Search content
Sort by
Showing 14 of 14 results by Martobtc
Re: ETH Mining with NVIDIA - cuda error - cannot allocate big buffer for dag
Use "-eres 0"
-eres setting is related to Ethereum mining stability. Every next Ethereum epoch requires a bit more GPU memory, miner can crash during reallocating GPU buffer for new DAG. To avoid it, miner reserves a bit larger GPU buffer at startup, so it can process several epochs without buffer reallocation. This setting defines how many epochs miner must foresee when it reserves GPU buffer, i.e. how many epochs will be processed without buffer reallocation. Default value of -eres is 2.
-eres setting is related to Ethereum mining stability. Every next Ethereum epoch requires a bit more GPU memory, miner can crash during reallocating GPU buffer for new DAG. To avoid it, miner reserves a bit larger GPU buffer at startup, so it can process several epochs without buffer reallocation. This setting defines how many epochs miner must foresee when it reserves GPU buffer, i.e. how many epochs will be processed without buffer reallocation. Default value of -eres is 2.
Hi, I have two Founder Edition 1060 6gb cards with Samsung memory. Following some other guides I have seen I have mine set in Afterburner as:
Power limit - 65
Temp limit - 87
Core Clock - -150
Memory Clock - +700
Fan - 65
If I increase the memory clock Claymore is showing GPU errors and today I am finding what I thought were these stable settings are causing one of the GPU's to throttle. My electricity cost of $0.18 so I have to be a bit careful. Has anyone got any ideas how I can get more out of these as the general consensus seems to be that with Samsung ram the memory clock should be able to go much higher.
Thank you
Power limit - 65
Temp limit - 87
Core Clock - -150
Memory Clock - +700
Fan - 65
If I increase the memory clock Claymore is showing GPU errors and today I am finding what I thought were these stable settings are causing one of the GPU's to throttle. My electricity cost of $0.18 so I have to be a bit careful. Has anyone got any ideas how I can get more out of these as the general consensus seems to be that with Samsung ram the memory clock should be able to go much higher.
Thank you
All my samsung memory cards go to +900 and I run them at +850 on 62-64% power. Try it with the core clock at 0. I never adjust my core clock despite what others say because I find there is no gain by doing so. I also don't set the fan at 65 as I prefer the AfterBurner curve but that shouldn't affect anything.
There is a chance your specific card may require more power. As a general rule I keep my power at 75 when testing to make sure it can handle +850. When I know there are no errors, I slowly decrease the power until I find the error point and then run it at 1 or 2 points higher than the minimum. My Founders edition run at 64% and my Zotacs can handle 62%.
afaik power limit in Nvidia cards only refers to how much voltage is given to the core, not the ram. In fact, when you increase that power limit, the core clock goes up aswell. If that was the case that more power limit gives more ram "room" to hit higher frecuencies, ram speed would also run faster as you move the slider, but that does not occur. Moreover, i can garantee that even increasing to 116% the power limiit, will make no difference in how much can i push the ram.
Thanks for this. You were right when I reduced the core to 0 and increased the memory clock the hash rate increased. Interestingly though one card will take +850 and the other +700 at most. This seems odd.
Yeah, I tried forcing constant voltage on my 1060s, changed the default volgate curve in AB also. It doesn't affect your memory stability in any way. It only affects the core clock in a smart way, you better don't play with voltages on nVidia GPUs, power limit does the job for you.
Let's say your GPU runs at 1500 Mhz core clock at stock (+0 core) and reduced power limit to 60%, that's at about 0.8mv. By increasing your core to +100 you don't force the GPU to draw more power, you only force it to run at 100 Mhz higher - 1600 at the same voltage and power draw. If your GPU runs 1400 stock then +100 will make it run at 1500 etc. You basically set some extra Mhz at the same voltage and power levels. Simply said you get the full potential of your GPU kind of free.
Some GPUs can handle +200 even 250 others can't take even +100. You should carefully play with your settings and see whats the maximum stable speed.
