• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

AMD Q1 2015 Earnings - 23 cents a share loss, to exit dense server (SeaMicro)

Page 21 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Unless AMD will degrease its margins for those SOCs going to 20nm, then there is no way Revenue from those sales will decrease.

Sony and MS are definitely going to pay less for each chip. This is likely the time when the Slim models will be introduced along with a price cut. As you said, TSMC 20 nm isn't any cheaper per transistor than TSMC 28 nm.

So yeah expect margins to go down considerably.
 
Sony and MS are definitely going to pay less for each chip. This is likely the time when the Slim models will be introduced along with a price cut. As you said, TSMC 20 nm isn't any cheaper per transistor than TSMC 28 nm.

So yeah expect margins to go down considerably.

Even without a price cut you can still reduce the console BOM by reducing things like the heatsink requirements, complexity of PCBs due to lower power requirements, and using a smaller (cheaper) power supply.
There are knock on cost savings from just a reduction in power consumption of the SoC even if the cost of the chip itself isn't the full discount.
While they would want a discount, they would also be aware of the commercial realities of the cost of production.
 
Even without a price cut you can still reduce the console BOM by reducing things like the heatsink requirements, complexity of PCBs due to lower power requirements, and using a smaller (cheaper) power supply.
There are knock on cost savings from just a reduction in power consumption of the SoC even if the cost of the chip itself isn't the full discount.
While they would want a discount, they would also be aware of the commercial realities of the cost of production.

The CPU in the consoles cost 100$ and 110$. I am quite sure both Sony and MS have it in their contracts to get lower price over time. And specially with a shrink. The CPU is the main part to reduce cost on.

The rest is peanuts compared. Expect the PS4 GDDR5.

AMD also specify in their own statements that semicustom will cost more (for AMD) but royalties is going down at the same time.
 
Last edited:
Sony and MS are definitely going to pay less for each chip. This is likely the time when the Slim models will be introduced along with a price cut. As you said, TSMC 20 nm isn't any cheaper per transistor than TSMC 28 nm.

So yeah expect margins to go down considerably.

I dunno. Why would they even bother porting if it gives them no benefit. Why waste precious engineer hours to port it to 20nm only to give all profits to sony/ms.

AMD may lower the soc price later, when 20nm matures but probably not from the start.

Sony and MS profits come from lower TDP SoC. They can cheap out on cooling, PSU, VRM, PCB, platic covers, packaging and so on.
 
I dunno. Why would they even bother porting if it gives them no benefit. Why waste precious engineer hours to port it to 20nm only to give all profits to sony/ms.

Sustain higher volumes. Lower cost to Sony/MS = price cuts to consumers = more sales. Same deal with lower power use (smaller, more appealing form factor).
 
I dunno. Why would they even bother porting if it gives them no benefit. Why waste precious engineer hours to port it to 20nm only to give all profits to sony/ms.

Because thats how supply contracts work. They are always more profitable in the beginning.
 
The CPU in the consoles cost 100$ and 110$. I am quite sure both Sony and MS have it in their contracts to get lower price over time. And specially with a shrink. The CPU is the main part to reduce cost on.

The rest is peanuts compared. Expect the PS4 GDDR5.

AMD also specify in their own statements that semicustom will cost more (for AMD) but royalties is going down at the same time.

I would say that the contract should stipulate that the price would only go down if commercially reasonable, however AMD have an agreement with GlobalFoundries that has now been amended 5 times due to how crappily it was written, and then re-written time after time, and still causes them problems, so I would be inclined to agree that the contract probably does require price drops even if not commercially reasonable, because it seems AMD suck at contracts.
 
Yes, they did. Check out this May 2005 article from Tom's Hardware. Note the last paragraph.



I disagree, it really is a steaming pile of crap. For the vast majority of desktop applications, a simple die-shrink of Thuban to 32nm would have outperformed FX-8150. Heck, the FX-8150 fell behind the actually existing Phenom II X6 1100T in many benchmarks. When you can't even beat your own previous product, that's hard to describe as anything but a failure.

Incremental improvements to the K10 architecture would have been far, far cheaper than Bulldozer to develop, and would almost certainly have had better results. AMD bet everything on high integer throughput, hoping to win the server market, and lost big.



Not if it's just a CPU. But for total system TDP of a PC with good graphics capabilities, 300W is not at all unreasonable. Look at the Mac Pro, with a big Intel Xeon (4 to 12 cores) and two Tahiti GPUs - all cooled by a single large heatsink and fan. That has to be over 400W when fully loaded. And OEMs had no compunction about shipping GK110-based cards with TDPs of 200-250W. Add in the CPU and you're easily up to 300 or more. The fact that all the heat in a 300W APU would be coming from one chip makes the design of the cooling system easier, not harder.

