Vertminer causes alot of HW errors on my GPU with I>12, at I=18 it's 1-5 HW per second and I>18 crashes drivers.... With normal scrypt CGminer i can mine at I=20 and it's stable with HW=0
What are you mining with?
I'm running 5x 290 non x, 3 hynix and 2 elpdia.
The hynix i've had stable for a couple days, no HW error, at i20 and are hashing at 473k# happily.
The elpdia I've had stable for a day, no HW error at I19 and is hashing at 423k#.
All 5 cards use less power than mining regular scrypt. Decreasing TC has a major impact on stability here I've seen.
Edit: Looked at your prev. posts and see 7790. This unit only has 2Gb of ram. With adaptive n-factor, you're asking the card to use way more memory than you have available for it and this causes it to die on you. This is part of how it is asic resistant. . by requiring more memory allocation. The fact that a 2Gb card is limited so soon is great news actually. How much ram do you have on your MOBO?
Comparison for you: Regular Scrypt n factor is 2^10 = 1024. This coin, starting at 2^11 = 2048. This basically means that rather than allocating 128 * r * N(2^10) = 131072 bytes in Scrypt, you're beginning with a memory allocation of 128 * r * N(11) = 262144 bytes. So if you were almost fully utilizing your card before at anywhere from I15-20, then you will now need to be < I14 to be fully using your card. The 290 non x's I've had running up to I23 for at least an hour or two stable before i got tired of it in sgminer. Anything past that, my card is overutilized and starts producing HW errors.
So when you have to lower the intensity to get your card to work. . you don't need to worry that you cannot use I20 anymore. Worry more that your card is either over/underutilized. It seems your card is being fully utilized at I12.
Someone please correct me if I'm wrong.