step_of_1; i++) {
+        HVX_Vector v1 = v_src[i];
+        HVX_Vector v2 = Q6_Vqf32_vsub_VsfVsf(v1, max_vec);
+
+        HVX_Vector v3 = hvx_vec_exp_f32(Q6_Vsf_equals_Vqf32(v2));
+
+        sum_vec = Q6_Vqf32_vadd_VsfVsf(Q6_Vsf_equals_Vqf32(sum_vec), v3);
+
+        v_pad[i] = v3;
+    }
+
+    sum_vec = hvx_vec_reduce_sum_f32(Q6_Vsf_equals_Vqf32(sum_vec)); // replicated over all lanes
+
+    HVX_VectorPred pos_sum   = Q6_Q_vcmp_gt_VwVw(sum_vec, zero_v);
+    HVX_Vector     v4        = hvx_vec_inverse_f32(sum_vec);
+    HVX_Vector     scale_vec = Q6_V_vmux_QVV(pos_sum, v4, one_v);
+
+    #pragma unroll(4)
+    for (int i = 0; i < step_of_1; i++) {
+        HVX_Vector v1 = v_pad[i];
+        HVX_Vector v2 = Q6_Vqf32_vmpy_VsfVsf(v1, scale_vec);
+        v_dst[i]      = Q6_Vsf_equals_Vqf32(v2);
+    }
+}
+
+static float hvx_softmax_f32(const uint8_t * restrict src,
+                             uint8_t * restrict dst,
+                             uint8_t * restrict spad,
+                             const int   num_elems,
+                             const float max) {
+    hvx_sub_scalar_f32(spad, src, max, num_elems);
+
+    hvx_exp_f32(spad, dst, num_elems, false);
+
+    float sum = hvx_reduce_sum_f32(dst, num_elems);
+
+    return sum;
+}
+
+static void softmax_job_f32(unsigned int nth, unsigned int ith, void * data) {
+    struct htp_softmax_context * smctx = (struct htp_softmax_context *) data;
+    struct htp_ops_context * octx = smctx->octx;
+
+    const struct htp_tensor * src0 = &octx->src0;
+    const struct htp_tensor * src1 = &octx->src1;
+    struct htp_tensor *       dst  = &octx->dst;
+
+    htp_softmax_preamble3;
+
+    const uint32_t src0_nrows            = ne01 * ne02 * ne03;  // src0 rows
+    const uint32_t src0_nrows_per_thread = smctx->src0_nrows_per_thread;
+
+    const uint32_t src0_start_row = src0_nrows_per_thread * ith;
+    const uint32_t src0_end_row   = MIN(src0_start_row + src0_nrows_per_thread, src0_nrows);
+
+    // no work for this thread
+    if (src0_start_row >= src0_end_row) {
+        return;
+    }
+
+    uint64_t t1, t2;
+    t1 = HAP_perf_get_qtimer_count();
+
+    int is_aligned = 1;
+    int opt_path   = 0;
+    if (!hex_is_aligned((void *) src0->data, VLEN) || !hex_is_aligned((void *) dst->data, VLEN)) {
+        is_aligned = 0;
+        FARF(HIGH, "softmax-f32: unaligned addresses in elementwise op, possibly slower execution\n");
+    }
+    if ((1 == is_aligned) && !(nb01 & (VLEN - 1))) {
+        opt_path = 1;
+    }
+
+    uint8_t * src0_spad_data = octx->src0_spad.data + (ith * smctx->spad_stride);
+    uint8_t * src1_spad_data = octx->src1_spad.data + (ith * smctx->spad_stride);
+    uint8_t * dst_spad_data  = octx->dst_spad.data + (ith * smctx->spad_stride);
+
+    float * wp0 = (float *) src0_spad_data;
+    float * wp1 = (float *) src1_spad_data;
+    float * wp2 = (float *) dst_spad_data;
+
+    uint32_t prev_i2 = (uint32_t)-1;
+    float slope = 1.0f;
+
+    for (uint32_t r = src0_start_row; r < src0_end_row; ++r) {
+        uint32_t i1 = fastmodulo(r, ne01, &smctx->fastdiv_ne01);
+        uint32_t r_div_ne01 = fastdiv(r, &smctx->fastdiv_ne01);
+        uint32_t i2 = fastmodulo(r_div_ne01, ne02, &smctx->fastdiv_ne02);
+        uint32_t i3 = fastdiv(r_div_ne01, &smctx->fastdiv_ne02);
+
+        // Map to original logic indices
+        // i01 = i1
+        // i02 = i2
+        // i03 = i3
+
+        const uint32_t i11 = i1;
+        // const uint32_t i12 = i2 % ne12;
+        // const uint32_t i13 = i3 % ne13;
+
+        uint32_t i12, i13;
+        if (ne12 == ne02) {
+             i12 = i2;
+        } else {
+             i12 = fastmodulo(i2, ne12, &smctx->fastdiv_ne12);
+        }
+
+        if (ne13 == ne03) {
+             i13 = i3;
+        } else {
+             i13 = fastmodulo(i3, ne13, &smctx->fastdiv_ne13);
+        }
+
+        // ALiBi
+        if (i2 != prev_i2) {
+            const uint32_t h = i2;  // head
+
+            slope = (smctx->max_bias > 0.0f) ?