If you are concerned about electricity then work with the power limit as well and see how low you can get it without losing too much performance.
My 1060s run stable at 60% power, around 0.800mv, some of them run at 0.762 even. Losing about 1% hash rate compared to 100% power limit.
Palit gtx 1060 3gb @ 1974 gpu /+600 mem and I'm getting ~17 MH/s DaggerHashimoto ~832 MH/s Decred in dual mining.
What could be the reason for such low hashrate? Mining with NiceHash miner 2.1.0.2 beta.
Thanks for any help.
What could be the reason for such low hashrate? Mining with NiceHash miner 2.1.0.2 beta.
Thanks for any help.
Consider mining on the NiceHash Legacy Miner 1.8.x.x version, it gave me at least 5-10% more profit on my 1060 rig last time I compared both.
Also install GPU-Z and check the type of your memory, it most likely is Hynix with such low Dagger hash rate. Your card might give you better result on Equihash coins.
Hey fellow miners, my 1060 rig has been runing for almost 2 months now. I'm happy to say I was lucky with the memory and all my cards are with Samsung, 3 GB tho.
I've been tuning them up and down, playing a lot with the clocks. The final settings I found to be optimal are:
Power limit - 60%
Temp limit - 69c
Core +150
Memory +880
With 100% power limit I was getting a solid 25 Mh/s , up to 25.2 - 25.3 per card with Claymore Dual Miner. I dropped the power limit all the way down to 60% at the cost of about 0.5 Mh/s per card. My temps don't go above 38-42c with custom fan curve (mostly blowing at 40-50%) in a 15c room temp.
If I bump the power limit a bit higher at 80-85% I'm also getting solid 305-310 Sol/s on equihash. But atm the profit is less, even if same the extra electricity cost and higher temps on the cards are not worth imo.
I just wanted to share with you my personal opinion about 1060s 3 GB as I've read a ton of info about them and experimented a lot with mine.
About the settings:
Overclock your memory first, I think +500 or even +600 is a perfectly safe to start with if you've got Samsung memory, I started mine from +700 on the very first boot. Don't increase with more than +50 on each step. When you reach the maximum stable clock, start with the core - any card should be able to handle +100, but you can start from +50 and increase by +10 until you find the maximum stable setting. Then start lowering your power limit by 5% and see how low you can go without your hash rate droping like crazy. It depends on the algo you desire to optimize it for of course. You should be able to get your power down to 60-70% at the cost of 1-2% performance on Ethash and about 80-85% for Equihash or other algos.
Micron and Elpida memory cards are decent on Ethash as well as Equihash. Depends what clocks your card can handle and what algo you prefer (as well as what's most profitable atm).
Hynix sadly is bad at Ethash, so you better work your settings around Equihash algo. Find the highest stable core clock first then get to memory and power. Cards are not useless, they are quite decent just a bit limited in the choice of algo/coins.
My advice is also pay attention to the power draw and your temps as that's really important. 1-2 mh/s or a few sols are not worth it if your power is 20% more. Also the lower the power - the lower the temps. Less noise and longer life of the cards.
Note that I have a second rig - exactly the same. I can't get the power limit under 66-67% without my hash rate going to crap, but the memory clocks are stable at a bit higher clocks +910/+930. Cards are from the same batch, in fact all the 12 cards have consequent serial numbers. I guess not only cards matter but PSU, Mobo, risers as well. Don't try to copy settings from somebody else even if they have same hardware. Just play around with your settings for a couple of minutes/hours and find what's best for you.
If you are just getting into mining, don't be too picky on hardware. Just see what cards you can afford and find then build your rig around them. The sooner you start mining - the better. You really lose more while waiting than building a rig with a bit lower ROI. Also don't go with more than 8 GPUs if it's your first rig.
