Sign in to follow this  
Followers 0
IAmARock

Nvidia's Road to perdition

1 post in this topic

Well, here is the link: http://www.semiaccurate.com/forums/showthread.php?t=7499

Here is the whole post made:

 

 

I think for the last two years there's been a kind of unstated 'gentlemens agreement' in place between AMD and Nvidia, which is why the prices of the 7xxx line started so high and went down so slow. AMD had nothing to gain with a price war because they weren't positioned to substantially hurt Nvidia with one and were in desperate need of the increased cash flow.

They are positioned to very substantially hurt Nvidia with a price war now though, and will be far more able to absorb a short term profit loss. Hence the low low price points of $99, $199 and $299 for the boards released so far. Now it's time to sacrifice some short term profits for the larger stragetic objective - ripping Nvidia's consumer GPU throat out. Nvidia is already being driven to substantially lower their prices for their competing cards. As Mantle games start appearing, and there's looking to be a veritable river of them in the pipeline, Nvidia will be forced to lower their prices further and further to stay cost/performance competitive. 

This same dynamic will be on nitrous for the R9 290 and R9 290x, a space where far more informed buyers play. I'm guessing an initial 'hurt' price from AMD, $449-499 for the R9 290 and $549-599 for the R9 290x. Low enough to force Nvidia to slash $100 to $400 per board off their profit margins and low enough to keep buyers happy they are getting a bargain yet high enough to generate healthy profits. 

As Mantle kicks in on these boards, the informed crowd that buys them will force Nvidia to slash their prices still further. As more and more Mantle games come on line there will come a point Nvidia cannot sell their high end boards at a profit. 

As it looks like Maxwell won't be coming until Q4 2014, Nvidia is in for long hot summer of hurt. But will Maxwell save them? .. probably not. By then AMD can have a full line of GCN 2/3.0 cards @20nm out keeping the squeeze on Nvidia and Kaveri will have been lead piping Intel with it's HSA/hUMA/Mantle advantages. As Kaveri starts becoming the gaming CPU componenet of choice, the natural companion will be AMD GPUs. Kaveri 2.0 + DDR4 in 2015 will continue the assault on Nvidia and Intel and by then AMD discrete GPUs will essentially own the market. 

APU13 is likely to see a number of professional graphics players make public their HSA plans. Maybe all of them. HSA will only get more compelling over time. 2015 could well see a major shift to camp AMD. Additionally the professional players will be watching the situation in Nvidia's retail markets and if it starts becoming a massacre the handwriting will be on the wall. AMD/HSA is where the future lies and efforts to transition to AMD/HSA will intensity. 

1H 2014 sees an array of enhanced HSA Jaguar/GCN embed processors that will share the HSA/hUMA/Mantle optimization advantage on PC games. That will provide for highly playable games on tablets -> laptops not before possible at extremely competitive prices. Nvidia will lose a big chunk of it's embed/OEM market and what it retains will be far less profitable. 

Whether or not the rumored 2013 consumer ARMv8/GCN comes to pass, AMD is certain to have a line of ARM/HSA/GCN chips ready to go in 2015 with the same HSA/Mantle optimization advantages it has in the x86 market. This will be about the time Nvidia's Tegra division finally starts becoming profitable.

Execution will be an important factor. 

AMD has been working for years on an SOC mix-n-match multiple IP from multiple vendors model. It's the heart of their custom business unit. It's what they used for the PS4 and Xbox one SOCs. GCN was engineered to be highly modular and scalable. They have considerable experience successfully combining CPUs and GPUs on a single chip. 

Also note: SCEA President ""But we can pre-sell every unit we can manufacture and the good news, production yields have been phenomenal ...". AMD obviously has world class engineering -> fabrication chops to have yields at that point so early in the game. 

They are more than capable of hitting an ARM based SOC grand slam their first time at bat.

They will initially be targeting the high end Android gaming market where publishers can leverage HSA/hUMA/Mantle optimized Android ports to provide high end/hardcore gaming experiences on that platform. One might expect this to be accompanied by a release of joystick/trigger equipped phablets, tablets and purpose built handheld gaming machines. The specific market Nvidia is attempting to leverage to ARM SOC success.

They certainly have world class CPU architectural and engineering experience. They could have any number of highly customized ARM based SOCs in the pipeline for 2015 and beyond though their initial foray can be stock ARM as maximizing battery life will be far less important in the larger form factors they will initially target and their GCN GPU module will be their primary edge.

But as nodes shrink high end gaming will move to the smartphone market, and AMD's HSA/Mantle advantage will still apply, so they probably do have highly customized ARM/GCN SOCs in the pipeline. It's a very big market and that would get them a substantial chunk of it. 

Having an AMD based smartphone you could plug into a MOGA controller or purposed gaming keyboard/mouse and play on your computer monitor or big screen TV in graphics intense 'hardcore' mode would be pretty awesome and nobody will be better positioned than AMD to make that happen. EA's seemingly boundless enthusiasm for AMD/Mantle and the possibilities it's opening is starting to make more and more sense. 

AMD is also working with several partners on their own GRID style implementation where they will be able to leverage their coming ARM/Jaguar/GCN based Sea Micro servers and general server experience and knowledge. Another Nvidia market segment in which AMD's console based HSA/Mantle optimization advantage might give AMD the decisive edge

Considering their extreme cash flow problems and all the high level execs and broad swaths of people that were fired Rory Read appears to be pulling the rabbit out of the hat ... an almost incredible feat of creating a clear grand strategy, pinpoint prioritization of funds and manpower and world class management. 

Consoles -> PC gaming -> ARM gaming/Internet server market/Professional market -> HPC Market. 

AMD's console advantage goes on for 8 years at an absolute minimum and they are far and away the likely winner if there is a further generation of consoles, or a 'turbo' edition of the PS4 and Xbox One, all the while HSA/hUMA/Mantle optimized games continue to grow in quantity and quality. There is no light at the end of the tunnel for Nvidia. 

AMD is in the beginning stages of dominating Nvidia in every market they compete in and will doing so for the next several years and I'm not seeing squat Nvidia can do about it. Or how they survive it. 

On the road to perdition indeed.

There are parts I agree with, and parts I disagree with (crazy, right?) which I will go over later in a post/reply of my own.


Discuss please ^^

-Q

0

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  
Followers 0