gnfs-lasieve4I14e benchmark
 

Using the polynomial and parameters in [url]http://www.mersenneforum.org/showpost.php?p=360526&postcount=12[/url]

I get

Avoton C2750 (2.4GHz)
total yield: 1981, q=50001037 (0.14874 sec/rel)

(I suspect this is because the machine has 24kb L1 cache)

Running eight jobs on the eight cores gives me runtimes around 0.1575 sec/rel

msieve benchmark
 

This is running msieve-1.52 on the msieve.dat from [url]http://www.chiark.greenend.org.uk/~twomack/i7test.zip[/url]

[code]
Fri Dec 13 23:25:59 2013 matrix is 1067896 x 1068126 (312.4 MB) with weight 82099651 (76.86/col)
Fri Dec 13 23:25:59 2013 sparse part has weight 71206234 (66.66/col)
Fri Dec 13 23:25:59 2013 using block size 8192 and superblock size 98304 for processor cache size 1024 kB
Fri Dec 13 23:26:08 2013 commencing Lanczos iteration (8 threads)
Fri Dec 13 23:26:08 2013 memory use: 244.1 MB
Fri Dec 13 23:26:13 2013 linear algebra at 0.1%, ETA 0h58m
Fri Dec 13 23:26:14 2013 checkpointing every 1220000 dimensions
Sat Dec 14 00:20:07 2013 lanczos halted after 16889 iterations (dim = 1067894)
Sat Dec 14 00:20:09 2013 recovered 29 nontrivial dependencies
Sat Dec 14 00:20:09 2013 BLanczosTime: 3461
[/code]
[code]
Sat Dec 14 00:45:29 2013 matrix is 1067896 x 1068126 (312.4 MB) with weight 82099651 (76.86/col)
Sat Dec 14 00:45:29 2013 sparse part has weight 71206234 (66.66/col)
Sat Dec 14 00:45:29 2013 using block size 8192 and superblock size 98304 for processor cache size 1024 kB
Sat Dec 14 00:45:38 2013 commencing Lanczos iteration
Sat Dec 14 00:45:38 2013 memory use: 243.9 MB
Sat Dec 14 00:45:59 2013 linear algebra at 0.1%, ETA 3h53m
Sat Dec 14 00:46:06 2013 checkpointing every 280000 dimensions
Sat Dec 14 04:46:35 2013 lanczos halted after 16891 iterations (dim = 1067894)
Sat Dec 14 04:46:37 2013 recovered 29 nontrivial dependencies
Sat Dec 14 04:46:37 2013 BLanczosTime: 14679

[/code]

For comparison, BLanczosTime is 1219 for 4 threads on i7/4770 and 3436 for one.

So in most senses this is actually a rather slow computer.

I wonder how it would compare in power consumption and throughput to our first box with a 486 DX-33. Sure has been a lot progress in a bit over 20 years.

I just recently bought an Intel NUC D54250WYKH and installed 8 GB of 1600 Mhz CL11 RAM. I wanted something small. portable, quiet and efficient and this suited my wants (I really didn't need it yet as my old workhorse, an Intel Core2 Duo E6300 @ 1.86 GHz built back in Q4 of 2006 still works relatively well with some hiccups, however consumes on average 90Wh/h, running mprime this jumps to 150Wh/h).

So far less than 24 hours of work has completed at least 18 trial factors in the M35732xxxx, 2^65 to 2^66 range (its capacity is greater as I had not all four cores running, for 12 of these hours, only the one was operating, quietly). All four are running now and if I mount this on the wall, it is virtually inaudible and warm to the touch.

If anyone has any suggestions as to best measuring the output performance of mprime on this machine, I'd be glad to run it (I'll even hook up my Kill-A-Watt equivalent device to measure energy consumption).

My Gigabyte Brix (GB-BXi5-4200) showed up from Amazon today. Tomorrow I will build it and attach it to the TV. then I will run Prime95 on it :)