I think that's an enforced radio silence as Raja got on his usual big marketing spiel (like he has done regularly with AMD products and they don't live up to expectations, i.e. Fury, Vega, R7) with all the odyssey stuff, marketing events etc. I think someone in Intel yanked a chokechain.
As someone looking to get a new card but not wanting to splurge for a 3070 tier card - I do wonder if the used market will ever drop to match performance, because right now the 2080 seems to be going strong. Surely at some point they have to drop and at that point I'm unsure which of a used card and a 3060 ti would be a better deal.
On the subject of Intel and others... Y'know what might be a dark horse in a couple of years... Apple. I've been reading the latest Anandtech article on their M1 chip (actual one, not preview) and they have a seriously good foundation. They have an integrated GPU which pretty much nips at the heels of the low end discrete. And all in a decent power envelope.
Or any version outside one that supports their own 'proprietary' metal framework for that matter.
While the m1 seems to perform ok in tests (beats a 1050ti etc) the comparisons being made are versus things that are apparently in emulation, and as such slower, from what I'm seeing. I wonder how well it would actually perform in opengl (basically being killed off on mac os-x) or a more cross platform graphics engine like unreal.
M1 chip integrates DRAM on CPU though. I'm not sure I'd like to be stuck at 16 gig max ram for a serious product cos that's all you're getting currently with an m1 chip. No upgrade path ever for ram. Also they are comparing an octa-core cpu with a quad-core from Intel. And realistically in the Air that was actually a 2c4t chip
Old puter - still good enuff till I save some pennies!
1:1 price VS US market,,,,, i wish i could be so lucky
Old puter - still good enuff till I save some pennies!
That Intel 14nm++++ is close to TSMC 10nm in specs, so we are looking at 2 nodes away from 5nm with from memory (fact check welcome) something like a 1.25 performance boost *if you stay at the same power usage level* between nodes. So compound improvement would be about 56%. That's significant, but I don't think detracts from M1 being a pretty good first outing and make me wonder what the heck M2 will be like.
There are currently 1 users browsing this thread. (0 members and 1 guests)