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.
Cellular Ram news
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?t=96
Cheers
Cor
Rambus Announces Resetting of Infineon Patent Trial Date
http://www.rambus.com/news/pressrelease.cfm?id=135
LOS ALTOS, Calif. -- April 27, 2004 -- Rambus Inc. (Nasdaq:RMBS), a leading provider of chip interface products and services, today announced that the retrial of Rambus patent claims against Infineon has been reset from June 10, 2004 to a new date in the Fall of 2004. The new date is expected to be set before the end of May 2004. The date change was at the initiative of the trial court judge, who, among other things, cited the number of pretrial issues that remain to be resolved, and the possible need for time for Rambus to appeal certain of those rulings prior to trial.
Well legal is not my thing as you all know but this looks as if Rambus bit themselves in the tail (at least temporarily), tell me if I got this wrong pse.
Cheers
Cor
Memory types at chipstocks.net
I have at my site on the message board a chapter about memory types. As I am doing this singlehanded it is not complete but I add from time to time. Today started a thread on embedded memory (a dram competitor in some areas) at:
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?p=133#133
All memory types at:
http://chipstocks.net/semi-subjects/phpBB2/viewforum.php?f=6
and the main boards including Infineon and Rambus at:
http://chipstocks.net/semi-subjects/phpBB2/index.php
Feel free to visit, register and by all means post (anything reasonable, HA:)
Cheers
Cor
The Infineon capacity increase in Richmond and my commentary (for the time being;) at:
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?t=90
Cheers
Cor
About IBM ASICs using Denali memory controllers (Databahn)
I posted this at the Fool, but decided to repost here (quieter, the Fool rambus board it's difficult to find Rambus material at the moment, the politico bug has struck again;(
Hey Sabatino,
I seem to be late in responding to this post of yours, but this is simply because I was connecting it up through search with recent news. I have not completely figures out the angles yet, so no opinion(yet:), but let me share the pieces.
In your link:
http://tinyurl.com/2yg6c
Denali's Databahn(TM) memory controller IP supports DDR1, DDR2, and dual mode DDR1-2 operation, and is licensed by over 40 leading semiconductor and system companies for use in commercial chip designs. The silicon-proven Databahn IP is library independent and covers solutions from .18-micron to .08-micron technologies, and supports DRAM device frequencies from 100-400MHz (200-800MHz data rate).
This at first sight to me looks like so called "soft core IP library cells" (I hope I did not commit a "terminological inexactitude" there [W. Churchill], if so pse somebody correct me)
Now this is about configurable memory controllers which talk to DDR, DDR2 and other types of memory, see for more:
http://www.denali.com/products_databahn.html
snip:
Databahn is a comprehensive solution that provides a safe and efficient way to configure the optimal memory controller core for your design, validate the performance within the context of your system, and implement the solution in Silicon. The entire process is managed by our online application infrastructure, and our browser-based interface gives you complete visibility and control over each step of the process:
* IP configuration
* Memory selection/compatibility
* Performance analysis
* IP verification/delivery
* System integration
* I/O design
* Physical implementation
snip2:
Whether you choose DDR-SDRAM, RLDRAM, FCRAM, or SDRAM, Databahn's direct integration with over 5000 memory components on eMemory.com ensures that your memory system is compatible with the latest vendor-specific devices.
Connection 1: possible DDR memory controller infringement; how can Denali sell this IP w/o license or would they require their customers to take on eventual liability?
In the list of partners, Rambus is mentioned:
http://www.denali.com/partners_soc.html
Connection 3: This press release, but this is about PCI-Express:
http://www.rambus.com/news/pressrelease.cfm?id=106
(and not so much soft IP core, but more the verification step in the design, which appears to be Denali's main (software) product.)
The reason which prompted this outburst from me was this release now from IBM'Asic design department:
IBM is pleased to announce that it has licensed the Denali Databahn memory controller core offering for Cu-11 and Cu-08 technologies.
The Databahn core offering consists of memory controllers for DDR1 and DDR2 applications. The cores have been developed by Denali thru the Silicon Horizons Plus program, allowing IBM to license these cores directly to its ASIC customers.
Supporting technical documentation will be available shortly on the IP Development database and the ASIC Cores web page at ibm.com/chips/products/asics/products/cores/corelist.html. Further information on Denali IP offerings can be found at www.denali.com.
Connection 4:Here is a list of the Denali partners "on the other side", leading memory vendors.
http://www.denali.com/partners_memory.html
(alphabetically: AMD, CY, edram, Fujitsu, GSI, Hynix, IBM Micro, IDT, Infineon, Intel, Micron, Nec, Renesas, Samsung, Sandisk, Sharp, SST, Toshiba)
and from this page also:
Denali is an active member of JEDEC, the semiconductor engineering standardization body of the Electronics Industries Alliance (EIA), a national trade organization representing manufacturers in all areas of the electronics industry for over 74 years. ...
