Originally posted by Mike Frett
View Post
Announcement
Collapse
No announcement yet.
NVIDIA RTX 30 Series OpenCL / CUDA / OptiX Compute + Rendering Benchmarks
Collapse
X
-
Originally posted by Jumbotron View Post4: dGPUs are NOT for us anymore. They're for Big Data, Coin Mining, Streaming Game Services, Banking, etc. PC gaming is not where the money is. Gaming is now firmly in the hands of phones, consoles and streaming. All platforms where there are ZERO local PCIe slots. The Age of Cheap Performant dGPUs is over.
Putting my Mr. Cynical hat on that means the solution is to go back to low quality gaming, ie GPUs which are not suitable for mining due to low performance, low memory capacity or low memory bandwidth.
The open question is whether it's worth increasing production for products like that (eg Polaris cards with 4GB or less) given that (a) it would be 6 months minimum before any additional volume could appear and (b) those cards would still be competing for production capacity and shipping capacity with all of the higher end GPUs.
The only real solution I can see (other than a mining/cryptocurrency crash) is breaking the link between cryptocurrencies and proof-of-work on GPUs, either by switching more currencies to proof-of-stake or by finding some other way to make GPUs ineffective at mining while continuing to be effective for other GPU compute tasks.
There is a nuclear option which is breaking GPU compute in general, but I don't think anyone wants to go there.Last edited by bridgman; 16 April 2021, 11:29 AM.Test signature
- Likes 2
Comment
-
Originally posted by bridgman View PostUnfortunately it's going to be worse than that - anything that can mine will not be for you any more as long as the mining boom lasts. First they came for the PC dGPU boards, now they're starting to use dGPUs on laptops. Consoles will probably be next - AFAIK the only thing that keeps phones from being impacted is their limited RAM. APUs probably remain safe as long as they have limited memory bandwidth.
Putting my Mr. Cynical hat on that means the solution is to go back to low quality gaming, ie GPUs which are not suitable for mining due to low performance, low memory capacity or low memory bandwidth.
The open question is whether it's worth increasing production for products like that (eg Polaris cards with 4GB or less) given that (a) it would be 6 months minimum before any additional volume could appear and (b) those cards would still be competing for production capacity and shipping capacity with all of the higher end GPUs.
The only real solution I can see (other than a mining/cryptocurrency crash) is breaking the link between cryptocurrencies and proof-of-work on GPUs, either by switching more currencies to proof-of-stake or by finding some other way to make GPUs ineffective at mining while continuing to be effective for other GPU compute tasks.
There is a nuclear option which is breaking GPU compute in general, but I don't think anyone wants to go there.
First the Infinity Cache does a very nice Job in smear the ETH etherium people Alway they dont Like it.
Also Miners Need to Run 24/7 all day all night because of this they buy 3/5/7nm node Chips to save Energy
A Gamer Just don't Need this Kind of efficiency.
Build a 12nm 600-700mm² RDNA2 Card with 600-800watt Power consuming
A Gamer Like me who Play 1-3 hours a day Dont Care much About efficiency it will hurt me Less than a miner who Run it 24/7.
So If is Very simple to build a Card to smear all the ETH etherium people Alway
800watt dont Hurt a Gamer who Play 1 hour a day
But it will Ruin the Profit of the miner who Run it 24/7
Phantom circuit Sequence Reducer Dyslexia
Comment
Comment