* [patch, fortran-dev] Use fixed variable sizes for stride calculations
@ 2012-03-24 14:14 Thomas Koenig
2012-03-28 8:12 ` Janne Blomqvist
2012-04-11 7:51 ` Thomas Koenig
0 siblings, 2 replies; 5+ messages in thread
From: Thomas Koenig @ 2012-03-24 14:14 UTC (permalink / raw)
To: fortran, gcc-patches
[-- Attachment #1: Type: text/plain, Size: 16962 bytes --]
Hello world,
this patch uses division by known sizes (which can usually be replaced
by a simple shift because intrinsics have sizes of power of two) instead
of division by the size extracted from the array descriptor itself.
This should save about 20 cycles for a single calculation.
I'll go through the rest of the library to identify other possibilities
for this.
Regression-tested, no new failures.
OK for the branch?
Thomas
2012-03-24 Thomas König <tkoenig@gcc.gnu.org>
* libgfortran.h (GFC_DESCRIPTOR_SIZE_TYPEKNOWN): New macro.
(GFC_DESCRIPTOR_STRIDE_TYPEKNOWN): New macro.
* m4/cshift0.m4: Use GFC_DESCRIPTOR_STRIDE_TYPEKNOWN.
* m4/in_pack.m4: Likewise.
* m4/pack.m4: Likewise.
* m4/spread.m4: Likewise.
* m4/transpose.m4: Likewise.
* m4/iforeach.m4: Likewise.
* m4/eoshift1.m4: Likewise.
* m4/eoshift3.m4: Likewise.
* m4/shape.m4: Likewise.
* m4/cshift1.m4: Likewise.
* m4/in_unpack.m4: Likewise.
* m4/matmull.m4: Likewise.
* m4/bessel.m4: Likewise.
* m4/unpack.m4: Likewise.
* m4/reshape.m4: Likewise.
* m4/ifunction_logical.m4: Likewise.
* m4/ifunction.m4: Likewise.
* m4/matmul.m4: Likewise.
* generated/all_l16.c: Regenerated.
* generated/all_l1.c: Regenerated.
* generated/all_l2.c: Regenerated.
* generated/all_l4.c: Regenerated.
* generated/all_l8.c: Regenerated.
* generated/any_l16.c: Regenerated.
* generated/any_l1.c: Regenerated.
* generated/any_l2.c: Regenerated.
* generated/any_l4.c: Regenerated.
* generated/any_l8.c: Regenerated.
* generated/bessel_r10.c: Regenerated.
* generated/bessel_r16.c: Regenerated.
* generated/bessel_r4.c: Regenerated.
* generated/bessel_r8.c: Regenerated.
* generated/count_16_l.c: Regenerated.
* generated/count_1_l.c: Regenerated.
* generated/count_2_l.c: Regenerated.
* generated/count_4_l.c: Regenerated.
* generated/count_8_l.c: Regenerated.
* generated/cshift0_c10.c: Regenerated.
* generated/cshift0_c16.c: Regenerated.
* generated/cshift0_c4.c: Regenerated.
* generated/cshift0_c8.c: Regenerated.
* generated/cshift0_i16.c: Regenerated.
* generated/cshift0_i1.c: Regenerated.
* generated/cshift0_i2.c: Regenerated.
* generated/cshift0_i4.c: Regenerated.
* generated/cshift0_i8.c: Regenerated.
* generated/cshift0_r10.c: Regenerated.
* generated/cshift0_r16.c: Regenerated.
* generated/cshift0_r4.c: Regenerated.
* generated/cshift0_r8.c: Regenerated.
* generated/cshift1_16.c: Regenerated.
* generated/cshift1_4.c: Regenerated.
* generated/cshift1_8.c: Regenerated.
* generated/eoshift1_16.c: Regenerated.
* generated/eoshift1_4.c: Regenerated.
* generated/eoshift1_8.c: Regenerated.
* generated/eoshift3_16.c: Regenerated.
* generated/eoshift3_4.c: Regenerated.
* generated/eoshift3_8.c: Regenerated.
* generated/iall_i16.c: Regenerated.
* generated/iall_i1.c: Regenerated.
* generated/iall_i2.c: Regenerated.
* generated/iall_i4.c: Regenerated.
* generated/iall_i8.c: Regenerated.
* generated/iany_i16.c: Regenerated.
* generated/iany_i1.c: Regenerated.
* generated/iany_i2.c: Regenerated.
* generated/iany_i4.c: Regenerated.
* generated/iany_i8.c: Regenerated.
* generated/in_pack_c10.c: Regenerated.
* generated/in_pack_c16.c: Regenerated.
* generated/in_pack_c4.c: Regenerated.
* generated/in_pack_c8.c: Regenerated.
* generated/in_pack_i16.c: Regenerated.
* generated/in_pack_i1.c: Regenerated.
* generated/in_pack_i2.c: Regenerated.
* generated/in_pack_i4.c: Regenerated.
* generated/in_pack_i8.c: Regenerated.
* generated/in_pack_r10.c: Regenerated.
* generated/in_pack_r16.c: Regenerated.
* generated/in_pack_r4.c: Regenerated.
* generated/in_pack_r8.c: Regenerated.
* generated/in_unpack_c10.c: Regenerated.
* generated/in_unpack_c16.c: Regenerated.
* generated/in_unpack_c4.c: Regenerated.
* generated/in_unpack_c8.c: Regenerated.
* generated/in_unpack_i16.c: Regenerated.
* generated/in_unpack_i1.c: Regenerated.
* generated/in_unpack_i2.c: Regenerated.
* generated/in_unpack_i4.c: Regenerated.
* generated/in_unpack_i8.c: Regenerated.
* generated/in_unpack_r10.c: Regenerated.
* generated/in_unpack_r16.c: Regenerated.
* generated/in_unpack_r4.c: Regenerated.
