Update benchmarks with faster hardware (Core i5-4590)

It is approximately 4x faster than the Core Q8200.

benchmarks/bench

@@ -1,17 +1,20 @@
 # Benchmark of combined FFT and math/big (using threshold)
-BenchmarkMul_1kb	 1000000	      1500 ns/op
-BenchmarkMul_10kb	   50000	     50777 ns/op
-BenchmarkMul_50kb	    5000	    658534 ns/op
-BenchmarkMul_100kb	    1000	   2127534 ns/op
-BenchmarkMul_1Mb	     100	  22391830 ns/op
-BenchmarkMul_5Mb	      10	 133550600 ns/op
-BenchmarkMul_10Mb	       5	 318595800 ns/op
-BenchmarkMul_20Mb	       5	 671512800 ns/op
-BenchmarkMul_50Mb	       1	2451476000 ns/op
-BenchmarkMul_100Mb	       1	5228991000 ns/op
-BenchmarkMul_1x5Mb	      20	  78091550 ns/op
-BenchmarkMul_1x10Mb	      10	 164484500 ns/op
-BenchmarkMul_1x20Mb	       5	 366846400 ns/op
-BenchmarkMul_1x50Mb	       1	1040516000 ns/op
-BenchmarkMul_5x20Mb	       5	 457571200 ns/op
-BenchmarkMul_5x50Mb	       1	1040664000 ns/op
+# Run on a Core i5-4590 (Go 1.8)
+BenchmarkMul_1kb      	 3000000	       525 ns/op
+BenchmarkMul_10kb     	  100000	     19274 ns/op
+BenchmarkMul_50kb     	    5000	    213375 ns/op
+BenchmarkMul_100kb    	    2000	    651794 ns/op
+BenchmarkMul_1Mb      	     200	   8546244 ns/op
+BenchmarkMul_5Mb      	      30	  49127283 ns/op
+BenchmarkMul_10Mb     	      10	 109888838 ns/op
+BenchmarkMul_20Mb     	       5	 227088971 ns/op
+BenchmarkMul_50Mb     	       2	 731298339 ns/op
+BenchmarkMul_100Mb    	       1	1480340166 ns/op
+
+BenchmarkMul_1x5Mb    	      50	  28872973 ns/op
+BenchmarkMul_1x10Mb   	      20	  58841416 ns/op
+BenchmarkMul_1x20Mb   	      10	 124189252 ns/op
+BenchmarkMul_1x50Mb   	       3	 349402586 ns/op
+BenchmarkMul_5x20Mb   	      10	 153528843 ns/op
+BenchmarkMul_5x50Mb   	       3	 348753322 ns/op
+

benchmarks/bench.big

@@ -1,24 +1,19 @@
-BenchmarkMul_1kb	 1000000	      1599 ns/op
-BenchmarkMul_10kb	   50000	     61533 ns/op
-BenchmarkMul_50kb	    2000	    833693 ns/op
-BenchmarkMul_100kb	    1000	   2567995 ns/op
-BenchmarkMul_1Mb	      20	 105237800 ns/op
-BenchmarkMul_5Mb	       1	1272947000 ns/op
-BenchmarkMul_10Mb	       1	3834354000 ns/op
-BenchmarkMul_20Mb	       1	11514488000 ns/op
-# these values are extrapolated using Karatsuba theoretical
-# complexity:
-# time(10Mb) = 3*time(5Mb)
-# time(20Mb) = 3*time(10Mb)
-# time(50Mb) = 4.27*time(20Mb)
-# time(100Mb) = 3*time(50Mb)
-BenchmarkMul_50Mb	       1        49199945000 ns/op
-BenchmarkMul_100Mb	       1	147599836000 ns/op
+# Benchmarks run on a Core i5-4590 (Go 1.8)
+BenchmarkMul_1kb      	 5000000	       298 ns/op
+BenchmarkMul_10kb     	  100000	     14562 ns/op
+BenchmarkMul_50kb     	   10000	    204698 ns/op
+BenchmarkMul_100kb    	    2000	    636230 ns/op
+BenchmarkMul_1Mb      	      50	  25950594 ns/op
+BenchmarkMul_5Mb      	       5	 314799939 ns/op
+BenchmarkMul_10Mb     	       2	 943065686 ns/op
+BenchmarkMul_20Mb     	       1	2837283743 ns/op
+BenchmarkMul_50Mb     	       1	14329431306 ns/op
+BenchmarkMul_100Mb    	       1	42590328264 ns/op
 
