No Problem,
i this metod working lhr 1 (74) my first card and i find memory 1700 not bad mh if 1800+ i get lower mh and some times reset system.
I found memory oc it by increasing 200.
step by step memory raise OC stop when mh starts to drop
my advice use lock clock
try setting the core clock around 1500-1600, lhr cards are much more stable with locked cores instead of negative offsets in my experience
Thought I was going crazy… Bought 2 3080TIs this week and Zotac is hitting 90mh/s But eVGA hits limiter at 64mh/s Been trying everything drivers, miners, OC settings. Nothing works.
NovaMiner
Without overclocking how much mh?
some times nbminer locked oc, stop miner Core: 0 Memory: 0 later start miner try firtst find miner options for evga later find oc settings, Each card needs different settings.
just set evga 3080ti to default and restarted miner getting 70.21-70.64mh/s @ 256w in miner
Update: So card is stable at default (70.5mh/s) but as soon as I input any OC setting the card drops to 64-65mh/s. I was hoping to hit 86mh+ on card but seems to be odd card.
ok that good, without oc 70 not bad, you can start without oc your evga, later step by step memory raise.
Try increasing it by 200 memory oc (hiveos) (windows increasing 100),
increasing by 200 wait 5-10 minutes mining, again increasing by 200.
you need to find where get lower mh
Note: First find best memory oc for your evga later find lock clock. Each card needs different settings.
I have 2x 3080ti. When running them on windows using nbminer 39.7. At moderate OC settings, the miner never trigger LHR lock and been stable for hours at LHR 73.5. My 3080ti hash 89MH and 86MH.
As soon as I move them to my hiveos rig, together with a bunch of non-LHR cards, nbminer keeps triggering LHR lock every 0.4 or 0.5 mins, and hashrate jumps between 65MH to 82MH, the number of accepted shares are about the same as my 2080.
Just FYI, running 2 3080ti on their own did not solve the problem.
Is there a trick I can do, or just nbminer isn’t designed for Linux?
i use 2x3080ti 1x3070ti 5x3070 on hiveos 1 rig and stable. maybe from nb miner maybe from hiveos some times oc settings locked. there is usually an unlock loop. if see unlock circle long time (20-25 minutes 30sec re unlock try) (because 15 minutes standart unlock time, maybe 4-6 times). delete oc settins and start mining in 15 minutes you can stable mh (dont stop miner) Increase step by step increment by 200 memory.
Note: if change miner again delete oc on lhr cards, start miner and re oc your cards
no difference when changing values:
Core-0 Mem-0 69.54-70.25mh/s 256w in miner
Core-0 Mem-+200 69.5-70mh/s 256w in miner
Core-0 Mem-+400 68.23-69.25mh/s 256w in miner
Core-0 Mem-+600 68.18-69.10mh/s 256w in miner
Core- Mem- -200 69.89-70.25mh/s 256w in miner
Core-0 Mem- -400 69.89-70.25mh/s 256w in miner
can you try evga on windows for memory temperture, my idea that card reaches thermal limit. It’s helpful to check.
how many did you set power limit ?
I can. But I’ll have to pull card from system and hook up to test bench. Card is running at 59c in Hive. Topping out at 72.2mh/s at default setting. PL is set to 0. I have tried to set PL to 310-350w but card never goes past 256-257w. If I change memory even 100 mhz it drops. Looking at the card I can see memory chips are making good contact with thermal pads and cooler. But like I said I’ll have to pull card to test temps. Very picky eVGA card. The Zotac has been running at 62c(-200/+2600/PL-290w) locked at 88-89mh/s since I installed it.
i understant. very interesting situation. there must be a reason. power limit only guard. i tried “lhr-mode”: “2” lock clock 1600 memory:1700 pl:285w 48 hours test now 80 mh stable (my cards suprimx). but use power waves 190-275w. i see some 3080ti users use lock clock:1200 memory:3000 - some users lock clock:1500 memory :2600 that bad this settings for my cards. my cards usable maximum memory:2300 (if it doesn’t increase mh, it doesn’t make sense to increase it that much.i test memory:1700)
test temps thelpful to check.
I always say that a separate setting is required for each card. our goal is to find the right setting for you evga
We try to solve it without seeing it theoretically. A lot of information would be helpful as I haven’t been able to test it personally. I don’t have the same card
Heres some info:
eVGA 3080Ti- 12G-P5-3955-KR
HiveOS 0.6-205@210715
nVidia driver- 495.44
PS 2x1400w(2800x total) Plat. HP server with breakout board.
2x8 pins for power on card. (being split off one line) could try and run two.
Rig has 2x3060Ti, 3070non LHR, 2x3080Ti (adding another 3060Ti and 3080LHR this week)
Card runs @ 59-62c Stable at 72mh/s default setting.
Update: ran two separate pwr lines to 8 pin connectors. Didnt do anything. Peaked at 74 then dropped to 72 stable
try again oc
Didnt work… just drops hashrate.
Here is the card:
https://eu.evga.com/products/product.aspx?pn=12G-P5-3955-KR
i cant find EVGA 3080 Ti XC3 ULTRA temperture info, but i find same model 3080 lhr info 60/88 C memory:0, that info I think it’s a thermal limit for your card. ti series use more power and high temperture than 3080.
Changed to “lhr-mode”: 2
Now getting 74.25-74.60mh/s default settings
%90 you card get thermal limit. i more try fix same problem. lhr-mode 2 low power = low temperture.
Ya… looking that way. Just surprised so bad thermal wise. Really will have to pull card and check in Windows. Dont what to have to replace pads on new card…
It is not difficult to change, but you need to measure the thicknesses, you just need to be slow and careful, of course, choose a quality thermal pad. i used extreme odyssey for gigbayte Aorus 3080 (not lhr), memory temp maximum 74C now (i never see 75C, memory oc 2800).