George's dream build

[Prime95]

Originally Posted by bgbeuning

Results for HP 8300 (Small Form Factor) with DDR3-1600, cost $200 on ebay, power usage = 80W (Measured by Kill-A-Watt)

TPD for used HP 8300 (Small Form Factor) with DDR3-1600, $200 on ebay, power usage = 80W
3 year TPD = 1 CPUs * 129.8 * 3 years / (200+80*3) = 0.885
4 year TPD = 1 CPUs * 129.8 * 4 years / (200+80*4) = 0.9985

A good purchase!

[axn]

Originally Posted by Prime95

TPD for used HP 8300 (Small Form Factor) with DDR3-1600, $200 on ebay, power usage = 80W
3 year TPD = 1 CPUs * 129.8 * 3 years / (200+80*3) = 0.885
4 year TPD = 1 CPUs * 129.8 * 4 years / (200+80*4) = 0.9985

A good purchase!

I would like to point out that i5-3570 has a TDP of 77W, but an observed system power draw of only 80W. So, perhaps your calculation of 500W for your hypothetical system is too conservative? It could be closer to 350W rather than 500W?! How would that affect your calculations?

[Prime95]

Is it time to retire my 5 year old 4GHz DDR3-1866 i5-2500K Sandy Bridge? Let's do the calculation. 4 worker throughput is 137.0. CPU is burning 100W, I'll assume 120 watts at the wall, though I need to measure this with a kill-a-watt.

TPD for existing Sandy Bridge system:
3 year TPD = 1 CPUs * 137.0 * 3 years / (0+120*3) = 1.142
4 year TPD = 1 CPUs * 137.0 * 4 years / (0+120*4) = 1.142

Not time for retirement! 5 years old and still more productive than a new build.

Next up for analysis is the SandyBridge video card. Does anyone have info on the power draw of a GTX-580 running CudaLucas and how many iterations per second one gets running a 4M FFT?

[Dubslow]

Thank goodness. I've only got 2 Sandy Bridges and nowhere near the sort of money to replace them...

[Madpoo]

Not that anyone asked, but I was curious what the 4M FFT benchmark looked like for my dual 14-core system:

