使用 SIMD 管理将累积(单个)值打包成两个值的清理代码循环的方法是什么?
What the method to manage Cleanup Code loop for a cumulative (single) value packed into two values using SIMD?
假设我管理一个名为 v_phase
的 __m128d
变量,计算为
index 0 : load prev phase
index 1 : phase += newValue
index 2 : phase += newValue
index 3 : phase += newValue
index 4 : phase += newValue
...
这是基本代码:
__m128d v_phase;
// load prev cumulated mPhase to v_phase (as mPhase, mPhase + nextValue)
for (int sampleIndex = 0; sampleIndex < blockSize; sampleIndex += 2, pValue += 2) {
// function with phase
// update pValue increment (its not linear)
// phase increment: v_phase += newValue
}
// cleanup code
if (blockSize % 2 == 0) {
mPhase = v_phase.m128d_f64[0];
}
事实是:如果 blockSize
是偶数,它工作正常:它将在最后一个循环迭代中求和 另外两个相位值 ,并取 v_phase.m128d_f64[0]
(即新的两个加法中的第一个)。
但是如果 blockSize
是奇数呢?我只需要最后一次迭代 的 v_phase.m128d_f64[1]
而不求和另外两个相位值 .
我可以使用 sampleIndex < blockSize - 1
,但这会在 // cleanup code
中移动逻辑 // function with phase
(我不太喜欢它)。
在循环中放置一个 if 是我会避免的事情(分支预测;因为我使用的是 SIMD,我正在优化代码,这会变慢)。
有什么建议吗?
这里有一个更 "complete" 的例子:
double phase = mPhase;
__m128d v_pB = _mm_setr_pd(0.0, pB[0]);
v_pB = _mm_mul_pd(v_pB, v_radiansPerSampleBp0);
__m128d v_pC = _mm_setr_pd(0.0, pC[0]);
v_pC = _mm_mul_pd(v_pC, v_radiansPerSample);
__m128d v_pB_prev = _mm_setr_pd(0.0, 0.0);
v_pB_prev = _mm_mul_pd(v_pB_prev, v_radiansPerSampleBp0);
__m128d v_pC_prev = _mm_setr_pd(0.0, 0.0);
v_pC_prev = _mm_mul_pd(v_pC_prev, v_radiansPerSample);
__m128d v_phaseAcc1;
__m128d v_phaseAcc2;
__m128d v_phase = _mm_set1_pd(phase);
// phase
v_phaseAcc1 = _mm_add_pd(v_pB, v_pC);
v_phaseAcc1 = _mm_max_pd(v_phaseAcc1, v_boundLower);
v_phaseAcc1 = _mm_min_pd(v_phaseAcc1, v_boundUpper);
v_phaseAcc2 = _mm_add_pd(v_pB_prev, v_pC_prev);
v_phaseAcc2 = _mm_max_pd(v_phaseAcc2, v_boundLower);
v_phaseAcc2 = _mm_min_pd(v_phaseAcc2, v_boundUpper);
v_phase = _mm_add_pd(v_phase, v_phaseAcc1);
v_phase = _mm_add_pd(v_phase, v_phaseAcc2);
for (int sampleIndex = 0; sampleIndex < blockSize; sampleIndex += 2, pB += 2, pC += 2) {
// code that will use v_phase
// phase increment
v_pB = _mm_loadu_pd(pB + 1);
v_pB = _mm_mul_pd(v_pB, v_radiansPerSampleBp0);
v_pC = _mm_loadu_pd(pC + 1);
v_pC = _mm_mul_pd(v_pC, v_radiansPerSample);
v_pB_prev = _mm_load_pd(pB);
v_pB_prev = _mm_mul_pd(v_pB_prev, v_radiansPerSampleBp0);
v_pC_prev = _mm_load_pd(pC);
v_pC_prev = _mm_mul_pd(v_pC_prev, v_radiansPerSample);
v_phaseAcc1 = _mm_add_pd(v_pB, v_pC);
v_phaseAcc1 = _mm_max_pd(v_phaseAcc1, v_boundLower);
v_phaseAcc1 = _mm_min_pd(v_phaseAcc1, v_boundUpper);
v_phaseAcc2 = _mm_add_pd(v_pB_prev, v_pC_prev);
v_phaseAcc2 = _mm_max_pd(v_phaseAcc2, v_boundLower);
v_phaseAcc2 = _mm_min_pd(v_phaseAcc2, v_boundUpper);
v_phase = _mm_add_pd(v_phase, v_phaseAcc1);
v_phase = _mm_add_pd(v_phase, v_phaseAcc2);
}
// cleanup code
if (blockSize % 2 == 0) {
mPhase = v_phase.m128d_f64[0];
}
else {
??? if odd?
}
除了最后一个,您还可以从循环中输出 previous v_phase
。也就是说,在更新您的 v_phase
之前,存储上一个:
__m128d prev_v_phase;
for (...) {
...
prev_v_phase = v_phase;
v_phase = _mm_add_pd(v_phase, v_phaseAcc1);
v_phase = _mm_add_pd(v_phase, v_phaseAcc2);
}
// cleanup code
if (blockSize % 2 == 0) {
mPhase = v_phase.m128d_f64[0];
}
else {
mPhase = prev_v_phase.m128d_f64[1];
}
如果循环根本不执行任何迭代,这将失败(然后prev_v_phase
将未初始化),但这是一种性能不重要的情况,所以很容易处理。
假设我管理一个名为 v_phase
的 __m128d
变量,计算为
index 0 : load prev phase
index 1 : phase += newValue
index 2 : phase += newValue
index 3 : phase += newValue
index 4 : phase += newValue
...
