Are used mining GPUs a good buy?

>>61239328
no. >>>/g/sqt NEXT

>>61239360
Not stupid. People on /v/ have been debating this.

>>61239328
You are looking for >>>/biz/
That's where all the coin threads are located.

Only if the guy that used it for mining wasn't a moron and undervolted it. So probably not.

>>61241479
You forgot to mention increased DRAM clockspeed, you mong.

Ya, they're a steal. Morons don't buy them, decreasing demand, making a bunch of great gaming cards cheap af. I've bought dozens of mining cards and put them in client PC builds. I've never had a compliant.

>>61241479
>(wasn't a moron) and undervolted it
>wasn't (a moron and undervolted it)
explain

>>61239328
This is probably case by case. Miners run cards into the ground, hot and they use insane PCI-E splitters with weird power supplies that often catch fire.

I remember reading Google doing a study on HDDs lifespans, and noting that if a HDD lasted beyond an initial break-in period it was statistically very likely for it to last a long time.

If a card isn't destroyed via mining, does it mean it's an especially resilient card? Also consider hardware mods miners do to cards.

>>61242027
He means to say:
If the guy who was using it was a moron he would've not undervolted it.

>>61242060
>does it mean it's an especially resilient card?
So long as there is no power outage or sudden power spike, yes. These are machines you're talking about, running them 24/7 is the true test of a machine.

>>61242060
I'd say sans modifications they're a safe enough bet assuming the discount is particularly steep. The constant load really isn't the biggest issue, it's the expansion and contraction that comes with variable loads and powering cards on and off. When you're building transistors on the nanometer level they're far more liable to open or short when they're repeatedly heated and cooled or are in states of constant flux.

>>61242060
>hardware mods
Like? If I found a really good deal what kind of should I expect?

>>61242126
People underclock their cards to get better power to hashing ratios, which negatively affects you as a gamer.

>>61242137
>underclock
You mean overclock? Or undervolt? Wouldnt underclocking give less performance or am I missing something about the mining process here?

>>61242172
sure undervolt, or underclock. It really depends, I'm actually a casual miner and I know that different algos have different requirements. Underclocking memory would actually negatively effect mining of Monero since it's actually a memory-hard algo that is primarily bottlenecked by the video memory. Whereas other algos make use of raw FLOPS of OpenCL/CUDA processing.

The point is that most of these modifications are bad for end-users of graphics cards, like gamers.

>>61242204
How will undervolting harm the components over time?

>>61239328
Absolutely not

>>61242240
Not him but there's a reason manufacturers select a proper voltage range.

>>61242204
In your experience are the performance gains worth it or are they being cheapfucks? If I occasionally mined with my own card (havent bought one yet but will soon, even I dont make any money itd be an interesting thing to try) should I try said mods or leave it all stock?

>>61239328
If the miner isn't an idiot then sure. Make sure he reverses the card to it's original bios though. Expect a flood of gpus around next year when Eth goes POS.

>>61242279
I only mine with one GPU, which I use to mine zcash (equihash). I don't mod because it's my primary desktop GPU.

I primarily mine Monero using underutilized CPU resources on my servers. CryptoNight algo is memory-hard which means CPUs aren't at a disadvantage to GPUs because of their relatively fast L2 cache memory (it was intentionally designed this way).

>>61242279
I forgot to mention all of mining depends on your cost of electricity, if you can lower your power usage it's worth it to just buy additional GPUs. It's not about total hashing power, it's about power to hashing ratio. That's your profit.

I don't pay for electricity when it comes for any of my mining (included in rent, datacenter doesn't charge for electricity), so underclocking/undervolting makes zero sense for me, anyways.

>>61242355
>>61242393
I see, thanks anon.

>>61242063
Oh my fucking G-d this doesn't make it any easier.

>>61239498
>Not stupid
>People on /v/

>>61242747
>censoring yourself
stupid newfag, you can say GOD

>>61242763
Jews do that

>>61242240
My only guess is that it could cause overdraw from certain parts of the card and stress them. Thing is they would shit themselves immediately if this happened so undervolting them to that point would cause instantly recognizable instability.

>>61242765
Probably some Christians too

> Thou shalt not take the name of the LORD thy God in vain; for the LORD will not hold him guiltless that taketh his name in vain.

>mining gpu
>not asic

what is current year?

