Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commite3d41d0

Browse files
committed
Apply auto-vectorization to the inner loop of div_var_fast().
This loop is basically the same as the inner loop of mul_var(), whichwas auto-vectorized in commit8870917, but the compiler will onlyconsider auto-vectorizing the div_var_fast() loop if the assignmenttarget div[qi + i] is replaced by div_qi[i], where div_qi = &div[qi].Additionally, since the compiler doesn't know that qdigit isguaranteed to fit in a 16-bit NumericDigit, cast it to NumericDigitbefore multiplying to make the resulting auto-vectorized code moreefficient (avoiding unnecessary multiplication of the high 16 bits).While at it, per suggestion from Tom Lane, change var1digit inmul_var() to be a NumericDigit rather than an int for the samereason. This actually makes no difference with modern gcc, but itmight help other compilers generate more efficient assembly.Dean Rasheed, reviewed by Tom Lane.Discussion:https://postgr.es/m/CAEZATCVwsBi-ND-t82Cuuh1=8ee6jdOpzsmGN+CUZB6yjLg9jw@mail.gmail.com
1 parent6b04abd commite3d41d0

File tree

1 file changed

+11
-2
lines changed

1 file changed

+11
-2
lines changed

‎src/backend/utils/adt/numeric.c

Lines changed: 11 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -8323,7 +8323,7 @@ mul_var(const NumericVar *var1, const NumericVar *var2, NumericVar *result,
83238323
*/
83248324
for (i1=Min(var1ndigits-1,res_ndigits-3);i1 >=0;i1--)
83258325
{
8326-
intvar1digit=var1digits[i1];
8326+
NumericDigitvar1digit=var1digits[i1];
83278327

83288328
if (var1digit==0)
83298329
continue;
@@ -8908,13 +8908,22 @@ div_var_fast(const NumericVar *var1, const NumericVar *var2,
89088908
* which would make the new value simply div[qi] mod vardigits[0].
89098909
* The lower-order terms in qdigit can change this result by not
89108910
* more than about twice INT_MAX/NBASE, so overflow is impossible.
8911+
*
8912+
* This inner loop is the performance bottleneck for division, so
8913+
* code it in the same way as the inner loop of mul_var() so that
8914+
* it can be auto-vectorized. We cast qdigit to NumericDigit
8915+
* before multiplying to allow the compiler to generate more
8916+
* efficient code (using 16-bit multiplication), which is safe
8917+
* since we know that the quotient digit is off by at most one, so
8918+
* there is no overflow risk.
89118919
*/
89128920
if (qdigit!=0)
89138921
{
89148922
intistop=Min(var2ndigits,div_ndigits-qi+1);
8923+
int*div_qi=&div[qi];
89158924

89168925
for (i=0;i<istop;i++)
8917-
div[qi+i]-=qdigit*var2digits[i];
8926+
div_qi[i]-=((NumericDigit)qdigit)*var2digits[i];
89188927
}
89198928
}
89208929

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp