more optimizations...
calculate the offsets from the phase differently, this happens
to reduce the register pressure in the main loop, which in turns
allows the compiler to generate much better code (doesn't need
to spill a lot of stuff on the stack).
this gives another 15% performance increase
Change-Id: I2ce3479dd48b9e6941adb80e6d443d6e14d64d96
diff --git a/services/audioflinger/AudioResamplerSinc.cpp b/services/audioflinger/AudioResamplerSinc.cpp
index 952abb4..d68b839 100644
--- a/services/audioflinger/AudioResamplerSinc.cpp
+++ b/services/audioflinger/AudioResamplerSinc.cpp
@@ -637,17 +637,20 @@
void AudioResamplerSinc::filterCoefficient(
int32_t* out, uint32_t phase, const int16_t *samples, uint32_t vRL)
{
+ // NOTE: be very careful when modifying the code here. register
+ // pressure is very high and a small change might cause the compiler
+ // to generate far less efficient code.
+ // Always sanity check the result with objdump or test-resample.
+
// compute the index of the coefficient on the positive side and
// negative side
const Constants& c(*mConstants);
+ const int32_t ONE = c.cMask | c.pMask;
uint32_t indexP = ( phase & c.cMask) >> c.cShift;
- uint32_t indexN = (-phase & c.cMask) >> c.cShift;
uint32_t lerpP = ( phase & c.pMask) >> c.pShift;
- uint32_t lerpN = (-phase & c.pMask) >> c.pShift;
- if ((indexP == 0) && (lerpP == 0)) {
- indexN = c.cMask >> c.cShift;
- lerpN = c.pMask >> c.pShift;
- }
+ uint32_t indexN = ((ONE-phase) & c.cMask) >> c.cShift;
+ uint32_t lerpN = ((ONE-phase) & c.pMask) >> c.pShift;
+
const size_t offset = c.halfNumCoefs;
indexP *= offset;
indexN *= offset;
@@ -677,7 +680,8 @@
asm (
"vmov.32 d2[0], %[lerpP] \n" // load the positive phase
"vmov.32 d2[1], %[lerpN] \n" // load the negative phase
- "veor q0, q0 \n" // result, initialize to 0
+ "veor q0, q0, q0 \n" // result, initialize to 0
+ "vshl.s32 d2, d2, #16 \n" // convert to 32 bits
"1: \n"
"vld1.16 { d4}, [%[sP]] \n" // load 4 16-bits stereo samples
@@ -727,8 +731,8 @@
[coefsN1] "+r" (coefsN1),
[sP] "+r" (sP),
[sN] "+r" (sN)
- : [lerpP] "r" (lerpP<<16),
- [lerpN] "r" (lerpN<<16),
+ : [lerpP] "r" (lerpP),
+ [lerpN] "r" (lerpN),
[vLR] "r" (mVolumeSIMD)
: "cc", "memory",
"q0", "q1", "q2", "q3",
@@ -742,8 +746,9 @@
asm (
"vmov.32 d2[0], %[lerpP] \n" // load the positive phase
"vmov.32 d2[1], %[lerpN] \n" // load the negative phase
- "veor q0, q0 \n" // result, initialize to 0
- "veor q4, q4 \n" // result, initialize to 0
+ "veor q0, q0, q0 \n" // result, initialize to 0
+ "veor q4, q4, q4 \n" // result, initialize to 0
+ "vshl.s32 d2, d2, #16 \n" // convert to 32 bits
"1: \n"
"vld2.16 {d4,d5}, [%[sP]] \n" // load 4 16-bits stereo samples
@@ -802,8 +807,8 @@
[coefsN1] "+r" (coefsN1),
[sP] "+r" (sP),
[sN] "+r" (sN)
- : [lerpP] "r" (lerpP<<16),
- [lerpN] "r" (lerpN<<16),
+ : [lerpP] "r" (lerpP),
+ [lerpN] "r" (lerpN),
[vLR] "r" (mVolumeSIMD)
: "cc", "memory",
"q0", "q1", "q2", "q3", "q4",