Author: post
Date: 2012-11-07 19:14:26 +0100 (Wed, 07 Nov 2012)
New Revision: 4308
Modified:
trunk/plugins/denoise/complexfilter-x86.cpp
Log:
Utilize the amazing fact that sqrt(x) = x * (1 / sqrt(x)). Faster processing on
older machines.
Modified: trunk/plugins/denoise/complexfilter-x86.cpp
===================================================================
--- trunk/plugins/denoise/complexfilter-x86.cpp 2012-11-07 17:59:24 UTC (rev
4307)
+++ trunk/plugins/denoise/complexfilter-x86.cpp 2012-11-07 18:14:26 UTC (rev
4308)
@@ -103,12 +103,12 @@
"rcpps %%xmm13, %%xmm13\n" // 1 / (psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax) (stall)
"movaps 16(%4), %%xmm4\n" // Load wsharpen
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
- "mulps %%xmm13, %%xmm15\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
+ "mulps %%xmm13, %%xmm15\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm13 free
+ "rsqrtps %%xmm7, %%xmm5\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "rsqrtps %%xmm15, %%xmm13\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm5, %%xmm7\n" // sqrt(x)
+ "mulps %%xmm13, %%xmm15\n" //
"movaps 64(%1), %%xmm5\n" // Load "1.0"
- "rsqrtps %%xmm7, %%xmm7\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rsqrtps %%xmm15, %%xmm15\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt (..)
- "rcpps %%xmm15, %%xmm15\n" // sqrt (..)
"mulps %%xmm6, %%xmm7\n" // multiply wsharpen
"mulps %%xmm4, %%xmm15\n" // multiply wsharpen
"addps %%xmm5, %%xmm7\n" // + 1.0 xmm7 = sfact
@@ -178,11 +178,11 @@
"addps %%xmm13, %%xmm5\n" //xmm5 = psd + sigmaSquaredSharpenMin
//xmm6 free
"mulps %%xmm4, %%xmm5\n" // (psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax) xmm4 free
- "movaps (%4), %%xmm6\n" // load wsharpen[0->4]
"rcpps %%xmm5, %%xmm5\n" // 1 / (psd + sigmaSquaredSharpenMin)*(psd
+ sigmaSquaredSharpenMax) (stall)
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
- "rsqrtps %%xmm7, %%xmm7\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt (..)
+ "rsqrtps %%xmm7, %%xmm6\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm6, %%xmm7\n" // sqrt
+ "movaps (%4), %%xmm6\n" // load wsharpen[0->4]
"mulps %%xmm6, %%xmm7\n" // multiply wsharpen
"addps %%xmm15, %%xmm7\n" // + 1.0 xmm7 = sfact
"movaps %%xmm7, %%xmm5\n"
@@ -256,12 +256,12 @@
"addps %%xmm6, %%xmm5\n" //xmm5 = psd + sigmaSquaredSharpenMin
//xmm6 free
"mulps %%xmm4, %%xmm5\n" // (psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax) xmm4 free
- "movaps (%4), %%xmm6\n" // load wsharpen[0->4]
"rcpps %%xmm5, %%xmm5\n" // 1 / (psd + sigmaSquaredSharpenMin)*(psd
+ sigmaSquaredSharpenMax) (stall)
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
+ "rsqrtps %%xmm7, %%xmm6\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm6, %%xmm7\n" // sqrt(x)
+ "movaps (%4), %%xmm6\n" // load wsharpen[0->4]
"movaps 64(%1), %%xmm5\n" // Load "1.0"
- "rsqrtps %%xmm7, %%xmm7\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt (..)
