29.2 benchmark help - Page 4

mackerel · 2017-04-29, 18:03

i3-4360 at 3.7 GHz all cores active, 1600 dual channel, 2 rank ram. HT was turned on
Note I made a typo in filename, cpu is 4360 at 3.7 GHz not 4370 which is fractionally faster.

I tried running this twice on a mildly OC'd i7-6700k system but it rebooted twice, so I'm giving up. I find it curious, as the same system had participated in 2 weeks of multi-thread LLR prior to this without fuss, and had been doing single thread work for much longer. It's not overheating and it's not Windows Update either!

mackerel · 2017-04-29, 21:17

The final results are in. I do not intend to run further testing unless requested.

i5-5675C at 3.5 GHz, 1600 dual channel 2R ram (and 128MB L4 cache)
i3-6100 at 3.7 GHz, 2666 dual channel 1R ram
i5-4570S at 3.2 GHz, 2400 dual channel 2R ram (rather fast for DDR3...)

GP2 · 2017-04-29, 22:48

Quote:

Originally Posted by ATH

I used the special 29.2 mprime. Mine took 14h15m for just 1 + all workers, so yours must have taken a lot longer than 15 hours.

Mine started at 16:54:10 and ended just past 08:02:25 the next day. The timestamps are in the results.txt file.

Your file has the "FFTlen=" format and you seem to be running four separate tests with different parameters at each combination of FFT length and no. of workers. My file has only one test for each combination.

It seems we ran rather different tests, but I don't know what is the explanation.

ATH · 2017-04-29, 23:00

I think you were missing AllBench=1 in prime.txt then. I started a run by mistake without that at first as well.

Mark Rose · 2017-04-30, 19:03

An early look at my benchmarks shows there's a 10% improvement to be had by better FFT selection in at least one case.

tului · 2017-05-01, 10:27

Got a Ryzen 1800X DDR4-2666 14-15-15-15 I just started it on. I'll post the results.txt when it completes.

tului · 2017-05-01, 15:22

1800X with DDR4 2666 @ CL 14

Mark Rose · 2017-05-01, 15:27

Quote:

Originally Posted by tului

1800X with DDR4 2666 @ CL 14

I believe you forgot to put AllBench=1 in prime.txt before starting Prime95. It will take several times longer to run (also, there is no need to test hyperthreading).

tului · 2017-05-01, 18:59

Quote:

Originally Posted by Mark Rose

I believe you forgot to put AllBench=1 in prime.txt before starting Prime95. It will take several times longer to run (also, there is no need to test hyperthreading).

kek. I put allbench=1 in prime.txt before doing anything. looks like it overwrote the file.

Mark Rose · 2017-05-01, 19:15

Quote:

Originally Posted by tului

kek. I put allbench=1 in prime.txt before doing anything. looks like it overwrote the file.

You have to edit it when Prime95/mprime isn't running.

tului · 2017-05-02, 05:07

Quote:

Originally Posted by Mark Rose

You have to edit it when Prime95/mprime isn't running.

i made the file before I moved the .exe into the folder

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
29.2 benchmark help #2 (Ryzen only)	Prime95	Software	10	2017-05-08 13:24
Benchmark Variances	Fred	Software	5	2016-04-01 18:15
LLR benchmark thread	Oddball	Riesel Prime Search	5	2010-08-02 00:11
Does anyone have i7 920? for Benchmark?	cipher	Twin Prime Search	2	2009-04-14 20:16
Benchmark Weirdness	R.D. Silverman	Hardware	2	2007-07-25 12:16

2017-04-29, 23:00	#37
ATH Einyen Dec 2003 Denmark 2×1,579 Posts	I think you were missing AllBench=1 in prime.txt then. I started a run by mistake without that at first as well.

2017-04-30, 19:03	#38
Mark Rose "/X\(‘-‘)/X\" Jan 2013 2×5×293 Posts	An early look at my benchmarks shows there's a 10% improvement to be had by better FFT selection in at least one case.

2017-05-01, 10:27	#39
tului Jan 2013 2²×17 Posts	Got a Ryzen 1800X DDR4-2666 14-15-15-15 I just started it on. I'll post the results.txt when it completes.