* generated/in_unpack_r8.c: Regenerated.
* generated/iparity_i16.c: Regenerated.
* generated/iparity_i1.c: Regenerated.
* generated/iparity_i2.c: Regenerated.
* generated/iparity_i4.c: Regenerated.
* generated/iparity_i8.c: Regenerated.
* generated/matmul_c10.c: Regenerated.
* generated/matmul_c16.c: Regenerated.
* generated/matmul_c4.c: Regenerated.
* generated/matmul_c8.c: Regenerated.
* generated/matmul_i16.c: Regenerated.
* generated/matmul_i1.c: Regenerated.
* generated/matmul_i2.c: Regenerated.
* generated/matmul_i4.c: Regenerated.
* generated/matmul_i8.c: Regenerated.
* generated/matmul_l16.c: Regenerated.
* generated/matmul_l4.c: Regenerated.
* generated/matmul_l8.c: Regenerated.
* generated/matmul_r10.c: Regenerated.
* generated/matmul_r16.c: Regenerated.
* generated/matmul_r4.c: Regenerated.
* generated/matmul_r8.c: Regenerated.
* generated/maxloc0_16_i16.c: Regenerated.
* generated/maxloc0_16_i1.c: Regenerated.
* generated/maxloc0_16_i2.c: Regenerated.
* generated/maxloc0_16_i4.c: Regenerated.
* generated/maxloc0_16_i8.c: Regenerated.
* generated/maxloc0_16_r10.c: Regenerated.
* generated/maxloc0_16_r16.c: Regenerated.
* generated/maxloc0_16_r4.c: Regenerated.
* generated/maxloc0_16_r8.c: Regenerated.
* generated/maxloc0_4_i16.c: Regenerated.
* generated/maxloc0_4_i1.c: Regenerated.
* generated/maxloc0_4_i2.c: Regenerated.
* generated/maxloc0_4_i4.c: Regenerated.
* generated/maxloc0_4_i8.c: Regenerated.
* generated/maxloc0_4_r10.c: Regenerated.
* generated/maxloc0_4_r16.c: Regenerated.
* generated/maxloc0_4_r4.c: Regenerated.
* generated/maxloc0_4_r8.c: Regenerated.
* generated/maxloc0_8_i16.c: Regenerated.
* generated/maxloc0_8_i1.c: Regenerated.
* generated/maxloc0_8_i2.c: Regenerated.
* generated/maxloc0_8_i4.c: Regenerated.
* generated/maxloc0_8_i8.c: Regenerated.
* generated/maxloc0_8_r10.c: Regenerated.
* generated/maxloc0_8_r16.c: Regenerated.
* generated/maxloc0_8_r4.c: Regenerated.
* generated/maxloc0_8_r8.c: Regenerated.
* generated/maxloc1_16_i16.c: Regenerated.
* generated/maxloc1_16_i1.c: Regenerated.
* generated/maxloc1_16_i2.c: Regenerated.
* generated/maxloc1_16_i4.c: Regenerated.
* generated/maxloc1_16_i8.c: Regenerated.
* generated/maxloc1_16_r10.c: Regenerated.
* generated/maxloc1_16_r16.c: Regenerated.
* generated/maxloc1_16_r4.c: Regenerated.
* generated/maxloc1_16_r8.c: Regenerated.
* generated/maxloc1_4_i16.c: Regenerated.
* generated/maxloc1_4_i1.c: Regenerated.
* generated/maxloc1_4_i2.c: Regenerated.
* generated/maxloc1_4_i4.c: Regenerated.
* generated/maxloc1_4_i8.c: Regenerated.
* generated/maxloc1_4_r10.c: Regenerated.
* generated/maxloc1_4_r16.c: Regenerated.
* generated/maxloc1_4_r4.c: Regenerated.
* generated/maxloc1_4_r8.c: Regenerated.
* generated/maxloc1_8_i16.c: Regenerated.
* generated/maxloc1_8_i1.c: Regenerated.
* generated/maxloc1_8_i2.c: Regenerated.
* generated/maxloc1_8_i4.c: Regenerated.
* generated/maxloc1_8_i8.c: Regenerated.
* generated/maxloc1_8_r10.c: Regenerated.
* generated/maxloc1_8_r16.c: Regenerated.
* generated/maxloc1_8_r4.c: Regenerated.
* generated/maxloc1_8_r8.c: Regenerated.
* generated/maxval_i16.c: Regenerated.
* generated/maxval_i1.c: Regenerated.
* generated/maxval_i2.c: Regenerated.
* generated/maxval_i4.c: Regenerated.
* generated/maxval_i8.c: Regenerated.
* generated/maxval_r10.c: Regenerated.
* generated/maxval_r16.c: Regenerated.
* generated/maxval_r4.c: Regenerated.
* generated/maxval_r8.c: Regenerated.
* generated/minloc0_16_i16.c: Regenerated.
* generated/minloc0_16_i1.c: Regenerated.
* generated/minloc0_16_i2.c: Regenerated.
* generated/minloc0_16_i4.c: Regenerated.
* generated/minloc0_16_i8.c: Regenerated.
* generated/minloc0_16_r10.c: Regenerated.
* generated/minloc0_16_r16.c: Regenerated.
* generated/minloc0_16_r4.c: Regenerated.
* generated/minloc0_16_r8.c: Regenerated.
* generated/minloc0_4_i16.c: Regenerated.
* generated/minloc0_4_i1.c: Regenerated.
* generated/minloc0_4_i2.c: Regenerated.
* generated/minloc0_4_i4.c: Regenerated.
* generated/minloc0_4_i8.c: Regenerated.
* generated/minloc0_4_r10.c: Regenerated.
* generated/minloc0_4_r16.c: Regenerated.
* generated/minloc0_4_r4.c: Regenerated.
* generated/minloc0_4_r8.c: Regenerated.