Hoping to connect this up a bit more later. In 2000 (and also later afaik) IBM (MicroElectronics) was the world's leading ASIC supplier.
Cheers
Cor
As for packing, remember those sage words...when in doubt, throw it out.
Hey Threejack,
I seem to have been living this for te past 15 years, first moving from a large house to a large apartment (but a lot smaller than the house), then to storage as I went off sailing and then from storage to a smaller apartment. Now the house I am moving to is small in some places but a bit larger in total, main diff. being that it is a house (with grounds, garage etc, so more flexibility)
I have been throwing out and giving away lots of things over those years. Trouble is I have stuff nobody wants and I think it is too good for the bin ;)
Anyway, picking up things which have not been picked up for 10 years make you wonder. Looking at a 300 MB hd which is left over also makes you wonder, could I do something useful with this inside one of my computers? Problem with giving away items like that is that people expect lots of advice, compatibility with WIN-XP and what have you with the present, so maybe it's the bin after all:(
Cheers
Cor
Hey Threejack
Thanks for posting some stuff on my message board.
Sandy would be very good for that, she ferrets out the connections. (I wish she would report all AMD patents... joke:))
I am always looking for more participation. Just back from a 2 week trip to finalize my Spanish home buying and back in Holland today (after 16.5 hours drive, should go to bed now;)
So the catching up will take several days, maybe weeks as I am moving permanently to Spain in two weeks, which means packing and all that stuff;(
Cheers
Cor
Thanks Threejack and also thanks for your two posts there today on the Rambus board.
Everybody is welcome to post, be it a news byte, an industry thing, a particular type of chip/interface/memory or just say hello.
And kudos to the guys of phpBB who made all this possible with very little setup work for me to do.
Cheers
Cor
Ok that is a (big) difference then with hypertransport afaik.
The Buffers don't "re-transmit" data to the next node, they use pass-through logic. Maybe some additional delay, but minor compared to the DRAM core latency.
Cheers
Cor
Elixe, are you sure about this?
The flight time to the furthest module/buffer is almost insignificant compared to the DRAM core latency.
You are talking flight timebut i thought the devices acted like "tunnels" and retransmit the data to the next node, causing a delay per node?
Cheers
Cor
The FB-DIMM, the more I look at it the more it looks like a Hypertransport connection. Very similar. LVDS, separate read and write paths, each controller acts as a bridge... the clocking is slightly different I believe.
The worry would be the latency on the farther away modules. Maybe latency is not such an issue in a server as it is in a desktop/workstation. Capacity is the main thing. Good thing Intel can address these Gigabytes when they have the AMD64 instructions working, LOL:)
I suppose in the server you can make sure in software that the most accessed stuff resided in the front DIMM, for example tables and directories.
All in all is going to be interesting.
Cheers
Cor
PS look in at my message board and for heaven's sake post something....
http://chipstocks.net/semi-subjects/phpBB2/index.php
The Rambus clock is pretty ingenious.
Cal I could not agree more, it was a beauty.
Cheers
Cor
Thanks Paul,
lost track of all the claims;)
The recent patent I referenced in my last post does not include first and second external clock signal terms in the claims. They reference an external clock signal that is used for the internal clock signal. The new patent should have no problem being asserted against DDR with respect to the claim terms. The claims being broader makes them more vulnerable to prior art problems. Nevertheless, the new patent could be used in new litigation without claim term problems.
Cheers
Cor
Hi Paul,
on the clocks, it is useful to look at the pdf's published on the rambus website on the hynix case.
There is a difference between the two clock schemes for rdram and DDR (and we know te patents were based on rdram).
In rdram the clocks were used to determine the delay to the particular chip in the chain by having one clock going directly and one via the "back way". In DDR the clock signals are CLK and CLK_ (the reverse of CLK) both going one way.
Naturally having those two versions of the same clock help in dteremine the correct crossing point for data, but it is not the same as what was happenign with rdram and possibly also in the patent language.
Cheers
Cor
But maybe a problem with PS3 here, see:
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?t=58
Elixe thanks for your seriously addressing my questions on this. I have meanwhile read the cell patents (or at least some five of them) and am getting a better idea.
What really interests me, is if IBM is putting new life into their PowerPC architecture, which has been aroudn for quite a while. (I believe it was in those XS6000s I played with some 15 years ago, wrote some comms protocol emulation for it)
Are they going to give ARMHY some trouble?
Cheers
Cor
Oh btw, don;t take that link too seriously, look at the Sony guy's name:)
Interesting announcement from IBM (Power5)
see:
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?t=55
Also a new China connection here.
Do we still expect PS3 with "Cell" or will it be Power5 now that Sony licensed Power?
