Ethereum GPU mining?

[Aikouka]

nothing like that. Although ive had no issues with Corsair HX and AX (never used an RM) in the past, I went with all EVGA G2 Superflower based units for my mining rigs based on recommendations of others and I haven't been disappointed.

I own a Corsair AX and two RM units, and they work fine. I stopped using Corsair because I could never figure out what was wrong with one of my PCs as it kept BSOD'ing about once every few months. Given that the PSU was something I never changed, I decided to swap over to EVGA. I did purchase a G3 earlier today, as I need to build another machine.

I tried to stuff a 1080 Ti and a 1080 into a Fractal Design Define R5, which as you can guess, it didn't work out too well for me. 😛 Heck, even only mining on the 1080 Ti with both cards in the system saw about +6-8C over mining with just the 1080 Ti in the system. (I did move the 1080 Ti from the middle PCI-E slot to the top one.) I have a Z170 board and a Pentium G4400 lying around, so all I really needed was a power supply, and I'll borrow some memory from a HTPC that I'm not using right now. I'm also going to test a new case out -- the Phanteks Enthoo Evolv ATX. Unfortunately, I was mostly planning on using air, which means I didn't consider buying cards with a reference design and such. So, going with water cooling became a tad bit more expensive than I'd like (EKWB's ROG Strix blocks are $150 + $50 backplate).

 

[IEC]

I've got multiple eVGA G2 and P2 units which are rock solid. I had one defective unit out of the box, but have never had one fail in 4+ years of mining.

I also have five Corsair RM650x factory refurbished units, which have been fine for months but I do not have long term reliability data for.

 

[casiofx]

How are you guys with 1080Ti's in your main rig finding the experience? With my previous dual 390 system, I didn't really notice any real effect on system usability while mining other than gaming. Using two 1080Ti's I've noticed a lot of lag, especially while trying to watch videos.

I faced the same issue too. For me its either using claymore eth miner with reduced intensity or equihash. I prefer equihash though

 

[Elfear]

I've got multiple eVGA G2 and P2 units which are rock solid. I had one defective unit out of the box, but have never had one fail in 4+ years of mining.

I also have five Corsair RM650x factory refurbished units, which have been fine for months but I do not have long term reliability data for.

+1 to everyone recommending the G2's. I've got two 1300W units that have been mining for months (one for almost 2yrs) and they are rock solid.

Picked up a 1000W Rosewill Quark that had great reviews. Little disappointed in the fan noise and the loud electrical switching sound when the unit powers on. The fan noise sounds like the blades are hitting something rather than just being a loud fan so I may RMA it.

 

[Charlie22911]

I've got my stuff built on wire shelving at the moment...

As for power supplies, my 5x 1070 is using the EVGA 1200 P2 and my 6x 1080ti is using the 1600 T2.

Those 6x 1080ti at 100% power target hit ~1700w at the wall which is pretty bonkers, I have it at 75% and draw ~1250w or so. I'm pulling about as much power through my breaker as I'm comfortable with.

 

[Madpacket]

I've got my stuff built on wire shelving at the moment...

As for power supplies, my 5x 1070 is using the EVGA 1200 P2 and my 6x 1080ti is using the 1600 T2.

Those 6x 1080ti at 100% power target hit ~1700w at the wall which is pretty bonkers, I have it at 75% and draw ~1250w or so. I'm pulling about as much power through my breaker as I'm comfortable with.

Playing it pretty safe on the power supplies. 1200P2 or a 1300 G2 should be plenty for a properly undervolted 6x 1080Ti setup.

5x1070's only need an 850W G2, or 1000W if you plan on adding a 6th card.

 

[DrMrLordX]

I have a single 1300W G2 that has been rolling along for about a year non-stop. Very solid unit. My Seasonic is flaky.

 

[Aikouka]

How are you guys with 1080Ti's in your main rig finding the experience? With my previous dual 390 system, I didn't really notice any real effect on system usability while mining other than gaming. Using two 1080Ti's I've noticed a lot of lag, especially while trying to watch videos.

Given the graphics-heavy nature of applications these days, it's either usable or pretty bad. I pretty much use every PC that I have to mine since they usually stay on anyway. One of them is a gaming HTPC in the living room, and Plex was fine even when mining, but any video played through a browser (live streaming or YouTube) was pretty bad. I ended up turning on multi-GPU mode in the BIOS, and switching back to the internal GPU. I can still game on the machine if I want to by turning off the miner and moving the cable back to the dGPU. My only complaint right now is that I use NiceHash, which switches algorithms depending on which will be more lucrative. The problem is that each time it switches, the new window pops up on top of everything else. This will happen if I'm watching Plex, YouTube, Twitch, etc., and it's really annoying.