* generated/minloc0_8_i16.c: Regenerated.
* generated/minloc0_8_i1.c: Regenerated.
* generated/minloc0_8_i2.c: Regenerated.
* generated/minloc0_8_i4.c: Regenerated.
* generated/minloc0_8_i8.c: Regenerated.
* generated/minloc0_8_r10.c: Regenerated.
* generated/minloc0_8_r16.c: Regenerated.
* generated/minloc0_8_r4.c: Regenerated.
* generated/minloc0_8_r8.c: Regenerated.
* generated/minloc1_16_i16.c: Regenerated.
* generated/minloc1_16_i1.c: Regenerated.
* generated/minloc1_16_i2.c: Regenerated.
* generated/minloc1_16_i4.c: Regenerated.
* generated/minloc1_16_i8.c: Regenerated.
* generated/minloc1_16_r10.c: Regenerated.
* generated/minloc1_16_r16.c: Regenerated.
* generated/minloc1_16_r4.c: Regenerated.
* generated/minloc1_16_r8.c: Regenerated.
* generated/minloc1_4_i16.c: Regenerated.
* generated/minloc1_4_i1.c: Regenerated.
* generated/minloc1_4_i2.c: Regenerated.
* generated/minloc1_4_i4.c: Regenerated.
* generated/minloc1_4_i8.c: Regenerated.
* generated/minloc1_4_r10.c: Regenerated.
* generated/minloc1_4_r16.c: Regenerated.
* generated/minloc1_4_r4.c: Regenerated.
* generated/minloc1_4_r8.c: Regenerated.
* generated/minloc1_8_i16.c: Regenerated.
* generated/minloc1_8_i1.c: Regenerated.
* generated/minloc1_8_i2.c: Regenerated.
* generated/minloc1_8_i4.c: Regenerated.
* generated/minloc1_8_i8.c: Regenerated.
* generated/minloc1_8_r10.c: Regenerated.
* generated/minloc1_8_r16.c: Regenerated.
* generated/minloc1_8_r4.c: Regenerated.
* generated/minloc1_8_r8.c: Regenerated.
* generated/minval_i16.c: Regenerated.
* generated/minval_i1.c: Regenerated.
* generated/minval_i2.c: Regenerated.
* generated/minval_i4.c: Regenerated.
* generated/minval_i8.c: Regenerated.
* generated/minval_r10.c: Regenerated.
* generated/minval_r16.c: Regenerated.
* generated/minval_r4.c: Regenerated.
* generated/minval_r8.c: Regenerated.
* generated/norm2_r10.c: Regenerated.
* generated/norm2_r16.c: Regenerated.
* generated/norm2_r4.c: Regenerated.
* generated/norm2_r8.c: Regenerated.
* generated/pack_c10.c: Regenerated.
* generated/pack_c16.c: Regenerated.
* generated/pack_c4.c: Regenerated.
* generated/pack_c8.c: Regenerated.
* generated/pack_i16.c: Regenerated.
* generated/pack_i1.c: Regenerated.
* generated/pack_i2.c: Regenerated.
* generated/pack_i4.c: Regenerated.
* generated/pack_i8.c: Regenerated.
* generated/pack_r10.c: Regenerated.
* generated/pack_r16.c: Regenerated.
* generated/pack_r4.c: Regenerated.
* generated/pack_r8.c: Regenerated.
* generated/parity_l16.c: Regenerated.
* generated/parity_l1.c: Regenerated.
* generated/parity_l2.c: Regenerated.
* generated/parity_l4.c: Regenerated.
* generated/parity_l8.c: Regenerated.
* generated/product_c10.c: Regenerated.
* generated/product_c16.c: Regenerated.
* generated/product_c4.c: Regenerated.
* generated/product_c8.c: Regenerated.
* generated/product_i16.c: Regenerated.
* generated/product_i1.c: Regenerated.
* generated/product_i2.c: Regenerated.
* generated/product_i4.c: Regenerated.
* generated/product_i8.c: Regenerated.
* generated/product_r10.c: Regenerated.
* generated/product_r16.c: Regenerated.
* generated/product_r4.c: Regenerated.
* generated/product_r8.c: Regenerated.
* generated/reshape_c10.c: Regenerated.
* generated/reshape_c16.c: Regenerated.
* generated/reshape_c4.c: Regenerated.
* generated/reshape_c8.c: Regenerated.
* generated/reshape_i16.c: Regenerated.
* generated/reshape_i4.c: Regenerated.
* generated/reshape_i8.c: Regenerated.
* generated/reshape_r10.c: Regenerated.
* generated/reshape_r16.c: Regenerated.
* generated/reshape_r4.c: Regenerated.
* generated/reshape_r8.c: Regenerated.
* generated/shape_i16.c: Regenerated.
* generated/shape_i4.c: Regenerated.
* generated/shape_i8.c: Regenerated.
* generated/spread_c10.c: Regenerated.
* generated/spread_c16.c: Regenerated.
* generated/spread_c4.c: Regenerated.
* generated/spread_c8.c: Regenerated.
* generated/spread_i16.c: Regenerated.
* generated/spread_i1.c: Regenerated.
* generated/spread_i2.c: Regenerated.
* generated/spread_i4.c: Regenerated.
* generated/spread_i8.c: Regenerated.
* generated/spread_r10.c: Regenerated.
* generated/spread_r16.c: Regenerated.
* generated/spread_r4.c: Regenerated.
* generated/spread_r8.c: Regenerated.
* generated/sum_c10.c: Regenerated.
* generated/sum_c16.c: Regenerated.
* generated/sum_c4.c: Regenerated.
* generated/sum_c8.c: Regenerated.
* generated/sum_i16.c: Regenerated.
* generated/sum_i1.c: Regenerated.
* generated/sum_i2.c: Regenerated.