"mulps %%xmm6, %%xmm7\n" // multiply wsharpen
"addps %%xmm5, %%xmm7\n" // + 1.0 xmm7 = sfact
"movaps %%xmm7, %%xmm5\n"
@@ -340,9 +340,9 @@
"movaps (%4), %%xmm6\n" // load wsharpen[0->4]
"rcpps %%xmm5, %%xmm5\n" // 1 / (psd + sigmaSquaredSharpenMin)*(psd
+ sigmaSquaredSharpenMax) (stall)
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
+ "rsqrtps %%xmm7, %%xmm5\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm5, %%xmm7\n" // sqrt(x)
"movaps 64(%1), %%xmm5\n" // Load "1.0"
- "rsqrtps %%xmm7, %%xmm7\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt (..)
"mulps %%xmm6, %%xmm7\n" // multiply wsharpen
"addps %%xmm5, %%xmm7\n" // + 1.0 xmm7 = sfact
"movaps %%xmm7, %%xmm5\n"
@@ -432,8 +432,8 @@
"mulps %%xmm4, %%xmm5\n" // (psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax) xmm4 free
"rcpps %%xmm5, %%xmm5\n" // 1 / (psd + sigmaSquaredSharpenMin)*(psd
+ sigmaSquaredSharpenMax) (stall)
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
- "rsqrtps %%xmm7, %%xmm7\n" // 1 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt(...)
+ "rsqrtps %%xmm7, %%xmm5\n" // 1 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm5, %%xmm7\n" // sqrt(x)
"mulps (%4), %%xmm7\n" // multiply wsharpen
"addps %%xmm9, %%xmm7\n" // + 1.0 xmm7 = sfact
"mulps %%xmm6, %%xmm7\n" // *= Wienerfactor
@@ -522,9 +522,9 @@
"mulps %%xmm4, %%xmm5\n" // (psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax) xmm4 free
"rcpps %%xmm5, %%xmm5\n" // 1 / (psd + sigmaSquaredSharpenMin)*(psd
+ sigmaSquaredSharpenMax) (stall)
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
+ "rsqrtps %%xmm7, %%xmm5\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm5, %%xmm7\n" // sqrt(x)
"movaps 64(%1), %%xmm5\n" // Load "1.0"
- "rsqrtps %%xmm7, %%xmm7\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt (..)
"mulps (%4), %%xmm7\n" // multiply wsharpen
"addps %%xmm5, %%xmm7\n" // + 1.0 xmm7 = sfact
"mulps %%xmm6, %%xmm7\n" // *= Wienerfactor
@@ -617,9 +617,9 @@
"mulps %%xmm4, %%xmm5\n" // (psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax) xmm4 free
"rcpps %%xmm5, %%xmm5\n" // 1 / (psd + sigmaSquaredSharpenMin)*(psd
+ sigmaSquaredSharpenMax) (stall)
"mulps %%xmm5, %%xmm7\n" // psd*sigmaSquaredSharpenMax/((psd +
sigmaSquaredSharpenMin)*(psd + sigmaSquaredSharpenMax)) - xmm5 free
+ "rsqrtps %%xmm7, %%xmm5\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
+ "mulps %%xmm5, %%xmm7\n" // sqrt(x)
"movaps 64(%1), %%xmm5\n" // Load "1.0"
- "rsqrtps %%xmm7, %%xmm7\n" // 1.0 / sqrt(
psd*sigmaSquaredSharpenMax/((psd + sigmaSquaredSharpenMin)*(psd +
sigmaSquaredSharpenMax))
- "rcpps %%xmm7, %%xmm7\n" // sqrt (..)
"mulps (%4), %%xmm7\n" // multiply wsharpen
"addps %%xmm5, %%xmm7\n" // + 1.0 xmm7 = sfact
"mulps %%xmm6, %%xmm7\n" // *= Wienerfactor
@@ -878,6 +878,7 @@
}
#else // 32 bits
+
void ComplexWienerFilterDeGrid::processNoSharpen_SSE3( ComplexBlock* block )
{
fftwf_complex* outcur = block->complex;
_______________________________________________
Rawstudio-commit mailing list
[email protected]
http://rawstudio.org/cgi-bin/mailman/listinfo/rawstudio-commit