New Licensee
* Sony today disclosed that it has licensed the Power Architecture from IBM. Sony said Power offers versatility and a unique combination of low power and high performance, making it optimal for a wide range of consumer devices.
Note: Power5 could also use Xdram imo.
Cheers
Cor
Mobile is another target. DDR2 will provide significant improvement in battery life.
And Hynix is playing for that market right now:
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?t=47
Hynix Announces Industry First 1GB DDR2 SODIMM
Quote:
Seoul, Korea, March 29, 2004 - Hynix Semiconductor Inc. (www.hynix.com) today announced the launch of its industry-first 1GB DDR2 SODIMM (Small Out Dual In-line Memory Module) manufactured on its leading edge 0.11-micron technology.
The newly launched 1GB DDR2 SODIMM, supporting 400 and 533MHz, were developed to meet the projected DDR2 demand for notebook applications in second half of this year. Hynix is planning to begin mass production of 1GB DDR2 SODIMM next quarter to coincide with the release of Intel DDR2 chipset.
Cheers
Cor
Og btw I would not call it technical journalism, but technical satire;)
Jeff I have added a piece on this using all german sources, they are cross and misquoting happily.
Partial translations and commentary from me, including this one:
Some background on von Pierer:(with picture)
http://www.manager-magazin.de/magazin/artikel/0,2828,druckbild-207067-245815,00.html
The Siemens results have been under pressure these past few years. They have been shored up (amongst others) by smaller payments to the Siemens pension funds by putting IFX shares instead of cash and more recently by profitable sales of IFX shares.
see:
http://chipstocks.net/semi-subjects/phpBB2/viewforum.php?f=28
enjoy;)
Cheers
Cor
Hi Threejack, replied to your post there.
I also have posted a transcript of this morning's CC at Infineon and some articles the union had about their relationship with Infineon.
Cheers
Cor
Infineon's Schumacher resigns suddenly
all details here:
http://chipstocks.net/semi-subjects/phpBB2/viewforum.php?f=28
Cheers
Cor
Legal, technical and business....
Hey Threejack,
Well I am the legal nitwit. I have no problem with people posting legal stuff in the proper threads, but USJeff is a repository of legal info, I have no desire to try to duplicate that. Maybe some posts referring to what is going on and linking that's always great.
I stay with technical (mostly lots of links, my site has over a thousand links) and business/financial.
In the case of business, where I analyse I will certainly take legal in as risks, w/o commenting on my ideas of probability of success.
But a repository, as you say, is nice.
The Hynix pre-trial activity is heating up with the claim construction hearings, though. Will be interesting to see the outcome.
I have been following. There are some misconceptions abotu that, see also:
http://boards.fool.com/Message.asp?mid=20531755
http://boards.fool.com/Message.asp?mid=20532936
http://boards.fool.com/Message.asp?mid=20533450
http://boards.fool.com/Message.asp?mid=20533568
Cheers
Cor
Location of Hynix sub board on my site:
http://chipstocks.net/semi-subjects/phpBB2/viewforum.php?f=31
Cheers
Cor
Hi Jeff
There are an awful lot of patents in the first dicket entry as per your site (the original hynix complaint: invalid or unenforceable patents..), but I am not sure if that has not changed to less.
Anyway will keep searching. Your site in page selection has become a lot faster than it used to be btw.
Cheers
Cor
Jeff, re: Hynix patents in suit
Hey Jeff,
Do you know where I can find this list? I have it somewhere but cannot find it at the moment. Weren't there some 5 patents in that suit?
tia
Cheers
Cor
Threejack, saw your message :)
I am totally swamped by the number of features that thing has. For example I am putting in a thread for every of the semi companies, which I track on my site (about 30 companies, will soon be 40). If I put them in as boards, they all get on the front page and you realy have to scroll down a lot to get for example to Semiconductor Technologies.
So I am downgrading them to "topics" so as to keep the first page compact.
Next: to make it look and feel like the rest of my site.
Those guys of the open source community really have done a great job in making this BB. I have also looked at blog software but so far saw too many problems and much less "grown up" software.
Cheers
Cor
Hey Threejack re my message board
This is not intended as a competition to TMF or these boards, but I am adding a message board to my site www.chipstocks.net
This to be used as a repository of semiconductor industry interesting stuff to keep.
I have put a few posts and subjects up and this is purely in a test phase. Would appreciate if some of you would fine the time to visit and try to post. Do not post something which takes lots of time now, because it may have to change yet.
The direct patch to the main index is:
http://chipstocks.net/semi-subjects/phpBB2/index.php
I would be interested to know afore you go there if you can read one specific post directly w/o regsitering, try this one:
http://chipstocks.net/semi-subjects/phpBB2/viewtopic.php?t=8
The bulletin board system is from the open source community and it is really incredible what this community, which gives the use away for free, makes of these products nowadays.
This one is from phpBB, using servers side scripts.