* generated/sum_i4.c: Regenerated.
* generated/sum_i8.c: Regenerated.
* generated/sum_r10.c: Regenerated.
* generated/sum_r16.c: Regenerated.
* generated/sum_r4.c: Regenerated.
* generated/sum_r8.c: Regenerated.
* generated/transpose_c10.c: Regenerated.
* generated/transpose_c16.c: Regenerated.
* generated/transpose_c4.c: Regenerated.
* generated/transpose_c8.c: Regenerated.
* generated/transpose_i16.c: Regenerated.
* generated/transpose_i4.c: Regenerated.
* generated/transpose_i8.c: Regenerated.
* generated/transpose_r10.c: Regenerated.
* generated/transpose_r16.c: Regenerated.
* generated/transpose_r4.c: Regenerated.
* generated/transpose_r8.c: Regenerated.
* generated/unpack_c10.c: Regenerated.
* generated/unpack_c16.c: Regenerated.
* generated/unpack_c4.c: Regenerated.
* generated/unpack_c8.c: Regenerated.
* generated/unpack_i16.c: Regenerated.
* generated/unpack_i1.c: Regenerated.
* generated/unpack_i2.c: Regenerated.
* generated/unpack_i4.c: Regenerated.
* generated/unpack_i8.c: Regenerated.
* generated/unpack_r10.c: Regenerated.
* generated/unpack_r16.c: Regenerated.
* generated/unpack_r4.c: Regenerated.
* generated/unpack_r8.c: Regenerated.
[-- Attachment #2: p1.diff --]
[-- Type: text/x-patch, Size: 20338 bytes --]
Index: libgfortran.h
===================================================================
--- libgfortran.h (Revision 185261)
+++ libgfortran.h (Arbeitskopie)
@@ -364,6 +364,11 @@
#define GFC_DESCRIPTOR_TYPE(desc) (((desc)->dtype & GFC_DTYPE_TYPE_MASK) \
>> GFC_DTYPE_TYPE_SHIFT)
#define GFC_DESCRIPTOR_SIZE(desc) ((desc)->dtype >> GFC_DTYPE_SIZE_SHIFT)
+
+/* This is for getting the size of a descriptor when the type of the
+ descriptor is known at compile-time. Do not use for string types. */
+
+#define GFC_DESCRIPTOR_SIZE_TYPEKNOWN(desc) (sizeof((desc)->base_addr[0]))
#define GFC_DESCRIPTOR_DATA(desc) ((desc)->base_addr)
#define GFC_DESCRIPTOR_DTYPE(desc) ((desc)->dtype)
@@ -391,6 +396,13 @@
#define GFC_DESCRIPTOR_STRIDE(desc,i) \
(GFC_DESCRIPTOR_SM(desc,i) / GFC_DESCRIPTOR_SIZE(desc))
+/* This is for getting the stride when the type of the descriptor is known at
+ compile-time, to avoid expensive divisions. Do not use for string
+ types. */
+
+#define GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(desc,i) \
+ (GFC_DESCRIPTOR_SM(desc,i) / GFC_DESCRIPTOR_SIZE_TYPEKNOWN(desc))
+
/* Macros to get both the size and the type with a single masking operation */
#define GFC_DTYPE_SIZE_MASK \
Index: m4/cshift0.m4
===================================================================
--- m4/cshift0.m4 (Revision 185261)
+++ m4/cshift0.m4 (Arbeitskopie)
@@ -70,10 +70,10 @@
{
if (dim == which)
{
- roffset = GFC_DESCRIPTOR_STRIDE(ret,dim);
+ roffset = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,dim);
if (roffset == 0)
roffset = 1;
- soffset = GFC_DESCRIPTOR_STRIDE(array,dim);
+ soffset = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,dim);
if (soffset == 0)
soffset = 1;
len = GFC_DESCRIPTOR_EXTENT(array,dim);
@@ -82,8 +82,8 @@
{
count[n] = 0;
extent[n] = GFC_DESCRIPTOR_EXTENT(array,dim);
- rstride[n] = GFC_DESCRIPTOR_STRIDE(ret,dim);
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,dim);
+ rstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,dim);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,dim);
n++;
}
}
Index: m4/in_pack.m4
===================================================================
--- m4/in_pack.m4 (Revision 185261)
+++ m4/in_pack.m4 (Arbeitskopie)
@@ -60,7 +60,7 @@
for (n = 0; n < dim; n++)
{
count[n] = 0;
- stride[n] = GFC_DESCRIPTOR_STRIDE(source,n);
+ stride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,n);
extent[n] = GFC_DESCRIPTOR_EXTENT(source,n);
if (extent[n] <= 0)
{
Index: m4/pack.m4
===================================================================
--- m4/pack.m4 (Revision 185261)
+++ m4/pack.m4 (Arbeitskopie)
@@ -127,7 +127,7 @@
extent[n] = GFC_DESCRIPTOR_EXTENT(array,n);
if (extent[n] <= 0)
zero_sized = 1;
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,n);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
}
if (sstride[0] == 0)
@@ -188,7 +188,7 @@
}
}
- rstride0 = GFC_DESCRIPTOR_STRIDE(ret,0);
+ rstride0 = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
if (rstride0 == 0)
rstride0 = 1;
sstride0 = sstride[0];
@@ -241,7 +241,7 @@
nelem = ((rptr - ret->base_addr) / rstride0);
if (n > nelem)
{
- sstride0 = GFC_DESCRIPTOR_STRIDE(vector,0);
+ sstride0 = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(vector,0);
if (sstride0 == 0)
sstride0 = 1;
Index: m4/spread.m4
===================================================================
--- m4/spread.m4 (Revision 185261)
+++ m4/spread.m4 (Arbeitskopie)
@@ -91,7 +91,7 @@
{
count[dim] = 0;
extent[dim] = GFC_DESCRIPTOR_EXTENT(source,dim);
- sstride[dim] = GFC_DESCRIPTOR_STRIDE(source,dim);
+ sstride[dim] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,dim);
rstride[dim] = rs;