这是基本代码:
__m128d v_phase;
// load prev cumulated mPhase to v_phase (as mPhase, mPhase + nextValue)
for (int sampleIndex = 0; sampleIndex < blockSize; sampleIndex += 2, pValue += 2) {
// function with phase
// update pValue increment (its not linear)
// phase increment: v_phase += newValue
}
// cleanup code
if (blockSize % 2 == 0) {
mPhase = v_phase.m128d_f64[0];
}
事实是:如果 blockSize
是偶数,它工作正常:它将在最后一个循环迭代中求和 另外两个相位值 ,并取 v_phase.m128d_f64[0]
(即新的两个加法中的第一个)。
但是如果 blockSize
是奇数呢?我只需要最后一次迭代 的 v_phase.m128d_f64[1]
而不求和另外两个相位值 .
我可以使用 sampleIndex < blockSize - 1
,但这会在 // cleanup code
中移动逻辑 // function with phase
(我不太喜欢它)。
在循环中放置一个 if 是我会避免的事情(分支预测;因为我使用的是 SIMD,我正在优化代码,这会变慢)。
有什么建议吗?
这里有一个更 "complete" 的例子:
double phase = mPhase;
__m128d v_pB = _mm_setr_pd(0.0, pB[0]);
v_pB = _mm_mul_pd(v_pB, v_radiansPerSampleBp0);
__m128d v_pC = _mm_setr_pd(0.0, pC[0]);
v_pC = _mm_mul_pd(v_pC, v_radiansPerSample);
__m128d v_pB_prev = _mm_setr_pd(0.0, 0.0);
v_pB_prev = _mm_mul_pd(v_pB_prev, v_radiansPerSampleBp0);
__m128d v_pC_prev = _mm_setr_pd(0.0, 0.0);
v_pC_prev = _mm_mul_pd(v_pC_prev, v_radiansPerSample);
__m128d v_phaseAcc1;
__m128d v_phaseAcc2;
__m128d v_phase = _mm_set1_pd(phase);
// phase
v_phaseAcc1 = _mm_add_pd(v_pB, v_pC);
v_phaseAcc1 = _mm_max_pd(v_phaseAcc1, v_boundLower);
v_phaseAcc1 = _mm_min_pd(v_phaseAcc1, v_boundUpper);
v_phaseAcc2 = _mm_add_pd(v_pB_prev, v_pC_prev);
v_phaseAcc2 = _mm_max_pd(v_phaseAcc2, v_boundLower);
v_phaseAcc2 = _mm_min_pd(v_phaseAcc2, v_boundUpper);
v_phase = _mm_add_pd(v_phase, v_phaseAcc1);
v_phase = _mm_add_pd(v_phase, v_phaseAcc2);
for (int sampleIndex = 0; sampleIndex < blockSize; sampleIndex += 2, pB += 2, pC += 2) {
// code that will use v_phase
// phase increment
v_pB = _mm_loadu_pd(pB + 1);
v_pB = _mm_mul_pd(v_pB, v_radiansPerSampleBp0);
v_pC = _mm_loadu_pd(pC + 1);
v_pC = _mm_mul_pd(v_pC, v_radiansPerSample);
v_pB_prev = _mm_load_pd(pB);
v_pB_prev = _mm_mul_pd(v_pB_prev, v_radiansPerSampleBp0);
v_pC_prev = _mm_load_pd(pC);
v_pC_prev = _mm_mul_pd(v_pC_prev, v_radiansPerSample);
v_phaseAcc1 = _mm_add_pd(v_pB, v_pC);
v_phaseAcc1 = _mm_max_pd(v_phaseAcc1, v_boundLower);
v_phaseAcc1 = _mm_min_pd(v_phaseAcc1, v_boundUpper);
v_phaseAcc2 = _mm_add_pd(v_pB_prev, v_pC_prev);
v_phaseAcc2 = _mm_max_pd(v_phaseAcc2, v_boundLower);
v_phaseAcc2 = _mm_min_pd(v_phaseAcc2, v_boundUpper);
v_phase = _mm_add_pd(v_phase, v_phaseAcc1);
v_phase = _mm_add_pd(v_phase, v_phaseAcc2);
}
// cleanup code
if (blockSize % 2 == 0) {
mPhase = v_phase.m128d_f64[0];
}
else {
??? if odd?
}
除了最后一个,您还可以从循环中输出 previous v_phase
。也就是说,在更新您的 v_phase
之前,存储上一个:
__m128d prev_v_phase;
for (...) {
...
prev_v_phase = v_phase;
v_phase = _mm_add_pd(v_phase, v_phaseAcc1);
v_phase = _mm_add_pd(v_phase, v_phaseAcc2);
}
// cleanup code
if (blockSize % 2 == 0) {
mPhase = v_phase.m128d_f64[0];
}
else {
mPhase = prev_v_phase.m128d_f64[1];
}
如果循环根本不执行任何迭代,这将失败(然后prev_v_phase
将未初始化),但这是一种性能不重要的情况,所以很容易处理。