I've been tuning them up and down, playing a lot with the clocks. The final settings I found to be optimal are:
Power limit - 60%
Temp limit - 69c
Core +150
Memory +880
With 100% power limit I was getting a solid 25 Mh/s , up to 25.2 - 25.3 per card with Claymore Dual Miner. I dropped the power limit all the way down to 60% at the cost of about 0.5 Mh/s per card. My temps don't go above 38-42c with custom fan curve (mostly blowing at 40-50%) in a 15c room temp.
If I bump the power limit a bit higher at 80-85% I'm also getting solid 305-310 Sol/s on equihash. But atm the profit is less, even if same the extra electricity cost and higher temps on the cards are not worth imo.
I just wanted to share with you my personal opinion about 1060s 3 GB as I've read a ton of info about them and experimented a lot with mine.
About the settings:
Overclock your memory first, I think +500 or even +600 is a perfectly safe to start with if you've got Samsung memory, I started mine from +700 on the very first boot. Don't increase with more than +50 on each step. When you reach the maximum stable clock, start with the core - any card should be able to handle +100, but you can start from +50 and increase by +10 until you find the maximum stable setting. Then start lowering your power limit by 5% and see how low you can go without your hash rate droping like crazy. It depends on the algo you desire to optimize it for of course. You should be able to get your power down to 60-70% at the cost of 1-2% performance on Ethash and about 80-85% for Equihash or other algos.
Micron and Elpida memory cards are decent on Ethash as well as Equihash. Depends what clocks your card can handle and what algo you prefer (as well as what's most profitable atm).
Hynix sadly is bad at Ethash, so you better work your settings around Equihash algo. Find the highest stable core clock first then get to memory and power. Cards are not useless, they are quite decent just a bit limited in the choice of algo/coins.
My advice is also pay attention to the power draw and your temps as that's really important. 1-2 mh/s or a few sols are not worth it if your power is 20% more. Also the lower the power - the lower the temps. Less noise and longer life of the cards.
Note that I have a second rig - exactly the same. I can't get the power limit under 66-67% without my hash rate going to crap, but the memory clocks are stable at a bit higher clocks +910/+930. Cards are from the same batch, in fact all the 12 cards have consequent serial numbers. I guess not only cards matter but PSU, Mobo, risers as well. Don't try to copy settings from somebody else even if they have same hardware. Just play around with your settings for a couple of minutes/hours and find what's best for you.
If you are just getting into mining, don't be too picky on hardware. Just see what cards you can afford and find then build your rig around them. The sooner you start mining - the better. You really lose more while waiting than building a rig with a bit lower ROI. Also don't go with more than 8 GPUs if it's your first rig.
Re: Gigabyte-GA-H110-D3A Mining board SUCKS! Can't get 6gpu up and running
at the end I got it to work, I got scammed with the pack of 4 risers, received them dead :/
the mayor problem of the H110-D3A that I had, is the Mining mode that didn't want to load windows, don't remember how I solved it.
the mayor problem of the H110-D3A that I had, is the Mining mode that didn't want to load windows, don't remember how I solved it.
Yeah I remember going through this as well. I think I had to load windows for the first time with Mining mode off, then enabling it after installing 1st GPU with drivers. Not sure exactly what I did but didn't take me long to go through. Even now after resetting BIOS, I have no problem with detecting GPUs or loading Windows. It's just my IGPU wont display anything for me.
Glad to hear. May I ask if you did something specific to get your IGPU going? Or it was working by default? Did you set everything in BIOS and boot windows first time without any GPUs plugged in PCIE? Also did you update bios to latest F25? I saw changes are minor and didn't bother as my mobos had the latest F24 version already.
Just trying to find the missing piece of the puzzle as I've assembled two of those rigs with same components and I have the same problem on both.
Just trying to find the missing piece of the puzzle as I've assembled two of those rigs with same components and I have the same problem on both.