ext = extent[dim];
@@ -126,7 +126,7 @@
ret_extent = GFC_DESCRIPTOR_EXTENT(ret,n);
if (n == along - 1)
{
- rdelta = GFC_DESCRIPTOR_STRIDE(ret,n);
+ rdelta = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
if (ret_extent != ncopies)
runtime_error("Incorrect extent in return value of SPREAD"
@@ -147,8 +147,8 @@
if (extent[dim] <= 0)
zero_sized = 1;
- sstride[dim] = GFC_DESCRIPTOR_STRIDE(source,dim);
- rstride[dim] = GFC_DESCRIPTOR_STRIDE(ret,n);
+ sstride[dim] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,dim);
+ rstride[dim] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
dim++;
}
}
@@ -159,7 +159,7 @@
{
if (n == along - 1)
{
- rdelta = GFC_DESCRIPTOR_STRIDE(ret,n);
+ rdelta = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
}
else
{
@@ -167,8 +167,8 @@
extent[dim] = GFC_DESCRIPTOR_EXTENT(source,dim);
if (extent[dim] <= 0)
zero_sized = 1;
- sstride[dim] = GFC_DESCRIPTOR_STRIDE(source,dim);
- rstride[dim] = GFC_DESCRIPTOR_STRIDE(ret,n);
+ sstride[dim] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,dim);
+ rstride[dim] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
dim++;
}
}
@@ -252,12 +252,12 @@
else
{
if (ncopies - 1 > (GFC_DESCRIPTOR_EXTENT(ret,0) - 1)
- / GFC_DESCRIPTOR_STRIDE(ret,0))
+ / GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0))
runtime_error ("dim too large in spread()");
}
dest = ret->base_addr;
- stride = GFC_DESCRIPTOR_STRIDE(ret,0);
+ stride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
for (n = 0; n < ncopies; n++)
{
Index: m4/transpose.m4
===================================================================
--- m4/transpose.m4 (Revision 185261)
+++ m4/transpose.m4 (Arbeitskopie)
@@ -87,13 +87,13 @@
}
- sxstride = GFC_DESCRIPTOR_STRIDE(source,0);
- systride = GFC_DESCRIPTOR_STRIDE(source,1);
+ sxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,0);
+ systride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,1);
xcount = GFC_DESCRIPTOR_EXTENT(source,0);
ycount = GFC_DESCRIPTOR_EXTENT(source,1);
- rxstride = GFC_DESCRIPTOR_STRIDE(ret,0);
- rystride = GFC_DESCRIPTOR_STRIDE(ret,1);
+ rxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
+ rystride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,1);
rptr = ret->base_addr;
sptr = source->base_addr;
Index: m4/iforeach.m4
===================================================================
--- m4/iforeach.m4 (Revision 185261)
+++ m4/iforeach.m4 (Arbeitskopie)
@@ -39,11 +39,11 @@
"u_name");
}
- dstride = GFC_DESCRIPTOR_STRIDE(retarray,0);
+ dstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
dest = retarray->base_addr;
for (n = 0; n < rank; n++)
{
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,n);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,n);
extent[n] = GFC_DESCRIPTOR_EXTENT(array,n);
count[n] = 0;
if (extent[n] <= 0)
@@ -160,11 +160,11 @@
else
runtime_error ("Funny sized logical array");
- dstride = GFC_DESCRIPTOR_STRIDE(retarray,0);
+ dstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
dest = retarray->base_addr;
for (n = 0; n < rank; n++)
{
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,n);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
extent[n] = GFC_DESCRIPTOR_EXTENT(array,n);
count[n] = 0;
@@ -272,7 +272,7 @@
"u_name");
}
- dstride = GFC_DESCRIPTOR_STRIDE(retarray,0);
+ dstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
dest = retarray->base_addr;
for (n = 0; n<rank; n++)
dest[n * dstride] = $1 ;
Index: m4/eoshift1.m4
===================================================================
--- m4/eoshift1.m4 (Revision 185261)
+++ m4/eoshift1.m4 (Arbeitskopie)
@@ -145,7 +145,7 @@
rstride[n] = GFC_DESCRIPTOR_SM(ret,dim);
sstride[n] = GFC_DESCRIPTOR_SM(array,dim);
- hstride[n] = GFC_DESCRIPTOR_STRIDE(h,n);
+ hstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(h,n);
n++;
}
}
Index: m4/eoshift3.m4
===================================================================
--- m4/eoshift3.m4 (Revision 185261)
+++ m4/eoshift3.m4 (Arbeitskopie)
@@ -149,7 +149,7 @@
rstride[n] = GFC_DESCRIPTOR_SM(ret,dim);
sstride[n] = GFC_DESCRIPTOR_SM(array,dim);
- hstride[n] = GFC_DESCRIPTOR_STRIDE(h,n);
+ hstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(h,n);
if (bound)
bstride[n] = GFC_DESCRIPTOR_SM(bound,n);
else
Index: m4/shape.m4
===================================================================
--- m4/shape.m4 (Revision 185261)
+++ m4/shape.m4 (Arbeitskopie)
@@ -53,7 +53,7 @@
ret->base_addr = internal_malloc_size (sizeof ('rtype_name`) * rank);
}
- stride = GFC_DESCRIPTOR_STRIDE(ret,0);
+ stride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
if (GFC_DESCRIPTOR_EXTENT(ret,0) < 1)
return;
Index: m4/cshift1.m4
===================================================================
--- m4/cshift1.m4 (Revision 185261)
+++ m4/cshift1.m4 (Arbeitskopie)
@@ -142,7 +142,7 @@
rstride[n] = GFC_DESCRIPTOR_SM(ret,dim);
sstride[n] = GFC_DESCRIPTOR_SM(array,dim);
