If he's cherry picking, so are you. You can't just call one piece of evidence cherry picking and then counter with the same, solitary number of sources.
In his defense, Crysis 2 produces similar behavior in that review.
-
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.
Richland & Kabini rumours
Page 39 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
- Status
In his defense, Crysis 2 produces similar behavior in that review.
NTMBK
Lifer
Battlefield 3 uses the Frostbite 2 engine, which is being pushed by EA as their primary next-gen engine. A lot of their games will use it going forwards (Need for Speed, Mass Effect, Medal of Honour, Battlefield, Dragon Age and Command & Conquer franchises are all confirmed as using it in the next game, if not already). It's a good indication of the way that game engines are heading.
NTMBK
Lifer
Looking good indeed! Interesting that even with twice the number of pixels the Samsung has, the UI performance is better. (Makes me sad that Vivek is still buying Intel's crap about a 7W Ivy Bridge, though. It's 10W, or 13W for a non-crippled one.)
inf64
Diamond Member
Frostbite2 is indeed a next gen engine. It not only looks great it performs great too (even on mid range hardware!). Compared to some other engines out there it's quality and performance are on another level so no wonder EA is pushing it(and evolving it further).
@On topic of Hondo
If Hondo chip alone has 1.72W of "sdp" power draw in typical web browsing usage,what do you guys think Temash will get in similar scenario? Assuming similar number of threads and clock speed.
itsmydamnation
Diamond Member
1.337 would be my guess.
The Alias
Senior member
haha saw that coming
itsmydamnation
Diamond Member
28%/23% (depending how you want to count it) power reduction for SDP, it is possible 40nm to 28nm, uarch power saving measures, better clock gating.
Well, from personal experience of Atom, its a dog. It killed the 10.1 nettop market by its utterly dismall performance. From experience of e450 bobcat i wouldnt count on 4.6 TDP to be more than just barely acceptable for a tablet if you are not bussy, and dont care to about battery life 🙂 - bobcat on the old leaking 40nm is not for tablets, 28nm hkmg and new powerstates is needed.
I am sure kabini and temash will have the x86 tablet market for themselves. How many expensive haswell tablets or ib 7w, with predictable weaker battery life than temash because of the fatter front end, seperate southbridge, end and perhaps even weaker gpu, will be sold for tablets? no f.....g way its going to be 0.000567% share - for nerds, CEOs, and nerd-reviewers. AMD is going to own the market with kabine and temash. Excactly as they have done with bobcat for the lowend laptop market. The problem is, does the volume give any sizable profit to finance GF and kaveri adventures? - i have my doubts.
What Intel needs is a new Atom, with 100% more cpu horsepower and 800% more gpu power - or be extremely effective in the sub 1W mobile segment. I dont beliewe they can do both in this market. That is going to be the interesting product for low end laptop and tablet market. But if its going to fight a big-little solution, i dont think the brute techlology force will outweight the smart, scalable and extremely costeffective A7/A15 solution. I think Intel will have far a better chance beating AMD kabini on the lowend. What Intel have chosen is very interesting.
If Windows need something to get win8 of the ground its kabini and temash, and perhaps the new atom. Absolutely the most interesting x86 products, and future variations of it, for this and next year. Mobile x86 computing is starting to rock!
Ventanni
Golden Member
Don't underestimate Atom. I'm no fanboy of the damn thing, but I've never gotten the impression that Intel has ever put much more than secondary thought into it - designed mostly to keep older fab units online and running, and to generally see what they can make out of it - until now. Compared to the Core architecture and its array of iterations that have that have impressed us since their introduction (Core2, Nehalem, Sandy Bridge, etc), Atom has typically always failed to produce. I mean, it wouldn't surprise me if the original folder that Atom's beginning started from was called "Project Backburner".
But I think Intel is now putting a lot more focus and resources into its Atom line. Unlike 5 years ago, there's a much more stable, defined market that these products can break into.
Your claim that Intel needs more CPU/GPU power will be addressed with Intel's 22nm shrink (Merrifield and Silvermont I believe). I don't for a moment believe their GPU prowess will rival AMD's iterations, but its CPU performance should be substantially improved.
But I think Intel is now putting a lot more focus and resources into its Atom line. Unlike 5 years ago, there's a much more stable, defined market that these products can break into.
Your claim that Intel needs more CPU/GPU power will be addressed with Intel's 22nm shrink (Merrifield and Silvermont I believe). I don't for a moment believe their GPU prowess will rival AMD's iterations, but its CPU performance should be substantially improved.
Designing GPU, and writing good drivers, have proven to be a v e r y difficult task. Intel have sort of used the Buldozer x2 method for it.
Intel needs to catch up here, because its very obvious qualities for the consumers - and thats why Apple is going that direction. To what degree can you play the new games on your tablet / small laptop?
