1 of 1 people found this helpful
I would love to see a dedicated AMD powered mining-solution as well - not just for the potential financial benefit it could have for AMD, but to see AMD lead the way in more efficient solutions. This would align with AMD’s Steadfast Commitment to Energy and Environment
One thing that comes to mind however, is the current volatility of HBM2 in regards to availability; if that situation does not improve, delving into the crypto-market could be a potential disaster for AMD! So considering the risks of crypto vs value fluctuations, AMD could make a killing in the short term, but at the same time risk losing it all! From the miner's standpoint, they don't stand to lose much unless they're just now getting in the ballgame. For those who have been active in crypto since the beginning like you and I have - losing a 50 or so BTC investment wouldn't matter much - I don't see others willing to risk that!
The timing of your post couldn't be more perfect - the rumor-mill on the "Videocardz" website shows a listing of an Intel/Nvidia's supposed 8 card miner!
Just form a "An AMD Mining Division" and mine for the cryptocurrency instead of selling GPU's to anyone at all, if it is really that profitable.
I dont know what the profit on GPU's is but I would guess they have to aim for at least 30% gross profit per GPU sold including all operating expenses by selling them.
Maybe someone out there who knows more about Mining than me can take the time to write the business case/analsys for doing that, including the added profits made by the increasing cost of GPU's due to lack of supply amongst gamers and external miners looking for cards.
Just based off the thermal profile and energy efficiency of a single Vega GPU, I'm sure any 8 chip card would burn like a demon's heart. Not sure how you would keep such a card at a thermal equilibrium that would greatly reduce the lifespan of the chips.
I too would love to see AMD lead the way in efficiency. It is something they were at the forefront of when the HD 5870 released compared to NVidia's Fermi and Kepler GPUs. Polaris was the beginning of a return to that, after the R9 290/390 series smoldered. But Vega certainly is not. After a lot of hubbub about Vega continuing and expanding on the efficiency of Polaris, the result GPU is actually the opposite.
While it does perform fine, it's energy efficiency and thermal profile are a step backward from Polaris, and certainly behind NVidia's Pascal line. Due to that inefficiency, powering and cooling multi GPU configurations in Vega will be far more challenging than with Pascal.
However, it may be that, similar to Fiji, Vega downclocks without loosing much performance. The R9 Fury Nano operated at a 100W lower thermal profile than the Fury X, but at 90% of the performance. It will be interesting to see how the Vega Nano performs when it launches, and that should give us a better idea of the feasibility of creating multi GPU cards.
I wouldn't write Vega off just yet.
Faster HBM2, Same or lower GPU Clock, Some Undervolting, Optimized Drivers. Maybe DSBR. It could all add up to improve things lots.
With all that has happened this past year, you would think that AMD would be listening to the words in my post. Especially considering that they are now concerned about loosing money because of lack of demand.
I Told you so.
The cost in efficiency for continuing to validate crytocurrency will likely make it unsustainable in the long run. Probably not what you want to build your company around long term.