-BenchmarkMul_1x5Mb         20         507268000 ns/op
-BenchmarkMul_1x10Mb         1        1002801000 ns/op
-BenchmarkMul_1x20Mb         1        1988546000 ns/op
-BenchmarkMul_1x50Mb         1        4940572000 ns/op
-BenchmarkMul_5x20Mb         1        5011810000 ns/op
-BenchmarkMul_5x50Mb         1        12449508000 ns/op
+BenchmarkMul_1x5Mb    	      10	 126106007 ns/op
+BenchmarkMul_1x10Mb   	       5	 248876061 ns/op
+BenchmarkMul_1x20Mb   	       3	 492849546 ns/op
+BenchmarkMul_1x50Mb   	       1	1249673962 ns/op
+BenchmarkMul_5x20Mb   	       1	1261943492 ns/op
+BenchmarkMul_5x50Mb   	       1	3098019651 ns/op
 

benchmarks/bench.fft

@@ -1,19 +1,20 @@
 # Benchmarks using the mulFFT function only.
-BenchmarkMul_1kb       100000             29659 ns/op
-BenchmarkMul_10kb       10000            292823 ns/op
-BenchmarkMul_50kb        1000           1673191 ns/op
-BenchmarkMul_100kb       1000           2609174 ns/op
-BenchmarkMul_1Mb           50          28511840 ns/op
-BenchmarkMul_5Mb           10         168610100 ns/op
-BenchmarkMul_10Mb           5         403888600 ns/op
-BenchmarkMul_20Mb           5         849141200 ns/op
-BenchmarkMul_50Mb           1        2922353000 ns/op
-BenchmarkMul_100Mb          1        5970592000 ns/op
+# Run on a Core i5-4590 (Go 1.8)
+BenchmarkMul_1kb      	  200000	      9737 ns/op
+BenchmarkMul_10kb     	   10000	    105408 ns/op
+BenchmarkMul_50kb     	    3000	    584090 ns/op
+BenchmarkMul_100kb    	    2000	    973130 ns/op
+BenchmarkMul_1Mb      	     200	   8622463 ns/op
+BenchmarkMul_5Mb      	      30	  48602728 ns/op
+BenchmarkMul_10Mb     	      10	 109184721 ns/op
+BenchmarkMul_20Mb     	       5	 227053895 ns/op
+BenchmarkMul_50Mb     	       2	 727421044 ns/op
+BenchmarkMul_100Mb    	       1	1550029484 ns/op
 
-BenchmarkMul_1x5Mb        100          99174500 ns/op
-BenchmarkMul_1x10Mb        10         207445000 ns/op
-BenchmarkMul_1x20Mb         5         463875000 ns/op
-BenchmarkMul_1x50Mb         1        1341762000 ns/op
-BenchmarkMul_5x20Mb         5         575246400 ns/op
-BenchmarkMul_5x50Mb         1        1326256000 ns/op
+BenchmarkMul_1x5Mb    	      50	  28827150 ns/op
+BenchmarkMul_1x10Mb   	      20	  58097775 ns/op
+BenchmarkMul_1x20Mb   	      10	 124998246 ns/op
+BenchmarkMul_1x50Mb   	       3	 350045770 ns/op
+BenchmarkMul_5x20Mb   	      10	 160220847 ns/op
+BenchmarkMul_5x50Mb   	       3	 350824154 ns/op
 

