The AMD Execution Thread [2007 - 2017]

Status
Not open for further replies.
http://wsw.com/webcast/rj87/amd/

Today's webcast at Raymond James...focused on leveraging IP...John Byrne Senior VP and CSO

From 95% PC market business model when Rory became CEO transition company to higher growth markets to over 50% as they exit 2015

Non-PC now 30%

Currently in acceleration phase of restructuring...focus on the "5 pillars" ultra low power, professional graphics, embedded, dense servers, semi-custom

Won consoles based on IP...x86 and graphics

Cycle 5-7 years....peak year 3...40 million units per year (???I thought 20???)....$60-$100 asp....

Semi custom opportunities to announce 2 more designs of $250-$500 million in 2014...revenue starting in 2015

Semi-custom not just console...thin client, other gaming opportunities, servers

Sticking to down 10% PC market...hedged slightly to 7-10%

Overall market has gone from 90 million units to 80 million units per quarter

Commercial decline low single digits...consumer PC down "double digits"

Desktop stable...laptop getting squeezed by tablet..."the challenge is in consumer notebook"

AMD 20% share consumer...largely "entry level"...Intel focal about taking that share from us...that's OK...we welcome that..APU strength in this entry level

A8-A10 mix shift moving up the stack...APUs strong

Commercial PC market...."platforms in 2014 are 3x from 2013"...China market share is 40%...up 2 points in a few quarters

Kabini in new socket in first half 2014

Bay Trail being aggressive...Kabini first to market..over 100 platforms Kabini/Temash....why?...superior graphics...better battery life...battle on x86 yes BUT APU resonates

ARM...18 months since SeaMicro buy...dense servers "one of 5 pillars of our strategy"...single thread = x86...dense = Google, Facebook, Amazon...lower cost, lower space, more compute...can AMD win against 8-9 competitors...leverage our experience in server market and new SeaMIcro IP to "be the leader in ARM 64 bit products"...Verizon

2014 is about marketing and design wins...that space by 2017 (ARM) will be over 20% of server market

GPU share to rise from 35% to over 50% "in the coming quarters"

Professional graphics...we are attacking...great hardware but sketchy software reputation...we are changing that perception...high gross margins...one of the "5 pillars"

GF and TSMC...we are on track to meet our 2013 WSA and pay $200 million in 2014 for breaking 2012 WSA.,..negotiating 2014 WSA now..."flexibility on consoles both GF and TSMC"

APUs at GF?...we are now building IP to be "consistently reusable"...i.e. easier to move products because 60% of IP is consistent from gen to gen

Deflected process node and FinFET discussion...next move to 20nm..."then we will get to FinFET"

Embedded is a $7 billion market with very "sticky revenue"...casino gaming...industrial controls...we are doubling down...design cycle 12-18 months...cycle is 7-8 years
 
Last edited by a moderator:
http://cc.talkpoint.com/bmoc001/121013a_dy/?entity=29_8EOTKV0

John Byrne at BMO Capital Wednesday December 11

AMD embedded share is currently 3-4%...Our pipeline products are applicable to $3 billion of the $7 billion market moving forward

AMD is the "graphics leader"...we specialize in visualization

ARM could take 20% of the total dense server market...Q1 2014 is the first fully integrated 64 bit ARM product based on SeaMicro..."We believe we will be the leader in this space"

Desktop 1/3...notebook 2/3 of 80 million units per quarter...Consumer PC down 13-15% in 2013

Mullins and Beema will do well...A8 and A10 big reliance on GPU power...Kaveri 47% GPU

Mantle is currently console first...then PC (a poster earlier in thread thought Mantle was PC only)

Reaffirmed moving to "over 50% in discrete graphics share in next few quarters based on R7 and R9"

Professional graphics is a $1 billion market...18 months ago began investing in this space...resulted in 5 consecutive quarters of growth....from "single digits to 19%"

Reaffirmed 2013 WSA...for third time in 2 weeks

The new AMD "you have to execute"

T
 
He is either accidentally or intentionally wrong. Mantle has no console presence (it would make little to no sense there).

You are correct...

http://gearnuke.com/dice-mantle-playstation-4-will-drive-frostbite-development-going-forward/

Microsoft openly hostile to Mantle...DX11

"AMD has provided the GPU for both the next-generation consoles this time. This means that PlayStation 4′s libraries are much more accessible and compatible with Mantle than Xbox One’s."
 