Compare your results to other computers at http://www.mersenne.org/report_benchmarks
Intel(R) Xeon(R) CPU E5-2697 v3 @ 2.60GHz
CPU speed: 2597.05 MHz, 28 hyperthreaded cores
CPU features: Prefetch, SSE, SSE2, SSE4, AVX, AVX2, FMA
L1 cache size: 32 KB
L2 cache size: 256 KB, L3 cache size: 35 MB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
TLBS: 64
Prime95 64-bit version 28.7, RdtscTiming=1
Timings for 4096K FFT length (1 cpu, 1 worker): 22.60 ms.  Throughput: 44.25 iter/sec.
Timings for 4096K FFT length (2 cpus, 1 worker): 11.99 ms.  Throughput: 83.41 iter/sec.
Timings for 4096K FFT length (2 cpus, 2 workers): 24.89, 24.15 ms.  Throughput: 81.59 iter/sec.
Timings for 4096K FFT length (3 cpus, 1 worker):  7.92 ms.  Throughput: 126.34 iter/sec.
Timings for 4096K FFT length (3 cpus, 3 workers): 25.41, 25.06, 24.80 ms.  Throughput: 119.58 iter/sec.
Timings for 4096K FFT length (4 cpus, 1 worker):  6.00 ms.  Throughput: 166.66 iter/sec.
Timings for 4096K FFT length (4 cpus, 2 workers): 12.72, 12.72 ms.  Throughput: 157.27 iter/sec.
Timings for 4096K FFT length (4 cpus, 4 workers): 25.83, 24.89, 24.99, 24.72 ms.  Throughput: 159.35 iter/sec.
Timings for 4096K FFT length (5 cpus, 1 worker):  4.78 ms.  Throughput: 209.06 iter/sec.
Timings for 4096K FFT length (5 cpus, 5 workers): 25.47, 25.11, 25.12, 25.02, 24.79 ms.  Throughput: 199.20 iter/sec.
Timings for 4096K FFT length (6 cpus, 1 worker):  4.00 ms.  Throughput: 249.86 iter/sec.
Timings for 4096K FFT length (6 cpus, 2 workers):  8.77,  8.49 ms.  Throughput: 231.92 iter/sec.
Timings for 4096K FFT length (6 cpus, 3 workers): 13.18, 12.84, 12.85 ms.  Throughput: 231.55 iter/sec.
Timings for 4096K FFT length (6 cpus, 6 workers): 25.96, 25.26, 24.85, 25.16, 25.19, 25.19 ms.  Throughput: 237.49 iter/sec.
Timings for 4096K FFT length (7 cpus, 1 worker):  3.47 ms.  Throughput: 288.18 iter/sec.
Timings for 4096K FFT length (7 cpus, 7 workers): 26.36, 25.64, 25.36, 25.22, 25.66, 25.62, 25.78 ms.  Throughput: 272.80 iter/sec.
Timings for 4096K FFT length (8 cpus, 1 worker):  3.17 ms.  Throughput: 315.68 iter/sec.
Timings for 4096K FFT length (8 cpus, 2 workers):  6.61,  7.09 ms.  Throughput: 292.43 iter/sec.
Timings for 4096K FFT length (8 cpus, 4 workers): 13.65, 13.21, 13.19, 14.19 ms.  Throughput: 295.22 iter/sec.
Timings for 4096K FFT length (8 cpus, 8 workers): 26.69, 25.75, 25.50, 25.35, 25.31, 25.83, 25.78, 22.98 ms.  Throughput: 315.49 iter/sec.
Timings for 4096K FFT length (9 cpus, 1 worker):  2.86 ms.  Throughput: 349.79 iter/sec.
Timings for 4096K FFT length (9 cpus, 3 workers):  9.02,  8.74,  9.35 ms.  Throughput: 332.19 iter/sec.
Timings for 4096K FFT length (9 cpus, 9 workers): 25.78, 25.58, 25.39, 25.58, 25.46, 25.55, 26.17, 24.93, 24.69 ms.  Throughput: 353.61 iter/sec.
Timings for 4096K FFT length (10 cpus, 1 worker):  2.60 ms.  Throughput: 385.23 iter/sec.
Timings for 4096K FFT length (10 cpus, 2 workers):  5.28,  5.47 ms.  Throughput: 371.96 iter/sec.
Timings for 4096K FFT length (10 cpus, 5 workers): 13.75, 13.11, 13.47, 15.71, 12.58 ms.  Throughput: 366.33 iter/sec.
Timings for 4096K FFT length (10 cpus, 10 workers): 26.44, 26.19, 25.21, 25.30, 25.21, 25.48, 26.13, 25.32, 24.91, 24.86 ms.  Throughput: 392.25 iter/sec.
Timings for 4096K FFT length (11 cpus, 1 worker):  2.36 ms.  Throughput: 422.88 iter/sec.
