Hi Navy, how the heck do you keep mem clock from dropping to 500? my 64's are stuck there.
-cvdcc 1000 too low?
I'm having some power supply issues so I've had to stop mining until my replacement PSU came in. Which means I'm not entirely trusting some of my settings and power consumption reading.
However, with the Vega FE public beta these were my settings:
setx GPU_FORCE_64BIT_PTR 0
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100
EthDcrMiner64.exe -epool us2.ethermine.org:4444 -ewal 0xd909bB6EbAbBE96F46B07836267B7700C48894d2.C_VEGA_FE -epsw x -cclock 1137,1137 -mclock 1050,1100 -powlim 50,50 -tt 55,55 -fanmax 97,97 -fanmin 50,50 -cvddc 950,950
When I moved to the blockchain drivers I had to bump -cvddc to 1000,1000. I don't completely trust that anymore as looking back I was getting lots of evidence that my powersupply was struggling. I should have figured that, it being 9 years old and having served in two gaming/mining rigs for a good portion of it's lifespan. For the first two weeks I had the Vega FE's it wasn't possible to pull back the core clock effectively, so power consumption must have been huge but I didn't measure it.
I'll report back when I have the new PSU in place, but I think a new PSU will let me get back to having both cards at -mclock 1100,1100.
I will mention that it can take multiple restarts, and gradually working the cards up to whatever memory overclock they would support. Example, if I changed a driver, upon starting mining mem clock would fall to 167 or 500. So I would close the miner, restart the computer. Then I would change my .bat to -mclock 945, 945. Start the miner. If it went to 945, I would close the miner and try my 1100MHz .bat. Sometimes this would require another restart, other times it didn't. Once I got them to happily stay at 1050,1100 they would endure restarts without issue.
I have a friend who now has 4 vega 64's and they're all over the place in terms of what they'll support for memory overclock. I think the FE's must have got the best binning because both of mine do 1100 without breaking a sweat. Going back to the PSU issue, the longer I ran the cards the more artifacts and occasional hard crashes that would occur. Now that GPUZ properly supports Vega I can see that the -cvddc voltage falls off hard as soon as a real load is put onto the cards which in turn caused the memory frequency to fall off. So I'll have to update you once the new PSU is in.
So you could get 120 MH with 4 1070s for the same power usage.
That's easy to say without considering my situation. My computer is doubling as a gaming/mining rig. I can only run two video cards currently, so I wanted to get the cards that have the highest hashrate per slot. I don't care about power consumption because we have cheap electricity. I did think hard about going 2x Titan XP, or 1070, or 1080 Ti, but I decided to take a gamble and ordered Vega FE in July when they were released. I don't regret it.
