Ethereum GPU mining?

[Muadib]

I'm not too impressed by Nicehash so far. Their profitability calculator said that I should be making about $8 a day, while in reality I'm making about 5 Euros ($6) worth of Bitcoin a day. Also, their software doesn't seem to have a lot of "tweak ability"... I had to shut off the Monero CPU miner because it was pegging it at 100% and making the system run uncomfortably hot. If I could tell it not to use all 8 CPU cores, I'd imagine that it would be fine.

Like some other mining pools, they also don't let you take your profits until you have about $50 worth of cryptocurrency mined. I can't help but think that they do that on purpose, as many miners probably give up before they hit that threshold.

I'm new to this (2 days in), and I'm using Nicehash too. My biggest issue so far, is that Nicehash is using Excavator for my GPU. Excavator is saying that the GPU is running too hot, and performance is degraded. I use Afterburner to monitor my GPU temps, and my temp is 64-66 C, which I thought was good. I'm thinking that it could be the memory on the GPU, but I don't know how to monitor that.

My CPU is also running at 100%, and it's temps go from 68 -72C, so I'm good there. I'd be curious to know what your CPU temps are?

 

[ultimatebob]

I'm new to this (2 days in), and I'm using Nicehash too. My biggest issue so far, is that Nicehash is using Excavator for my GPU. Excavator is saying that the GPU is running too hot, and performance is degraded. I use Afterburner to monitor my GPU temps, and my temp is 64-66 C, which I thought was good. I'm thinking that it could be the memory on the GPU, but I don't know how to monitor that.

My CPU is also running at 100%, and it's temps go from 68 -72C, so I'm good there. I'd be curious to know what your CPU temps are?

It was registering 84C on my Ryzen 7 3700X after 30 minutes of mining. It was running a steady boost block speed of 4.1 GHz, which is impressive but scary at the same time. I don't want to burn out a brand new $330 CPU and $100 waterblock cooler to make 50 cents a day doing CPU mining.

The GPU (An MSI GeForce 2060 Super) seems to be unfazed by running at 100% all the time, and is holding at a steady 67C.

 

[Muadib]

It was registering 84C on my Ryzen 7 3700X after 30 minutes of mining. It was running a steady boost block speed of 4.1 GHz, which is impressive but scary at the same time. I don't want to burn out a brand new $330 CPU and $100 waterblock cooler to make 50 cents a day doing CPU mining.

The GPU (An MSI GeForce 2060 Super) seems to be unfazed by running at 100% all the time, and is holding at a steady 67C.

Wow, your CPU is hot! I'm using a 3900X, and had heat issues when I first got it. Turns out my thermal grease was the problem.

 

[aleader]

It was registering 84C on my Ryzen 7 3700X after 30 minutes of mining. It was running a steady boost block speed of 4.1 GHz, which is impressive but scary at the same time. I don't want to burn out a brand new $330 CPU and $100 waterblock cooler to make 50 cents a day doing CPU mining.

The GPU (An MSI GeForce 2060 Super) seems to be unfazed by running at 100% all the time, and is holding at a steady 67C.

I don't believe that mining with your CPU is worth it...something like a few pennies a day. I like to keep my Nvidia GPUs under 50C as there is no way to know what the memory temps are. The 5700XT has a sensor, so I can see that one. It's probably fine, but I know that the 3080/3090 memory is said to run at 110C even though the GPU registers 49C. My Sapphire Pulse 5700XT is one of the better ones for cooling, but the memory runs at 80-82C with the card at 45C and the fans at 35% (1190RPM). That obviously bumps up the higher I o/c the memory.

 

[Leeea]

How the hell am I supposed to mine ETH with rolling blackouts now in Texas?

... mining during an electrical emergency is immoral!

( I know your not actually trying to mine during an emergency, just being silly )

 

[Leeea]

I don't believe that mining with your CPU is worth it...something like a few pennies a day.

