I thought this was significant information so here is the thread to discuss. I listened to the call and the linked article explains it pretty well. If they can pump out 45nm in 1h of 08 things will get pretty interesting to say the least.
On a side note, they discussed yields, saying that 65nm yields are as good for K10 as they are for K8. They said the defect density for Barcy is the same per^2" of silicon as the rest of 65nm(K8). What they said is that they are having to tweak Barcy to get the clock speed up not that they were having problems making functionals dies.
I am rooting for AMD, not a fanboy though. I want AMD to do well so that we will have more choices and more innovation.
I have always been an Nvidia guy, but still pull for ATI to push out some good products. I normally use Nvidia for one reason, drivers. That might change in the future if they can prove the drivers are as good or better than Nvidia's.
Im the same guys... I have bought AMD in the past but my purchases are based on performance at the time, and lately intel/nvidia have been on top. Hopefully AMD will be the next purchase for me. As suggested if they go well = more innovation, competitive pricing and performance to boot.
I really do hope AMD pulls out of this. They improved on their quarterly earnings as well. Mind you they still lost 350 million but it's better then last quarters 500-600 million loss.
Intel and AMD have a extremely interesting relationship. AMD always aiming to push the envelope further with a very limited R&D/Fab/Workforce....While Intel has a very large R&D/Fab/Workforce. Intel always seems to follow AMD's routes. Although it makes complete sense. If your rival in business is taking things a step further to provide a new product. You need to match them so you don't lose market share. Makes absolute sense...With this relationship though, at the rate AMD is trying to innovate...It's really pushing intel to step up their products. With their R&D funding it actually benefits them quite a bit.
AMD purchased ATI and are attempting a fusion processor?...Well let's enter the graphics market next year. They also are investing into a CPU/GPU.... Intel finally adopts a IMC in their upcoming Nehalem as well as a similar fashion to AMD's Hyper Transport called "QuickPath Interconnect".
Now I'm not saying Intel is copying AMD step for step...Just stating that if it wasn't for AMD..We'd have a much slower progress towards new innovative products. I just hope AMD can get out of this deficit and get back into the profit margins..The x86 world would not be very bright without AMD...
I will believe it when AMD actually has a working 45nm CPU running. Other than that, it's just the same 45nm SRAM wafer they showed earlier in the year, and that's not manufacturing 45nm working die. Sorry, but after the 65nm and Barcelona fiascoes, I will believe it when I see it.
As for tweaking the process - again, I will believe it when there isn't a shortage of Barcelona CPUs. Otherwise, it's just talk to soothe the analysts. When Tier 1 OEMs have servers available with the highest clocked Barcelona available to purchase, it's just all talk.
Oh, and can we quit the "without AMD" stuff. Without Intel, there would be no AMD. Without IBM, there would be neither. Without sand there would be no silicon. Etc, etc, etc....
It would be really nice for AMD to get their 45nm manufacturing going. I hope they get it in time.
The only problem is and i have said that every time some AMD employee babels about catching up to intel (in regards to the manufacturing process) or shortening the gab by a year, that they have failed every single time during the last years. That goes hand in hand with all those missed product launches and paper launches.
Given that history i firmly believe that we won't see any 45nm chips sold with an AMD logo in Q1 or Q2 2008. Maybe in Q4.
I hope, like every time in the past, that i'm wrong though.
So AMD claims they have basic production of 45nm chips. They further expect to start production ramp Q1 2008.
AMD has barely tangoed with the 65nm process. They AMD hasn't even reached a full year on that process and has not fully converted to the 65nm process on all their CPU's. For example, the Athlon X2 6000+ is still 90nm. If you recall, AMD's switch to 65nm started with the bottom rung CPU's on the desktop front, and currently stands at the X2 5000+ - which may represent the end of the 65nm Athlon X2 speed bumps - so what's next.
Well, AMD, much like Intel, migrated processors that were 90nm to the 65nm process without much technical changes to the actual chip - in essence it was just a shrink. One would expect that is the plan with the 65nm - 45nm process.
So the question is, which CPU's will come out first as 45nm parts? Well, based on AMD's way of doing things, it would seem that Barcelona gets the first crack at the shrink. It will be the slower parts that come out first in a die shrink - much like the 90nm to 65nm move.
People question AMD's capabilities of pulling off the 45nm move. They can do it. How do I know? Well, another 3 letters says so - IBM. IBM claims to have had the 45nm process figured out since early this year. In fact IBM, et al., are working on 32nm for cell processor right now.
We all know by now that IBM was working with AMD on the 45nm node. And if AMD was smart - which I hope they are/were - then they did not use the “Help Us IBM” card on Barcelona development but instead use it on the Barcelona die shrink + tweaks process at 45nm. Much like IBM did when it was called in on the 11th hour to help with the Opteron, which was essentially broken until Big Blue flexed a few brain cells.
What worries me though is that now Intel is playing the money game. Intel is spending money to make money. They are the playground bully, taunting the weaker kind in the playground: How fast can you convert your fabs to this new process? Can you do it every 2 years? How many of them can you make a month? Can you close one fab to upgrade it for new process? Well we can! Yup - Intel with their tic - toc model in full swing will be on the 14nm process by 2012 and AMD will still be trying to convert to 45nm.
Think about it - AMD is converting their fabs right now to 65nm - whilst Intel is converting to 45nm. By this time next year, 90-95% of Intel chips will be 45nm - and perhaps as many as 5-10% will be 32nm. AMD will have about 90-95% 65nm and 5-10% 45nm.
