mersenneforum.org - View Single Post

mackerel · 2019-07-09, 23:23

So much testing... so much more to do.

Summary test results at https://linustechtips.com/main/topic...umber-finding/

Attempted interpretation on how it relates to PrimeGrid work: http://www.primegrid.com/forum_threa...ap=true#130991

4096k FFT iter/s for 1, 2, 6 workers tested with 29.8b5:
8086k 187.59 180.16 170.94
3600 309.71 214.21 193.38

5120k FFT:
8086k 145.74 141.85 135.79
3600 193.86 155.41 149.88

8086k seems to run around 4.3 GHz not that it matters as it is ram bandwidth starved. I had dual channel 3000C15 in the tested system.

The 3600 system wouldn't boot with my 4000 rated ram, even after dropping it down to 3400. I gave up at that point and swapped in 3000C14 modules which were used for test. 3600 clock fluctuated with temperature but was in the region of 3.9 GHz.

2019-07-09, 23:23	#61
mackerel Feb 2016 UK 2³·5·11 Posts	So much testing... so much more to do. Summary test results at https://linustechtips.com/main/topic...umber-finding/ Attempted interpretation on how it relates to PrimeGrid work: http://www.primegrid.com/forum_threa...ap=true#130991 4096k FFT iter/s for 1, 2, 6 workers tested with 29.8b5: 8086k 187.59 180.16 170.94 3600 309.71 214.21 193.38 5120k FFT: 8086k 145.74 141.85 135.79 3600 193.86 155.41 149.88 8086k seems to run around 4.3 GHz not that it matters as it is ram bandwidth starved. I had dual channel 3000C15 in the tested system. The 3600 system wouldn't boot with my 4000 rated ram, even after dropping it down to 3400. I gave up at that point and swapped in 3000C14 modules which were used for test. 3600 clock fluctuated with temperature but was in the region of 3.9 GHz.