Again, this will only be palatable if the performance is really competitive, which will require 16nm FinFET plus a good performance from the Zen CPU architecture plus HBM shared memory.

Please see my elaboration in this post, in regards to your Tom's link: http://forums.anandtech.com/showpost.php?p=37345903&postcount=430
 
Your elaboration is wrong. Intel would not have launched Desktop Conroe in 2006 if they had made this decision in 2005.

I already explained myself once.

Crap on shingle, keep cherry picking and ignoring everything.

http://forums.anandtech.com/showpost.php?p=37346028&postcount=444

The elaboration is that no one back in ~2000 thought the new P4 arch was going to be replaced with beefed up P3 tech.

Get over your semantic BS nitpicking.
 
Last edited:
I dunno. Why would they even bother porting if it gives them no benefit. Why waste precious engineer hours to port it to 20nm only to give all profits to sony/ms.

I would not be surprised if there are clauses in the contracts w/ Sony and MS that say that they get to choose when the die shrinks occur (and what fab to use). They probably pay AMD some amount of money (aka an NRE) for their engineers' time/effort to port the chip to that process.
 
The elaboration is that no one back in ~2000 thought the new P4 arch was going to be replaced with beefed up P3 tech.

Get over your semantic BS nitpicking.

In 2000? No, but in 2003 Intel was full steam ahead with Conroe. AMD OTOH started to work on Bulldozer on 2005, when it was already clear that fireball designs were a no-go. AMD (mis)management team basically threw away a solid foundation and all the lessons learned that culminated on the K8/K10 in order to work in a fireball design by the time Intel itself was distancing itself from it. SUN, which also had a Fireball design, also saw that this path was a no-go and canned the project (SUN's Rock) in 2009.

There is no excuse for it, AMD did not have to develop Bulldozer, it wasn't an inescapable conclusion to develop it. AMD had the chance of can the product circa 2007 when they saw what Conroe was doing with them, and circa 2009 when they had to can the 45nm Bulldozer because it sucked even more than the 2011 version that crashed their sales. Only AMD management team could be incompetent enough to proceed with the development of such a bad family of products, and triple down their bet after all signs pointed to a disaster of epic proportions.

But hey, I think I'm going to buy a FX processor, it might become a cult item just like the C64 or the Z80 a few years down the road.
 
BD was a disaster as was NB but Intel can afford it, AMD could not plus AMD failed to fail fast. They had a winner (Bobcat) and dropped the ball due to the WSA (Bobcat was fabbed at TSMC) commitments. A double whammy of failures - BD and WSA multiplied by mobile/tablet trend. Enough to put a company bankrupt (which they would be if CPU is all they had as there wouldn't be any console revenue).

The biggest issue short term for AMD is channel inventory. They have got to clean out all the old skus of processors, motherboards, desktop and notebooks that are "lying around" in the channel. Stuff needs to be traded - out with the old, in with the new (Carrizo). Scary times because I said the same thing 5 years ago about Server.
 
Sony and MS are definitely going to pay less for each chip. This is likely the time when the Slim models will be introduced along with a price cut. As you said, TSMC 20 nm isn't any cheaper per transistor than TSMC 28 nm.

So yeah expect margins to go down considerably.

Anyone thinking that AMD will sell 20nm Console SOCs at lower margins than they sell 28nm SOCs is greatly mistaken.

AMD will go to 20nm only when they will be able to sell at lower price than what they sell today(28nm) but at the same time at least keeping those low margins they have today. This way both AMD and MS/Sonny will benefit, AMD will get higher volume lower GODs(thus higher income) and MS/Sonny will get lower GODs and increase Volume (thus higher Income).

What you and others suggest here is that AMD will go to 20nm and in the process they will loose both revenue and income when they should increase both. No company will do that, you dont transfer to a new node unless you get higher margins than what you had before, or higher volumes but keeping the same margins or both.

People here are talking about Moore's Law constantly but they dont seem to really understand what it is all about.
 
I dont think AMD decides when a shrink happens. MS and Sony would never leave a long term contract so unspecified.

AMD already have to allocate more R&D to consoles today, meaning the work on a shrink is already in place. All while the royalties from the consoles is decreasing.

You make it sound like AMD sat with all the goodies at the table. They didnt.

Its nothing new that the consoles is always most profitable for the suppliers in the beginning, while that profit and revenue diminishes over time.
 
