aleqx

I'm using yours, cloned git from yesterday, compiled against cuda 8. As I explained, memory consumption is definitely not the same for the same intensity. Please see my post above....

Yep, it says `GPU #0: TITAN Xp` For intensity 18.2 it says `Adding 52224 threads to intensity 18, 314368 cuda threads` Edit: both 1080Ti and TitanXp report the same number...

I'm also getting about 10% rejected shares with the message "invalid job id" (neoscrypt, on the altminer.net pool) which is way too much ... that pool is reliable and well...

I lowered intensity to 15 for all GPUs and now I get 0 rejected shares ... previously I was getting rejected shares immediately after starting mining. With intensity 15 I...

Did you see my 2nd message above? Lowering the intensity eliminates the rejected shares completely (same pools, same job frequency). This is definitely not a pool or job frequency issue.

> Of course when you are mining with 28% less speed you don't get many stale shares. The _fraction_ of stale shares should be the same. Going from 10% to...

Recompiled and tried again. Many thanks for fixing the intensity issue for TitanXp vs 1080Ti -- they are now more consistent (still not exactly the same memory consumption for the...

Is it expected that memory overclock even on a gtx1070 makes no difference to ccminer for neoscrypt? It doesn't, at all. I thought neoscrypt is a memory intensive algo, isn't...

I was just about to come and post again but your above post may fix it (really hope so). Will pull and retry. Here's what I had discovered on the...

Still getting about 15% lower hashrate reported by the pool after 12h compared to the rate reported by ccminer (10.5 MH/s vs 12.2 MH/s) ... something is just not right....