benchmarks/bench.gmp

@@ -1,20 +1,21 @@
 # These benchamrks were realised using gmpbench at
 # http://gmplib.org/gmpbench.html
 # and converted to the go test output format.
-BenchmarkMul_1kb        18460846	       536 ns/op
-BenchmarkMul_10kb	  374920	     26669 ns/op
-BenchmarkMul_50kb	   38936	    252270 ns/op
-BenchmarkMul_100kb	   14505	    686813 ns/op
-BenchmarkMul_1Mb	     836	  12100000 ns/op
-BenchmarkMul_5Mb	      88	 111731843 ns/op
-BenchmarkMul_10Mb	      48         212314000 ns/op
-BenchmarkMul_20Mb	      21	 490196000 ns/op
-BenchmarkMul_50Mb	       8	1280000000 ns/op
-BenchmarkMul_100Mb	       4	2673000000 ns/op
+# Numbers are for a Core i5-4590 with GMP 6.1.2
+BenchmarkMul_1kb        47143107	       175 ns/op
+BenchmarkMul_10kb	 1321291	      7573 ns/op
+BenchmarkMul_50kb	  125645	     79693 ns/op
+BenchmarkMul_100kb	   47298	    211500 ns/op
+BenchmarkMul_1Mb	    2950	   3344500 ns/op
+BenchmarkMul_5Mb	     413	  23920000 ns/op
+BenchmarkMul_10Mb	     164          60606000 ns/op
+BenchmarkMul_20Mb	      78	 127700000 ns/op
+BenchmarkMul_50Mb	       8	 352100000 ns/op
+BenchmarkMul_100Mb	       4	 746270000 ns/op
 
-BenchmarkMul_1x5Mb        151          66225165 ns/op
-BenchmarkMul_1x10Mb       111          90909000 ns/op
-BenchmarkMul_1x20Mb        59         176678000 ns/op
-BenchmarkMul_1x50Mb        23         434782000 ns/op
-BenchmarkMul_5x20Mb        35         289855000 ns/op
-BenchmarkMul_5x50Mb        14         746269000 ns/op
+BenchmarkMul_1x5Mb        884          11670000 ns/op
+BenchmarkMul_1x10Mb       337          27174000 ns/op
+BenchmarkMul_1x20Mb       195          52630000 ns/op
+BenchmarkMul_1x50Mb        70         131000000 ns/op
+BenchmarkMul_5x20Mb       134          74188000 ns/op
+BenchmarkMul_5x50Mb        49         207770000 ns/op

fft_test.go

@@ -237,9 +237,8 @@ func BenchmarkMulBig_1Mb(b *testing.B)   { benchmarkMulBig(b, 1e6, 1e6) }
 func BenchmarkMulBig_5Mb(b *testing.B)   { benchmarkMulBig(b, 5e6, 5e6) }
 func BenchmarkMulBig_10Mb(b *testing.B)  { benchmarkMulBig(b, 10e6, 10e6) }
 func BenchmarkMulBig_20Mb(b *testing.B)  { benchmarkMulBig(b, 20e6, 20e6) }
-
-// 50Mb multiplication takes about 1 minute.
-//func BenchmarkMulBig_50Mb(b *testing.B)  { benchmarkMulBig(b, 50e6, 50e6) }
+func BenchmarkMulBig_50Mb(b *testing.B)  { benchmarkMulBig(b, 50e6, 50e6) }
+func BenchmarkMulBig_100Mb(b *testing.B) { benchmarkMulBig(b, 100e6, 100e6) }
 
 func BenchmarkMulFFT_1kb(b *testing.B)  { benchmarkMulFFT(b, 1e3, 1e3) }
 func BenchmarkMulFFT_10kb(b *testing.B) { benchmarkMulFFT(b, 1e4, 1e4) }