Customers are customers - they can't afford to be picky, especially if Vega isn't quite all it was hyped to be. If AMD can't meet demand, they should increase production. Mining isn't as likely to crash like last time since so many coins use ASIC-resistant algorithms now.
beside double price of pretty much all GPUs on market?Only 1080TI is here for like + 30% of official MSRP.Exactly. I don't understand the hate for mining
That is the fault of AMD and Nvidia for having inadequate supply. Mining isn't just a fad anymore that's destined to burn out after a few months of popularity, now that a large and growing proportion of mineable cryptocurrency is specifically designed to prevent ASIC miners from being developed. This isn't a new revelation - it's been public knowledge for years that the latest and greatest cryptos were headed this way. AMD and Nvidia should have paid better attention.beside double price of pretty much all GPU on market?
Yeah why is so much hate on miners they only double prices on all GPUs that nothing
beside double price of pretty much all GPUs on market?Only 1080TI is here for like + 30% of official MSRP.
Yeah why is so much hate on miners they only double prices on all GPUs.Thats nothing
Customers are customers - they can't afford to be picky, especially if Vega isn't quite all it was hyped to be. If AMD can't meet demand, they should increase production. Mining isn't as likely to crash like last time since so many coins use ASIC-resistant algorithms now.
Supply and demand. It's NVIDIA and AMD's problem for not understanding and responding properly to their market, not the miners or the GPU distributors fault.
That's right, I had heard Vega was great for mining Monero. If it can in fact be undervolted significantly then it may not be too bad at ether or other currencies either.Vega is apparently great for mining XMR. With some tuning I'll bet Claymore can get some serious performance out of it for something.
If Nvidia and AMD understood the market, they'd mine the hell out of crypto themselves with custom ASICs linked to fat memory controllers, have their own in-house mining software, and rake in billions just off that. We'd all be pushed to the side. Why they ever let some ratty-arsed ASIC producers in China beat them to the punch on BTC is beyond me.
If Nvidia and AMD understood the market, they'd mine the hell out of crypto themselves with custom ASICs linked to fat memory controllers, have their own in-house mining software, and rake in billions just off that. We'd all be pushed to the side. Why they ever let some ratty-arsed ASIC producers in China beat them to the punch on BTC is beyond me.
Vega is apparently great for mining XMR. With some tuning I'll bet Claymore can get some serious performance out of it for something.
If Nvidia and AMD understood the market, they'd mine the hell out of crypto themselves with custom ASICs linked to fat memory controllers, have their own in-house mining software, and rake in billions just off that. We'd all be pushed to the side. Why they ever let some ratty-arsed ASIC producers in China beat them to the punch on BTC is beyond me.
That is the fault of AMD and Nvidia for having inadequate supply. Mining isn't just a fad anymore that's destined to burn out after a few months of popularity, now that a large and growing proportion of mineable cryptocurrency is specifically designed to prevent ASIC miners from being developed. This isn't a new revelation - it's been public knowledge for years that the latest and greatest cryptos were headed this way. AMD and Nvidia should have paid better attention.
Supply and demand. It's NVIDIA and AMD's problem for not understanding and responding properly to their market, not the miners or the GPU distributors fault.
If Nvidia and AMD understood the market, they'd mine the hell out of crypto themselves with custom ASICs linked to fat memory controllers, have their own in-house mining software, and rake in billions just off that. We'd all be pushed to the side. Why they ever let some ratty-arsed ASIC producers in China beat them to the punch on BTC is beyond me.
it doesnt work that way in business.
even assuming crypto is a marginally steady market segment, things like ETH ProofOfStake means that segment is likely a dead end. the whole point of POS is to eliminate the wasted electricity from gpu mining. any new gpu-centric alt-coin or ICO will be utterly unproven and require years to demonstrate viability and the confidence of speculators.
Bollocks, various ASIC manufacturers smelled blood in the water and were generally able to prototype and crank out BTC ASICs in 3-6 months, tops. They'd build the things quickly, give themselves something like a 3-month lead time to do their own mining, and then sell to the pre-orders and general market well into the development cycle of their next ASIC revision.
AMD had every advantage they needed to beat those people to the punch and/or just move on to another coin where they knew the BTC/LTC ASIC-builders could offer little competition. All they had to do was take existing memory controllers and cut down the GPU circuitry or just disable bits that a code profiler said wasn't in use. Huge money, minimal effort (relatively speaking).
It's not like we're talking about producing an entirely new GPU or other IC design here.