So I'm using the said mobo for minign with 6x 1060s, I sorted out how to plug the cards one by one in order to properly recognise them. I have set the BIOS and Windows settings exactly as described in Gigabyte's official post about setting this mobo for mining. My rig is runing and mining already, the only issue I have is I can't get my Integrated Graphics to work while I have GPU plugged in the PCI-E slot. No matter what settings I choose in BIOS Integrated wont display video. So I'm stuck at using one of the GPUs for display and it affects mining performance. Might affect stability as well, as I got lucky with Samsung memory and I'm pushing the limits of the cards.
I've heard some motherboards have 6 PCI slots and only support UP to 6 GPUs, therefore you can't get 6 + integrated working at the same time because integrated also counts as GPU.
Can anyone using this same motherboard confirm that it works just fine with 6 GPUs mining and using the Integrated VGA for display?
Thanks.
I've heard some motherboards have 6 PCI slots and only support UP to 6 GPUs, therefore you can't get 6 + integrated working at the same time because integrated also counts as GPU.
Can anyone using this same motherboard confirm that it works just fine with 6 GPUs mining and using the Integrated VGA for display?
Thanks.
Re: Gigabyte-GA-H110-D3A Mining board SUCKS! Can't get 6gpu up and running
For me works like a charm, I have two of them and no problem at all, have 6 pcs of Gv-1070 gtx on them.
Used uefi win.
try this https://www.gigabyte.com/MicroSite/462/mining.html
Used uefi win.
try this https://www.gigabyte.com/MicroSite/462/mining.html
That's what I followed when I got it runing, but I was more focused on recognising and getting the 6 GPUs to work. I still couldn't get it to use integrated VGA instead. If I enable integrated graphics in bios and choose IGFX as primary display output, I lose both integrated and gpu and have to reset bios after that to get video signal. Have the same issue on 2nd rig with same config as I said. Doesn't seem like a deffective product.
Other than that, I use GPU0 for display, rig is working fine, mining is going well, it just bothers me that it's not right.
Re: Gigabyte-GA-H110-D3A Mining board SUCKS! Can't get 6gpu up and running
I got this board and at first I had some trouble getting it running properly but this was my first ever rig. Had the same issue as many other people, after loading windows it got stuck on loading screen. But that's somewhat normal, it is detecting GPUs in the meantime, give it a few seconds up to a minute instead of hitting numlock capslock or trying different method to check if PC has frozen.
I plugged 1 GPU at first and set the bios settings properly (enable 4g, mining mode, gen2 etc etc) then installed video drivers and then restarted with 1 more GPU (wait to recognise it), then restart and plug one more (wait to recognise) and so on until you plug all 6 one by one.
The only issue I'm having is it won't force the video on the integrated graphics as long as there is GPU plugged in the PCI-E slot. I enabled integrated gpu in bios and chose it as primary display output, but still nothing. It either displays through PCI-E main slot, or just doesn't display video anywhere. Bios is updated. Also I got 2nd rig recently - exactly same config and mobo - same issue.
So do you have any suggestions what might be the problem? I really don't want to have my video output on one of the GPUs as it affects performance a little bit. Recently heard it might be causing instability with higher overclock which I encountered, it's not all the time, sometimes it's stable for hours, sometimes unstable after couple of minutes. Maybe it's because of the video being on one of the GPUs?
I plugged 1 GPU at first and set the bios settings properly (enable 4g, mining mode, gen2 etc etc) then installed video drivers and then restarted with 1 more GPU (wait to recognise it), then restart and plug one more (wait to recognise) and so on until you plug all 6 one by one.
The only issue I'm having is it won't force the video on the integrated graphics as long as there is GPU plugged in the PCI-E slot. I enabled integrated gpu in bios and chose it as primary display output, but still nothing. It either displays through PCI-E main slot, or just doesn't display video anywhere. Bios is updated. Also I got 2nd rig recently - exactly same config and mobo - same issue.