+                        h < smctx->n_head_log2 ?
+                        powf(smctx->m0, h + 1) :
+                        powf(smctx->m1, 2 * (h - smctx->n_head_log2) + 1) :
+                        1.0f;
+            prev_i2 = i2;
+        }
+
+        float * sp = (float *) ((char *) octx->src0.data + i1 * nb01 + i2 * nb02 + i3 * nb03);
+        float * dp = (float *) ((char *) octx->dst.data + i1 * nb1 + i2 * nb2 + i3 * nb3);
+
+        // broadcast the mask across rows
+        __fp16 * mp_f16 = (smctx->use_src1) ?
+                              (__fp16 *) ((char *) octx->src1.data + i11 * nb11 + i12 * nb12 + i13 * nb13) :
+                              NULL;
+        float *  mp_f32 = (smctx->use_src1) ?
+                              (float *) ((char *) octx->src1.data + i11 * nb11 + i12 * nb12 + i13 * nb13) :
+                              NULL;
+
+        if ((1 == opt_path) && (mp_f32) && !(smctx->use_f16)) {
+            hvx_fast_softmax_prep_f32((const uint8_t *) sp, (uint8_t *) wp0, ne00, smctx->scale,
+                                      (const uint8_t *) mp_f32, slope);
+        } else {
+            hvx_scale_f32((uint8_t *) wp0, (const uint8_t *) sp, ne00, smctx->scale);
+            if (mp_f32) {
+                if (smctx->use_f16) {
+                    for (int i = 0; i < ne00; ++i) {
+                        wp0[i] += slope * (float) mp_f16[i];
+                    }
+                } else {
+                    for (int i = 0; i < ne00; ++i) {
+                        wp0[i] += slope * mp_f32[i];
+                    }
+                }
+            }
+        }
+
+        if (1 == opt_path) {
+            hvx_fast_softmax_f32((const uint8_t *) wp0, (uint8_t *) dp, (uint8_t *) wp1, ne00);
+        } else {
+            float max = hvx_reduce_max_f32((const uint8_t *) wp0, ne00);
+            float sum = hvx_softmax_f32((const uint8_t *) wp0, (uint8_t *) wp2, (uint8_t *) wp1, ne00, max);
+            sum       = sum > 0.0 ? (1.0 / sum) : 1;
+            hvx_scale_f32((uint8_t *) dp, (const uint8_t *) wp2, ne00, sum);
+        }
+    }
+
+    t2 = HAP_perf_get_qtimer_count();
+
+    FARF(HIGH, "softmax-f32 %d/%d/%d/%d: %ux%ux%ux%u (%u:%u) x %ux%ux%ux%u -> %ux%ux%ux%u usec %u\n", ith, nth,
+         smctx->use_f16, opt_path, ne00, ne01, ne02, ne03, src0_start_row, src0_end_row, ne10, ne11, ne12, ne13,
+         ne0, ne1, ne2, ne3, (unsigned) HAP_perf_qtimer_count_to_us(t2 - t1));
+}
+
+static int execute_op_softmax_f32(struct htp_ops_context * octx) {
+    int err = HTP_STATUS_OK;
+
+    const struct htp_tensor * src0 = &octx->src0;
+    const struct htp_tensor * src1 = &octx->src1;
+    struct htp_tensor *       dst  = &octx->dst;
+
+    struct htp_softmax_context smctx;
+    const char * op_type = "softmax-f32";
+
+    switch (octx->op) {
+        case HTP_OP_SOFTMAX:
+            init_softmax_ctx(&smctx, octx);
+            break;
+
+        default:
+            FARF(ERROR, "Unsupported Op %u\n", octx->op);
+            return HTP_STATUS_NO_SUPPORT;
+    }
+
+    const uint32_t n_threads = octx->n_threads;
+
+    const size_t src0_row_size = src0->nb[1];
+    const size_t src1_row_size = src0_row_size;
+    const size_t dst_row_size  = dst->nb[1];
+
+    // VTCM scratchpads for all tensors
+    // N rows per thread, padded to HVX vector size
+    octx->dst_spad.size  = hex_round_up(dst_row_size, 128) * n_threads;
