Cryptocurrencies.. whos in it?

[Tazzi]

Wow... talk about overpriced vega! Shiiiiiiiiit!

[vlad01]

lol yeah. Some of the issues with the price also has to do with AMD and the partners not being able to produce many. They had supply issues from get go, yields and clock stability/power issues etc...

All other factors included they went bananas on price.

[immortality]

My boys are waiting for the next gen GPU's to come out, the apparent trend seems to be that card prices are rising after release rather than dropping as the design ages and get superseded etc

And yes, supply and demand, currently supply isn't meeting demand so price increases are inevitable..

[vlad01]

Thats because they have the next gen but are withholding release and production due to the vast volumes they are shifting of the current get. This is nvidia strictly speaking.

there is 2 problems causing this, fanboys blindly buying anything nvidia without any research in the products and brand loyalty etc.. and 2ndly is there is no competition since RTG (AMD's graphic division) are at least 2 generations behind nvidia in graphics architecture.

nvidia have already the new gen like a year ago for the scientific, AI and professional market, but don't plan giving this to general consumers any time soon.

Basically just like what intel were doing except at least when nvidia did release a new gen it would be typically 30-40% faster than the previous gen, Intel on the other hand were so complacent they only uplift 5% on average gen to gen almost purely from lifting clock for the last........ fuck I don't know anymore its been that long, maybe 10 years already?


Once upon a time where competition was fierce between nvidia and radeon there were performance leaps at its peak of 180% gen to gen, yes that much. Remember geforce 5 (FX) series vs geforce 6 series? that was it. I have the 6800 Ultra in my retro PC as it was truly an iconic and historically significant GPU and quite expensive almost 15 years later. Around 150-200 bucks if you can find one.

[antus]

NVidia are smart. They always have the next 2 generations under development so if one fails to work as expected they dont have some insurance ready. And the're platforms overlap, so the bottom end cards really arnt much better from generation to generation - they're all crap. But it keeps people who dont do enough research updating.

Interesting what you say about the next gem being only scientific. What is the name of the chipset? I saw they changed the licensing on newer drivers to exclude using it on consumer GPUs in scientific applications. I know from the high end computing we have at work the consumer GPUs dont last when you smash them at 100% load for a couple of years, but the scientific versions which cost 4 times as much are built to higher physical quality and the reliability in those conditions is far greater. I expect they would have much lower yields on the next gen and want to sell them at 4x the price to scientific community only for as long as they can.

[vlad01]

Its volta, its been out around a year, while the general consumers still only get pascal, launched first half 2016

nvidia themselves said some time later last year they don't plan to release volta to gamers as of current plans, pretty vague info . Some rumors have been circulating that they plan to give us ampere which is the gen after that.

Either case we are waiting a long time, they don't have any reason to release anything new as their sales just keep growing regardless.

[immortality]

Its volta, its been out around a year, while the general consumers still only get pascal, launched first half 2016

nvidia themselves said some time later last year they don't plan to release volta to gamers as of current plans, pretty vague info . Some rumors have been circulating that they plan to give us ampere which is the gen after that.

Either case we are waiting a long time, they don't have any reason to release anything new as their sales just keep growing regardless.

My boy was saying the same thing, they intend to skip a generation but no idea where he got that from.

[vlad01]

Probably from leaks that the next geforce is ampere. Volta architecture is primarily for scientific application and AI as it s fundamentally based around tensor cores perfect for AI and stuff but those parts are completely useless for graphics. This is why volta has a ginormous die, packed full of the non graphics AI and computational circuity. That said the general graphics side of stuff has been proven faster than the current gen, but only marginally.

There is potential to produce a cut down graphics only version of volta and significantly increase the clock to suit. Hard to say if this will happen or not but its certainly possible. But so is the possibility they will skip it.

Volta is used in the tesla compute cards as well as self driving cars as the core processor, actually 2 of them. Funny thing was many years ago I was wondering if they would ever use GPUs in cars to handle large data input in parallel, mostly I envisioned like in F1 or similar with all the sensor inputs. I got the car part right but never thought it would be self driving AI applications.

[vlad01]

This is the pro-sumer compute card Titan V

The memory is actually built into the same package as the GPU. So there is actually 5 pieces of silicon on there with the mammoth GPU die in the middle.

Those would be extremely expensive to produce due to the sheer size and transistor count. Something like 21 billion of them.

I don't know if anyone has mined on these but I recon it would be the fastest by far, only down side is the cost of the card heavily outweighs the benefit at "cheap" 3k USD. The Tesla version of this card is like 10k so 3k is cheap by comparison

[immortality]

People might start stealing tesla' s just for those card soon. Better keep that info quiet. ...