So do you have any suggestions what might be the problem? I really don't want to have my video output on one of the GPUs as it affects performance a little bit. Recently heard it might be causing instability with higher overclock which I encountered, it's not all the time, sometimes it's stable for hours, sometimes unstable after couple of minutes. Maybe it's because of the video being on one of the GPUs?
I'm mining with 6x Gigabyte Windforce OC 3GB (Samsung)
Runing them on:
TDP 67%
Temp 69c
Core -100
Memory +950
They seem stable so far keeping just above 25 Mh/s eth on claymore dual miner. I thiink it's not quite the sweetspot, but I like the consistent 150 Mh/s combined with total draw from the wall 560 watts. They also run quite cool - under 50c with 60-65% fan.
I was wondering if it's worth using different settings for auto-switching algos like a bit higher core and less memory as I notice equihash is also a good option according to my benchmarks in nicehash miner. Potentially more profitable from time to time, but these settings are far from ok for equi.
So any suggestions? Keep the good work on eth or sacrifice a bit for potential auto-switch profits?
Edit: Working full time so auto-switch is probably worth it.
Runing them on:
TDP 67%
Temp 69c
Core -100
Memory +950
They seem stable so far keeping just above 25 Mh/s eth on claymore dual miner. I thiink it's not quite the sweetspot, but I like the consistent 150 Mh/s combined with total draw from the wall 560 watts. They also run quite cool - under 50c with 60-65% fan.
I was wondering if it's worth using different settings for auto-switching algos like a bit higher core and less memory as I notice equihash is also a good option according to my benchmarks in nicehash miner. Potentially more profitable from time to time, but these settings are far from ok for equi.
So any suggestions? Keep the good work on eth or sacrifice a bit for potential auto-switch profits?
Edit: Working full time so auto-switch is probably worth it.
are people still buying 3gig cards or are those no longer a safe option?
They are, few posts back you'll see I just said I bought 12 of them, luckily all with samsung memory and they mine pretty good so far. They'll pay back investment before they become less profitable or unusable... and with the shortage of cards on the market, you shouldn't be that picky Hi, I'm mining zcash with dstminer with 6 gtx 1060 Msi gaming x plus 6gb and getting about 340sols with settings: power limit 70%, core clock +100, mem clock +450. I just bought power meter to measure power consumption from the wall and its showing about 830w which is way too high for that power limit. I then tried to change power limit (higher and lower) with MSI AB, and power consumption barely changes even when I go to 100%. Then I figured no matter what power limit I set, voltage remains about 1050mv, so I guess I'm not changing anything with that power limit slider.
Any suggestion?
Thanks
Any suggestion?
Thanks
Your power slider should be affecting the watts your cards drain. Maybe it's the bios of the cards not being affected by changes you make in the OC software? I had similar issue on a 970 evga with dual bios, with voltage control though. It also depends on your PSU efficiency. Try lowering power limit to 60% or 55% even and see if then power from wall drops below 800w. Just so you know if the slider is really not working. Make sure to apply settings to all cards.
I just bought 12x Gigabyte 1060 3GB GPUs at lowest price possible in my country (lately prices are going crazy). I wanted some 1070s at first but couldn't get my hands on any for quite some time, so I wasn't really satisfied with the 1060s especially 3GB. But when I assembled the rigs last night, I found out all of them have got Samsung memory, yeey! Can't wait to set things up, OC and stuff and see results.
edit: Did some sloppy overclocking this morning on one of the rigs. The one at stock is mining at 20.5 - 21 mh/s each. The other one I got the power limit to 80% and memory to +750, didn't touch core and cards are hitting 24 mh/s on eth. I'm pretty sure I can take the memory a bit higher even. Seems like samsung memory really makes huge difference.
edit: Did some sloppy overclocking this morning on one of the rigs. The one at stock is mining at 20.5 - 21 mh/s each. The other one I got the power limit to 80% and memory to +750, didn't touch core and cards are hitting 24 mh/s on eth. I'm pretty sure I can take the memory a bit higher even. Seems like samsung memory really makes huge difference.