I remember a few years back there used to be alt-coins* that were only mined with cpu, an those could be profitable back then. However I am thinking intel avx-512 was required to do well with that stuff. I could be confused.

edit:
*I had to convert them to ETH, and then convert ETH to USD. Left me with the impression that eth is the only real future in the crypto-ecosystem, with bitcoin being way to expensive transaction wise.

 

[aleader]

I remember a few years back there used to be alt-coins that were only mined with cpu, an those could be profitable back then. However I am thinking intel avx-512 was required to do well with that stuff. I could be confused.

Yah, I haven't tried it, I'm just going by the recent videos on Son of a Tech and Mining Chamber. They both say don't bother. I plan on keeping my 5600x for many years, so I want to be nice to it

 

[VirtualLarry]

Here's TechDeals' viewpoint on "wearing out" hardware from mining.

TLDR: It's a non-issue.

 

[SteveGrabowski]

... mining during an electrical emergency is immoral!

( I know your not actually trying to mine during an emergency, just being silly )

LOL. I'm really just trying to run my heater. My neighborhood has had rolling blackouts applied to it nine times now since 2AM this morning. Texas blows, what a garbage electric grid requiring a ton of rolling blackouts because it gets a little cold.

 

[Leeea]

LOL. I'm really just trying to run my heater. My neighborhood has had rolling blackouts applied to it nine times now since 2AM this morning. Texas blows, what a garbage electric grid requiring a ton of rolling blackouts because it gets a little cold.

ouch.

That makes me wonder if I should go buy one of those propane heaters that can be placed indoors without a chimney. Awful, but way better then nothing. That said, I live next to a massive wind mill farm so probably not necessary.

 

[MrTeal]

No offense, but saying it's a non-issue isn't supported by that video.
I'm not saying mining will wear wear out a card excessively, but all they said was that it's not any worse than gaming and maybe it's even better without any evidence whatsoever, and then saying even if it does who cares because it'll be paid off anyway.

 

[Leeea]

Here's TechDeals' viewpoint on "wearing out" hardware from mining.

TLDR: It's a non-issue.

Did not watch the video, but it is definitely an issue. Thermal paste only lasts so long before degrading. Fans only spin so many revolutions. Most high end caps are 10,000 hours @ 105 C on graphics cards. VRMs are frequently rated around 5000 hrs @ 105 C.

10,000 hrs = 417 days
5000 hrs = 208 days

Drop the temperature below 105 C, and lifespan increases. Cooling is usually focused on the gpu, the rest of most graphics cards run hotter. Miners tend to run cards hot, as the greater the temperature difference the cheaper to cool.

Most miners burn through most of the life of the cards before selling. Usually selling because it is no longer stable at their mining undervolt/overclock, and indication most of the life is burnt out of the card. Yea, it still works at stock, but for how long?

In order to avoid burning a card out prematurely when mining, the operator must run the card at 80% or less capacity and keep all of the components below 80C. An approach completely opposite to what nearly all of them do.

 

[Shmee]

The issue is generally the fans, or sometimes the memory? Fans are usually turned up for cooling, but keep in mind the core is generally undervolted and underclocked, so it is possibly more stress on some parts, and less on others. As far as I know, most miners run the cards cooler, with possible exception of memory. Depends on the card of course.

That said, I wouldn't avoid buying used cards just because they were mined on. In some cases they may be in better condition, except for the fans, but definitely check those. Also, sometimes there are great deals on used mining cards, especially if there are a bunch being sold off in bulk.

 

[IEC]

LOL. I'm really just trying to run my heater. My neighborhood has had rolling blackouts applied to it nine times now since 2AM this morning. Texas blows, what a garbage electric grid requiring a ton of rolling blackouts because it gets a little cold.

That's what happens when you invest in wind and solar without the necessary non-renewable co-generation facilities. Everything's just peachy when the sun is shining and the wind is blowing, but when your wind turbines literally freeze stuck and the sun is nowhere in sight due to blizzard conditions... lol.

