Hi, I have 2 Vega 64 Nitros. I tried flashing Vega 56 Pulse BIOS (because of the lower TDP I thought it would be e better idea) - it killed it. The card cannot be seen neither in Safe mode (no win driver) neither in DOS with bootable flash drive with ATI flash on it. Luckily there's a BIOS switch. I downloaded the Nitro+ version for 56 bios that has the same device ID 1002 687F - 1DA2 E37F and 180W TDP (socket and battery) the other ones had 220W as far as I remember. Guess what - I killed the second 64 as well. So I wonder which BIOS did you use to flash your Vega 64 nitro to Vega 56? I read somewhere else that another user killed his gigabyte 64 when attempting to flash 56 BIOS on it, so I guess this is not universal and there is trial and error involved.

On your other question. memory performance is a combination of Mhz and timings. It is really complicated to measure which exact parameter makes what difference. If you look at the timing numbers from your 2 cases - they are VERY different. I assume ethash uses the memory in a very specific way, but some timings compensate for others. So you can get similar results with very different combinations. I believe this could end up being so complicated that you actually need an AI to test and evaluate all the 100000s of combinations.

55 is close to a record high! If you get that stable I guess it makes only sense to look for ways to lower power consumption. Or if you dont care - simply dont mess around with it anymore Having reached the max of a rig is a moment where I'd be pleased to open up a beer or two, sit back and relax 

This website's information is inaccurate. You can check with claymore miner in benchmark mode where will the DAG file generation fail. Mine failed at DAG #176 and I have GTX 1060 3GBs , which will be ruled out sooner than I though.. about 9.6 months from now if I did the math right.

This website shows that 3GB cards will drop at DAG #256 which is not true.

Also it is worth mentioning that DAG allocation is a matter of software configuration. What I mean is that each vendor decides just how much his Video card model will allow its memory to be filled to full capacity. At dag 175 my cards were about 2650 MB used as per MSI afterburner. Which means that in theory they should be able to do DAG 176 as well (very small MB difference maybe 20-30) but they can't, because they have some small reserve that has to be left (I guess) The card that had the monitor on it had 2800ish of its memory used so I guess it is the culprit (if one fails maybe all fail)

A little OFF but couldn't help it: Sorry mate, but those of you in western Europe are pretty behind in terms of electricity bills. Here in Bulgaria the average cost of Kw/h is 0.09 EUR or $0.11 (similar in USA/Canada, just a little bit more than Russia and China), so we have tons of R9 280/290 still hashing away. And they are being sold second hand pretty good.

ON the topic I confirm the observations of everyone:

6 x MSI 1060 3GB (hynix) - 114 - 115 top (19 each) with eth miner. Claymore 9.7 is 110-111 + 1200 pascal (sia wasn't stable for some reason) . I run them at core +100, mem +550 PL 60% , power draw is 500W from wall with GOLD PSU

1 x Palit dual fan 1060 3GB (samsung) - 23-25 top with eth miner (mem + 800, PL 60%, fan at 55%) Don't know the power draw but i guess it's not so good given it's just 1 card - this is my desktop PC in the room where I sleep and I have no problems with noise. I thought I may need to buy accelero cooling for silence but not necessary

It's funny how the cheapest card is actually the fastest

Exactly! Once I installed v12 I got stability problems on all 3 rigs (mixed 280s and 290s) I guess from the heavier Vram utilization, which is clocked to 1500 on all cards (no bios mod, just MSI aft)  So I read here about the li mode, so I switched it on and noticed immediate 20% drop in consumption and same hashrate in miner. Thought hm... too good to be true ?? But ok, let's give it a try. I mined in coinmine.pl pool for 1 night and the reported by pool speed was more than 30% lower over this period. I switched to flypool - the same. Finally removed li mode and speeds in pool improved (and payouts consequently) but are still quite far from speeds in miner. In miner I get 5300 H/s total in the pool is like 4300 - 4700 which is a whopping 17% difference. They say pool luck.. ok, but then I should be getting speeds ABOVE what I see in miner from times to times, right? So if this is due to variance the 24h average should be pretty close to what I see in miner. Why is such a big difference ?

Anyone else have such a big gap in reported speed in miner vs pool? Have you tried several pools and noticed which one is closest to speeds in miner?

Cheers

Two questions from me:

1. I watched this clip https://www.youtube.com/watch?v=Bjs6m35zLUs where the guy says your wallet name has to be 8 characters. Mine is 10... I didn't get any warning or notification when inputting it. Also this is not mentioned in readme / wiki files, so I wonder where he got it from. Now I don't want to close the miner and change as it is already mining for 4h approx. Well, if I close it will it start all over again? From round 1 that is..

2. My second question is about hashrate. I have a mix of cards on 4 machines (one being my personal PC with 2 cards only) On one rig I got 600 M/s from R9 290 with OC, on my PC - 500 M/s from 280x again with OC but then on the 3rd , which is 6 x 280x I get avg 270 M/s per card, which is almost twice lower than my personal PC. Same win (10) , same drivers (16.6) because I was foolish enough to install win 10 which forces driver updates... The rest of components are different of course. I just don't have a clue..  

Thanks in advance