To illustrate your point, my old GPU, a GTX 1080 from 2016 (basically ancient history - Obama was still president back then) remains a very useful for ML-applications today - and this isn’t even their oldest card that is still relevant for AI. This card was never meant for this, but thanks to Nvidia investing into CUDA and CUDA being useful for all sorts of non-gaming applications, the API became a natural first choice when ML tools that run on consumer hardware started to get developed.
My current GPU, an RTX 2080, is just two years younger and yet it’s so powerful (for everything I throw at it, including ML) that I won’t have to upgrade it for years to come.
fuckwit_mcbumcrumble@lemmy.dbzer0.com
on 20 Jun 03:31
collapse
Whatever makes RTX work is what accelerations a lot of AI tasks. I’d argue the 1080 is bordering on irrelevant if it wasn’t for the 8 gigs of ram to save it. The 2060 should be much faster despite for gaming being about in par.
itslilith@lemmy.blahaj.zone
on 19 Jun 10:45
nextcollapse
At the moment it feels like the proverb about gold rushs and shovels
What’s missing in the shield? I have a gen 1 and it still plays anything I throw at it. High bitrate HDR 4k video etc.
Scrath@lemmy.dbzer0.com
on 19 Jun 14:39
nextcollapse
I believe its missing h265 and av1 hardware support and while it probably has enough performance to handle those codecs in software, I wasn’t willing to drop more than 100 euros on a 5 year old device without hardware decoding for them
That’s untrue. I specifically bought it for h265 decoding. Which it does have in hardware. I play h265 content exclusively and never had an issue. Even 80~100mbps blue ray UHD rips. And because it has gigabit ethernet there’s never any buffering issues either from my NAS.
Even first gen ones from 2015 are still being used. I don’t think these die all that often. They will be obsolete at some point, but even this takes far longer than with other tech. As long as you make sure it doesn’t overheat, it should last for a while longer.
I mean, one of the core ideas behind these things is that these are highly capable devices that are receiving updates for several times as long as normal tech, so you can just keep using them for ages.
Apart from the very latest codecs, what else should they do that they aren’t already doing?
algorithmae@lemmy.sdf.org
on 19 Jun 13:33
nextcollapse
I wish it were as easy to make money on stock prices going down as it is to make money on stock prices going up
technocrit@lemmy.dbzer0.com
on 19 Jun 17:07
nextcollapse
Pretty sure the entire system is based around making the line go up regardless of everything else.
algorithmae@lemmy.sdf.org
on 19 Jun 17:59
collapse
I can’t reliably predict when line go up, but I have a pretty good idea of when line go down. Let me make money on that lol
pineapple_pizza@lemmy.dexlit.xyz
on 19 Jun 17:15
collapse
You could buy options
algorithmae@lemmy.sdf.org
on 19 Jun 17:58
collapse
I’m not employed by Nvidia though? Unless I’m misunderstanding
pineapple_pizza@lemmy.dexlit.xyz
on 19 Jun 19:30
collapse
threaded - newest
They were years ahead of the curve with AI hardware, and they’re well placed to benefit from the AI craze.
Regardless of whether a company’s AI product is useful, or profitable, they need lot of hardware to make it run.
To illustrate your point, my old GPU, a GTX 1080 from 2016 (basically ancient history - Obama was still president back then) remains a very useful for ML-applications today - and this isn’t even their oldest card that is still relevant for AI. This card was never meant for this, but thanks to Nvidia investing into CUDA and CUDA being useful for all sorts of non-gaming applications, the API became a natural first choice when ML tools that run on consumer hardware started to get developed.
My current GPU, an RTX 2080, is just two years younger and yet it’s so powerful (for everything I throw at it, including ML) that I won’t have to upgrade it for years to come.
Whatever makes RTX work is what accelerations a lot of AI tasks. I’d argue the 1080 is bordering on irrelevant if it wasn’t for the 8 gigs of ram to save it. The 2060 should be much faster despite for gaming being about in par.
At the moment it feels like the proverb about gold rushs and shovels
Except the gold is actually poop and the shovels require burning several trees per dig
And still no updated Shield TV. 😭
What’s missing in the shield? I have a gen 1 and it still plays anything I throw at it. High bitrate HDR 4k video etc.
I believe its missing h265 and av1 hardware support and while it probably has enough performance to handle those codecs in software, I wasn’t willing to drop more than 100 euros on a 5 year old device without hardware decoding for them
That’s untrue. I specifically bought it for h265 decoding. Which it does have in hardware. I play h265 content exclusively and never had an issue. Even 80~100mbps blue ray UHD rips. And because it has gigabit ethernet there’s never any buffering issues either from my NAS.
Ah, I guess I was only looking at AV1 support in that case. I only remembered it was missing something I wanted due to its age
I got most of my Plex library on h265 and there is no problem with my shield,
Nothing is missing - but mine’s 5 years old and I want to make sure it can be replaced before it finally dies.
Even first gen ones from 2015 are still being used. I don’t think these die all that often. They will be obsolete at some point, but even this takes far longer than with other tech. As long as you make sure it doesn’t overheat, it should last for a while longer.
Yeah, I was surprised there were still lots of people using the 2015 version. It’s a good sign so hopefully mine will be good for a long while still.
I mean, one of the core ideas behind these things is that these are highly capable devices that are receiving updates for several times as long as normal tech, so you can just keep using them for ages.
Apart from the very latest codecs, what else should they do that they aren’t already doing?
I wish it were as easy to make money on stock prices going down as it is to make money on stock prices going up
Pretty sure the entire system is based around making the line go up regardless of everything else.
I can’t reliably predict when line go up, but I have a pretty good idea of when line go down. Let me make money on that lol
You could buy options
I’m not employed by Nvidia though? Unless I’m misunderstanding
Anyone can trade options. See www.investopedia.com/terms/p/putoption.asp
This is very risky though, I wouldn’t recommend doing this without some training and research.
I’m Not a financial advisor
That goes back to my original point, “I wish it were as easy…”
<img alt="" src="https://lemmy.ml/pictrs/image/e1aca104-ac71-48b5-b846-9d4ef612e993.jpeg">