Timings for 4096K FFT length (11 cpus, 11 workers): 26.54, 25.63, 25.56, 25.39, 25.51, 25.42, 25.97, 25.47, 25.05, 24.83, 25.12 ms.  Throughput: 431.51 iter/sec.
Timings for 4096K FFT length (12 cpus, 1 worker):  2.19 ms.  Throughput: 456.84 iter/sec.
Timings for 4096K FFT length (12 cpus, 2 workers):  4.27,  4.49 ms.  Throughput: 456.81 iter/sec.
Timings for 4096K FFT length (12 cpus, 3 workers):  6.75,  7.18,  6.35 ms.  Throughput: 444.73 iter/sec.
Timings for 4096K FFT length (12 cpus, 4 workers):  9.26,  8.89, 10.46,  8.36 ms.  Throughput: 435.81 iter/sec.
Timings for 4096K FFT length (12 cpus, 6 workers): 13.68, 13.26, 13.55, 15.76, 12.91, 12.68 ms.  Throughput: 442.02 iter/sec.
Timings for 4096K FFT length (12 cpus, 12 workers): 26.47, 25.47, 25.47, 25.33, 25.43, 25.65, 25.98, 25.26, 25.08, 25.06, 24.78, 24.96 ms.  Throughput: 472.37 iter/sec.
Timings for 4096K FFT length (13 cpus, 1 worker):  2.04 ms.  Throughput: 489.00 iter/sec.
Timings for 4096K FFT length (13 cpus, 13 workers): 26.39, 25.30, 25.47, 25.59, 25.48, 25.87, 26.19, 25.43, 25.64, 25.31, 25.39, 25.10, 25.11 ms.  Throughput: 508.72 iter/sec.
Timings for 4096K FFT length (14 cpus, 1 worker):  1.92 ms.  Throughput: 520.82 iter/sec.
Timings for 4096K FFT length (14 cpus, 2 workers):  3.51,  3.52 ms.  Throughput: 568.75 iter/sec.
Timings for 4096K FFT length (14 cpus, 7 workers): 13.81, 13.42, 13.20, 15.70, 13.03, 13.07, 13.43 ms.  Throughput: 514.12 iter/sec.
Timings for 4096K FFT length (14 cpus, 14 workers): 26.49, 25.77, 25.59, 25.21, 25.45, 25.79, 26.32, 26.13, 25.90, 25.69, 25.74, 25.54, 25.61, 26.39 ms.  Throughput: 542.11 iter/sec.
Timings for 4096K FFT length (15 cpus, 1 worker):  1.91 ms.  Throughput: 523.35 iter/sec.
Timings for 4096K FFT length (15 cpus, 3 workers):  5.49,  6.37,  5.51 ms.  Throughput: 520.28 iter/sec.
Timings for 4096K FFT length (15 cpus, 5 workers):  9.13,  9.25, 10.80,  8.71,  9.54 ms.  Throughput: 529.76 iter/sec.
Timings for 4096K FFT length (15 cpus, 15 workers): 28.14, 26.91, 26.90, 26.75, 26.59, 26.83, 27.32, 26.27, 26.20, 26.09, 25.74, 25.74, 25.85, 25.97, 27.99 ms.  Throughput: 563.90 iter/sec.
Timings for 4096K FFT length (16 cpus, 1 worker):  1.77 ms.  Throughput: 564.87 iter/sec.
Timings for 4096K FFT length (16 cpus, 2 workers):  3.35,  3.33 ms.  Throughput: 597.96 iter/sec.
Timings for 4096K FFT length (16 cpus, 4 workers):  6.87,  7.52,  6.55,  7.57 ms.  Throughput: 563.35 iter/sec.
Timings for 4096K FFT length (16 cpus, 8 workers): 14.90, 14.91, 14.90, 17.12, 13.08, 13.21, 13.39, 15.09 ms.  Throughput: 552.79 iter/sec.
Timings for 4096K FFT length (16 cpus, 16 workers): 29.35, 28.41, 28.12, 28.34, 28.07, 28.70, 28.69, 26.36, 25.93, 25.69, 25.77, 25.94, 26.23, 26.26, 29.26, 28.94 ms.  Throughput: 583.16 iter/sec.
Timings for 4096K FFT length (17 cpus, 1 worker):  1.70 ms.  Throughput: 588.48 iter/sec.
Timings for 4096K FFT length (17 cpus, 17 workers): 31.10, 35.34, 29.88, 29.00, 29.12, 29.68, 30.34, 26.13, 25.94, 25.86, 25.91, 25.81, 25.85, 26.20, 30.20, 35.12, 29.80 ms.  Throughput: 594.23 iter/sec.
Timings for 4096K FFT length (18 cpus, 1 worker):  1.66 ms.  Throughput: 601.29 iter/sec.
Timings for 4096K FFT length (18 cpus, 2 workers):  3.24,  3.20 ms.  Throughput: 620.79 iter/sec.
Timings for 4096K FFT length (18 cpus, 3 workers):  5.06,  5.18,  5.30 ms.  Throughput: 579.22 iter/sec.