PoW has been in use since what, 2009 for BTC? Even if they had no confidence in the value of the coins themselves, they could have sold hashrate to people which is what firms are already doing.
it doesnt work that way in business.
the timelines and projected sales are already set in years before. the capital set aside for the wafers much less scheduling time at the fabs is not something that can be done without huge lead up time(at least not without overpaying for priority.) even if amd/nv saw sustained demand(ie over a year) they couldnt just ramp up the output. if they do set up another run at the time it could be 6 months before it hits retailers, and any cryptocurrency crash in that time could easily cause a sell off glut of used cards and worse tons of new inventory on the shelves that will never sell.
and even if either gpu maker wanted to commit to adding miners to their long term plans, there is no way to predict the crypto market fluctuations. the main spikes in mining are typically a result from a spike in coin market cap from speculators. the first bitcoin spike to $1000 was a result of greece/eu debt crisis and changes to banking policy in china. people looked to park their assets somewhere other than banks and btc was happenstance. eth spikes and crashes are equally the result of market manipulators, fomo speculators, and hacks that diminish confidence. There is NO way to RELIABLY predict these things.
if amd/nv did bump production numbers based on unprovable CRcurrency assumptions or mere speculation of a bull market and the market crashed causing financial loss, then amd/nv would be liable to the shareholders(and more likely the shareholder's lawyers). no CEO is going on the witness stand and stating that they risked millions of dollars of company assets on an unproven/un-forecastable market segment.
even assuming crypto is a marginally steady market segment, things like ETH ProofOfStake means that segment is likely a dead end. the whole point of POS is to eliminate the wasted electricity from gpu mining. any new gpu-centric alt-coin or ICO will be utterly unproven and require years to demonstrate viability and the confidence of speculators.
That's right, I had heard Vega was great for mining Monero. If it can in fact be undervolted significantly then it may not be too bad at ether or other currencies either.
Just that report I read in one of the Vega threads of someone dropping the voltage of his Vega FE by more than 100mV and it causing the card to stop throttling and stay at 1600 MHz stable.Nice. Do you have any references on this? I'm curious to know how Vega compares to Fiji when undervolted etc.
The difference is that private companies (especially smaller ones) can be quite a bit more agile. I have worked for some private and public corporations (including some Very Large™ ones) and there is a marked difference in what you can do, and how you do things. On average, there is significantly less freedom to take large risks in a public company. Not to mention being subject to policies 123.45, 234.56, 345.67, etc., in the off-chance that your project is approved. At a certain (more agile) private company, I could directly approach the CEO or one of the Executive VPs with an idea, get it approved, and even hire people to pursue that idea to fruition. I could purchase anything (within reason) as long as there was sufficient justification for it. No annual budget.
Just that report I read in one of the Vega threads of someone dropping the voltage of his Vega FE by more than 100mV and it causing the card to stop throttling and stay at 1600 MHz stable.
The difference is that private companies (especially smaller ones) can be quite a bit more agile.
Nice. Do you have any references on this? I'm curious to know how Vega compares to Fiji when undervolted etc.
To all those wondering how the GPU makers could actually use mining to their advantage, the bolded is about the only thing which really could do it. As many of the ASIC resistant algorithms are very depend on memory speeds, an actual mining card could use that to it's advantage so a tailored Polaris card could, for example, run ETH at 40+ while regular cards max out at 28 or so.a real mining card should have better cooling (for memory and vrms as well as the gpu), better vrms (with more power phases) and really tweaked memory timings. maybe even special higher speed memory. put in a dual bios, one for higher power efficiency with lower hashrate, one with balls out settings that max hash and power be damned. THAT would be a mining card.
To all those wondering how the GPU makers could actually use mining to their advantage, the bolded is about the only thing which really could do it. As many of the ASIC resistant algorithms are very depend on memory speeds, an actual mining card could use that to it's advantage so a tailored Polaris card could, for example, run ETH at 40+ while regular cards max out at 28 or so.
Yup, put cards like that on store shelves for better $/hash and hashes/watt, and the mining demand for gaming cards will dry up, returning their prices to normal levels.Exactly. Not a new GPU whatsoever, just more memory bandwidth with timings that work best for mining. This is an extremely simple card for AMD or NVIDIA to build which requires next to no additional work to build.
Yup, put cards like that on store shelves for better $/hash and hashes/watt, and the mining demand for gaming cards will dry up, returning their prices to normal levels.
If Nvidia and AMD already aren't providing enough supply, how will it help? Also, once the miners buy the majority of the mining-oriented cards and they're in short supply, they'll just turn back to the normal gaming cards... especially if the difference isn't that much.
Honestly, what'll likely happen is that the return won't be enough anymore to make buying new hardware worth it, and most people will stop buying more or stop buying as much. The ROI is already quite long compared to only a month or two ago.
Are AMD and NVIDIA running at 100% capacity to meet demand? We honestly don't know but I think it's safe to say they weren't a few months ago despite the demand still being high.