More models and more choice is ultimately better.
You could continue to buy every other year.
The unnecessary need to have the best is what will be causing any unnecessary grief
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Blackwell gpus
- Thread starter scubes
- Start date
More models and more choice is ultimately better.
You could continue to buy every other year.
The unnecessary need to have the best is what will be causing any unnecessary grief
The unnecessary need to have the best is what will be causing any unnecessary grief
Lucky for me am not one them kinds of people i only have a 4090
Used to be 7 or 8 months in some intances back in the day for releases, though we weren't paying a grand plus back then..
when did it change to 2 year cycles ?
i can't remember for sure ,But thinking it was somewhere around the 9800gtx or 1080gtx time
The gamers nexus video from the other day had the launch cadence of both companies in a chart. I think the 2 year thing kicked in when they were stuck on 28nm for a few years. Around 2015 or so maybe?
Associate
This is fantastic news.
Soldato
Would not be a problem if they had good price to performance in the mid/higher end, give us a card with 80-90% performance of the highest end card but at a much cheaper price.
if they change to yearly cycle how will the process nodes support such a cycle? Usually advances are reliant not only on architecture design but also process node advances these go hand in had usually. We are already down to 3nm very soon 2 and then 1 ect, where the advancement in node technology is set to slow down barring some transistor material breakthrough or stackable designs with some miracle cooling throw in there. Up top now there has been steady process in manufacturing technology but from my understanding things are about to slow down. Does leather jacket man know something we don't? The only solution I can think of is something similar to what AMD did with the Ryzen CPU. Multi die GPU with a fast interconnect between them this will certainly work better with AI compute type work but gaming might have drawbacks like SLI did. I wonder if SLI will make a comeback again?
if you were siting at Jensen table and made this suggestion I have feeling he would slapped you. YOU WANT WHAT, how dare you, just for that ill increase prices even more next year
You mean you want something a 5080 for 2k with 85% of the performance of a 3k 5090
Soldato
He deserves a good booing on stage at some point thoughif you were siting at Jensen table and made this suggestion I have feeling he would slapped you. YOU WANT WHAT, how dare you, just for that ill increase prices even more next year
Prices can't just keep going up and up though, can they?
Caporegime
I want a 5080 for £899 that is around 4090 performance. Does not need 24GB. 16 is enough
Soldato
But when will the "is 16GB vram enough" thread turn up?I want a 5080 for £899 that is around 4090 performance. Does not need 24GB. 16 is enough
Caporegime
But when will the "is 16GB vram enough" thread turn up?
Bring it on. I have a build up of popcorn
But when will the "is 16GB vram enough" thread turn up?
Hopefully Nvidia GPUs come with more vram this time and someone can create the "is 24gb vram enough" thread
It'll be interesting to see what Nvidia do this time round though as vram is very important for these ai/ml workloads so I imagine they'll probably want to provide considerably more across the board or more likely they'll keep it for the 5090 and charge some ridiculous sum for it.....
crazy money as usal knowing nvidia..Hopefully Nvidia GPUs come with more vram this time and someone can create the "is 24gb vram enough" thread
It'll be interesting to see what Nvidia do this time round though as vram is very important for these ai/ml workloads so I imagine they'll probably want to provide considerably more across the board or more likely they'll keep it for the 5090 and charge some ridiculous sum for it.....
Yup probably, I've got a budget up to 1k and won't budge, in no real rush so if Nvidia can't meet that then I'll just hold onto 3080 for longer or look for a second hand 4090.
Last edited:
Soldato
- Joined
- 14 Aug 2009
- Posts
- 2,872
A 5090 coming a year after 4090 would not make the 4090 obsolete since it still remains a strong card.
You'll need software go push things quite far, which I doubt it will happen.
I don't see how the announcement that AI development getting a yearly cadence suggests in any way that the gaming segment will also get changed to 1 year. They don't use the same chips/designs.
edit: I've just gone and searched for further sources. Some articles only mention AI, but The Verge and Business Today claim all chips including gaming are changing.
edit: I've just gone and searched for further sources. Some articles only mention AI, but The Verge and Business Today claim all chips including gaming are changing.
Last edited: