Read more.Quote:
Mike Rayfield is SVP and general manager, David Wang becomes SVP graphics engineering.
Printable View
Read more.Quote:
Mike Rayfield is SVP and general manager, David Wang becomes SVP graphics engineering.
Saw this elsewhere yesterday and wondered why it wasn't on Hexus yet...
Lets hope their first actions aren't to start making mining cards...
2 > 1?
1 > Raja?
:tomato:
There's a reasonable argument to be made that heading up RTG needed more than one person, given the circumstances. Splitting the general management from the technical and engineering management should allow a better focus in both areas. My only real concern is that both are responsible for 'strategy', and you really need the business strategy and technical strategy to be fully aligned if you're going to make successful products.Quote:
Originally Posted by Ozaron
Of course, the next couple of years is just going to be a case of delivering the existing roadmap as smoothly as possible - it'll be a while before we see the direction the new management will take RTG in....
Actually, let us hope that is exactly what they do, neuter gaming cards to be ineffective for mining, but great for gaming. Then make a specific mining card that is great for mining, but lousy for gaming.Quote:
Originally Posted by shaithis
You can't make "Gaming" cards not good at mining, you need that otherwise it'd be terrible for gaming, ASIC's will eventually replace ordinary cards.Quote:
Originally Posted by IotaActually, let us hope that is exactly what they do, neuter gaming cards to be ineffective for mining, but great for gaming. Then make a specific mining card that is great for mining, but lousy for gaming.Quote:
Originally Posted by shaithis
I'm sure they could do something to the GPU itself to prevent it being used for mining if they chose to.Quote:
Originally Posted by Fury559
In the AT article about this,it seems semi-custom is now part of RTG and RTG will also get more funding too:
https://www.anandtech.com/show/12363...new-leadership
I don't see what. The thing that makes a card good for gaming is the thing that also makes it good for mining; being able to process lots of parallel vector arithmetic. That's why pretty much all GPUs have been hit by mining demand.Quote:
Originally Posted by Iota
I'd much rather see AMD just make as good GPUs as possible than split their already stretched engineering and financial resources just to cater for the mining community. It's [b]expensive[/i] to produce dedicated silicon...
Firmware tweak?Quote:
Originally Posted by scaryjim
Why would AMD conceive "gimping" (for a lack of a better word) their GPU technology so that a specific use case that generates them a lot of revenue can no longer be done.
Frankly they did the best thing by using lower binned silicon and removed the output components and classed them as "mining cards".
Miners won't buy mining cards. Part of the attraction of GPUs is that they can be sold on to gamers afterwards, recouping some of the initial investment. That's not possible with mining cards.Quote:
Originally Posted by Tabbykatze
The news story about PowerVR's new GPU design did trigger one thought - since perf/watt is potentially the key factor if you could make a GPU that produce 1/4 of the hash rate of an RX550 but at /10th the power. No real interest to gamers, but you slap four of them and a PCIe switch on an x1 card to sell to miners.
That's only true to the extent that the mining cards are literally just rehashed full GPU designs. If you could provide the same hashing performance at either a fraction of the price or a fraction of the power draw (and ideally, both), the resale value would become less important. For many of the popular mining cards running them 24/7 for 2 years costs as much in electricity as buying the card up front (not to mention the cost of running the underlying server). It's almost unheard of to get much more than half your initial investment back on a second hand card. If an alternative card comes along, at the same price, but costs less than half as much in electricity over its lifespan, it's going to be attractive. If a card comes along that uses the same power but costs half as much up-front, it's going to be attractive. That's why people eventually moved to ASICs for Bitcoin - they offered a better performance:cost ratio than slogging away with GPUs. All it needs is for someone to hit that market spot. Cheaper upfront, or much lower power. Either will do.Quote:
Originally Posted by spacein_vader
As lots are buying ASICs, I doubt that is the case. Pro miners only care about hash rate / wattQuote:
Originally Posted by spacein_vader