Although I personally agree cryptocurrencies have major appeal, we must be realistic here. Bitcoin is probably here to stay, etherium has its benefits, and maybe two or three other altcoins will survive in the long haul. The rest will die a quiet death.
Additionally banks and other big (groups of) companies are already actively developing their own blockchain technology, so the classic minable cryptocurrencies are already getting competition from the establishment.
So lets say 4 or 5 cryptocurrencies are here to stay, they all have a finite number of coins to mine and some (like BTC) are not interesting to mine with gpus.
So even if there are some cryptocurrencies to stay, the mining will decrease in the long run.
So if they believed the increased demand would be sustainable they would push foundries to invest in additional production capacity (which they havenât). The production roadmaps are planned so far in advance because foundries have to allocate production capacity for customers and donât like downtime.
There will always be another new coin⌠each cycle will convince less people, but people donât want stable, that want growth, and sticking with major coins limits that. Thus I see a good few more cycles to come, a few dozen coins pop up, ~3 get good attention, 1-2 crash of which die and the remaining on stays basically just as a public share to be traded, not a mined currency. Something like half the people who did well want a repeat of the results of rapid growth or mining, so they attract a few others each and a few dozen new cryptocurrenices pop upâŚ
Long run I agree, the chance of each new currency being stable long term drop with each on the market, but just as long after people thought the few chain based fast food giants where stable, bam subway or dominos shows up late, and is a huge hit due to some unique feature or unfair advantage.
As for drop in card sales, its economical to mine on cards long after payback times are not reasonable to buy more. Ie as long as power use is below earnings you mine, you only buy cards if you are getting earnings high enough to pay back the purchase cost and your estimation of risk. Also mining is a much lower portion of all card sales than you may expect, which is why you donât see many dedicated mining hardware. Stuff like a software and drivers are cheaper to roll out, and adding features for mining rarely reduces utility to anybody else (eg >8 card multiprocessing or a peak efficiency preset, not just regular and some factory OC).
Iâd go one step further and say because used cards (especially mining cards) have low used value, plus you typically totally loose the cost of risers, dedicated mining boards and whatever low end CPU you had, miners have a vested interest in new things to mine. It always seems like a waste of power to me though, as other non-gold or USD backed currencies maintain agreed value without wasting work.
These foundries are at top load for some time now as far as Iâve heard. Building new ones is rather expensive. Furthermore, the minerâs demand for GPU is very erratic and right now unpredictable. I think these manufacturers would very much like to be on the safe side and increase their output by smaller numbers.
Vega56 is the best power efficiency per flop of reasonable priced cards in the world, so I wouldnât count on their being much in the way of gains down clocking. Rather I think AMD has pushed power efficiency on the 56 and total output on the 64 to try differentiate them and hit more markets, so downvolting the 64 will be very interesting and rewarding, overclocking the 56 will also be interesting and potentially rewarding (while the 64 has very limited extra headroom).
Keep in mind the 175w vega56 is only clocked at 780mhz already vs the 945mhz on its identical process and only 14% more shaders, compute and texture units 300w brother. I see no reason why the vega56 wonât go right up to the same speed with just a software tweak (thus being only 14% lower performance than vega64), or that the vega64 wonât reach the same efficiency if you go down to the same speeds (thus only 200w) while still offering 14% more performance than the vega56. Still just speculation at this stage, but nothing Iâve seen tells me it shouldnât work this way.
Have to be a bios tweak, AMD has a power lock which will prevent it. The HBM2 will not clock as high either. Whether that is bios or because it is Hynix I donât know. It has been reported that 56 has Hynix HBM2.
So far the only way to order Vega56 in Germany seems to be Caseking, and they are listing it for 509,- Euro. Vega64 sells for 609,- for the XFX card at mindfactory, everything else is 649,- or more, liquid goes up to 800,- Euro.
huh? Caseking, MF and Alternate all had Vega56 at (roughly, CK being the most expensive) MSRP today. After about 30 mins the price went up by ~e100, ofc, but whatever.
OK, so it seems to be the same placebo amount of MSRP cards as it was with Vega64. Wasnât aware of that, thanks. Right now I canât even find a Vega56 listing at alternate.
Just out of curisoity: Whatâs the reason you guys bought the 56/64 and didnât wait for aftermarket cards? (Iâm exclding the water cooled one for obvious reasons^^)