As for my main desktop, I usually just turn off the miner if I want to play a game. I can still do most normal, non-multimedia things while the miner is running, but doing things like scrolling can have noticeably abysmal redraw rates. I'm going to be rebuilding my main desktop in another case, and I'm curious to see if I can do something like mine on one card and game on another. The only problem that I assume that I'll run into is the algorithm swapping that I mentioned earlier. I'm mostly going to accomplish this through using a hybrid card and a normal card, which should be fine. Although, amusingly enough, I see that EVGA just released their 1080 Ti HydroCopper card, which would've been even better (+$10 over the Hybrid). 😛

 

[thilanliyan]

I have 3x1070 + 1x1080 with 70% power limit running on an EVGA G2 850w PSU. Been humming along for about a month now, no issues.

 

[[DHT]Osiris]

The problem is that each time it switches, the new window pops up on top of everything else. This will happen if I'm watching Plex, YouTube, Twitch, etc., and it's really annoying.

As a note, if you use the old (non 2.0) miner, you can turn off those window popups. The new miner doesn't play as nicely though so I've got both in my existing game system. The 'old' miner runs for my 960gtx 24/7, minimized, the 2.0 miner runs my 1080 when I'm not gaming/doing stuff. Headless 1080ti's run the new miner 24/7.

 

[[DHT]Osiris]

Total tangent. 1080ti miners, notice driver crashes? nvlddmkm.sys errors, short halt in mining followed by recovery? Trying to read up/fix these, was going through downclocking my cards one by one before I found a post from someone stating that it was caused by ccminer halting/nicehash firing up another miner, but upon returning to oc'd settings on all my cards and eliminating those miners, they started happening more frequently, so I'm wondering if it's a combination of more than one factor. Need more testing to verify...

 

[Charlie22911]

Playing it pretty safe on the power supplies. 1200P2 or a 1300 G2 should be plenty for a properly undervolted 6x 1080Ti setup.

5x1070's only need an 850W G2, or 1000W if you plan on adding a 6th card.

Perhaps, however come the winter I will likely be running these cards at 100% power target and forgo the heater. If folding isn't very profitable I will also likely just try to crush the leader board in Folding@Home provided I'm not severely hampered by the PCIe 3.0 x1 interface...

 

[Madpacket]

Perhaps, however come the winter I will likely be running these cards at 100% power target and forgo the heater. If folding isn't very profitable I will also likely just try to crush the leader board in Folding@Home provided I'm not severely hampered by the PCIe 3.0 x1 interface...

Sure more power, more flexibility, more fun 🙂

For dedicated miners, you don't need such oversized power supplies, the money is better spent on buying more cards. Running Geforce 1060/70/80/80Ti's at a 100% power target is wasteful and could reduce the lifespan of your cards due to the added heat. You won't get any more hashes out of it compared to running at 75 - 80%. In fact, it's possible you may get less hashes as Geforce cards are very temperamental when trying to force specific GPU core speeds and memory speeds. Running 100% power target limits your ability to reduce the core GPU speed.

You only need enough GPU Mhz speed to properly feed the memory controller (at least for Ether). Other coins react differently but each card has a different ratio (GPU:MEM) due to a number of factors, but I haven't run into a situation (yet) where overclocking the GPU core much past its default (non-boosted) clocks makes sense. Geforce cards by default overclock themselves with 100% power targets, heck depending on the ASIC quality they'll continue to boost GPU core speeds even at significantly reduced power targets. You generally want to lower GPU speeds and use the available power target to increase the memory speed as much as possible (without introducing errors).

Also, the waste heat generated from the extra power required for that last 20% would cost more in hydro than the heat produced by a natural gas furnace. Unless you have electric heating, you're likely wasting money.

For reference, I run most of my miners at ~+-80 -85% loads (slightly little less after losses from the wall). I have several 6x480 rigs running off 1KW P2 power supplies, and two 5x480 rigs running off 850W G2's. All of them have been running almost 24/7 for about a year now. AFAIK 1070's don't use much more (if any) energy than a tuned 480 when mining. Also, good quality gold or platinum rated power supplies generally stay within an ideal power curve all the way up to 90% loads (before measurable efficiency losses kick in that are actually noticeable on a hydro bill).