Timings for 4096K FFT length (18 cpus, 6 workers): 10.98, 10.98, 12.47,  8.67,  9.89, 11.01 ms.  Throughput: 569.64 iter/sec.
Timings for 4096K FFT length (18 cpus, 9 workers): 17.08, 16.64, 16.56, 18.76, 13.12, 13.08, 13.34, 17.24, 16.61 ms.  Throughput: 578.22 iter/sec.
Timings for 4096K FFT length (18 cpus, 18 workers): 32.91, 31.45, 32.12, 32.43, 31.58, 32.10, 32.03, 26.36, 26.35, 25.95, 25.81, 25.56, 26.02, 26.43, 33.22, 32.28, 31.65, 32.08 ms.  Throughput: 610.60 iter/sec.
Timings for 4096K FFT length (19 cpus, 1 worker):  1.64 ms.  Throughput: 608.80 iter/sec.
Timings for 4096K FFT length (19 cpus, 19 workers): 35.27, 34.10, 34.86, 34.26, 33.87, 33.91, 34.82, 26.53, 26.24, 25.98, 26.06, 25.92, 25.78, 26.42, 35.15, 34.37, 34.51, 33.85, 33.69 ms.  Throughput: 616.93 iter/sec.
Timings for 4096K FFT length (20 cpus, 1 worker):  2.70 ms.  Throughput: 369.85 iter/sec.
Timings for 4096K FFT length (20 cpus, 2 workers):  3.11,  3.15 ms.  Throughput: 638.71 iter/sec.
Timings for 4096K FFT length (20 cpus, 4 workers):  7.67,  8.02,  5.70,  7.62 ms.  Throughput: 561.76 iter/sec.
Timings for 4096K FFT length (20 cpus, 5 workers):  8.47,  9.02,  6.58,  8.05,  8.61 ms.  Throughput: 621.33 iter/sec.
Timings for 4096K FFT length (20 cpus, 10 workers): 19.41, 19.26, 19.44, 20.68, 13.10, 13.11, 13.10, 19.44, 19.30, 19.07 ms.  Throughput: 588.00 iter/sec.
Timings for 4096K FFT length (20 cpus, 20 workers): 35.57, 35.92, 36.41, 37.04, 37.37, 41.83, 35.76, 26.69, 26.38, 26.08, 25.99, 25.89, 26.15, 26.45, 36.66, 35.75, 36.45, 37.40, 37.11, 41.55 ms.  Throughput: 616.34 iter/sec.
Timings for 4096K FFT length (21 cpus, 1 worker):  3.04 ms.  Throughput: 329.47 iter/sec.
Timings for 4096K FFT length (21 cpus, 3 workers):  5.17,  3.58,  5.36 ms.  Throughput: 659.34 iter/sec.
Timings for 4096K FFT length (21 cpus, 7 workers): 13.75, 13.40, 13.92,  8.80, 10.47, 13.99, 13.62 ms.  Throughput: 573.26 iter/sec.
Timings for 4096K FFT length (21 cpus, 21 workers): 39.68, 39.84, 38.91, 38.48, 38.48, 39.27, 40.40, 26.49, 26.33, 26.06, 25.84, 25.79, 25.93, 26.53, 39.87, 39.06, 38.90, 38.74, 39.34, 39.04, 39.94 ms.  Throughput: 624.30 iter/sec.
Timings for 4096K FFT length (22 cpus, 1 worker):  2.81 ms.  Throughput: 355.48 iter/sec.
Timings for 4096K FFT length (22 cpus, 2 workers):  2.99,  3.06 ms.  Throughput: 661.55 iter/sec.
Timings for 4096K FFT length (22 cpus, 11 workers): 22.31, 21.85, 21.50, 23.33, 13.64, 13.40, 13.76, 22.52, 22.23, 22.40, 22.90 ms.  Throughput: 578.26 iter/sec.
Timings for 4096K FFT length (22 cpus, 22 workers): 38.55, 38.71, 39.38, 38.64, 38.87, 38.85, 40.93, 27.82, 27.03, 27.19, 26.88, 27.18, 27.08, 27.66, 40.32, 38.58, 39.66, 38.65, 38.69, 39.75, 40.15, 27.58 ms.  Throughput: 649.71 iter/sec.
Timings for 4096K FFT length (23 cpus, 1 worker):  2.48 ms.  Throughput: 402.87 iter/sec.
Timings for 4096K FFT length (23 cpus, 23 workers): 41.37, 39.47, 38.48, 38.31, 39.56, 39.80, 40.92, 29.10, 29.17, 28.49, 28.17, 28.40, 28.30, 29.34, 38.71, 38.52, 38.43, 38.69, 40.39, 38.63, 40.04, 29.23, 29.13 ms.  Throughput: 668.16 iter/sec.
Timings for 4096K FFT length (24 cpus, 1 worker):  1.80 ms.  Throughput: 556.65 iter/sec.
Timings for 4096K FFT length (24 cpus, 2 workers):  2.83,  3.00 ms.  Throughput: 687.32 iter/sec.
Timings for 4096K FFT length (24 cpus, 3 workers):  4.93,  3.93,  5.06 ms.  Throughput: 654.84 iter/sec.
