I've updated speed so we can do comparisons of combined functions easier.
rev 1941+ in trunk
a new option colsum=A+B+...+Z
which sums the speed columns A,B, ... (MAX 4 columns for time being)
Note: the summed columns A,B, etc are excluded from the fastest # thing
-r and -d now take a parameter to specify which column we want the
ratio/difference against.
Documented the scale factor option eg where 2*mpn_.. can be any float
eg
./speed -c -s 1-10 mpn_addlsh1_n mpn_add_n mpn_lshift1 colsum=2+3
overhead 5.71 cycles, precision 1000000 units of 3.75e-10 secs, CPU freq
2664.74 MHz
mpn_addlsh1_n mpn_add_n mpn_lshift1 colsum=2+3
1 #13.33 19.05 5.71 24.77
2 #19.05 22.62 9.52 32.15
3 #24.76 25.86 20.00 45.87
4 #19.05 29.23 18.10 47.32
5 #24.95 27.72 22.86 50.58
6 #30.48 31.43 24.76 56.20
7 #36.27 35.64 28.57 64.22
8 #32.43 38.34 27.62 65.96
9 #38.10 37.15 31.43 68.58
10 #43.81 41.50 34.29 75.79
./speed -c -s 1-10 mpn_mul_2 colsum=3+4 mpn_mul_1.33 mpn_addmul_1.33
overhead 5.71 cycles, precision 1000000 units of 3.75e-10 secs, CPU freq
2664.74 MHz
mpn_mul_2 colsum=3+4 mpn_mul_1.33 mpn_addmul_1.33
1 n/a #20.00 8.57 11.43
2 #21.87 30.54 13.33 17.21
3 #28.57 39.38 16.24 23.15
4 #37.59 50.89 21.91 28.98
5 #44.29 57.48 22.86 34.62
6 #51.40 67.53 26.91 40.63
7 #58.67 75.41 29.69 45.72
8 #66.67 86.17 34.29 51.89
9 #73.84 93.15 36.00 57.15
10 #82.61 103.66 40.63 63.03
./speed -c -s 1-10 2*mpn_add_n mpn_addadd_n
overhead 5.71 cycles, precision 1000000 units of 3.75e-10 secs, CPU freq
2664.74 MHz
2*mpn_add_n mpn_addadd_n
1 38.10 #14.29
2 45.24 #20.00
3 51.73 #25.72
4 58.45 #24.77
5 55.43 #27.62
6 62.86 #33.53
7 71.44 #39.05
8 76.67 #38.10
9 74.29 #41.04
10 83.02 #46.81
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"mpir-devel" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/mpir-devel?hl=en
-~----------~----~----~----~------~----~------~--~---