In order for Microsoft to be openly hostile to Mantle, it would need to be saying things or doing things to damage it.
Just doing things on its own for its own platform, which it would have been developing before Mantle was announced, isn't an open expression of hostility.

What in that link would indicate that?
 
"AMD has provided the GPU for both the next-generation consoles this time. This means that PlayStation 4′s libraries are much more accessible and compatible with Mantle than Xbox One’s."

This is another misguided view and a vacuous statement. What does it even mean? Mantle on consoles makes little to no sense, end of story. It is an interesting way for AMD to lever their console wins in their core market, by exposing an API that makes porting (arguably) easy. Conveniently, some of the facilities it exposes also tend to (colaterally) help weak CPUs, which is something that AMD is selling.
 
Why does Mantle on consoles make little sense? It sounds like you're making the assumption that they launch with very low level interfaces in the first place, which is not necessarily the case, especially in tight launch schedules. AFAIK, PS3, for example, launched with a relatively high level OpenGL like (if not OpenGL) API and its wasn't until a few years later that the more low level API became an option for developers.
 
AMD is the "graphics leader"

Is that so? AMD is no longer the leader in integrated graphics (that would be Intel Iris Pro), nor are they the leader in discrete graphics (that would be NVIDIA GTX 780 Ti), nor are they the leader in ultra-mobile graphics (that would be Qualcomm and ImgTech). AMD is currently the leader in console graphics, but their focus on console computing and away from ultra mobile computing over the last two years will hurt them to some extent over the next two years.

Reaffirmed moving to "over 50% in discrete graphics share in next few quarters based on R7 and R9"

And how exactly will that happen when R7/R9 are virtually the same architecture as HD 7xxx other than higher end R9 290? When the competitive landscape changes starting in March 2014 with the launch of GTX 8xx series, will AMD be able to keep increasing discrete graphics share? Does AMD intend to crater their gross margins in order to increase discrete graphics share? Time will tell.
 
Actually, the R9 290 architecture and feature set is already implemented in other products in the R7 Series as well.
 
Actually, the R9 290 architecture and feature set is already implemented in other products in the R7 Series as well.

HD 7730 has support for the same feature set as R7 250, including lack of support for TrueAudio/XDMA. They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

HD 7790 has support for the same feature set as R7 260X, including support for TrueAudio/XDMA. They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

HD 7870 does not have support for the same feature set as R9 270X (ie. does not support TrueAudio/XDMA). They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

HD 7970 has support for the same feature set as R9 280X, including lack of support for TrueAudio/XDMA. They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

R7 240, R7 250, R9 270X, R9 280X do not support TrueAudio/XDMA.

R9 and R7 series products (other than R9 290) are light refreshes at best compared to HD 7xxx.
 
Last edited by a moderator:
HD 7730 has support for the same feature set as R7 250, including lack of support for TrueAudio/XDMA. They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

HD 7790 has support for the same feature set as R7 260X, including support for TrueAudio/XDMA. They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

HD 7870 does not have support for the same feature set as R9 270X (ie. does not support TrueAudio/XDMA). They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

HD 7970 has support for the same feature set as R9 280X, including lack of support for TrueAudio/XDMA. They each have the exact same number of pixel shader execution units, texture filtering units, and ROP's.

R7 240, R7 250, R9 270X, R9 280X do not support TrueAudio/XDMA.

R9 and R7 series products (other than R9 290) are light refreshes at best compared to HD 7xxx.

Are you somehow implying that NVIDIA is better in this regard? You know that the GTX700 series is literally identical to the 600 series right?
 
In order for Microsoft to be openly hostile to Mantle, it would need to be saying things or doing things to damage it.
Just doing things on its own for its own platform, which it would have been developing before Mantle was announced, isn't an open expression of hostility.

What in that link would indicate that?

Sorry, that is my impression. I imagine Microsoft saying "DX11 is all we need thank you very much" and they appear to be avoiding any attachment to any vendor specific api. "Hostile" was the wrong choice of words perhaps. I would say "resistant" is better...
 
Is that so? AMD is no longer the leader in integrated graphics (that would be Intel Iris Pro), nor are they the leader in discrete graphics (that would be NVIDIA GTX 780 Ti), nor are they the leader in ultra-mobile graphics (that would be Qualcomm and ImgTech). AMD is currently the leader in console graphics, but their focus on console computing and away from ultra mobile computing over the last two years will hurt them to some extent over the next two years.