I hope to hear either here or there what you think about it.
Cheers
Cor
Interesting chipset that SiS Northbridge, Elixe.
As they are expecting some extremely high prices for DDR2 initially this year, plus the fact that DDR is fast enough anyway for the fsb now it may be a good one for transitional period.
And at the low speed DDR400, I believe, DDR2 has more disadvantages than advantages to DDR (mostly latency related afaik).
Cheers
Cor
I will change it to USJeff.com.
Cheers
Cor
Jeff, I suddenly realise I still have a link to rambusite on my site (chipstocks.net).
Should I change it or delete it, I mean how invisible do you want to be?
Cheers
Cor
Hey Jeff,
welcome back, and mum is the word:)
Cheers
Cor
" PS3 will use four 512Mb XDR DRAMs"
As you say:
The PS2 uses two 128Mb RDRAMs, 32MB of memory.
Rumors are the PS3 will use four 512Mb XDR DRAMs, 256MB of memory. So I would expect higher income per PS3.
That is a factor of eight, so I agree it is probably more (especially as cdr will be more expensive than rdram initially). Otoh we are talking comparing today's (and yesterday's) rdram prices with tomorrow's xdr prices. We see more or less the same price per chip (order of magnitude $ 5 I believe), whether it is 128 Mb, 256 Mb or 512 Mb. (when the lifecycle is ripe)
I did not realise the difference in play "behaviourism" between PS2 and PS3 to be honest I have never seen the products, don't need to see them any more than I need a line of coke;)
Thanks for the info.
Cheers
Cor
Rambusite, curtains
Threejack, I decided to throw my two cents on the yahoo board (something I seldom do, because "Dopie" is seen as a personification of the antichrist there by many, LOL;)
Here is the post:
http://finance.messages.yahoo.com/bbs?.mm=FN&action=m&board=4687909&tid=rmbs&sid=468...
Cheers
Cor
On PS3 main memory
Hi Elixe,
The claims seem a bit overdone;)
I agree with your cell perspective, the broadband / internet connected family of cells. It is a bit similar to the distributed computing used for some radiotelescope data crunching and the cancer research project Intel sponsors. (I did do that some time and ut in some years of cpu time, but stopped when the organization started doing things on their own, like switching to anthrax, w/o telling or leaving a choice; anyway that behind the scenes access to one's PC is a bit scary these days;)
PS3 will use a relatively small DRAM capacity.
Do you have an idea of the amount of memory in the PS3? Did the PS2 have some 64 MB rdram?
To come back to rambus, if the PS3 follows on the PS2, eliminating a major user of rdram, but transferring it to xdram for all we know (not 100% sure imo) wouldn't the income to rambus be similar before and after PS3? (there would be a dip in rdram as the PS2 gets "osbourned", which I expect from later this year until late next year.
There is the one hard pr here:(I cannot find any others)
http://www.rambus.com/news/pressrelease.cfm?id=78
(Sony and Toshiba sign up for redwood and yellowstone 2003-01-06, note the logic to logic part)
Two new interfaces selected for logic-to-memory and logic-to-logic connectivity
Cheers
Cor
Elixe re Playstation 3
I have been spending some time reading one of Sony's patent applications, # 20020138637
Computer architecture and software cells for broadband networks
I guess this is the one you have perused yourself.
(also read # 20040039861, which is interesting in that it mentions the Transmeta Crusoe)
I was wondering whether you think that memory switch in figures 12 etc is particularly suited to xdram?
Cheers
Cor
Re Rambusite
Threejack I think we agree on what happened. The points is that all those documents which are not available as pdfs were there (VA1 a.o) and even though navigation was tedious, it served a purpose.
I don't know if those older docs ever came later as pdfs.
Cheers
Cor
PS calroofer now also calpainter? Maybe he should take a vacation in Spain later this year at my new house; I am sure I want to change some things;)
Hey Threejack, re Rambusite, have you noticed that it is not up any more? Not only that, the domain name is for sale.
http://www.rambusite.com/
Any idea why this is?
I could imagine reasons of cost or time (I know from firsthand experience how that works:)
Cheers
Cor
Do you have a point?
No. Other than I saw them prominently featured.
Thanks for the list, I had not dug that deep yet.
Cheers
Cor
Re:MIF
And co-sponsored by Dell, Intel and HP.
Cheers
Cor
Elixe, re:I believe that PCI-Express supports simultaneous bi-directional transfers.
Are you talking about simultaneous transfer on the individual data lines or lanes in one direction and a different set of lanes in the other direction? (as with Hypertransport)
Cheers
Cor
Intel's new memory bus architecture?
Hi Elixe,
Did you see the article on this type of simultaneous bidirectional data transfer, of which "our" Horowitz is the co-author?
Have to search for the link, and is bedtime now, but maybe you saw it.
Cheers
Cor