are the straps in the OP still the best for samsung memory?
If not what are the best straps now? I don’t really understand them so I just blindly copy/paste.
My SSD died and I am having to re-do all my tuning. I was getting a solid 196MH reliable before it died. Now with I think the same settings I am only seeing 183mh at the same wattage
I am using windows at the moment as I am dual mining Chia on this system. Are there any driver changes I need to make? I seem to remember chasing some driver issues during my tuning the first time around but forgot what they were.
hey
I couldn´t hit that 1404 spot for Strix (gpu3), at one point i had it @1405 but to keep gpu up, it required higher vdd…so my as allway, each gpu is different.
Ended up on these for now, did not feel comfortable to push any above “160W”… ill try to play later with those few that are under 190mh
Other Hynix (gpu6) can stay as is, couldnt find now any combo that would yield signigally better result.
hi.
Starps looks ok if gpu is stable with those, that “missing” 13mh must have some other explanation.
Dont know about Hive in Win, related to mining, usually gpu BIOS is related to gpu performance but if you have flashed your bios once, that is not required to repeat…cant really say anything about any drivers.
If your gpu is running fine with those oc´s, vdd 950 should handle even higher core or you should be able to go down with vdd for lower wattage. Hope u find your missing hash
Well, every Vega is different so a copy paste may get you a basic good hashrate if you’re lucky but if you want to get the best out of them you’ll need to finetune every one of them.
I’m going to try ergo in the next few weeks too, so I’m starting with reading this entire topic to see the other users experiences.
And sorry, no support for Windows here, but the advantage of HiveOS is that even if your SSD dies, your settings (straps, OC’s and much more) aren’t stored on the SSD but on the hive server, so just install hiveos on a new SSD, enter your old credentials and up and running you go again.
yeah, I want to use hiveos but I can’t farm chia on it so in a bit of a pickle. If I get a few more cards it would be worth it to split the systems up but right now it doesn’t make sense to setup a whole separate system just for 2 cards.
Considered setting up a VM and passing the GPU to it but not sure it is worth the hassle. This is simply the only thread I could find on Vega 56’s and ERGO.
I am about to setup a hiveos drive for some troubleshooting and see what I can get that way though.
The worst part is that in the morning it was doing 196 stable without an issue. Now it won’t go past 183 without seriously jacking the clocks and power.
Ok, I have hive running now and besides being VERY un-intuitive for someone used to working directly with the miner. I finally got it running.
Biggest issue I am having is I cant test my settings since there seems to be a power limit at 165w that I can’t get past. I was getting the best performance at around 175w before. So not sure if my hash is being stopped short due to that.
How are people getting higher then 165w? I put in 200w+ in the power limit but no change?
I have done everything but my vega 64 running 85°C can u tell me is it normal? And also my power consumption is 205W … I am using your OC and Straps… tell me good OC and strap it has Samsung memory i have flashed vega 56 bios…
I am running mine at around 85c mem temps but they are in a 35c garage. If you are getting those temps in cooler weather looks like it is time to re-pad and paste (in my case I also could use a re-paste but pretty sure it would only drop temps a few degrees so waiting a bit longer)
Also, I ended up sorting out the low hash in windows, seems that I had to revert to AMD drivers 21.3.1. I am now back up to 196mh per card.
Thanks for all the work here, this is the only place I have found good overclock info for Ergo and vega.
Okay, i will give them a try, high end is done…too high wattage on whole rig total when +190mh each.
My psu´s are red hot allready ! lol …for the looks of those they work just fine with hynix allso…any experiences with them ?
Btw. Any idea why Soc jumps sometimes like this …?
ya thats the reason I use the 180mh settings, its like 20w less per card with effective core at 1300.
SOC is jumping for that card because core is so high, i’ve seen that before with really high core setting, even with lower mem, but have not figured out exactly how core is related to SOC level.
I had that when mem +930…could be thought that it was high Core, i was running way over 1400… first thought that i managed lock next soc state. Had nice high hash with vegas on erg then just pop into my head to check Wattage from the wall… man… i was pushing my other psu 1600w to the limits … that why all the sudden intrest to low hash, glad i didnt fry d psu… or burn d whole house down! … back on ETH again…
I got a Gigabyte Gaming OC Vega 56 with samsung HBM2 memory. I cannot for the life of me get it above 125MHs in NBminer (right now it is in a rig of rtx 3070 ti’s which dont play well with trex miner but does well in NBminer. But when I run NBminer and teamredminer, the vega56 refuses to do anything).
I haven’t tried out the 3070tis but I run 3070s and Vegas on the same rig, with TRM and NB Miner, in the flight sheet for NB miner I just had to add the extra config argument
“platform”: 1
so NB miner only runs on the Nvidia cards.
Gotta run TRM the other miners are not even close for the Vegas. OCs look fine, and I assume your are a pretty new OS version and myabe a beta image since your are running the 70tis?
Sometimes with TRM I have had to reboot the rig a couple times to get it running the first time but then it ran for weeks.
well, it is asking to update to 210822 but forget it.
So TRM is really the only option. But when I tried running it the way you suggested, it didn’t give any hash rates at all for the nvidia gpu’s and the Vega just kicked it…
I dont have much time today so I will give it a shot tomorrow. Thanks for the help so far, I appreciate it.