CPU Benchmarks for Lucas-Lehmer/Trial Factor

Great Internet Mersenne Prime Search
GIMPS
Finding World Record Primes Since 1996

PrimeNet CPU Benchmarks

Log in to see your benchmarks	Exclude version 25 benchmarks	Exclude overclocked and underclocked data
Show 32-bit benchmarks	Exclude version 26 benchmarks	Exclude benchmarks flagged by users as suspect
Show 64-bit benchmarks	Exclude version 27 benchmarks

CPU speed range:
CPU must have this many benchmarks reported:
Only display benchmarks reported since:	(yyyy-mm-dd)

Model	Intel(R) Core(TM) i7-8086K CPU @ 4.00GHz
Stock Speed	no stock speed set
Features	Hyperthreaded,Prefetch,SSE,SSE2,SSE4,AVX,AVX2,FMA
L1 Cache	32 KB
L2 Cache	256 KB
L3 Cache	12288 KB

Program	Speed (MHz)	29.69M to 34.56M (1792K)	34.56M to 39.50M (2048K)	39.50M to 49.10M (2560K)	49.10M to 58.52M (3072K)	58.52M to 68.13M (3584K)	68.13M to 77.91M (4096K)
Windows64, Prime95, v28.10, build 1	5085	5.57	6.05	7.90	10.14	11.79	13.09
Windows64, Prime95, v29.4, build 8	4986
Windows64, Prime95, v28.10, build 1	4984	6.10	6.56	8.60	10.42	11.93	14.36
Windows64, Prime95, v30.3, build 6	4886
Windows64, Prime95, v30.19, build 13	4788
Windows64, Prime95, v29.4, build 8	4787
Mac OS X 64-bit, Prime95, v29.8, build 4	4786
Windows64, Prime95, v29.4, build 8	4687
Windows64, Prime95, v29.4, build 8	4687
Windows64, Prime95, v29.4, build 8	4687
Windows64, Prime95, v30.8, build 10	4686
Windows64, Prime95, v29.4, build 8	4289
Windows64, Prime95, v29.4, build 8	4288
Windows64, Prime95, v30.8, build 17	4287
Mac OS X 64-bit, Prime95, v28.7, build 1	4008	5.46	6.10	7.91	9.35	11.49	12.58
Mac OS X 64-bit, Prime95, v28.7, build 1	4008	5.51	6.35	8.03	9.68	11.73	13.27
Mac OS X 64-bit, Prime95, v29.4, build 7	4008
Windows64, Prime95, v29.8, build 3	4008
Windows64, Prime95, v29.8, build 3	4008
Windows64, Prime95, v29.8, build 5	4008
Windows64, Prime95, v28.7, build 1	4007	12.91	12.45	15.92	20.90	23.41	30.86
Windows64, Prime95, v28.7, build 1	4007	10.72	14.58	18.38	21.20	25.03	27.04
Windows64, Prime95, v29.8, build 6	100
All timings are in milliseconds - lower is better

You can help improve the accuracy of our benchmark data by reporting benchmarks that are obviously wrong. This can happen when cpu speed is detected incorrectly which often occurs when SpeedStep is enabled. Bad data can also occur when other programs run during a benchmark -- you might see all timings are too high or you might see a smaller FFT taking more than a larger FFT. By flagging these bad benchmarks, the main benchmark page will report more accurate average timings.

Check to confirm you want to mark benchmarks suspect

If someone has inadvertantly marked good benchmark data suspect, you can vote for marking the benchmark valid.

If the stock CPU speed is not set correctly, you can set the stock CPU speed so that we correctly identify overclocked and underclocked benchmarks.

Stock CPU speed:

Username
Password