AMD - The Master Plan - Part 1, History

Seems fairly accurate to me as to whats happened so far. Was is really a master plan? lol More of a plan of desperation to save the company from disappearing.
 
Master Plan = Dropping to 15-20% of the dedicated GPU market.

Good history lesson though.
 
Last edited:
Master Plan = Dropping to 15-20% of the dedicated GPU market.

Good history lesson though.

since AMD have patented Zen Based APUs with Integrated FPGAs and HBM2 Memory... i bet that is going to be the "master plan".
 
Their master plan was a fail so far. losing so much market share in the GPU and CPU market in a failed attempt to conquer your enemies at a later date with less resources is a hail mary.

PS adorned has posted a few times at B3D, he takes alot of liberties in his videos and really doesn't understand how many things work technically and business wise. Its great that his videos are dumbed down but talking about the "master plan" the master plan will only work if certain things happened which haven't already happened. Lets see if he will get into that......
 
Last edited:
Their master plan was a fail so far. losing so much market share in the GPU and CPU market in a failed attempt to conquer your enemies at a later date with less resources is a hail mary.

Whats the master plan then?
 
It was talked about for close to a year ago now, if you don't know what it is, how much of an AMD fan do you think you are?

I stated it when AMD tried to push mantle out and said it would never work. Also I knew about their "plan" well before that, over 2 or 3 years ago......
 
since AMD have patented Zen Based APUs with Integrated FPGAs and HBM2 Memory... i bet that is going to be the "master plan".

When that fails due to AMD being so far behind on CPU tech, surely the next product will be a hit!
 
Yep, they've been talking about "APUs" for many, many years now, since shortly after purchasing ATI in like 2007.

The consumer market collectively ignored them for just that reason-- their CPUs have simply not been competitive since conroe. I guess you could order that current-gen consoles rendered AMD's APU ambitions not completely laughable, though.
 
Last edited:
A lot of people in the forums seem to be anti-AMD. Without competition, I'm sure that the Intel CPUs, technology wise, and price wise would look a lot different. I was Intel until the AMD 2500+ Barton. That was a great chip. For the most part I've been Intel since then. I'd love to see a third relevant competitor step up.
 
A lot of people in the forums seem to be anti-AMD. Without competition, I'm sure that the Intel CPUs, technology wise, and price wise would look a lot different. I was Intel until the AMD 2500+ Barton. That was a great chip. For the most part I've been Intel since then. I'd love to see a third relevant competitor step up.

I think a lot of the "anti-AMD" crowd are referring to the CPUs. Those people have simply come to grips with knowing AMD is light years behind Intel CPUs right now. Many of us were AMD fans in the past, and would love to see them be competitive again. However reality sinks in that AMD has hyped and then failed delivered for over five years now.

On the GPU side of things, there are plenty of reasons to be hopeful. Though AMD relaunching the 290/290X as the "new" 390 series and referring to Fury as an "overclocking monster" didn't help things :p
 
I think a lot of the "anti-AMD" crowd are referring to the CPUs. Those people have simply come to grips with knowing AMD is light years behind Intel CPUs right now. Many of us were AMD fans in the past, and would love to see them be competitive again. However reality sinks in that AMD has hyped and then failed delivered for over five years now.

On the GPU side of things, there are plenty of reasons to be hopeful. Though AMD relaunching the 290/290X as the "new" 390 series and referring to Fury as an "overclocking monster" didn't help things :p

I'm hoping that the whole Fury and 390 lineup was just a testament to AMD's being stuck at 28nm. I'm looking forward to what BOTH camps can offer in terms of the newer process nodes.
 
I just want to know where AMD said they'd be integrating FPGAs, because unless someone/something modified my memory it was Intel who sealed the deal on Altera a few months back

I think a lot of the "anti-AMD" crowd are referring to the CPUs. Those people have simply come to grips with knowing AMD is light years behind Intel CPUs right now. Many of us were AMD fans in the past, and would love to see them be competitive again. However reality sinks in that AMD has hyped and then failed delivered for over five years now.

On the GPU side of things, there are plenty of reasons to be hopeful. Though AMD relaunching the 290/290X as the "new" 390 series and referring to Fury as an "overclocking monster" didn't help things :p

They claimed Fiji was/is an overclocking monster ?
 