 

[MrTeal]

Wearing out cards through mining did used to be an issue, but it's much less so now IMO. Back when it was BTC being mined on GPUs, a 10% core overclock gave you exactly 10% more hashrate and cards were being really pushed. Combine that with the pretty poor stock blowers you'd see and it wasn't uncommon to have cards fail after 6 months to a year. I had 5970s where the VRMs would basically be 100°C 24/7.
Capacitor aging can be an issue, but it's probably not as big a one as you assume these days. Undervolted mining tends not to be super current intensive, and if temps are reasonable those lifespans increase drastically. A polymer cap that's rated 10k at 105°C will have a rated life of 1M hours at 65°C.

Big caveat to that is a card coming out of an open air mining rig is an entirely different beast than one coming out of some gamer's case where he was using default fan curves and running 100°c memory junction temps constantly.

 

[aleader]

No offense, but saying it's a non-issue isn't supported by that video.
I'm not saying mining will wear wear out a card excessively, but all they said was that it's not any worse than gaming and maybe it's even better without any evidence whatsoever, and then saying even if it does who cares because it'll be paid off anyway.

I don't put much stock in that guy's often looooong videos...although it is sometimes entertaining. He's the same one that told us not to buy a 3070, and pushes people to buy the most expensive CPUs with high core counts, even though the evidence is crystal clear that for gaming cores don't matter and a 3600/5600x is more than enough (see Hardware Unboxed, GN).

 

[aleader]

Wearing out cards through mining did used to be an issue, but it's much less so now IMO. Back when it was BTC being mined on GPUs, a 10% core overclock gave you exactly 10% more hashrate and cards were being really pushed. Combine that with the pretty poor stock blowers you'd see and it wasn't uncommon to have cards fail after 6 months to a year. I had 5970s where the VRMs would basically be 100°C 24/7.
Capacitor aging can be an issue, but it's probably not as big a one as you assume these days. Undervolted mining tends not to be super current intensive, and if temps are reasonable those lifespans increase drastically. A polymer cap that's rated 10k at 105°C will have a rated life of 1M hours at 65°C.

Big caveat to that is a card coming out of an open air mining rig is an entirely different beast than one coming out of some gamer's case where he was using default fan curves and running 100°c memory junction temps constantly.

All that is probably true for all but the 3080/3090, which have very hot memory and throttle almost instantly when pushed for mining (from the videos I've seen). It seems to be a given that you need to add thermal pads to a lot of those cards, which has to be a real kick in the nuts after spending such stupid money...especially on a 3090.

I am keeping my 3070 at 48C (mem at +1200/PL 52%/-502 core, Fan 980RPM). I know the 3070 has to run a lot cooler than the 3080/3090 just because it uses a lot less power, but there isn't enough data on these cards yet, and no memory sensor. The EVGA card I have also has the same cooler as the 3080/3090 version, so that is a bonus I guess. I could push the memory upwards of 1300 and it's stable, but I only get maybe another 1MH/s, which by my estimates after fees is around an extra $6 USD per month. Not worth the risk IMO, no matter how slight.

I'm not sure if it's the drivers or what, but my 5700XT is very sensitive to tuning. If I drop the memory lower, the power (at the wall and in the miner) jumps up by about 10W and takes a while to settle lower. The memory temp is also quite sensitive to ambient. When I put the memory to 1860 though, after an hour or so the wattage settles at 158W (wall) and memory temp at 80C, so I just keep it higher.

I have also had a few driver crashes that don't seem to be related to temps or settings at all. I haven't had any issues at all with the Nvidia cards. Not a single crash in 3 weeks, and the power is only affected by the power limit, less so by the core clock.

 

[aleader]

That's what happens when you invest in wind and solar without the necessary non-renewable co-generation facilities. Everything's just peachy when the sun is shining and the wind is blowing, but when your wind turbines literally freeze stuck and the sun is nowhere in sight due to blizzard conditions... lol.