Timings for 4096K FFT length (24 cpus, 4 workers):  6.59,  8.22,  6.75,  8.15 ms.  Throughput: 544.36 iter/sec.
Timings for 4096K FFT length (24 cpus, 6 workers): 10.57, 11.10,  7.36,  9.25, 10.26, 11.84 ms.  Throughput: 610.54 iter/sec.
Timings for 4096K FFT length (24 cpus, 8 workers): 13.64, 13.42, 14.45,  9.72, 11.42, 13.73, 13.96,  9.77 ms.  Throughput: 654.22 iter/sec.
Timings for 4096K FFT length (24 cpus, 12 workers): 22.21, 21.86, 22.05, 23.45, 15.32, 14.14, 14.49, 22.15, 21.96, 21.95, 23.11, 15.32 ms.  Throughput: 628.56 iter/sec.
Timings for 4096K FFT length (24 cpus, 24 workers): 38.18, 37.72, 36.90, 43.38, 37.45, 37.67, 46.32, 30.59, 33.50, 30.88, 29.61, 29.48, 29.66, 30.10, 38.96, 36.91, 36.71, 43.57, 37.86, 37.96, 46.17, 30.47, 32.89, 30.87 ms.  Throughput: 680.25 iter/sec.
Timings for 4096K FFT length (25 cpus, 1 worker):  1.96 ms.  Throughput: 511.40 iter/sec.
Timings for 4096K FFT length (25 cpus, 5 workers):  9.00, 12.37,  8.85,  9.10, 12.73 ms.  Throughput: 493.40 iter/sec.
Timings for 4096K FFT length (25 cpus, 25 workers): 40.57, 39.13, 39.14, 39.35, 38.96, 38.75, 40.64, 32.96, 32.42, 32.35, 32.23, 32.35, 32.05, 33.25, 38.52, 38.85, 39.09, 38.64, 38.88, 39.58, 40.03, 32.32, 32.05, 32.11, 32.03 ms.  Throughput: 696.21 iter/sec.
Timings for 4096K FFT length (26 cpus, 1 worker):  2.10 ms.  Throughput: 475.09 iter/sec.
Timings for 4096K FFT length (26 cpus, 2 workers):  2.85,  2.79 ms.  Throughput: 709.78 iter/sec.
Timings for 4096K FFT length (26 cpus, 13 workers): 22.08, 22.61, 22.06, 23.76, 15.22, 19.24, 15.51, 22.22, 22.35, 21.82, 23.43, 15.92, 18.92 ms.  Throughput: 653.04 iter/sec.
Timings for 4096K FFT length (26 cpus, 26 workers): 38.66, 37.85, 37.91, 37.47, 45.97, 39.17, 39.68, 33.77, 34.98, 36.74, 35.17, 34.92, 32.82, 34.01, 37.78, 37.94, 37.96, 37.09, 45.82, 39.89, 39.79, 34.03, 34.93, 36.66, 35.09, 34.72 ms.  Throughput: 700.94 iter/sec.
Timings for 4096K FFT length (27 cpus, 1 worker):  2.03 ms.  Throughput: 493.30 iter/sec.
Timings for 4096K FFT length (27 cpus, 3 workers):  4.67,  4.13,  5.33 ms.  Throughput: 643.58 iter/sec.
Timings for 4096K FFT length (27 cpus, 9 workers): 14.01, 13.18, 15.42, 11.21, 14.05, 13.63, 13.64, 11.42, 12.29 ms.  Throughput: 688.11 iter/sec.
Timings for 4096K FFT length (27 cpus, 27 workers): 40.91, 38.99, 38.90, 38.74, 38.42, 38.46, 40.73, 36.61, 36.16, 35.95, 35.74, 36.73, 40.22, 36.74, 39.01, 38.98, 38.93, 38.61, 39.84, 39.87, 40.50, 37.27, 35.70, 36.07, 35.82, 35.94, 39.63 ms.  Throughput: 709.54 iter/sec.
Timings for 4096K FFT length (28 cpus, 1 worker):  2.12 ms.  Throughput: 472.59 iter/sec.
Timings for 4096K FFT length (28 cpus, 2 workers):  2.99,  3.18 ms.  Throughput: 648.87 iter/sec.
Timings for 4096K FFT length (28 cpus, 4 workers):  5.37,  5.22,  5.27,  5.71 ms.  Throughput: 742.72 iter/sec.
Timings for 4096K FFT length (28 cpus, 7 workers): 10.54, 11.42,  9.46, 10.66, 10.37, 12.17,  9.44 ms.  Throughput: 666.42 iter/sec.
Timings for 4096K FFT length (28 cpus, 14 workers): 22.39, 22.05, 21.50, 24.47, 20.73, 16.83, 17.87, 21.88, 21.96, 21.37, 23.66, 20.31, 16.97, 17.49 ms.  Throughput: 686.66 iter/sec.
Timings for 4096K FFT length (28 cpus, 28 workers): 39.56, 40.26, 37.10, 38.40, 37.64, 47.57, 39.28, 37.62, 36.20, 36.27, 40.23, 36.54, 36.46, 63.56, 37.85, 40.32, 37.55, 38.02, 37.76, 46.63, 39.10, 37.57, 36.56, 36.58, 40.63, 36.55, 36.98, 63.16 ms.  Throughput: 707.14 iter/sec.