+    octx->src0_spad.size = hex_round_up(src0_row_size, 128) * n_threads;
+    octx->src1_spad.size = hex_round_up(src1_row_size, 128) * n_threads;
+
+    // Use stride for calculating offset
+    smctx.spad_stride = hex_round_up(src0_row_size, 128);
+
+    size_t spad_size = octx->src0_spad.size + octx->src1_spad.size + octx->dst_spad.size;
+
+    if (src1->ne[0]) {
+        FARF(HIGH,
+             "%s: %ux%ux%ux%u x %ux%ux%ux%u -> %ux%ux%ux%u : src0-spad-size %u src1-spad-size %u dst-spad-size %u\n",
+             op_type, src0->ne[0], src0->ne[1], src0->ne[2], src0->ne[3], src1->ne[0], src1->ne[1], src1->ne[2],
+             src1->ne[3], dst->ne[0], dst->ne[1], dst->ne[2], dst->ne[3], octx->src0_spad.size, octx->src1_spad.size,
+             octx->dst_spad.size);
+    } else {
+        FARF(HIGH, "%s: %ux%ux%ux%u -> %ux%ux%ux%u : src0-spad-size %u src1-spad-size %u dst-spad-size %u\n", op_type,
+             src0->ne[0], src0->ne[1], src0->ne[2], src0->ne[3], dst->ne[0], dst->ne[1], dst->ne[2], dst->ne[3],
+             octx->src0_spad.size, octx->src1_spad.size, octx->dst_spad.size);
+    }
+
+    // Make sure the reserved vtcm size is sufficient
+    if (octx->ctx->vtcm_size < spad_size) {
+        FARF(ERROR, "%s : current VTCM reservation %zu is too small, needed %zu\n", op_type, octx->ctx->vtcm_size,
+             spad_size);
+        return HTP_STATUS_VTCM_TOO_SMALL;
+    }
+
+    octx->src0_spad.data = octx->ctx->vtcm_base;
+    octx->src1_spad.data = octx->src0_spad.data + octx->src0_spad.size;
+    octx->dst_spad.data  = octx->src1_spad.data + octx->src1_spad.size;
+
+    uint32_t src0_nrows = src0->ne[1] * src0->ne[2] * src0->ne[3];
+
+    if (!(octx->flags & HTP_OPFLAGS_SKIP_COMPUTE)) {
+        uint32_t n_jobs             = MIN(n_threads, src0_nrows);
+        smctx.src0_nrows_per_thread = (src0_nrows + n_jobs - 1) / n_jobs;
+        worker_pool_run_func(octx->ctx->worker_pool, softmax_job_f32, &smctx, n_jobs);
+    }
+
+    return err;
+}
+
+int op_softmax(struct htp_ops_context * octx) {
+    int err = HTP_STATUS_OK;
+
+    switch (octx->src0.type) {
+        case HTP_TYPE_F32:
+            err = execute_op_softmax_f32(octx);
+            break;
+
+        default:
+            err = HTP_STATUS_NO_SUPPORT;
+            break;
+    }
+
+    return err;
+}

package/cpp/ggml/src/ggml-hexagon/htp/sum-rows-ops.c

@@ -0,0 +1,130 @@
+#pragma clang diagnostic ignored "-Wunused-variable"
+#pragma clang diagnostic ignored "-Wunused-function"
+#pragma clang diagnostic ignored "-Wunused-but-set-variable"
+
+#include <HAP_farf.h>
+#include <HAP_perf.h>
+
+#include <string.h>
+#include <math.h>
+
+#include "hex-dma.h"
+#include "hvx-utils.h"
+
+#define GGML_COMMON_DECL_C
+#include "ggml-common.h"
+#include "htp-ctx.h"
+#include "htp-msg.h"
+#include "htp-ops.h"
+
+#define sum_rows_preamble                       \
+    struct htp_tensor *src0 =  &octx->src0;\
+    struct htp_tensor *dst  = &octx->dst;  \
+                                           \
+    const uint32_t ne00 = src0->ne[0];     \
+    const uint32_t ne01 = src0->ne[1];     \
+    const uint32_t ne02 = src0->ne[2];     \
+    const uint32_t ne03 = src0->ne[3];     \
+                                           \
+    const uint32_t nb00 = src0->nb[0];     \
+    const uint32_t nb01 = src0->nb[1];     \
+    const uint32_t nb02 = src0->nb[2];     \
+    const uint32_t nb03 = src0->nb[3];     \