- hstride[n] = GFC_DESCRIPTOR_STRIDE(h,n);
+ hstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(h,n);
n++;
}
}
Index: m4/in_unpack.m4
===================================================================
--- m4/in_unpack.m4 (Revision 185261)
+++ m4/in_unpack.m4 (Arbeitskopie)
@@ -55,7 +55,7 @@
for (n = 0; n < dim; n++)
{
count[n] = 0;
- stride[n] = GFC_DESCRIPTOR_STRIDE(d,n);
+ stride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(d,n);
extent[n] = GFC_DESCRIPTOR_EXTENT(d,n);
if (extent[n] <= 0)
return;
Index: m4/matmull.m4
===================================================================
--- m4/matmull.m4 (Revision 185261)
+++ m4/matmull.m4 (Arbeitskopie)
@@ -165,13 +165,13 @@
`
if (GFC_DESCRIPTOR_RANK (retarray) == 1)
{
- rxstride = GFC_DESCRIPTOR_STRIDE(retarray,0);
+ rxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
rystride = rxstride;
}
else
{
- rxstride = GFC_DESCRIPTOR_STRIDE(retarray,0);
- rystride = GFC_DESCRIPTOR_STRIDE(retarray,1);
+ rxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
+ rystride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,1);
}
/* If we have rank 1 parameters, zero the absent stride, and set the size to
Index: m4/bessel.m4
===================================================================
--- m4/bessel.m4 (Revision 185261)
+++ m4/bessel.m4 (Arbeitskopie)
@@ -50,7 +50,7 @@
'rtype_name` last1, last2, x2rev;
- stride = GFC_DESCRIPTOR_STRIDE(ret,0);
+ stride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
if (ret->base_addr == NULL)
{
@@ -69,7 +69,7 @@
"(%ld vs. %ld)", (long int) n2-n1,
(long int) GFC_DESCRIPTOR_EXTENT(ret,0));
- stride = GFC_DESCRIPTOR_STRIDE(ret,0);
+ stride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
if (unlikely (x == 0))
{
@@ -117,7 +117,7 @@
'rtype_name` last1, last2, x2rev;
- stride = GFC_DESCRIPTOR_STRIDE(ret,0);
+ stride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
if (ret->base_addr == NULL)
{
@@ -136,7 +136,7 @@
"(%ld vs. %ld)", (long int) n2-n1,
(long int) GFC_DESCRIPTOR_EXTENT(ret,0));
- stride = GFC_DESCRIPTOR_STRIDE(ret,0);
+ stride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,0);
if (unlikely (x == 0))
{
Index: m4/unpack.m4
===================================================================
--- m4/unpack.m4 (Revision 185261)
+++ m4/unpack.m4 (Arbeitskopie)
@@ -96,7 +96,7 @@
rs * sizeof ('rtype_name`));
extent[n] = GFC_DESCRIPTOR_EXTENT(ret,n);
empty = empty || extent[n] <= 0;
- rstride[n] = GFC_DESCRIPTOR_STRIDE(ret,n);
+ rstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
rs *= extent[n];
}
@@ -111,7 +111,7 @@
count[n] = 0;
extent[n] = GFC_DESCRIPTOR_EXTENT(ret,n);
empty = empty || extent[n] <= 0;
- rstride[n] = GFC_DESCRIPTOR_STRIDE(ret,n);
+ rstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
}
if (rstride[0] == 0)
@@ -124,7 +124,7 @@
if (mstride[0] == 0)
mstride[0] = 1;
- vstride0 = GFC_DESCRIPTOR_STRIDE(vector,0);
+ vstride0 = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(vector,0);
if (vstride0 == 0)
vstride0 = 1;
rstride0 = rstride[0];
@@ -241,8 +241,8 @@
rs * sizeof ('rtype_name`));
extent[n] = GFC_DESCRIPTOR_EXTENT(ret,n);
empty = empty || extent[n] <= 0;
- rstride[n] = GFC_DESCRIPTOR_STRIDE(ret,n);
- fstride[n] = GFC_DESCRIPTOR_STRIDE(field,n);
+ rstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
+ fstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(field,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
rs *= extent[n];
}
@@ -257,8 +257,8 @@
count[n] = 0;
extent[n] = GFC_DESCRIPTOR_EXTENT(ret,n);
empty = empty || extent[n] <= 0;
- rstride[n] = GFC_DESCRIPTOR_STRIDE(ret,n);
- fstride[n] = GFC_DESCRIPTOR_STRIDE(field,n);
+ rstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,n);
+ fstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(field,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
}
if (rstride[0] == 0)
@@ -273,7 +273,7 @@
if (mstride[0] == 0)
mstride[0] = 1;
- vstride0 = GFC_DESCRIPTOR_STRIDE(vector,0);
+ vstride0 = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(vector,0);
if (vstride0 == 0)
vstride0 = 1;
rstride0 = rstride[0];
Index: m4/reshape.m4
===================================================================
--- m4/reshape.m4 (Revision 185261)
+++ m4/reshape.m4 (Arbeitskopie)
@@ -91,7 +91,7 @@
for (n = 0; n < rdim; n++)
{
- shape_data[n] = shape->base_addr[n * GFC_DESCRIPTOR_STRIDE(shape,0)];
+ shape_data[n] = shape->base_addr[n * GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(shape,0)];
if (shape_data[n] <= 0)
{
shape_data[n] = 0;
@@ -134,7 +134,7 @@
for (n = 0; n < pdim; n++)
{
pcount[n] = 0;
- pstride[n] = GFC_DESCRIPTOR_STRIDE(pad,n);
+ pstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(pad,n);
pextent[n] = GFC_DESCRIPTOR_EXTENT(pad,n);
if (pextent[n] <= 0)
{
@@ -197,7 +197,7 @@
for (n = 0; n < rdim; n++)
{
- v = order->base_addr[n * GFC_DESCRIPTOR_STRIDE(order,0)] - 1;
