Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
sgolds -
1) Since mas's post referenced a combined output of 45K 300mm WPM by the end of 2008, and Hans Deppe said he expects Fab36 to be at "20,000 wafers per month in the second half of 2007"( http://www.eetimes.com/news/latest/business/showArticle.jhtml?articleID=187000006), I guess it comes down to what the demand is and what they do with Chartered. Maybe they'll phase Chartered out as Fab38 comes fully on line? If they plan on phasing out Chartered by then, it sounds like they'd be approximately at parity; if not, then I guess Fab38 will be somewhat lower in production at that point in time than Fab36.
#2 2)Fab36 will go to 65nm soon, and I can't imagine Fab38 would start any larger than 65nm.
3) Purely by wafer size, two 300mm fabs would give you @4.5 times the capacity of current Fab30 output, and that doesn't take into account the 90nm->64nm transition or edge loss improvements. Pretty hard to try and be an oracle on what the product mix will look like from this point in time; heck, we don't even know what their next major product will completely look like, much less what the market may force by the time Fab38 comes online!
4)If and when Fab30 "closes", Chartered is also there with Fab36 to deal with the transition.
5) ? Your guess is surely as good as mine there.
drjohn - Really think so? Probably not, without OCing the FSB.
mas- Will be interesting to see what DDR2 1066 or 1100 will do.
mmoy - Let us know if you hear anything about Pacifica and virtualization, would you?
Now that the Turion X2s are out supporting it, and AM2 coming shortly, I imagine people will start playing around with it.
mmoy - Have you heard anything about Xen or VMware new versions coming and any info on their support of AMD's Pacifica or its performance?
wbmw - re: "enhancing electronic capability by up to 42%"
That can mean more than just speed. Notice that the 65nm roadmap at HKEPC didn't have power levels listed?
rlweitz - More and more, the period immediately following Christmas is becoming just as important. All those gift certificates and such. People waiting to see what they get for Christmas before they make their purchase decisions.
chipguy - I didn't think I said that, certainly didn't mean to. I meant to compare to main memory DRAM. I would be interested to know what you think the pros/cons would be for an L3 of either ZRAM or eDRAM. Assuming of course a process is in use which would allow ZRAM.
Paul
Should've guessed as much. :)
Tenchu - SRAM is faster than DRAM, and ZRAM is apparently between the two, not fast enough to work for L2, but a large(and much denser than SRAM L2) L3 is much faster than main memory. If it's still accessible, there was a 4-part series/interview on ZRAM at digitimes, if not there was an article on it at lostcircuits that is still accessible.
Jules - Yeesh, no need to get so touchy! Of course, you're entitled to your own taste, we all are. :)
Jules - I'd have to agree on the ugly Town car part(most of them, at least). Now they have had some quite nice Continentals, though. I'm afraid I'd probably still take a Lexus, though, which would be a more appropriate comparison than Toyota-Lincoln. Maybe Toyota-Ford or Toyota-Mercury.
Geode NX finding success.
Samsung using in MultiFunction Monitor(SiS741CX is a chipset for NX.)
http://www.digitimes.com/NewsShow/MailHome.asp?datePublish=2006/5/10&pages=PB&seq=203
Geode NX DB1500 development kit review
http://www.epiacenter.com/modules.php?name=Content&pa=showpage&pid=88
Another link re: Samsung MFM
http://www.digitimes.com/news/a20060509PR207.html
drjohn - Sure there is.
June 26th and June 19th
V
V
V
https://www.alienware.com/Configurator_Pages/aurora_m_alx.aspx?syscode=PC-LT-AURORA-M-ALX&subcode=SKU-DEFAULT
https://www.alienware.com/Configurator_Pages/aurora_m9700.aspx?syscode=PC-LT-AURORA-M-9700&subcode=SKU-DEFAULT
RE: "If anything, your links only underscore the amazing value you get from an Intel chip that AMD can't seem to deliver these days, given their limited manufacturing."
Funny that the marketplace doesn't seem to agree.
