• Competitor rules

    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.

Exclusive: The AMD Inside Story, Navi GPU Roadmap And The Cost Of Zen To Gamers

Even if it's only a shrink a Polaris chip would be a great mid-range contender for 1080p gaming.

It would be great timing to sew up the "low-end" before Nvidia release the 2030/2050/2060. Can't see it happening but equally don't believe that AMD will have no products at all until next year.

Edit:
Oems normally want to see a "new" product for back to school or christmas even if it is just a rebrand
 
Earliest I can see them releasing a new 7nm GPU is probably Q1 2019, think the only 7nm products they'll release in 2018 are low volume HPC market ones.
 
Consumer.

TSMC is in mass production of Kirin 980 7nm chip that isn't that small (est. ~90 sq. mm?) right now and has been in for at least a couple of months...
 
They must be desperate to replace vega, that HBM is probably costing them a fortune. I reckon they'll release this as soon as they possibly can, I'm still not expecting it to set the world alight though.
 
Don't think we'll see consumer 7nm this year, there's a reason 7nm HVM is starting with small mobile SoCs. Apple will most likely eat most of TSMC's production capacity anyway.
But I'd love to be wrong since the GPU market needs some competition.
 
Don't think we'll see consumer 7nm this year, there's a reason 7nm HVM is starting with small mobile SoCs. Apple will most likely eat most of TSMC's production capacity anyway.
But I'd love to be wrong since the GPU market needs some competition.
Margins for the PRO stuff are so much greater. It just wouldn't make sense to split your first production between PRO and consumer.
 
Can't AMD simply purchase the needed equipment, install it somewhere in the TSMC's factories and use it as much as they wish, without waiting for competitors to release some spare machines?! :confused:
 
Can't AMD simply purchase the needed equipment, install it somewhere in the TSMC's factories and use it as much as they wish, without waiting for competitors to release some spare machines?! :confused:
Not with costs running into tens of $billions for a single fab. Crazy expensive.

Hence why every fab operator needs multiple customers just to make any ROI at all; why GloFo backed out of 7nm completely.
 
Not with costs running into tens of $billions for a single fab. Crazy expensive.

Hence why every fab operator needs multiple customers just to make any ROI at all; why GloFo backed out of 7nm completely.

GloFo didn't back out completely but put it on hold without stating when they will start using it. If they don't go to 7nm and lower, then their PC business will be dead soon.
Of course, they can manufacture some less important things on older nodes but effectively, they will be out of the high performance computing.
 
Can't AMD simply purchase the needed equipment, install it somewhere in the TSMC's factories and use it as much as they wish, without waiting for competitors to release some spare machines?! :confused:

No a fab is built around the production equipment it isn't like you have a room with a bit of equipment in one corner, etc. AMD would have to put up the money for a complete end to end production installation.

For instance https://www.youtube.com/watch?v=4Q_n4vdyZzc

EDIT: Video has a few things wrong but it shows the infrastructure.
 
Last edited:
These sort of fabs are probably the most expensive manufacturing lines to setup, mega bucks.

In other news, had another one of those 'wut?' moments checking my portfolio this morning...AMD off like a rocket...again.
 
GloFo didn't back out completely but put it on hold without stating when they will start using it. If they don't go to 7nm and lower, then their PC business will be dead soon.
Of course, they can manufacture some less important things on older nodes but effectively, they will be out of the high performance computing.
100% of other fabs that did what GloFo have done never competed again at the leading edge.

GloFo also said that the leading edge represented something like <25% of their potential business (might have been less), so they felt confident they could drop it and concentrate on older processes.
 
These sort of fabs are probably the most expensive manufacturing lines to setup, mega bucks.

In other news, had another one of those 'wut?' moments checking my portfolio this morning...AMD off like a rocket...again.


They also take years to build.

AMD basically did build their fab with the investments in Global Foundaries. This almost bankrupted them.
 
They also take years to build.

AMD basically did build their fab with the investments in Global Foundaries. This almost bankrupted them.

I thought they were AMD's fabs and they spun them off, creating GloFo with heavy investment from Abu Dhabi who eventual took over completely in 2012 when AMD sold it's remaining stake.
 
They also take years to build.

AMD basically did build their fab with the investments in Global Foundaries. This almost bankrupted them.
That's a bit misleading, AMD did pour lots of money in fabrication R&D in the late 90's and early 2000's which caused them to run at loss for a number of years at time when they have a good technological advantage over Intel in chip design but that was to close the gap Intel had on AMD in terms of process manufacturing. The last process AMD developed when that part of the business was fully owned by AMD was 45nm, when AMD sold that part of the business in 2009 a part of the deal was Global Foundries had to take on a large chunk of the structural debt that AMD had on it's books which was paid by investors from the UAE. It was this along with the settlement AMD got from Intel and the console business that stopped AMD from going under.
 
Back
Top Bottom