+ v = order->base_addr[n * GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(order,0)] - 1;
if (v < 0 || v >= rdim)
runtime_error("Value %ld out of range in ORDER argument"
@@ -216,12 +216,12 @@
for (n = 0; n < rdim; n++)
{
if (order)
- dim = order->base_addr[n * GFC_DESCRIPTOR_STRIDE(order,0)] - 1;
+ dim = order->base_addr[n * GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(order,0)] - 1;
else
dim = n;
rcount[n] = 0;
- rstride[n] = GFC_DESCRIPTOR_STRIDE(ret,dim);
+ rstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(ret,dim);
rextent[n] = GFC_DESCRIPTOR_EXTENT(ret,dim);
if (rextent[n] < 0)
rextent[n] = 0;
@@ -243,7 +243,7 @@
for (n = 0; n < sdim; n++)
{
scount[n] = 0;
- sstride[n] = GFC_DESCRIPTOR_STRIDE(source,n);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(source,n);
sextent[n] = GFC_DESCRIPTOR_EXTENT(source,n);
if (sextent[n] <= 0)
{
Index: m4/ifunction_logical.m4
===================================================================
--- m4/ifunction_logical.m4 (Revision 185261)
+++ m4/ifunction_logical.m4 (Arbeitskopie)
@@ -127,7 +127,7 @@
for (n = 0; n < rank; n++)
{
count[n] = 0;
- dstride[n] = GFC_DESCRIPTOR_STRIDE(retarray,n);
+ dstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,n);
if (extent[n] <= 0)
return;
}
Index: m4/ifunction.m4
===================================================================
--- m4/ifunction.m4 (Revision 185261)
+++ m4/ifunction.m4 (Arbeitskopie)
@@ -48,11 +48,11 @@
len = GFC_DESCRIPTOR_EXTENT(array,dim);
if (len < 0)
len = 0;
- delta = GFC_DESCRIPTOR_STRIDE(array,dim);
+ delta = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,dim);
for (n = 0; n < dim; n++)
{
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,n);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,n);
extent[n] = GFC_DESCRIPTOR_EXTENT(array,n);
if (extent[n] < 0)
@@ -60,7 +60,7 @@
}
for (n = dim; n < rank; n++)
{
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array, n + 1);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array, n + 1);
extent[n] = GFC_DESCRIPTOR_EXTENT(array, n + 1);
if (extent[n] < 0)
@@ -113,7 +113,7 @@
for (n = 0; n < rank; n++)
{
count[n] = 0;
- dstride[n] = GFC_DESCRIPTOR_STRIDE(retarray,n);
+ dstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,n);
if (extent[n] <= 0)
return;
}
@@ -222,12 +222,12 @@
else
runtime_error ("Funny sized logical array");
- delta = GFC_DESCRIPTOR_STRIDE(array,dim);
+ delta = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,dim);
mdelta = GFC_DESCRIPTOR_SM(mask,dim);
for (n = 0; n < dim; n++)
{
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,n);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,n);
mstride[n] = GFC_DESCRIPTOR_SM(mask,n);
extent[n] = GFC_DESCRIPTOR_EXTENT(array,n);
@@ -237,7 +237,7 @@
}
for (n = dim; n < rank; n++)
{
- sstride[n] = GFC_DESCRIPTOR_STRIDE(array,n + 1);
+ sstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(array,n + 1);
mstride[n] = GFC_DESCRIPTOR_SM(mask, n + 1);
extent[n] = GFC_DESCRIPTOR_EXTENT(array, n + 1);
@@ -293,7 +293,7 @@
for (n = 0; n < rank; n++)
{
count[n] = 0;
- dstride[n] = GFC_DESCRIPTOR_STRIDE(retarray,n);
+ dstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,n);
if (extent[n] <= 0)
return;
}
@@ -460,7 +460,7 @@
for (n = 0; n < rank; n++)
{
count[n] = 0;
- dstride[n] = GFC_DESCRIPTOR_STRIDE(retarray,n);
+ dstride[n] = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,n);
}
dest = retarray->base_addr;
Index: m4/matmul.m4
===================================================================
--- m4/matmul.m4 (Revision 185261)
+++ m4/matmul.m4 (Arbeitskopie)
@@ -178,19 +178,19 @@
/* One-dimensional result may be addressed in the code below
either as a row or a column matrix. We want both cases to
work. */
- rxstride = rystride = GFC_DESCRIPTOR_STRIDE(retarray,0);
+ rxstride = rystride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
}
else
{
- rxstride = GFC_DESCRIPTOR_STRIDE(retarray,0);
- rystride = GFC_DESCRIPTOR_STRIDE(retarray,1);
+ rxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,0);
+ rystride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(retarray,1);
}
if (GFC_DESCRIPTOR_RANK (a) == 1)
{
/* Treat it as a a row matrix A[1,count]. */
- axstride = GFC_DESCRIPTOR_STRIDE(a,0);
+ axstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(a,0);
aystride = 1;
xcount = 1;
@@ -198,8 +198,8 @@
}
else
{
- axstride = GFC_DESCRIPTOR_STRIDE(a,0);
- aystride = GFC_DESCRIPTOR_STRIDE(a,1);
+ axstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(a,0);
+ aystride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(a,1);
count = GFC_DESCRIPTOR_EXTENT(a,1);
xcount = GFC_DESCRIPTOR_EXTENT(a,0);
@@ -214,7 +214,7 @@
if (GFC_DESCRIPTOR_RANK (b) == 1)
{
/* Treat it as a column matrix B[count,1] */
- bxstride = GFC_DESCRIPTOR_STRIDE(b,0);
+ bxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(b,0);
/* bystride should never be used for 1-dimensional b.