[bgbeuning]

Originally Posted by Madpoo

Not that anyone asked, but I was curious what the 4M FFT benchmark looked like for my dual 14-core system

How much did it cost and how much power does it draw?

[Madpoo]

Originally Posted by bgbeuning

How much did it cost and how much power does it draw?

Well... It's a pretty expensive HP Proliant DL380 Gen9 with dual Xeon E5-2697 v3 and DDR4 memory.

Average usage is 520-600 W, but this system also has a good amount of drives in there so it's not all CPU.

So, yeah, it's not practical for someone looking to pick one up solely for prime hunting. Interesting reference point though.

[ATH]

My 8 cores (5960X) reached a maximum throughput of 359 iter/sec vs your max of 742 iter/sec for 28 cores.

[Madpoo]

Originally Posted by ATH

My 8 cores (5960X) reached a maximum throughput of 359 iter/sec vs your max of 742 iter/sec for 28 cores.

The faster clock rate on the 5960X must help.

That's one downside of so many cores is that usually (to keep the TDP in check) they're clocked slower as more cores are added.

Works fine for my needs (virtual hosts with good parallel workloads). For some things that don't thread well (or at all) higher clocks and fewer cores is probably going to get better results.

[Cruelty]

Out of curiosity my home-built Xeon E5 v3 @ 2.5GHz 10c/20t (105W) + 4x4GB 2133 DDR4 SR ECC-R. Below excerpt from FullBench:

Compare your results to other computers at http://www.mersenne.org/report_benchmarks
Genuine Intel(R) CPU @ 2.20GHz
CPU speed: 2325.88 MHz, 10 hyperthreaded cores
CPU features: Prefetch, SSE, SSE2, SSE4, AVX, AVX2, FMA
L1 cache size: 32 KB
L2 cache size: 256 KB, L3 cache size: 25 MB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
TLBS: 64
Prime95 64-bit version 28.7, RdtscTiming=1
Timings for 4096K FFT length (1 cpu, 1 worker): 25.81 ms.  Throughput: 38.75 iter/sec.
Timings for 4096K FFT length (2 cpus, 2 workers): 27.71, 29.27 ms.  Throughput: 70.25 iter/sec.
Timings for 4096K FFT length (3 cpus, 3 workers): 28.33, 28.43, 28.43 ms.  Throughput: 105.64 iter/sec.
Timings for 4096K FFT length (4 cpus, 4 workers): 28.01, 27.97, 27.60, 27.31 ms.  Throughput: 144.30 iter/sec.
Timings for 4096K FFT length (5 cpus, 5 workers): 28.69, 29.04, 28.07, 28.28, 28.08 ms.  Throughput: 175.88 iter/sec.
Timings for 4096K FFT length (6 cpus, 6 workers): 30.30, 30.19, 29.20, 29.47, 29.10, 28.99 ms.  Throughput: 203.16 iter/sec.
Timings for 4096K FFT length (7 cpus, 7 workers): 31.55, 31.38, 30.25, 30.27, 30.01, 30.23, 30.24 ms.  Throughput: 229.13 iter/sec.
Timings for 4096K FFT length (8 cpus, 8 workers): 32.04, 31.92, 31.67, 31.96, 31.78, 31.68, 31.81, 31.79 ms.  Throughput: 251.34 iter/sec.
Timings for 4096K FFT length (9 cpus, 9 workers): 33.66, 33.60, 33.34, 33.58, 33.24, 33.28, 33.38, 33.35, 33.50 ms.  Throughput: 269.16 iter/sec.
Timings for 4096K FFT length (10 cpus, 10 workers): 35.26, 35.16, 35.15, 35.57, 34.93, 36.13, 35.20, 35.35, 35.32, 35.24 ms.  Throughput: 283.06 iter/sec.
Timings for 4096K FFT length (1 cpu hyperthreaded, 1 worker): 27.75 ms.  Throughput: 36.04 iter/sec.
Timings for 4096K FFT length (2 cpus hyperthreaded, 2 workers): 29.79, 29.48 ms.  Throughput: 67.50 iter/sec.
Timings for 4096K FFT length (3 cpus hyperthreaded, 3 workers): 30.22, 30.23, 30.61 ms.  Throughput: 98.84 iter/sec.
Timings for 4096K FFT length (4 cpus hyperthreaded, 4 workers): 29.79, 30.04, 29.84, 29.44 ms.  Throughput: 134.34 iter/sec.
Timings for 4096K FFT length (5 cpus hyperthreaded, 5 workers): 30.26, 30.35, 30.26, 30.61, 30.21 ms.  Throughput: 164.81 iter/sec.
Timings for 4096K FFT length (6 cpus hyperthreaded, 6 workers): 31.44, 31.34, 31.89, 31.04, 31.21, 31.43 ms.  Throughput: 191.15 iter/sec.
Timings for 4096K FFT length (7 cpus hyperthreaded, 7 workers): 32.73, 32.57, 32.84, 32.86, 32.52, 32.38, 32.83 ms.  Throughput: 214.22 iter/sec.
Timings for 4096K FFT length (8 cpus hyperthreaded, 8 workers): 34.12, 34.26, 34.30, 34.24, 34.19, 33.58, 34.20, 34.73 ms.  Throughput: 233.91 iter/sec.
Timings for 4096K FFT length (9 cpus hyperthreaded, 9 workers): 35.87, 36.36, 36.18, 35.65, 36.02, 35.44, 36.03, 35.79, 36.23 ms.  Throughput: 250.34 iter/sec.
Timings for 4096K FFT length (10 cpus hyperthreaded, 10 workers): 38.37, 38.45, 38.20, 37.64, 39.19, 37.55, 38.28, 37.55, 37.76, 37.94 ms.  Throughput: 262.55 iter/sec.

[Fred]

Originally Posted by Prime95

Mobo: ASRock Z170M-ITX/ac 5 * 130 = 650
Memory: DDR4-3200 2x4GB 5 * 60 = 300
CPU: I5-6600 (3.3GHz, 65W) 5 * 230 = 1150
SSD: Samsung 850 EVO 1 * 90 = 90
PicoPSU picoPSU-120 4 * 40 = 160

Another sample:

Mobo: H110M-ITX (can't OC RAM) 5 * 70 = 350
Memory: DDR4-2133 2x4GB 5 * 40 = 200
CPU: I5-6400 (2.7GHz, 65W) 5 * 190 = 950
SSD: Samsung 850 EVO 1 * 90 = 90
PicoPSU picoPSU-120 4 * 40 = 160

I'm considering a nearly identical build. My only sticking point left is deciding between a setup with DDR4-2133 and DDR4-3000. The system with DDR4-3000 would cost exactly 15% more (upgraded motherboard and ram). Both systems would have the i5-6600.

My question (which I'm not sure if anyone can answer) is would the difference in ram (2133 to 3000) yield 15% faster processing (for example on LL tests)? If so, then it seems it would be worth the faster ram. If not (for example if it only gave a 5% or 10% increase) then it seems it would be more cost effective to put the extra $ toward my next build.