+                                           \
+    const uint32_t  ne0 = dst->ne[0];      \
+    const uint32_t  ne1 = dst->ne[1];      \
+    const uint32_t  ne2 = dst->ne[2];      \
+    const uint32_t  ne3 = dst->ne[3];      \
+                                           \
+    const uint32_t  nb0 = dst->nb[0];      \
+    const uint32_t  nb1 = dst->nb[1];      \
+    const uint32_t  nb2 = dst->nb[2];      \
+    const uint32_t  nb3 = dst->nb[3];      \
+
+struct sum_rows_context {
+    const uint8_t * src_data;
+    uint8_t       * dst_data;
+    uint32_t        ne00;
+    size_t          src_stride;
+    size_t          dst_stride;
+    uint32_t        rows_per_thread;
+    uint32_t        total_rows;
+    bool            opt_path;
+};
+
+static void sum_rows_thread_f32(unsigned int nth, unsigned int ith, void *data) {
+    const struct sum_rows_context * smctx = (const struct sum_rows_context *) data;
+
+    const uint32_t rows_per_thread = smctx->rows_per_thread;
+    const uint32_t total_rows      = smctx->total_rows;
+
+    const uint32_t start_row = rows_per_thread * ith;
+    const uint32_t end_row   = MIN(start_row + rows_per_thread, total_rows);
+
+    if (start_row >= end_row) {
+        return;
+    }
+
+    const size_t   src_stride = smctx->src_stride;
+    const size_t   dst_stride = smctx->dst_stride;
+    const uint32_t ne00       = smctx->ne00;
+    const bool     opt_path   = smctx->opt_path;
+
+    const float * restrict src_th = (const float *) (smctx->src_data + (start_row * src_stride));
+    float       * restrict dst_th = (float *)       (smctx->dst_data + (start_row * dst_stride));
+
+    // Calculate actual number of rows for this thread
+    const uint32_t n_rows = end_row - start_row;
+
+    for (uint32_t ir = 0; ir < n_rows; ir++) {
+        const float * restrict src_local = src_th + (ir * (src_stride / sizeof(float)));
+
+        if (ir + 1 < n_rows) {
+            hex_l2fetch(src_local + (src_stride / sizeof(float)), src_stride, src_stride, 1);
+        }
+
+        if (opt_path) {
+            dst_th[ir] = hvx_reduce_sum_f32_a((const uint8_t *) src_local, ne00);
+        } else {
+            dst_th[ir] = hvx_reduce_sum_f32((const uint8_t *) src_local, ne00);
+        }
+    }
+}
+
+int op_sum_rows(struct htp_ops_context * octx) {
+    sum_rows_preamble;
+
+    if (octx->src0.type != HTP_TYPE_F32) {
+        return HTP_STATUS_NO_SUPPORT;
+    }
+
+    if (octx->flags & HTP_OPFLAGS_SKIP_COMPUTE) {
+        return HTP_STATUS_OK;
+    }
+
+    const int      n_threads  = octx->n_threads;
+    const uint32_t src0_nrows = ne01 * ne02 * ne03;
+
+    uint32_t n_jobs = MIN(n_threads, src0_nrows);
+    uint32_t rows_per_thread = (src0_nrows + n_jobs - 1) / n_jobs;
+
+    bool opt_path = false;
+    if ((0 == hex_is_aligned((void *) src0->data, VLEN)) && !(nb01 & (VLEN - 1))) {
+        opt_path = true;
+    }
+
+    struct sum_rows_context smctx = {
+        .src_data        = (const uint8_t *) src0->data,
+        .dst_data        = (uint8_t *) dst->data,
+        .ne00            = ne00,
+        .src_stride      = nb01,
+        .dst_stride      = nb1,
+        .rows_per_thread = rows_per_thread,
+        .total_rows      = src0_nrows,
+        .opt_path        = opt_path,
+    };
+
+    worker_pool_run_func(octx->ctx->worker_pool, sum_rows_thread_f32, &smctx, n_jobs);
+
+    return HTP_STATUS_OK;
+}