1. Exactly. People just need to put 2+2 together. AMD and Nvidia run full simulations of chips before they are ever taped out. They know to a fairly high level of confidence how their chip will perform. They did not regress so extremely from Polaris.
Right. Years ago it was impossible. Nobody can say something is never going to be possible. People were predicting years ago that AMD was going to do it years ago.
Actually haven't we read many times they predicted the slight IPC regression? Bulldozer was supposed to be able to scale to much higher clocks than it did, at better efficiencies. This was what really gave Global Foundries such a bad reputation.. that first gen 32nm process was pretty bad to start. There was also AMD's bet that CPU's would move to be more integer computation intensive and so then shared the floating point unit between "cores".The problem is, you've basically just argued that Bulldozer didn't exist.
So we will probably get a preview @ SIGGRAPH or a reference card only launch. Custom cards coming in September. The only other plausible place they could launch RX Vega is Quakecon in late August. I mean AMD was pushing Quake Champions so....
EDIT: I'll say it again, SIGGRAPH would be a weird trade show to launch a gaming card. It's for professionals, industry, etc.
No mention of RX.
https://pro.radeon.com/en-us/amd-capsaicin-siggraph-2017/
Experience our fieriest additions to the cutting-edge Radeon™ “Vega” architecture.
Although the latest driver is out and doesn't provide any mention of Vega, is it possible there could be Vega improvements? I'm sure we'll get a ton of rumors soon enough.
Are all driver improvements always mentioned in changelog?
No mention of RX.
Radeon said:We can't wait to announce our new Vega products, including RX at this year's #SIGGRAPH
1. Exactly. People just need to put 2+2 together. AMD and Nvidia run full simulations of chips before they are ever taped out. They know to a fairly high level of confidence how their chip will perform. They did not regress so extremely from Polaris.
2. Eh.. not really. If Vega cores themselves are solid then AMD will still be the only company on the planet that can provide a high performance APU. 1024 down clocked Vega cores with a 4C zen cluster will be an absolute monster.
Interesting video.
The problem is, you've basically just argued that Bulldozer didn't exist.
Undervolting doesn't work in the current drivers/BIOS. And even when downclocked to 1000 MHz, Vega can use as much or more power as Fury X despite the node shrink. None of this makes any sense.
Add this to the news that RX Vega AIBs won't be getting BIOSes until early August, and the respin hypothesis is starting to look a bit more credible. If it's true, I wonder how Frontier Edition buyers will feel about having paid $1000-$1600 for trash silicon that can't be fixed. AMD made a big mistake by releasing this instead of just admitting that they missed the H1 2017 milestone.
What would pulling off a GF110 mean?Since the volume will be small, the honest thing for AMD to do, is offer a free replacement. If there will be actual fixed silicon that is.
As said many times, the big why did they launch it pops to mind once more, in the hopes that shareholders satisfaction is the only possible explanation. Although I am not so sure how happy a shareholder would be with all this criticism and doubt lurking around.
Who knows maybe they plan to pull off a GF110.
AMD certainly knows how to create a hype train.
Sounds like a really stupid thing to do. They have a couple of PCs running cherry picked games, lots of people, unfinished drivers. Yeah, what a great experience for sure.