 

[Madpacket]

Total tangent. 1080ti miners, notice driver crashes? nvlddmkm.sys errors, short halt in mining followed by recovery? Trying to read up/fix these, was going through downclocking my cards one by one before I found a post from someone stating that it was caused by ccminer halting/nicehash firing up another miner, but upon returning to oc'd settings on all my cards and eliminating those miners, they started happening more frequently, so I'm wondering if it's a combination of more than one factor. Need more testing to verify...

Does the issue manifest when using Claymore?

 

[IEC]

Does the issue manifest when using Claymore?

I've never noticed any errors running Claymore on my 1080 Ti cards.

 

[ozzy702]

I've got multiple eVGA G2 and P2 units which are rock solid. I had one defective unit out of the box, but have never had one fail in 4+ years of mining.

I also have five Corsair RM650x factory refurbished units, which have been fine for months but I do not have long term reliability data for.

I have a dozen or so G2 and P2 units as well and they have all been 100% rock solid. Great products.

 

[[DHT]Osiris]

Does the issue manifest when using Claymore?

Never used a flat miner, only nicehash. As a note, i'm still having crashes with excluding all ccminer miners, so that's not it. I tried restricting to a single one (Lyra2REv2, highest earner at the time) to see if it happens with mining without switching. After that, I'll try a ccminer algo without switching. I have a feeling the algo switches themselves are what's doing it.

 

[casiofx]

Never used a flat miner, only nicehash. As a note, i'm still having crashes with excluding all ccminer miners, so that's not it. I tried restricting to a single one (Lyra2REv2, highest earner at the time) to see if it happens with mining without switching. After that, I'll try a ccminer algo without switching. I have a feeling the algo switches themselves are what's doing it.

I had crashes when switching miners (nicehash auto switch).
So I only run one type of miner at a time. I also use the miner directly using batch file instead of via nicehash gui too. Still got crashes sometime if I manually close the miner.

 

[[DHT]Osiris]

I had crashes when switching miners (nicehash auto switch).
So I only run one type of miner at a time. I also use the miner directly using batch file instead of via nicehash gui too. Still got crashes sometime if I manually close the miner.

Yeah, with no miner swaps I had no crashes last night. I think I'm just gonna let the driver crashes happen, since there's no issue with it (no hard lockups). Played with settings, all cards seem reliable and putting out good numbers @80% power, +100 clock, +550 memory. Some people seem to be able to get a lot higher on the memory but i'm immediately getting issues mining @+600, not sure what's up with that.

 

[Raduque]

Can you mine with mixed amd/nvidia cards? Thinking of throwing an unused gtx960 in with my 280x.

Edit: Nah, nevermind. My PSU isn't SLI ready and can't do two GPUs. I'm gonna have to build a second rig to mine with the 960.

 

[DrMrLordX]

Dunno what PSU you're using, but if the power draw is low enough I would think any PSU could handle two GPUs. Unless there's a shortage of PCI-e connectors or something?

 

[[DHT]Osiris]

Can you mine with mixed amd/nvidia cards? Thinking of throwing an unused gtx960 in with my 280x.

Edit: Nah, nevermind. My PSU isn't SLI ready and can't do two GPUs. I'm gonna have to build a second rig to mine with the 960.

960's don't earn much on their lonesome (I have one as well), there's not really a good reason to run an entire rig (~300w minimum) for one. There's no reason your existing PSU shouldn't be able to power both, as long as it's over say, 400w or so. It should have more than one PCIe power connector I imagine.

 

[Raduque]

960's don't earn much on their lonesome (I have one as well), there's not really a good reason to run an entire rig (~300w minimum) for one.

What's the Mh/s? I'm getting about 14 on my OCed 280x and if the 960 does at least that I'll be fine with it.

It should have more than one PCIe power connector I imagine.

It doesn't. Or rather it does, but the 280x uses both of them.

 

[casiofx]

What's the Mh/s? I'm getting about 14 on my OCed 280x and if the 960 does at least that I'll be fine with it.


It doesn't. Or rather it does, but the 280x uses both of them.

If your 960 is 2GB version then I think it can't run eth mining because of dag size

 

[DrMrLordX]

Yeah true enough. I know my 2GB 270 produces a bus error (or something similarly annoying) when I try to mine ETH on it.