Sounds like real shoddy wind turbines. Here in sunny, windy -42C Sask., the turbines run all year long, no issues like that. The biggest issue here is the government's reluctance to get with the times and cover the entire SW part of the province with them (about the size of Texas 😉). Instead they like subsidizing their farmer buddies (premier is a farmer) to basically plant crops in the desert so they can collect crop insurance (our tax dollars). Now that's just plain stupid.

 

[MrTeal]

All that is probably true for all but the 3080/3090, which have very hot memory and throttle almost instantly when pushed for mining (from the videos I've seen). It seems to be a given that you need to add thermal pads to a lot of those cards, which has to be a real kick in the nuts after spending such stupid money...especially on a 3090.

I am keeping my 3070 at 48C (mem at +1200/PL 52%/-502 core, Fan 980RPM). I know the 3070 has to run a lot cooler than the 3080/3090 just because it uses a lot less power, but there isn't enough data on these cards yet, and no memory sensor. The EVGA card I have also has the same cooler as the 3080/3090 version, so that is a bonus I guess. I could push the memory upwards of 1300 and it's stable, but I only get maybe another 1MH/s, which by my estimates after fees is around an extra $6 USD per month. Not worth the risk IMO, no matter how slight.

I'm not sure if it's the drivers or what, but my 5700XT is very sensitive to tuning. If I drop the memory lower, the power (at the wall and in the miner) jumps up by about 10W and takes a while to settle lower. The memory temp is also quite sensitive to ambient. When I put the memory to 1860 though, after an hour or so the wattage settles at 158W (wall) and memory temp at 80C, so I just keep it higher.

I have also had a few driver crashes that don't seem to be related to temps or settings at all. I haven't had any issues at all with the Nvidia cards. Not a single crash in 3 weeks, and the power is only affected by the power limit, less so by the core clock.

The bigger issue with the 3080s and hot memory IMO is that the fan curve is based on core temp (at least until the memory throttles) and most miners are running undervolted and downclocked cores with overclocked memory. At least prior to putting my waterblock on, with just my 3080 FTW3 installed I had no problem keeping memory cool just by manually setting a 70% fan speed. Wasn't crazy loud either.

 

[fleshconsumed]

In order to avoid burning a card out prematurely when mining, the operator must run the card at 80% or less capacity and keep all of the components below 80C. An approach completely opposite to what nearly all of them do.

Completely not true right now. Miners primary motivation is profitability, and it is far more profitable to reduce power limits/downclock GPU to reduce power consumption without affecting hashrate. According to reports the sweet spot for properly tuned 3060ti is 120W while mining which is about half of what it's rated for.

I'd feel totally fine buying mining card so long as I knew it was properly tuned. That is the only potential issue here - as someone mentioned above, I would be worried about buying mining card from some amateur who ran it in a regular case at full power without any tuning, aka buying a card from an idiot.

 

[Leeea]

That's what happens when you invest in wind and solar without the necessary non-renewable co-generation facilities. Everything's just peachy when the sun is shining and the wind is blowing, but when your wind turbines literally freeze stuck and the sun is nowhere in sight due to blizzard conditions... lol.

That is lies and part of a misinformation campaign.

https://www.npr.org/2021/02/16/9682...thern-mexico-as-blackouts-and-bitter-cold-con

The Electric Reliability Council of Texas said the grid lost some 34,000 megawatts of power. Energy sources powering the grid were knocked offline, most of which were powered by natural gas, coal or nuclear energy, according to Houston Public Radio.

https://fortune.com/2021/02/16/texas-power-outage-frozen-wind-turbines/

“The wind is not solely to blame,” said Wade Schauer, research director of Americas power and renewables at Wood Mackenzie. He estimates that about 27 gigawatts of coal, nuclear and gas capacity is unavailable, in part because the cold has driven up demand for natural gas for heating. “That’s the bigger problem.”

...

Even so, wind generation has actually exceeded the grid operator’s daily forecast through the weekend. Solar power has been slightly below forecast Monday.