And how exactly will that happen when R7/R9 are virtually the same architecture as HD 7xxx other than higher end R9 290? When the competitive landscape changes starting in March 2014 with the launch of GTX 8xx series, will AMD be able to keep increasing discrete graphics share? Does AMD intend to crater their gross margins in order to increase discrete graphics share? Time will tell.

It would appear that anything said positive about AMD will be met with open hostility from you. These quotes are from the CSO...not me. You underestimate the power of semi-custom APUs and the next gen APUs (Kaveri) are clearly superior graphically to Intel integrated...clearly. The R9 family while not as elegant as Nvidia from a power perspective are on par or superior to the 700 family. You are comparing a $700 TI card to a $500 290x...apples to watermelons?

This thread is about AMD as a company. They are healthy, they have superior IP in certain areas, they are migrating away from a losing business model and they are leveraging their advantage in APUs to drive sales in non-traditional markets. They are consolidating manufacturing of APUs at GloFo to increase yields, increase performance, and to increase efficiency by carrying "60% or more common IP from gen to gen" instead of taping out new products at different fabs. Throw in dense ARM 64 bit servers via SeaMicro, embedded, and the power of the APU to take share in the tablet, laptop, and low end desktop and you have a real winner. This AMD management team is far removed from Hector "we did what we said we would" Ruiz and Henri "we don't do paper launches" Richard...they e.x.e.c.u.t.e.

This AMD is not the x86 reject playing second fiddle to Intel and Nvidia...this AMD is carving out a new path via the APU, ARM, and GPU. Just watch...
 
It would appear that anything said positive about AMD will be met with open hostility from you. These quotes are from the CSO...not me. You underestimate the power of semi-custom APUs and the next gen APUs (Kaveri) are clearly superior graphically to Intel integrated...clearly. The R9 family while not as elegant as Nvidia from a power perspective are on par or superior to the 700 family. You are comparing a $700 TI card to a $500 290x...apples to watermelons?

This thread is about AMD as a company. They are healthy, they have superior IP in certain areas, they are migrating away from a losing business model and they are leveraging their advantage in APUs to drive sales in non-traditional markets. They are consolidating manufacturing of APUs at GloFo to increase yields, increase performance, and to increase efficiency by carrying "60% or more common IP from gen to gen" instead of taping out new products at different fabs. Throw in dense ARM 64 bit servers via SeaMicro, embedded, and the power of the APU to take share in the tablet, laptop, and low end desktop and you have a real winner. This AMD management team is far removed from Hector "we did what we said we would" Ruiz and Henri "we don't do paper launches" Richard… they e.x.e.c.u.t.e.

This AMD is not the x86 reject playing second fiddle to Intel and Nvidia...this AMD is carving out a new path via the APU, ARM, and GPU. Just watch...

Well, sort of. Mostly they tend to execute fairly well these days, but Kaveri is massively late and the R9 290(X) cooler debacle was another misstep. Still, this is an improvement over the Barcelona and Bulldozer episodes.
 
Well, sort of. Mostly they tend to execute fairly well these days, but Kaveri is massively late and the R9 290(X) cooler debacle was another misstep. Still, this is an improvement over the Barcelona and Bulldozer episodes.

OMG I remember Barcelona! I didn't realize Kaveri was late...I just got back reintegrated into AMD late October of this year. Was it to give time for the 28nm process to mature at GloFo or a design issue? The last management team I remember was Ruiz and Richard so I am very impressed with the commentary from the current CEO, CFO, CSO so far.

They are in essence trying to turn a large tanker in shallow water and so far appear to be pulling it off. The execution seems solid minus whomever said "yeah slap that reference cooler on and call it good". The Kaveri ramp and pulling all APUs (semi-custom and otherwise) out of GloFo at good margins and on time will be the true test for me.
 
Are you somehow implying that NVIDIA is better in this regard?

Did you even read the context of this thread? Rebrands and light refreshes are a fact of life in the GPU world, but going from 35% to > 50% discrete graphics market share based on a new lineup that is largely just a light refresh of the previous lineup doesn't make too much sense.
 
Do you mean R7 260X, lol :D That has the same feature set as HD 7790, as I mentioned earlier.
I'm well aware. But as I mentioned, this is the same IP set at R9 290.

Did you even read the context of this thread? Rebrands and light refreshes are a fact of life in the GPU world, but going from 35% to > 50% discrete graphics market share based on a new lineup that is largely just a light refresh of the previous lineup doesn't make too much sense.
Discrete market share comprises many segments (Channel, DT OEM, Notebook OEM). The product line you are mentioning is not addressing all those segments.
 
Status
Not open for further replies.
Back
Top