Rumors of AMD Polaris 30 coming out in October

Wich is a good move. AMD burned themself badly on hype.

It would make sense to release a “new” (= refresh)product before the next architecture launches at unknown date.

They certainly do, AMD can’t even compete with the 2070, three steps down from the halo product.

Technically, an Vega 56 at 210% power target and 1.7GHz core does compete…

2 Likes

That video got me thinking that 7nm vega should be able to come somewhat near a 2080…

but then 7nm nvidia is going to stomp all over that anyways…

heres hoping that amd have more than a process shrink up their sleeves

1 Like

RTX2080 die size is 575mm2 at 12nm; Vega64 is 510mm2 at 14nm. A die shrink would easily allow AMD to beat the RTX2080 and even the RTX2080ti, because Nvidia devoted so much of its silicon budget to ray-tracing. And ray-tracing has, at least for now, zero value.

And that’s what AMD should do. Take advantage of a process shrink to stuff as much traditional rasterizing performance in there as possible, as cheaply as possible, and obliterate Nvidia for games you can actually buy and play.

Then in the background, let R&D mess around with ray-tracing. Maybe it’ll be a player in 2020/2021 or so.

4 Likes

See, i did think of that but I can easily see sheep just saying

“but it doesnt do raytracing!”

“look at how badly it performs in this ‘benchmark’ which shows how future proof the gpu is”

and just carry on buying nvidia

I am really hoping you are right.

If AMD released a Vega 84 at $499 (roughly 1080ti/2080 rasterizing performance) and a Vega 108 (roughly 2080ti) at $699, they’d knock Nvidia right into the toilet. They would completely dominate the entire market.

Performance per cost, yes.
Going by the sad tradition, nothing would change.

2 Likes

Nah. Fanboys are a myth. When it comes down to it, people buy what’s fast and cheap. Look at Ryzen.

Performance like that would get me to buy one… I plan on getting a Ryzen 2, new motherboard and probably at least 3600 RAM to replace my current Ryzen 1 setup. Card… new monitor, sell the old shit and I would set for 5 years.

Funny thing is when Ryzen came out a good amount of the people I play games with jumped to Intel/NVidia. Over the last few months they have been complaining about something I have been telling them the whole time to keep in mind when doing a build. If one some or all of the parts fail or act of god happens they are out the cost to replace or repair it, or to factor in the cost of upgrades. Didn’t really sink in the cost of G-Sync vs FreeSync with them, but it has now. Funny how paranoid people can become when they go from around a $300 card to an $800 one, so then they go from 80+ bronze to Platinum PSU…

Yes and no.

Every time AMD has had more than dominant GPU hardware they’ve still struggled to get up to 50% market share, rather than attain a command lead as their hardware superiority at the time should have resulted in.

AMD need to get their GPU marketing shit together.

Their hardware is fine for what it is. But they can’t market their way out of a wet paper bag.

The thing is dominance takes years to build and many factors.

Even if you have the best card for 1 or even 2 years, at the best price, etc, most people don’t upgrade that often. You have to hit people when they upgrade over time, so you have to have something great for many years straight to overcome a major gap in market share.

On top of that, there is a factor of “well, last time I upgraded I knew this would be better, so I’ll go with something similar” to deal with.

Possibly the bigger factor though is dealing with the systems market. If you don’t manage to get enough deals with systems makers to push your card on enough systems, you are going to loose out on a huge chunk of the market. Many people just buy pre-built systems, and when they see 9/10 options as nvidia, they’re likely to get nvidia close to 9/10 times. There’s a bit of a feedback loop there too, as systems makers are more likely to put the card in that they already know sells and have a relationship with and see everyone is buying, which allows that status to continue.

1 Like

AMD/ATi had several dominant cards, almost as many as Nvidia. However marketing failed to capitalise, and management failed to negotiate game deals/software like Gameworks.

AMD has locked up the Mac discrete GPU market, which is probably a larger portion of the pre-built-including-GPU market than Nvidia get a slice of.

Well, technically they had TressFX to compete with HairWorks, and ReLive up against ShadowPlay. But, yeah, most people don’t even know about those two features.

Thing is tressfx was open source and not crippled on nvidia hardware… unlike hairworks on AMD hardware.

Which dominant cards are you thinking of, and when were they dominant?

This is for the last 10 years or so of descrete cards (if you include integrated, which includes Intel and AMD APUs, it’s a bit different)

You can see there were points when certain cards came out that AMD gains grounds, but there’s a lot to overcome and it needs to keep up. Right now, they’ve gained a lot of ground but are coming from a point of being very far behind 3 years ago.

But yea, marketing, business deals, etc matter. AMD’s Apple deals are probably pretty lucrative, but Apple actually has a pretty small share of the laptop/desktop market. It’s a good size chunk compared to other single companies, but remember there are many other companies in that market that are also big, and plenty of small companies (Dell, HP, Lenovo, etc).

If you were to talk about just all systems, Intel actually has by far the largest share of the graphics market due to their integrated graphics in so many systems, but that’s not exactly the same market.

AMD only really beat Nvidia at both price and performance at a couple points in the past 20 years or so. The 9700pro back in 2002, the 5850 in 2009 (the best GPU they ever released), and to a lesser extent the 7850/7950 in 2012. Polaris was a great product that launched at an amazing price, but it was a small chip and didn’t compete at the enthusiast level. Vega is not competitive.

That chart doesn’t go back to 2002, but you do see AMD catching up with Nvidia with the 5850. They just couldn’t do it for multiple generations in a row, so they lost their momentum. But my point stands; if AMD released an awesome product at a great price like they did in 2009 with the 5850, people would buy it.

I mean, yea, people would buy it. People are buying their stuff right now. 27% market share isn’t dominant, but that’s still a lot of people. I just think it’ll take more than 1 card or line to retake majority share, if for no other reason than that most people don’t upgrade every release.

I’m talking performance dominant, not sales. Thats the point.

AMD has had superior hardware several times in the past and did not gain the share they should have with that hardware.