either, or both.

He does.  It forces 64 bit pointers, and requires some variable settings and different dll's.  It ran slower on my cards out of the box than the published kernel.  It might be different on 280x/290/290x, and cards with horsepower to spare, but that's not what I run.

Yes, in-between two of your posts, in response to one of yours.  Thanks for using the "reading" skill (/sarcasm)

https://bitcointalk.org/index.php?topic=206577.msg11883061#msg11883061

You forgot to factor in a server-grade motherboard, and the price of the specialized intel compiler into that.  Lets say the device itself is your only cost.  Even then, you're getting 57 threads at 1.1 GHz with access to 8GB memory at $224 pulling 300 Watts.  Compare to a radeon r7 240 4GB, with 320 threads at 1.1 GHz with access to 4GB memory at $80 pulling 25 Watts.

Xeon Phi's have been available for at least a year, but were price prohibitive, even for the low-end products.  The likelihood that this product will be reasonably price for what it offers in mining is slim-to-none.  Now, if you get one anyway, let us know and we'll try to help you figure out how to mine on it.

I'm not sure why you would even think this a valid strategy.  Starting over on N IS making a new coin. 

We don't jump to NF=18 for a whole year, so don't start stocking your survival shelter yet.  By the time it arrives, YAC will have been on NF=17 for nearly 1/3 of its existence and it will just be "normal".  Is NF=17 hard on our GPU's? It is.  Is it any different from any NFactor change in the past?  No.  Every change in NFactor is met with confusion, frustration, and HW errors.  This difference is that we have years of understanding now instead of days, or weeks, or months. 

We've known for a while that CPU would become competitive to GPUs again at this NFactor, we just need to figure out how to keep the hashrate spread out so that there's no 51% network ownership.  We had this situation for a long, lone time with coinmine.pl having >51%, but everyone was okay with it because that's where most of the miners were.  During that period, solo mining was difficult because the pool would get most of the blocks and people not on the pool would see orphans.  Now, we have someone who has quite the mining capacity, but he's solo-mining, but only during specified hours, leaving the pools with orphans.  It's quite possible that this is not even malicious intent (other than likely using computers that aren't his or hers), but just that much hashpower that it dominates the network. 

The new AMD GPU's are coming with 8 GB of memory on the 390's.  The Fury's are only 4GB, but it's newer fabbed stuff called HBM (High bandwidth memory) and it has substantially more memory bandwidth of GDDR5 (I think it's double-ish).  Should see 8GB cards with the stuff in the future as AMD is really pushing 4K gaming, which necessitates more memory.  Don't despair about N - there's always the cpu miner

Actually, many pools have a cap on the amount you can have, and will actually suspend payments until you withdraw if you hit that cap.  I am struggling to believe that's not set on that pool.

buffer size too high for system ram
not enough swap file
could be many things

And what command line did the config generator I recommended to you generate?

something like...

yacminer --scrypt-chacha --worksize 128 -g 1 --lookup-gap 8 --buffer-size 3532 -R 1664

Your R is still WAY too high, that's what causes HW errors

probably lookup gap too

Post the new command line you're using and what you've tried changing

If what I provided doesn't help, and you're using YACMiner 3.5.0, then you're beyond the help I'm willing to provide for free (which is pretty far)

Check my reply to your post in the other thread you posted this exact same problem on.

Also, make sure you're using YACMiner 3.5.0 and not 3.3.1 - they are vastly different!  You can find binaries and source at the github link in my signature

Well, I'm guessing you're mining zccoin, which is NF 16 - I'm guessing you've never mined a scrypt-chacha coin before?

Your buffer size is too low
Your lookup gap is too low
Your raw intensity is too high

read the primer on how to tune your miner for SC mining here : https://github.com/Thirtybird/YACMiner/blob/master/SCRYPT-README

there is also a config generator for ultracoin, which will work for your coin here : http://cfg.ultracoin.net/
it doesn't guarantee it will generate the fastest hashrate for your card, but it will be in the ballpark.

In theory, that should suffice. 

You're kind of dreaming... Someone else may come in and fix/implement all the stuff, but this coin is going to be seriously fubar unless someone takes a serious look at fixing it before block 100,000....

Well, at this point, I think we're at the tipping point between CPUs and GPUs.  4GB R7 240 is only marginally faster than my laptop i7 3610.  With that said though, the GPUs still pull less power and I can have 4-6 of them in one machine vs. 1 CPU.  CPU mining is a real possibility again if that's all you've got. 

I wouldn't recommend older CPUs as they are lacking in efficiency so they will be much slower while still pulling more power (my laptop i7 3610 is 157% faster then my desktop i5 2500k), but the miner requires almost no configuration to use, so will be an easy jump-on point for inexperienced or non-technical people.

Not yet - check back in about 9.5 hours (6:54 PM EST, 10:54 PM UTC)

If it appears to be a botnet, block his miners, or disable his account, or report it to some authority?  Just the first couple things that come to my mind.