Seems fairly accurate to me as to whats happened so far. Was is really a master plan? lol More of a plan of desperation to save the company from disappearing.

I agree! AMD doesn't appear to have a master plan other than being relevant and in business 5 years from now. Their leadership is full of more blunders than they are of actual innovation.

For years now AMD always appears to be 1, 2, 10 steps behind their competition.

Hopefully AMD can turn it around with DX12 if it gives them a performance lead and eventually larger market share; however, to say its Keystone Cops leadership is following a master plan is foolish and silly.
 
They claimed Fiji was/is an overclocking monster ?
One person at AMD said one time that the Fury X would be an "overclocker's dream". Although the cooling system on it certainly fit the bill, the card itself ultimately didn't.
 
Their master plan was to corner the entire graphics card API market with Mantle and their semi custom design wins. Now it would have worked if MS and Sony didn't say no we aren't going to do things that way. Ultimately this is what has happened by going down this road.

AMD pushed low level API's and this is what we have gotten so far. While accelerating the push of DX12 and accelerated Vulkan's development as well, by doing all this it has created a bigger divide on coding paths, not just IHV's coding paths but on a per generation on each IHV's gen's. By doing so it has now created a whole set of new problems (actually old as explained later) that high level API's were trying to avoid. The same problems that we saw with early console development back with the original PS and Sega Genesis.

A third problem (without the abstraction layers) is that it created a divide in the PC eco system with Windows, and this gives MS leverage to force people to upgrade to Win 10 and new hardware. I don't think this is a bad thing (because this will limit the amount of work a developer has to do) but it gives more power to Microsoft as Vulkan hasn't and probably won't gain much traction in the game development industry as we can see Valve's steam boxes aren't that much of the rage in PC gaming so far. As time continues this is less of a problem as it will help developers ignore development on systems prior to Windows 10. Why do you think MS wanted to stop supporting Win 7 on new processors? This was a power play, which ultimately they took back, but rest assured they will try something like this again a future date. Now add to this the plans they have for Win 10 store...... yeah there is a very common goal for Microsoft, by exclusively pushing out developers, publishers and in store purchases, giving them a upper hand over steam boxes, other consoles and the entire gaming industry.

In the short term this might give AMD a boost (larger boost if nV doesn't change their strategies), but in the long run the only one to gain from this is MS. As it has always been with the advent of Direct X. Hence why MS went with DX12 instead of using mantle out right. There would have been no way they would have given away that much "power", money, control. Sony did the same thing because they don't want to be in a situation where they would be at a disadvantage later to change vendors if they feel like they are getting squeezed by AMD as using Mantle as the API for PS4 would have given AMD more control over software and of course already has control of the hardware over Sony.

The whole "master plan" hung one two threads, the adoption of Mantle as a unified API and the console wins. One of those threads has already failed, which makes the master plan tenuous at best. The second is not in AMD's hands any more as nV changes their design strategies to better fit with what AMD's consoles have that will mitigate that. And we have seen after a generation launch of a new console games do tend to be better for that IHV's graphics cards but the other company does adapt and change their design philosophies.

PS Creig who was the person who said that, CTO? I think he would have know what their card is capable of lol.
 
Last edited:
They claimed Fiji was/is an overclocking monster ?

One person at AMD said one time that the Fury X would be an "overclocker's dream". Although the cooling system on it certainly fit the bill, the card itself ultimately didn't.

“You’ll be able to overclock this thing like no tomorrow,” AMD CTO Joe Macri said at the card's unveiling. “This is an overclocker’s dream.”
 
“You’ll be able to overclock this thing like no tomorrow,” AMD CTO Joe Macri said at the card's unveiling. “This is an overclocker’s dream.”

Yes overclock like there's no tomorrow, 500W here we come ! xD

Yeah I guess if an overclocker's dream involves melting plastic this makes a lot of sense

On a more serious note, had AMD launched 300 series earlier, how much higher do you think nvidia would have clocked maxwell ?

Maxwell actually is an overclocker's dream
 
yeah he had high hopes! 500W?! isn't the rad basically an h80i?
only way ive seen to get good or any results is with modded bios.
 
yeah he had high hopes! 500W?! isn't the rad basically an h80i?
only way ive seen to get good or any results is with modded bios.