Where is Atom compared to kabini/temash and little-big?, what product and consumers are they targeting?
I have no doubt intel is bringing all their impressive tech, and we will see some impressive stuff, but will it translate into tangible consumer benefit?
Intel needs to catch up here, because its very obvious qualities for the consumers - and thats why Apple is going that direction. To what degree can you play the new games on your tablet / small laptop?
Where is Atom compared to kabini/temash and little-big?, what product and consumers are they targeting?
I have no doubt intel is bringing all their impressive tech, and we will see some impressive stuff, but will it translate into tangible consumer benefit?
Phynaz
Lifer
OEM's don't buy from the channel.
You said Trinity, and now say "in general". In General which chips is it that GLF isn't supplying enough of to AMD? And if that is the case, why would AMD pay to not take wafers?
piesquared
Golden Member
New specs posted here:
http://semiaccurate.com/forums/showthread.php?t=6837&page=7
Sounds credible. That's a pretty capable little package.
http://semiaccurate.com/forums/showthread.php?t=6837&page=7
Sounds credible. That's a pretty capable little package.
AtenRa
Lifer
Llano/Trinity (APUs), Athlon/Phenom II AND BD/PD(CPUs) are produced in GloFo ok so far ???
One quarter you may produce more APUs than CPUs and the next quarter you may have the opposite. So in short, you are production volume constrained.
Now, the total production volume may decline(you buy less wafers) but you may use 90% of the entire production volume for APUs and 10% for the CPUs. This way you can sell more APUs and have supply problems with CPUs. Buy doing this you have less production but you raise the APUs percentage.
Now, until we see Q4 2012 results (APU/CPU sales percentages) we may not know if Trinity sells more than Q3 2012 or not.
But, how Trinity is selling doesnt negate that fact that OEMs prefer APUs than CPU+Discrete 😉
The contract between GLF and AMD has a take-or-pay clause, which means that AMD will pay for a minimum X quantity of wafers whether they order it or not. AMD also has to pay for any orders above X.
Order less than X and you still have to pay for X wafers. Orders X and you have to pay for X wafers. Order X + Y and you have to pay X + Y wafers.
If AMD were supply constrained they wouldn't pay GLF to get out of the take or pay clause, they would simply order the wafers as they would be essentially be for free. The fact they didn't order means that they cannot find any customer willing to buy their processors above their break even point.
AMD could be money constrained, talent constrained, anything constrained, except supply constrained.
AtenRa
Lifer
Just ask your self this,
AMD sold 1M Bulldozer 4C CPUs in Q3 2012, you know the 315mm2 die at $100.00 on average and they couldn't sell more than 1M Trinity APUs with die size 245mm2 ???
AMD increased the Bulldozer based CPUs Sales in Q3 2012 but they saw a decline in APUs sales. Does that seams logical to you ??? in a market that goes for smaller, cheaper solutions ???
Either they made a terrible calculation and ordered more BD dies than they needed or they where production volume constrained and didnt have more Trinity APUs to sell.
Thats why i have said that we need to see the Q4 2012 CPU sales to understand what happened 😉
http://www.3dcenter.org/news/genauere-zahlen-zum-absatz-einzelnen-amd-prozessorenarchitekturen
No, it's not logical.
If AMD had overshot Bulldozer sales and undershot Trinity sales, they could just cut new Bulldozer orders, let Bulldozer inventory baloon a bit and then sell this inventory at a slower than forecast rate or cut prices to improve the rates. With Trinity, they would do the opposite, increasing new orders, eat into inventory and sell at a faster rates, or increase prices to diminish the flow a bit.
Under no circumstance they could be supply-constrained AND pay the under-commitment fee from the WSA. Once you decide to pay the take or pay charge in any of such contracts, the costs going from ordering any volume up to X is essentially $0. If you paid the fine, why wouldn't you order the chips that you are supply constrained?
You are so anti AMD that you re spreading falsehood
without even thinking a second about the ridiculousness
of you being so gullible by hate....
Who would pay 100mns to not produce CPUs while producing
them and selling them for 1$ each would be more profitable...?..
You think that at 1$ they wouldnt find buyers..?....
laputan machine
Member
Selling the chips that cheap would ruin their price structure for the future and just doom them.