Last edited:
Yeap, the manufacturer dosnt decides when and where to fab its products.
Those Aliens does. 🙄

Supply contracts are made with clauses and expected changes and delivery. Missed deadlines can give fines. Nobody is stupid enough to make them without. (Well, besides AMD and their WSA....)

I know in your world AMD sits with all the cards while MS and Sony are down kneeling to the CEO of AMD. Thats just not how the real world works.
 
Supply contracts are made with clauses and expected changes and delivery. Missed deadlines can give fines. Nobody is stupid enough to make them without. (Well, besides AMD and their WSA....)

I know in your world AMD sits with all the cards while MS and Sony are down kneeling to the CEO of AMD. Thats just not how the real world works.

Unless you have the Console Contract to display here for everyone to see, we are going to believe AMD operates as any other Semiconductor Manufacturer in the world. That is, AMD decides when and where they fabricate their products. 😉
 
Anyone thinking that AMD will sell 20nm Console SOCs at lower margins than they sell 28nm SOCs is greatly mistaken.

You are greatly mistaken. Long term supply agreements have annual price reductions written in them for the term on the contract. It is expected the manufactured will gain greater efficiency during the life of the product.

The console makers will pay less for chips each and every year.
 
You are greatly mistaken. Long term supply agreements have annual price reductions written in them for the term on the contract. It is expected the manufactured will gain greater efficiency during the life of the product.

The console makers will pay less for chips each and every year.

And production cost decreases each and every year.

I dont believe you people, do you honestly believe the end product cost is the same to AMD today in Q1 2015 than what it was in Q3 2013 ??
 
Anyone thinking that AMD will sell 20nm Console SOCs at lower margins than they sell 28nm SOCs is greatly mistaken.

I don't think the issue is margins. The fact is that at constant margins, overall income will shrink if the valuation of the product decreases. 40% of $100 = $40, 40% of 80% = $32.

AMD will go to 20nm only when they will be able to sell at lower price than what they sell today(28nm) but at the same time at least keeping those low margins they have today.

AMD may not have a choice depending on the contents of their contract with MS and Sony.

This way both AMD and MS/Sonny will benefit, AMD will get higher volume lower GODs(thus higher income) and MS/Sonny will get lower GODs and increase Volume (thus higher Income).

Almost always the value of these sort of contracts shrink as time goes on. Same was true for previous gen consoles.

What you and others suggest here is that AMD will go to 20nm and in the process they will loose both revenue and income when they should increase both. No company will do that, you dont transfer to a new node unless you get higher margins than what you had before, or higher volumes but keeping the same margins or both.

Again, depending on the contract with MS and Sony they may not have a choice if MS/Sony pushes for it.
 
I don't think the issue is margins. The fact is that at constant margins, overall income will shrink if the valuation of the product decreases. 40% of $100 = $40, 40% of 80% = $32.

As i have explained before, Console SOC volume increases in Q2 and peaks in Q3. So Volume will not stay the same but it will increase no matter if they will transition to 20nm or not.

AMD may not have a choice depending on the contents of their contract with MS and Sony.

Again, depending on the contract with MS and Sony they may not have a choice if MS/Sony pushes for it.

What you and others are suggesting here is that the customer have control over the manufacturer about fabrication schedules, node processes etc etc. No manufacturer will ever make such a deal, ever.
If people insist on that, i will only take those suggestions as legit when they will provide proof, until then im sorry but it seems you are making those out of thin air.
 
Market isn't falling at rates remotely close to what AMD has been reporting on their income statements.

Care to retract that?
http://www.dailytech.com/IDC+2014+Sales+Show+PC+Isnt+Dead+But+Desktop+May+be+Dying/article37087.htm

AMD's node disadvantage has pretty much locked them down to desktops almost exclusively..... And Desktops are a dying market. Carizzo should slow the bleeding for some laptop design wins, but AMD really needs 14nm.

Seriously, do you think Intel is giving away a billion dollars for contra-revenue for fun? Intel is burning its cash to fight back against ARM in tablets -- but AMD doesn't have their bank account.
 
Last edited:
Care to retract that?
http://www.dailytech.com/IDC+2014+Sales+Show+PC+Isnt+Dead+But+Desktop+May+be+Dying/article37087.htm

AMD's node disadvantage has pretty much locked them down to desktops almost exclusively..... And Desktops are a dying market.

Seriously, do you think Intel is giving away a billion dollars for contra-revenue for fun? Intel is burning its cash to fight back against ARM -- but AMD doesn't have their bank account.

Did desktops fall 70%+ like AMD CPU sales?
 
Back
Top