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.
480 to 590 should be something like a 20-25% improvement.
Probably a dumb question... But based on the pictures of the Vega 7nm chip it seems it is going to have a whole lot more transistors than a Vega64. Since GCN is limited to 4096 SPs and Vega already has that, what is the rest of that silicon real estate going to be doing?
Raytracing cores
I thought that GCN was never actually limited to 4096 CUs, it was hampered by a limit of 4 Shader Engines and keeping everything utilised properly? And adding more SEs would require a chunky amount of work? But haven't AMD actually resolved that 4 SE block with Vega, but just didn't implement more because the die was already huge?
Assuming I'm not talking utter rubbish, it's possible that Vega 20 has actually moved past the 4096 "limit" because AMD were able to build in more SEs.
Vega20 is going to support FP64 better than Vega10 and also include some new instruction sets (or how you want to call them). Lisa also said Zen designers did such a good job to increase hz that they are going to help gpu designers. Its all going to need more transistors.Probably a dumb question... But based on the pictures of the Vega 7nm chip it seems it is going to have a whole lot more transistors than a Vega64. Since GCN is limited to 4096 SPs and Vega already has that, what is the rest of that silicon real estate going to be doing?
New mask is going to cost tens or hundrends of millions of dollars, also sadly market has shown AMD they rather buy inferior or more expensive products from Nvidia. AMD also is going to make more money from Compute, AI or DL than they ever would from gaming. There is reason why even Nvidia is having additional cores on their GPU and its not to run couple of rays 30fps 1080p.Absolute spot on with what you're saying.
Myself was saying this before anyone looked into the architecture of Vega properly, I also identified the problems with Fiji and that the front end of Fiji was broken.
The disappointing part of Vega is that it's just a tweaked fiji with over 50% wasted hardware resources towards a half assed compute card.
They should have spun up two lines,
A real instinct like they are on 7mm, and Vega gaming without the hbbc and around 380mm2 die size.
This is what I said back in August 2017,
https://forums.overclockers.co.uk/posts/31072073/
And pray tell how do you propose they "upset the apple cart"? You've seen their roadmaps, you've seen how long they've been in trouble, you've seen how long it's taken for Zen to properly take hold. Suddenly AMD are capable of causing immediate mayhem in the GPU space?
Freesync monitor and an AMD gpu, at the prices you can get them for, IS able to cause mayhem
About the same as a 2080TiI hate to think how much juice a massive 700+ mm² Vega die would chew up