It suddenly paints a picture of how far behind AMD is from Intel – if die shrinks are important to you. Let's not forget that AMD leads in the design arena with their monolithic quads and Intel in the process arena with their impressive die shrinks. If AMD had a 45nm Barcelona, then it would be the talk of the town – even at 2.0Ghz! I understand why they released 1.7-20Ghz Barcelona first – they had to, to satisfy the loyal customers who wanted to upgrade at SAME POWER LEVELS, any higher than 2.0Ghz and the power level promise begins to look somewhat like a broken promise.
There has been chatter here and there about IBM buying AMD. If they do - it would make sense now. Big Blue could actually swallow AMD and just rename the whole operation IBM. They could rebrand all the AMD CPU's as IBM x2 or X4 etc. Which would be a marketing coup since IBM is a better known brand than AMD to the average Joe Schmo walking into BestBuy or CircuitCity.
The thing that may actually hinder such a deal is AMD's graphics acquisition - ATI -believe it or not! Right now ATI is a mess. It's simply not competitive with Nvidia. So perhaps it's better to wait until Fusion comes out before taking AMD over. Wait and for AMD to muddle through their original thinking of buying ATI. Let them come out with this new product that they are hyping. Even if it's a hot product - capacity constraints means it will not outsell Intel - so IBM can still go for AMD.
If I were IBM - I would make my move sometime next year - Q1 2008. Wait for AMD to have a competitive product out the gate - Barcelona/Phenom - yet, not wait so long that their fortunes change.
At the end of the day it's a volumes game. IBM is the only other player that can make these things and give Intel a run for their money. Intel has fabs, guess what - IBM has fabs too. Intel has money. IBM has money too. In fact IBM could go as far as making a true Apple competitor with their own OS to boot. Imagine, an IBM PC from the ground up!
Well - time will tell. Just hoe AMD lives through all this.
Oh, and can we quit the "without AMD" stuff. Without Intel, there would be no AMD. Without IBM, there would be neither. Without sand there would be no silicon. Etc, etc, etc....
There is a difference when speaking in regards of a company failing to a company never existing. With that case in mind it's a valid discussion. If AMD goes down...Processor prices will rise and innovation will bgin to crawl instead of the leaps it's currently making. Without one driving the other the x86 world would progress slower....That's a fact. Until the next competitor jumps in where amd left off to restart the process again.
There is a difference when speaking in regards of a company failing to a company never existing. With that case in mind it's a valid discussion. If AMD goes down...Processor prices will rise and innovation will bgin to crawl instead of the leaps it's currently making. Without one driving the other the x86 world would progress slower....That's a fact.
humm...
what happens if Intel's not around?
AMD charges you 300 bucks for an entry level processor.
what happens if AMD's not around?
Intel charges you 300 bucks for an entry level processor.
So the argument goes both ways. Without either one of them, we, as a consumer, are screwed.
Unless of course, you're BM, who believes CPU market is devalued, and is extremely happy to pay 500USD for X2 3800+
Message edited by yomamafor1 on 10-19-2007 at 04:10:22 AM
Last I heard though Intel isn't in danger of going under due to bleeding cash out it's eyes. AMD on the other hand has been getting hosed and has lost of $1 billion dollars over the last 3 quarters. That is why this is a valid argument. I want good healthy competition, therefore I will root for the underdog. If Intel were in this situation then I would be rooting for them.
Last I heard though Intel isn't in danger of going under due to bleeding cash out it's eyes. AMD on the other hand has been getting hosed and has lost of $1 billion dollars over the last 3 quarters. That is why this is a valid argument. I want good healthy competition, therefore I will root for the underdog. If Intel were in this situation then I would be rooting for them.
I'm not saying I'm rooting for Intel. I'm just merely saying that without one of them, or their competitions, we as consumers will lose.
This is the third consecutive quarter AMD has lost money. If all the taxation and interests are concerned, AMD lost about 1.6 billion this year alone.
I'm basically rooting for the company that has the best price / performance ratio on the market. At the moment, Intel fits my requirements, so I'm rooting for them. If AMD has a better product, as well as superior price / performance ratio, I'll be supporting them too.
I am rooting for AMD to be competitive because the market needs the competition. I will however buy whatever suits me best at the time I buy.
If Intel were in this situation I would be rooting for them because if AMD were the only X86 maker the market would suffer until competition were to re-appear.
I will believe it when AMD actually has a working 45nm CPU running. Other than that, it's just the same 45nm SRAM wafer they showed earlier in the year, and that's not manufacturing 45nm working die. Sorry, but after the 65nm and Barcelona fiascoes, I will believe it when I see it.
As for tweaking the process - again, I will believe it when there isn't a shortage of Barcelona CPUs. Otherwise, it's just talk to soothe the analysts. When Tier 1 OEMs have servers available with the highest clocked Barcelona available to purchase, it's just all talk.
Oh, and can we quit the "without AMD" stuff. Without Intel, there would be no AMD. Without IBM, there would be neither. Without sand there would be no silicon. Etc, etc, etc....
Rooting for AMD, but I would be surprised if any 45nm parts are really available in 2009.
They still cant' get 65nm working well.
As others have noted that they are still shipping lots of 90nm parts, including all of the upper end Athlons. Only the low clocked parts have made it to 65nm.
But for all of our sakes, I hope they can pull it off.
The cost per CPU is going to go way up with Barcey until they get to the 45nm process since right now the chip is MUCH larger than the X2 and far more expensive to produce.
Meanwhile, the per CPU cost for Intel is going to drop even further in Nov.
Thus I made the remark that AMD's only advantage for Barcelona is its scalability. It can't outperform in IPC, clockspeed, yield, as well as manufacturing cost.
AMD will try to advocate and speed up the development of multi-threaded programs, as it is their only way to make Barcelona more attractive.
It is very likely that IPC will improve with each revision as usual with AMD.
Also, the server pricing that Intel has will also make Barcelona attractive, as has been proven in the past, price is equally as important as performance.