I was actually thinking about the air-cooled Fury when I made that remark, but yeah... guy in AotS thread actually has highest clocked Fury X i've seen, I think it was 1200mhz
 
yeah he had high hopes! 500W?! isn't the rad basically an h80i?
The cooling system on the Fury X is similar to the AIO that came on the 295X2. If it can cool two Hawaii GPUs, I'm sure it can handle an overclocked Fiji.
 
Its not how much it can cool, the architecture itself is not made for overclocking. If we look at the Nano, and look at a 10% overclock, the power consumption jumps up 30%, Once ya start seeing that, you either have to keep the chip really cool to keep the leakage levels down otherwise you run into a vicious cycle where no amount of additional power will over come the leakage. Its very close to that limit. Now with FuryX it was even worse where most FuryX's couldn't even be overclocked more than 10% without issues.
 
Its not how much it can cool, the architecture itself is not made for overclocking. If we look at the Nano, and look at a 10% overclock, the power consumption jumps up 30%, Once ya start seeing that, you either have to keep the chip really cool to keep the leakage levels down otherwise you run into a vicious cycle where no amount of additional power will over come the leakage. Its very close to that limit. Now with FuryX it was even worse where most FuryX's couldn't even be overclocked more than 10% without issues.

Dude if everything from AMD is bad or sux for you why in the hell are you typing here? Holy crap do they pay you at least?
 
Whats bad about it? It is what it is wantapple, to try to sway away from what I just stated, it just blind as you always are. Come on how many people have shown that Fury X is just not able to overclock without issues? They were late because they needed to do something to get thier performance up to where it was competitive with nV's offereings, they weren't expecting just a difference with the same node. nV was smarter they changed their architecture to accommodate the loss of a node drop. AMD was too far into their design cycle when Maxwell 1 was released.

When ever you see more than a quarter difference from either IHV from launching new products, one or the other company has issues, its pretty simple.

FX from nV
r600 from AMD
Fermi from nV (not too bad as they were still able to get performance higher but power consumption was very high)
and now Fiji

And lets not forget ATi with the 8500, which had poor drivers.

Lets take a step back even more with 3dFx, they too when ever they were late the product they had just wasn't up to snuff with what was on the market.

You think if a company has a great product and knows they can compete with something that just comes out, they will keep in under wraps for 9 months, just so their competitor can grab a huge chuck of marketshare?

Logically I think AMD wasn't expecting such a large increase in perf/watt for the Maxwell 2, it was even greater than Maxwell by quite a bit in the neighborhood of 75% higher at that point they were still in safe territory as they didn't see Titan X yet. Once Titan X came out, they were ok with that too, because of the Titan X price they could charge less and still have good margins as it would have slotted right in there above the gtx 980 and below Titan X. The 980ti was a wrench in the machine that totally messed AMD's plans.
 
Last edited:
it wasn't a conspiracy it was planned out, and it was well know, even AMD employees at Rage3d hinted about it 2 to 3 years back.

Even Carmack stated the same thing about a year to a year and half ago when talking about Mantle with Sweeny on stage at dev conference.

Just because people don't stay abreast with history and what people in the know have been saying doesn't change the fact it was a plan, a plan that in my view never would have worked because you have MS and Sony as power players who will never give away their control. It was quite foolish for AMD to think it would have worked. But then again at the time we saw how foolish their management was........
 
Sorry, but it's hard for me to believe a company like AMD would be so retarded to believe they could "corner the entire graphics card API market with Mantle" and base their planning, and the whole future, on that. Maybe I'm just naive, but that just sounds beyond stupid.
 
Even if they get a fraction of the plan to work, it is still a positive its better than nothing at all right? if you are going to go for something go for it. At the time AMD also had enough resources too. It wasn't until the maxwell line up was released and the delay of Fiji and the r3xx line did AMD start loosing major ground. If they were able to get those cards out in time, I think we would have seen a much different picture with marketshare numbers. Developers would be more inclined to focus on AMD optimizations paths and this would have slowed down the use of Gameworks etc.

There are many factors that created the situation AMD is in right now, but their plans haven't change from a graphics market point of view, which is a good thing, because that means their resource limits haven't affected them enough to change those plans at least not yet.
 
Last edited:
Back
Top