Sounds like real shoddy wind turbines. Here in sunny, windy -42C Sask., the turbines run all year long, no issues like that. The biggest issue here is the government's reluctance to get with the times and cover the entire SW part of the province with them (about the size of Texas 😉). Instead they like subsidizing their farmer buddies (premier is a farmer) to basically plant crops in the desert so they can collect crop insurance (our tax dollars). Now that's just plain stupid.

What he posted is part of a false and malicious misinformation campaign. See the links above. Wind in Texas is actually outperforming expectations right now. Said poster is likely just repeating someone else's lies simply because it feels good to confirm preconceived notions of the world.

 

[Accord99]

What he posted is part of a false and malicious misinformation campaign. See the links above. Wind in Texas is actually outperforming expectations right now. Said poster is likely just repeating someone else's lies simply because it feels good to confirm preconceived notions of the world.

Only because the expectation for wind is low, yesterday it was <2 GW. Today it's still just <5 GW, out of nameplate capacity of more than 25 GW. And total system demand of over 45 GW (and probably more like 60+ GW without the rotating blackouts).

 

[Hitman928]

That is lies and part of a misinformation campaign.

https://www.npr.org/2021/02/16/9682...thern-mexico-as-blackouts-and-bitter-cold-con



https://fortune.com/2021/02/16/texas-power-outage-frozen-wind-turbines/

...





What he posted is part of a false and malicious misinformation campaign. See the links above. Wind in Texas is actually outperforming expectations right now. Said poster is likely just repeating someone else's lies simply because it feels good to confirm preconceived notions of the world.

Wind turbines freezing is still a big part of the problem. They lost 60% of the wind generation from the week before which is still 10% -20% of the total expected power generation. The overperforming comment is simply based on expecting a greater than 60% power loss. If Texas was entirely powered off of wind and solar, the amount without power would have been way higher. I'm not against getting more renewable energy, but I'm also very aware of the issues with current renewable solutions as well.

 

[blckgrffn]

So basically, I need to decide if I want to build a open-frame all purpose mining rig with mobos that are hard to find, but would be the cheapest option, or build a hyrbid gaming rig that can actually be used for other things AND mining. And then I would for some reason need 32GB and more gaming/higher end parts but would have a box that can be used for multiple things.

I have my own gaming rig, so this second rig wouldnt necessarily be used for "gaming" but I am thinking of things like, being a file server for streaming in house, mining, and anything else I might want. I suppose that is a moot point as being used for mining has the same GPU requirements as being used for gaming.

Costco has a Legion PC with 5700 for $1k. Just saying, ha.

For file serving and "the like" I've moved to Plex for videos and just hanging ~14TB drives off of USB. For home use it's good enough, and it's also full fledged PC that can do whatever as needed. I could RDP but that can be funny with displays, so I just use AnyDesk.

 

[Leeea]

Wind turbines freezing is still a big part of the problem. They lost 60% of the wind generation from the week before which is still 10% -20% of the total expected power generation. The overperforming comment is simply based on expecting a greater than 60% power loss. If Texas was entirely powered off of wind and solar, the amount without power would have been way higher. I'm not against getting more renewable energy, but I'm also very aware of the issues with current renewable solutions as well.

From the link I already provided:

The majority of outages overnight were plants fueled by natural gas, coal and nuclear, which together make up more than two-thirds of power generation during winter.

The inherent unpredictable fragility of fossil fuels is what causes the problem. The behavior of wind turbines is not only expected, but can be planned for and mitigated. If Texas had a majority wind power generation system this disaster would not be happening. Over provisioned wind power results in strong interconnections to other time zones to sell electricity to them when local demand is low and remote demand is high. Those same interconnections would be able to sell the necessary power back to Texas to allow it avoid this disaster entirely.


Nobody can plan for the unpredictable and unreliable behavior of fossil fuel, making disasters like this inevitable when relying on it.