Another thing that intrigues me is Polaris in 2018. Bigger Polaris implementation, or continuing with Radeon 400, 500, or even 600 series refresh with 2304 cores?
asks in a Vega thread; Question for a different day.
One of my biggest gripes with Vega isn't it's performance now , but that it will have to compete up until the entire year of 2018. That means that even against GTX 3xxx series. That's ridiculous!
If you don't have money and people, what can you do? At the financial analyst day Raja mentioned that till 2015 the AMD management thought GPUs are a shrinking market. Maxwell probably showed them that it's not the case. But till that time AMD probably reduced the r&d in gpus massively and thought that professional/datacenter are the only gpu markets where they'll be able to make money, so focused more on these parts. But reducing r&d won't work out if your rival is going on with full speed.
From 2015 on amd added more people and r&d to the graphics division, but this won't help vega much. Vega was only 1 year from tapeout, so the people could only help in implementing the architecture, but not change it much anymore. Vegas architecture was build at the time in which amd spend the lowest amout on gpu r&d and they added a ton of features. But unfortunately features can be implemented in efficient and non-efficient ways and when you're restricted with way too low ressources the latter is quite probable.
More proof of this is that they put a LOT of R&D into Ryzen. Which is going to pay off, but those resources had to come from somewhere. And its making more sense that it came at the expense of the Radeon devision.
If the speculation about the timing of Vega tapeout and the previous management design path is true, then Raja was more or less brought on board too late to address much of Polaris or even Vega, to truly "fix" whatever might be inherently wrong with it. I think it has been said before, many times, that Navi would be the first true Raja team design. Hell, it's entirely possible that Vega would have been an even greater disaster were it not for Raja's team eking out whatever improvements they could. Or not, no one knows (publicly, anyway). AMD knows, though. I think it's best to let them decide what is best for AMD going forward and clearly, new management and overall design is light-years ahead of what they had prior to Su.
Stuka87 is correct. The drummer boy video is right up there with Joe Macri's FuryX "overclocker's dream" comment.
Sad.
Ok, I was remembering wrong, it's only 13 months late to the game. If it's an 1080 competitor (it will cost AMD significantly more to produce) then it's over a year late to the game. AMD will have to sell this thing at near cost to sell units plain and simple. Now if it could hash decent at least they'd have the miners but unless custom ROMs can work miracles I don't see it being popular with how power hungry and hot it runs.
A 1070 mines almost the same for less than half the power. 1070's can also run off a single eight pin. Absolutely no reason to consider Vega unless custom ROMs bump the hash rate up to at least 40mh/s.
I have friends that have been waiting for Vega for a solid year because they were going to buy cheaper freesync monitors, I bet money they change their tunes once this disaster is finally released. At least AMD released a game changing product with Ryzen.
The only way to spur proper competition is by buying the product right for you. Pity buying a company's product doesn't push them to improve. It rewards poor performance.Think positive. If you want competition in the GPU market you can do that by buying an AMD CPU, give them some money to develop something better, and use the money you saved buying AMD over Intel to buy the nVidia GPU that's overpriced because of lack of competition.
Or maybe I'm just being cynical.
I'm actually hoping Vega is bad at mining so I can buy one and sell my 480 for more than I paid for it.
The only way to spur proper competition is by buying the product right for you. Pity buying a company's product doesn't push them to improve. It rewards poor performance.
I'm aware. I just clarified what he said into something easier to understand. Wasn't going against him.I think you guys are saying the same thing unless an AMD CPU wouldn't fit your needs which is a very small use case IMO.
You are misreading that chart.I think it's important to note that AMD is absolutely not going after or did not expect to match the 1080Ti with RX Vega. The 1080ish performance that Vega is showing was telegraphed by the below slide IMO.
You are misreading that chart.
That is for Total Annual Market (TAM), nothing to do with performance at all.
That is saying, since it will launch mid-year, they are short by ~6 months. Next year, they will have the full 12 months.
Did you forget about the "poor volta" rhetoric?
They absolutely were saying Vega is so powerful, that it can even beat Volta, and thus Vega could beat 1080ti, Titain Xp, and all the way down the product line.
Did you forget about the "poor volta" rhetoric?
They absolutely were saying Vega is so powerful, that it can even beat Volta, and thus Vega could beat 1080ti, Titain Xp, and all the way down the product line.
lmao. so much negativity. Can we have promises to do something humiliating from all the people so sure they know what rx vega is? if they are wrong of course.
Think positive. If you want competition in the GPU market you can do that by buying an AMD CPU, give them some money to develop something better, and use the money you saved buying AMD over Intel to buy the nVidia GPU that's overpriced because of lack of competition.
Or maybe I'm just being cynical.
I'm actually hoping Vega is bad at mining so I can buy one and sell my 480 for more than I paid for it.