X2 3800 has been 297-300 for quite a while now at newegg.
wbmw - re: "Just look at their prices right now. It's a rip-off for an AMD X2 processor. Their 2GHz, 512K L2 cache X2 has a list price of $303. For only $13 more, you can get a Pentium D 950 with much better performance at a reasonable 95W TDP. And AMD has nothing that can touch the price/performance of the 940 and 930 (at $241 and $209, respectively). These are the mainstream parts where all the volume is, and AMD has nothing more than slow single core processors to fill the gap. In the mainstream pricing segment, Intel has performance and price/performance already won."
Who do you think you're fooling? Yes, the 950 beats X2 3800 in some benchmarks, but the opposite is true as well. Much better performance? Hardly. The 950 costs about $50 more at newegg. And even Tom's(hardly a great AMD friend) interactive CPU system says AMD is the better value:
http://www23.tomshardware.com/cpu.html?modelx=33&model1=237&chart=118&model2=320
wbmw, Wouter - The difference being that Chartered has apparently come on line now, and Fab36 is slowly coming on with dual cores as well.
http://theinquirer.net/?article=31557
"...The 3800+ will work at 2.0GHz and you can expect 4400+, 4800+ and 5000+ to be available at launch as well. The only missing CPU is the 5200+, which is set to come in Q3 of this year (back to school period). The difference between 5000+ and 52000+ is of course, the size of the cache and the price.
But 5200+ is more or less the new name for FX-60 with the new processor socket and memory controller. In order to avoid a shortage, AMD did a lot of preparation and is sending a lot of CPUs down the channel. These are mostly products manufactured by Fab 30 and Chartered’s Fab 7. Of course, Fab 36 isn’t left out, and the factory is slowingly ramping up.
Launch date for the new Socket is set at May 23rd, when you can expect a plethorea of launches from partners and reviews. µ"
cg - Glad to hear it won't be a killer for you, at least.
Ouch for chipguy! :(
mmoy - They may have determined they're better served overall with the L2 they have, then adding ZRAM L3 which will be much denser(@2x/5x denser than SRAM/eDRAM), allowing a pretty large L3 with latency that is still far quicker than main memory. If they increase the size of the L2 much more, they may have to slow it down more than they feel will be effective for this architecture.
I wonder if maybe the "256 bit memory" means the data path to L2 is being increased to 256 bit from 128?
Tenchu - Y'all keep missing that AMD didn't say it, and they haven't given any reason anywhere that I'm aware of for the delay in Turion X2.
BUGGI - Great shots!
AAMMDD - I never thought H. Ruiz was a dreamer!
"Ruiz said he hopes Intel will get behind AMD's "50 by 15" effort.'
Yeah, that'll happen! :P
Keith - I like that! :)
wbmw - AMD didn't say anything about it there(digitimes), it was Taiwanese notebook OEM sources. As far as I've heard, AMD has been silent on the reason for the delay.
sgolds - Glad to hear things are cranking up for you!
chipguy - If Sun is the driving force behind it, you may have a point there; though it wouldn't surprise me if HP saw a good thing and beat Sun to the punch, assuming AMD puts it in a socket they are using or can quickly pick up.
chipguy - Do you agree with the general consensus that ZRAM is likely to be used in L3, and if so, how long do you think, realistically, it would be before it's put into a design for production? Would we be right in assuming that a good bit of work has been done checking it out and getting the ball rolling if they already have the agreement?
gollem - That's for sure!
mmoy - I'm sure there are some limits, but yes, you can re-program them to perform different tasks.
http://searchsmb.techtarget.com/sDefinition/0,290660,sid44_gci530571,00.html
A limited and dated definition, but the "field-programmable" part is what may wind up being the powerful part.
I suspected there might be something going on like that.
BUGGI - But then why are the monitors in the orginal image white, when everything else is yellow?
wbmw - Sorry, defective fingers! Bloomsfield is the name. The "s" is in the INQ article.
http://www.theinquirer.net/?article=31323
wbmw - Sounds like Intel has a monolithic coming up in Broomsfield, according to the INQ.
Keith - epscontest.com has one up.
wbmw - You were the one comparing 35W Turion to 31W Core as though that were a victory for Intel.
wbmw - So you think Intel's NB uses less than 4W? The other factors are arguable, but it remains to be seen which will be more successful in the market.
drjohn - Yes and I listened to AMD's, and Turion increased in units, sales and ASP's(even over Q405), probably gained market share in that segment. Core Duo was a relatively bright spot in a dim quarter for Intel, though.