ShintaiDK
Lifer
Oh please, get real for once and snap out of it.
http://www.amd.com/us/press-releases/Pages/amd-amends-wafer-2012dec06.aspx
http://online.wsj.com/article/SB10001424127887324001104578163841987546704.html?mod=googlenews_wsj
http://www.zdnet.com/amd-amends-globalfoundries-deal-to-pay-320-million-7000008443/
http://www.bit-tech.net/news/hardware/2012/12/07/amd-glofo-demand/1
http://www.bizjournals.com/albany/blog/2012/12/amd-pays-320m-to-modify.html?ana=yfcpc
http://hothardware.com/cs/forums/p/64894/442195.aspx
http://forums.anandtech.com/showthread.php?t=2288003
Last edited:
Abwx, I don't understand what you are attempting to communicate. Or rather, I think I understand what you are saying but what you are saying appears to only support what mrmt and ShintaiDK are saying.
(1) AMD did pay GloFo a take-or-pay fee in which they opted to pay-and-not-take
(2) Had they opted to pay-and-take then they would have had chips to sell, even if for only $1 profit per chip, if buyers were to be found for those chips
What other possible interpretations are there for the actions of a company that would rather pay-and-not-take versus pay-and-take with the opportunity to sell for $1 profit versus?
The fact AMD had to pay-and-not-take is proof that AMD is not supply constrained, GloFo is not production limited in any sense...it is proof that AMD is demand-constrained at all pricepoints extending all the way down the price curve to the point at which they would be selling the chips for $1 gross profit.
Think about it - AMD basically said "it is cheaper for us to give our money away versus giving our product away, when given the choice our accountants said it was better to just give money away for nothing in return rather than get chips in return and attempt to sell those chips for even $1 to recoup even some of the take-or-pay expenses"
Does it get any more dire than that?
I worked at TI long enough ago that we were making 486 chips and 8Mb dram chips...and the long-standing joke at the time was that the only way we could sell an 8Mb chip for a dollar was to wrap the 8Mb chip in a $1 bill. (i.e. we were taking $1 from the customer but giving them back $1 and a ram chip in exchange)
And even under those terms and conditions it made sense to keep selling the ram chips because at least the cash flow was partially offsetting the overhead expenses associated with being in business in the short-term.
AMD management is saying it isn't even worth attempting to do that. Their stock price reflects the reality of just how bad of a harbinger the take-or-pay choice was, argumentation in these forums does not negate the reality that pretty much everyone else acknowledges and accepts.
They have more than enough of both Trinity and BD dies. AMD literally paid GF to not make it chips. If they had wanted more of either kind, they could have had them from GF for free. AMD is not supply constrained.
It also didn't make a terrible calculation. Their chips just weren't selling.
There are other costs beyond manufacturing here: Q&A, packaging, freight, insurance, storage, customs, sales commissions, just to name a few. Once you add those to manufacturing costs, you'd need more than 1$ on each chip to break even.
We dont have the meaningfull details of this corporate arrangement
but given how it was released to the gullible press , and with AMD critics
always eager to spread whatever suit their propaganda , everybody took
thoses announcements at face value.
It s a little more complicated than this , for sure , for whom is in the knowledge
of some investment funds strategies.
Dont forget that at both ends there is the same investor.
but given how it was released to the gullible press , and with AMD critics
always eager to spread whatever suit their propaganda , everybody took
thoses announcements at face value.
It s a little more complicated than this , for sure , for whom is in the knowledge
of some investment funds strategies.
Dont forget that at both ends there is the same investor.
And how is Richland going to change that?
The only reason for that intermediate product must be the absolutely need to use some GF capacity because of wsa agreement. And i seriously doubt Kaveri will change the outlook.
If GF (read Atic), in anyway thinks AMD is going to help GF up from the .ppt slump, its the most pathetic thinking of the year. Ofcourse they take all the money the old agreements alow them to, writing off old 32nm equipment on AMD optimistic pre 2010 predictions about BD performance. After all, nothing but shortsighted greed control most business. But if GF dont start to deliver themselves, and for loads of customers, they are out of business. Not that i envy competing with Intel process tech or TSMC. And i havnt seen anything, indicating its just barely possible for them to do so.
If it wasnt for the WSA agreement, the mm2 and efficiency handicapped Richland will stand in stark contrast to the efficiency and market appeal of Kabini/temash.
The only reason for that intermediate product must be the absolutely need to use some GF capacity because of wsa agreement. And i seriously doubt Kaveri will change the outlook.
If GF (read Atic), in anyway thinks AMD is going to help GF up from the .ppt slump, its the most pathetic thinking of the year. Ofcourse they take all the money the old agreements alow them to, writing off old 32nm equipment on AMD optimistic pre 2010 predictions about BD performance. After all, nothing but shortsighted greed control most business. But if GF dont start to deliver themselves, and for loads of customers, they are out of business. Not that i envy competing with Intel process tech or TSMC. And i havnt seen anything, indicating its just barely possible for them to do so.
If it wasnt for the WSA agreement, the mm2 and efficiency handicapped Richland will stand in stark contrast to the efficiency and market appeal of Kabini/temash.
- Status
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
-
-