in case it is we want it to cause a segfault, rather than
@@ -224,8 +224,8 @@
}
else
{
- bxstride = GFC_DESCRIPTOR_STRIDE(b,0);
- bystride = GFC_DESCRIPTOR_STRIDE(b,1);
+ bxstride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(b,0);
+ bystride = GFC_DESCRIPTOR_STRIDE_TYPEKNOWN(b,1);
ycount = GFC_DESCRIPTOR_EXTENT(b,1);
}
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [patch, fortran-dev] Use fixed variable sizes for stride calculations
2012-03-24 14:14 [patch, fortran-dev] Use fixed variable sizes for stride calculations Thomas Koenig
@ 2012-03-28 8:12 ` Janne Blomqvist
2012-04-11 7:51 ` Thomas Koenig
1 sibling, 0 replies; 5+ messages in thread
From: Janne Blomqvist @ 2012-03-28 8:12 UTC (permalink / raw)
To: Thomas Koenig; +Cc: fortran, gcc-patches
On Sat, Mar 24, 2012 at 16:13, Thomas Koenig <tkoenig@netcologne.de> wrote:
> Hello world,
>
> this patch uses division by known sizes (which can usually be replaced
> by a simple shift because intrinsics have sizes of power of two) instead
> of division by the size extracted from the array descriptor itself.
>
> This should save about 20 cycles for a single calculation.
>
> I'll go through the rest of the library to identify other possibilities
> for this.
>
> Regression-tested, no new failures.
>
> OK for the branch?
Index: libgfortran.h
===================================================================
--- libgfortran.h (Revision 185261)
+++ libgfortran.h (Arbeitskopie)
@@ -364,6 +364,11 @@
#define GFC_DESCRIPTOR_TYPE(desc) (((desc)->dtype & GFC_DTYPE_TYPE_MASK) \
>> GFC_DTYPE_TYPE_SHIFT)
#define GFC_DESCRIPTOR_SIZE(desc) ((desc)->dtype >> GFC_DTYPE_SIZE_SHIFT)
+
+/* This is for getting the size of a descriptor when the type of the
+ descriptor is known at compile-time. Do not use for string types. */
+
+#define GFC_DESCRIPTOR_SIZE_TYPEKNOWN(desc) (sizeof((desc)->base_addr[0]))
#define GFC_DESCRIPTOR_DATA(desc) ((desc)->base_addr)
#define GFC_DESCRIPTOR_DTYPE(desc) ((desc)->dtype)
The comment should say "size of a type" rather than "size of a descriptor".
Otherwise Ok.
Though I think that longer term, maybe we should do the address
calculations in bytes directly? Now we have (on the branch) the stride
in bytes, which we convert to elements by dividing with the element
size, and then the compiler again converts it to bytes by multiplying
with the element size when we calculate the address of an array
element. So maybe instead we should have the address of the current
element as a variable of type char*, and then only when operating on
that element we cast it to the correct type?
--
Janne Blomqvist
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [patch, fortran-dev] Use fixed variable sizes for stride calculations
2012-03-24 14:14 [patch, fortran-dev] Use fixed variable sizes for stride calculations Thomas Koenig
2012-03-28 8:12 ` Janne Blomqvist
@ 2012-04-11 7:51 ` Thomas Koenig
2012-04-13 17:32 ` Janne Blomqvist
1 sibling, 1 reply; 5+ messages in thread
From: Thomas Koenig @ 2012-04-11 7:51 UTC (permalink / raw)
To: fortran, gcc-patches
Hi,
> this patch uses division by known sizes (which can usually be replaced
> by a simple shift because intrinsics have sizes of power of two) instead
> of division by the size extracted from the array descriptor itself.
>
> This should save about 20 cycles for a single calculation.
>
> I'll go through the rest of the library to identify other possibilities
> for this.
>
> Regression-tested, no new failures.
>
> OK for the branch?
Full patch at http://gcc.gnu.org/ml/fortran/2012-03/msg00120.html
Ping?
Thomas
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [patch, fortran-dev] Use fixed variable sizes for stride calculations
2012-04-11 7:51 ` Thomas Koenig
@ 2012-04-13 17:32 ` Janne Blomqvist
2012-04-13 22:50 ` Thomas Koenig
0 siblings, 1 reply; 5+ messages in thread
From: Janne Blomqvist @ 2012-04-13 17:32 UTC (permalink / raw)
To: Thomas Koenig; +Cc: fortran, gcc-patches
On Wed, Apr 11, 2012 at 10:51, Thomas Koenig <tkoenig@netcologne.de> wrote:
> Hi,
>
>
>> this patch uses division by known sizes (which can usually be replaced
>> by a simple shift because intrinsics have sizes of power of two) instead
>> of division by the size extracted from the array descriptor itself.
>>
>> This should save about 20 cycles for a single calculation.
>>
>> I'll go through the rest of the library to identify other possibilities
>> for this.
>>
>> Regression-tested, no new failures.
>>
>> OK for the branch?
>
>
> Full patch at http://gcc.gnu.org/ml/fortran/2012-03/msg00120.html
>
> Ping?
>
> Thomas
I thought I already approved this a few weeks ago
(http://gcc.gnu.org/ml/fortran/2012-03/msg00144.html)? Or if it has
been decided that I'm no longer a GFortran reviewer, it would be nice
if I'd be informed of this as well.
--
Janne Blomqvist
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2012-04-13 22:50 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-03-24 14:14 [patch, fortran-dev] Use fixed variable sizes for stride calculations Thomas Koenig
2012-03-28 8:12 ` Janne Blomqvist
2012-04-11 7:51 ` Thomas Koenig
2012-04-13 17:32 ` Janne Blomqvist
2012-04-13 22:50 ` Thomas Koenig
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).