risp4ml/isp_stages/dg_rvv.c

/*
 * Copyright 2023 Google LLC
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include <riscv_vector.h>

#include "risp4ml/common/utils.h"
#include "risp4ml/isp_stages/dg.h"

static const uint16_t kDgFractional = kRawPipelineFraction;
static const uint16_t kDgUnityGain = 1 << kDgFractional;
static DgParams dg_params = {
    .enable = true,
    .gains = {kDgUnityGain, kDgUnityGain, kDgUnityGain, kDgUnityGain}};

void set_dg_params(DgParams* params) { dg_params = *params; }

void dg_process(Image* img) {
  if (!dg_params.enable) return;

  size_t vl;
  size_t n = img->height * img->width * img->num_channels;
  uint16_t gain = dg_params.gains[0];
  uint32_t offset = 1 << (kDgFractional - 1);
  uint32_t max_val = kRawPipelineMaxVal << kDgFractional;
  // auxiliary variables
  vuint16m4_t vx;
  vuint32m8_t vy;
  uint16_t* x;
  for (size_t i = 0; i < n; i += vl) {
    x = img->data + i;
    vl = __riscv_vsetvl_e16m4(n - i);
    vx = __riscv_vle16_v_u16m4(x, vl);          // load
    vy = __riscv_vwmulu(vx, gain, vl);          // multiply
    vy = __riscv_vadd(vy, offset, vl);          // add
    vy = __riscv_vminu(vy, max_val, vl);        // clamp
    vx = __riscv_vnsrl(vy, kDgFractional, vl);  // bit shift
    __riscv_vse16(x, vx, vl);                   // save
  }
}