Quote from: jabbaah on December 03, 2013, 03:02:05 pm

For me it seems that the reject rate it going up for miners which do not have that my cores.

I am currently using 2x 32 core AWS spot instances for testing throughput, an 8 core i7
and a couple of 3 core virtual machines.

The reject rate for the 32 core monsters is below 1%
Reject rate for the 8 core is below 2%

And the reject rate for the 3 core things is at about 15% !!!
(I have 11 of those, and ALL of them have high reject rates)

I think I can see some pattern in this...

Unfortunately I don't know enough of the underlying work distribution software,
but maybe I am not the only one experiencing this situation and it would be
great if there might be some fine-tuning possible?

Who has the same experience?

I'm having the same observation mate. More cores - less rejections.

I would be more interested in the number of rejections vs the percentage. Presumably, a larger number of cores can process a larger number of shares, so if it's a fixed amount of rejects no matter the core number, it would appear to be a smaller percentage with more cores.

Thanks that helped, it is running faster!

You may also want to try it with 512MB per core. Depending on your exact configuration, you may see better performance.

ptsminer_avx_intel.exe PTS_address_here  4 27 avx
                                                                      |   |   |
                                                         # of cores |   |
                                   use 512MB of RAM/thread   |
                                                                   use AVX

@xolokram

Thanks ! Rough test showed an improved sh/m and c/m with 512mb.

512mb : 115 c/m + 5 sh/m
1024 mb: 113 c/m + 3.5 sh/m
4096 mb: 95.3 c/m + 3.2 sh/m

N00b question here, which is the most important, c/m or sh/m ?

I'm now testing how many thread I can put on without  slowing down my alienware too much ... 12 thread at 512mb seem to do it ...

Let see if my laptop catches fire during the night ...

If you're running 12 threads, then that means you probably have either a Sandy Bridge-E or Ivy Bridge-E i7 in your laptop. Those c/m numbers seem a little low for 12 threads. Are you using the AVX binaries? That chip will support it. Also, I find it helpful to set the process priority to low, so it doesn't interfere with whatever you're doing in the foreground. Also like Xolo said, make sure you have good ventilation for your laptop, as we're not responsible for frying hardware.

I'm noticing an increase from ~80 c/m to ~100 c/m on my 4 core Xeon E31220. I'm also experiencing the crashing problem on my i7 920. I'm wondering if it's cause it doesn't have AVX, but does have SSE4 and it's trying to default to AVX?