>>61242204
So why cant i just correct the voltage in my proprietor provided software?
>genuine question. I have the ability to underclock and overclock my card with asus software.
>also not same guy as you're replying to

>>61242804
Ethereum is "ASIC proof" or some dumb miner shit. Still don’t understand why

>>61242804
Making asics for some relatively obscure meme currencies is too expensive.

>>61242822
>"ASIC proof"
>proof
That almost makes it sound like if ASICs are somehow cheating or something. Any more info on the topic?

>>61242240
It won't, gpu will run cooler and you will use less electricity, that's it

>>61242810
I don't know, this type of discussion really gets into card specific, algo specific stuff. Electrical engineers fucking around. I honestly don't know that much, other than it's just a generic thing.

I've literally heard of people taking a graphite pencils to chips to lower resistances.

>>61242804
no one talking about mining with GPUs/CPUs are talking about SHA256/Scrypt coins. They're talking about Ethash, Equihash, CryptoNight, etc.

>>61242994
not a generic thing*

>>61242287
I don't know about you, but if I spent $10,000 on a mining rig I wouldn't just stop mining because Eth got harder. I'd find some new memecurrency to mine instead, telling myself that I'm getting in on it earlier THIS time.

>>61242822
>>61242894
wtf is asic proof

>>61243004
double wat

>>61242894
I have literally gotten into arguments about the concept of ASIC proof with other people in the cryptocurrency community. We start getting into discussions of scale of economy. But right now here's the issue.

>BTC is not actually that profitable to mine
>SHA256 ASICs are only made by a handful of mostly monopolized companies
>they keep the newest gen miners for themselves, and literally used secret "mining cheats" (look up ASICBoost and Antbleed)
>it's all based out of China, using cheap electricity they get through government corruption and shit
>no large companies are actually really bothering to compete with these relatively small companies
>SHA256 is a relatively easy algo to design an ASIC for in comparison to "ASIC-resistance coins"

Algos like CryptoNight intentionally make use of advanced CPU features, and the L2 cache on a CPU. It would be very hard to make an ASIC that does this without fundamentally making a CPU, or at least a stripped down x86 CPU. You essentially have to compete with AMD and Intel, which doesn't even exist in scale for SHA256 ASICs.

>>61242800
Only applies to oaths sworn to God, faggot. So every time you say DEUS VULT are you blaspheming? I think not. Stupid fucking protestants.

>>61243057
I want to specifically point out that centralization of hashing power is _bad_. ASIC-Resistant coins aren't just about "muh graphix card mining". It's about ensuring that the total network hashing power is coming from a diverse variety of sources and not under central control of a handful of ASIC manufacturers, which defeats the entire trustless aspect of cryptocurrency.

Having a million gamers mining your cryptocurrency is better than what is currently going on with Bitcoin.

>>61243057
>ASIC proof
>It would be very hard to make an ASIC

so the (very weak) claim is people hope no one bothers with their coin, which is in line with the fact that cpu/gpu is a ~subset of asic

>>61243172
Yes, the argument always is that if a certain coin becomes profitable enough an ASIC _will_ be made. That said the algos used in ASIC resistant coins aren't merely a mild variation of SHA256/scrypt just for the purpose of breaking ASIC compatibility. They are fundamentally designed to be vastly more complicated and expensive to design an ASIC for.

Part of my argument about algos like CryptoNight is that area between an ASIC and a stripped down x86 CPU become blurry. I honestly wouldn't be surprised if AMD started making stripped down GPUs that are made with Equihash/Ethash in mind. A company as large as AMD is utterly massive in comparison to the entire ASIC world of all cryptocurrency put together.

>>61243172
also it's my mistake for using ASIC proof, the correct term everyone actually uses it ASIC resistant. Nothing is ASIC proof.

>>61239328
Depends how much.
I bought one r9 280x 1.5 years ago that was used for mining. Before selling, the seller had them all refurbished.
Runs just fine and paid only 130 eurodollars. I am looking to upgrade again, but mining is driving up the pricing pretty bad. Considering waiting for a similar deal since the mining difficulty will probably be raised again in the near future.

>>61243238
You know "refurbished" just means dusting and wiping stuff down right?

No repair or maintenance actually goes into "refurbishment"

>>61243615
Reapplied thermal paste and a fan change. It's not much obviously but at least saves me the trouble of doing it.
As I said, it runs fine.

>>61243676
>Reapplied thermal paste and a fan change

Oh, thats good then.