|
|
|||||||||||||||
 |
|||||||||||||||
|
|
|||||||||||||||
|
|||||||||||||||
|
|
|||||||||||||||
|
|||||||||||||||
|
|||||||||||||||
|
|||||||||||||||
|
|||||||||||||||
|
|||||||||||||||
|
|
|||||||||||||||
|
|||||||||||||||
|
|
|||||||||||||||
| 2015 © KickAss Gear | |||||||||||||||
|
|
|||||||||||||||
|
|
| KickAss
Gear News Archive: March 2001 March 30th The New Thunderbirds? We have been intrigued by the sudden appearance of the 1.3 and 1.33GHz Athlon processors with the "Thunderbird" core. As you all may recall, the original Thunderbirds that came out last year typically overclocked to between 1.1GHz, and 1.2GHz. To achieve these OC speeds, the core voltage needed to be raised from 1.65 volts to 1.75 or 1.8volts. Now we are starting to get in the newer Thunderbirds that are rated at 1.2GHz, and have noticed some interesting things. First off, the L1 bridges which control the multiplier lock are intact! Normally they are cut at the factory with a laser, but these were untouched, and the chips could be run at any multiplier setting we threw at them. I don't know if this is a case where AMD just missed a batch, or if they are giving up on the lock concept altogether. (We should have some 1.3GHz chips in any day now, and I'll be interested to see if they have the multiplier lock or not). Next we noticed that the 1.2GHz chips that arrived (both 200MHz and 266MHz versions) would overclock to 1.33GHz (10 x 133MHz) at their default voltage (which happens to now be 1.75v). This is also contrary to our previous experience with overclocking Athlons. The chips would even run at 1.4GHz with a core voltage boost. That is far faster than the older Thunderbirds would go. I also noticed that the chips were running cooler than their older cousins, even at the higher clock speeds. To my knowledge, this has never happened in a chip line before without a change made to the design or packaging. So the question is... What did AMD do to the new Athlons that lets them run faster, and cooler, without raising the core voltage? Suggestions I've heard range from AMD improving their fabrication process (better etching), to making the chip packaging with pure Silicon 28, which improves the thermal characteristics of the chip (thanks Mike!). I don't know, but we will have more on this as we get more information. March 28th Mike Magee Leaves Register Mike Magee, the founder and soul of The Register resigned last Friday. Apparently, disagreements between Mike and several others at The Reg led to the falling out. I'm not entirely sure that the staff at The Reg fully understand what they have done. I found myself spending almost no time at their web site while Mike was on sabbatical. Mike has a new web site, which will probably become more popular than The Rag, as I now call it. Mike's new site, The Inquirer, can be found here. The site is just getting started, but several of the old Reg staff, like Pete Sherriff, have contributed articles already. Good luck Mike! We're all rooting for you. March 26th SiSoft's Sandra Updated SiSoft has released an updated version of their popular benchmarking and diagnostics program, Sandra. The new version, 2001se, offers better Windows 2K support, an improved interface, better multiprocessor support, and enhanced Pentium 4 support. Sadly, but not unexpectedly, Glide support has been removed :(. Micron PC... RIP It may sound like an early April fools joke, but it's not. Micron Electronics is selling it's PC manufacturing division, and getting out of the PC business. The reduction in competition should be good for other PC makers, but not for the consumer. Micron was a leader in producing high-end PCs, and even made their own high-performance chipsets. The operating loss last year for the PC division was $159 million. No wonder it's so hard to make a living selling computers, these bozos were obviously selling them below cost!! Consumers collectively got Micron computers last year at 159 million dollars less than it cost Micron to make, test, and ship them. Now we know why computers have gotten so cheap. Companies like Micron decided they don't need to make a profit on them anymore! This disturbing trend has infected Intel as well. A report over at EBN suggests Intel is subsidizing the high cost of P4 production to keep the chips rolling out the door. Pentium 4 discounting has cut into Intel's normally hefty profits, and may force them to move to the first P4 die shrink (0.13 microns) sooner than anticipated. March 22nd Rambus Case Takes Another Turn! EBN has a new piece on the upcoming Rambus trial. The court case was supposed to begin earlier this week, but was postponed until April 10th when new evidence was found on an old hard drive from Siemans Semiconductor. Apparently, the hard drive contained a document which indicated that in 1992, Siemans was well aware of the Rambus design, and were trying to make selective use of certain design aspects for the upcoming SDRAM standard. To me this seems a little far fetched, since it was JEDEC, not Siemans, that was working out the new SDRAM standard. In any case, I expect that the Rambus stock price may rise again on this break in the case. March 21st What Does Intel Really Think About Rambus? Jack Robertson over at EBN has written a piece on Intel's upcoming Brookdale chipset, which gives PC-133 SDRAM support to the Pentium 4 processor. Intel estimates that the Brookdale will account for 1/3 of all P-4 chipset sales this year. If true, it means Intel expects much greater things from Brookdale, than from the Rambus compatible i850 chipset. Considering that the Brookdale chipset has not even debuted yet, Intel obviously anticipates that Brookdale motherboards will be far more popular than current Pentium 4 motherboards. This probably means at least two things. Intel recognizes the threat that AMD poses with their high-performance, low-cost processors. And second, it indicates that Intel is losing confidence that it will be able to succeed with a Rambus-only platform. Prices have dropped so far on high-end processors and SDRAM in the last six months, that Intel's high-priced Pentium 4/Rambus combination seems out of touch. But the plain truth of the matter is, that clock for clock, the Pentium 4 is slower than the Athlon, even when the Pentium 4 is coupled with Rambus memory. Mating the Pentium 4 with SDRAM will further shift the equation in AMD's favor. Intel's only hope is to push the Pentium 4 higher in clock speed. They have plenty of head room left, and I suggest they start using it. March 20th It Depends on What The Meaning of "IC" Is. It may sound a little far-fetched, but the outcome of the upcoming Rambus court case may hinge on what the meaning of IC is. IC stands simply for integrated circuit. And yes, that's correct, the litigants in this case can't even agree on the definition of the term integrated circuit. And yet, much of the case depends on precise definitions for several terms. Here is an excerpt. "The parties agree that all the terms have the same meaning throughout with the exception of 'integrated circuit device'. The Defendants contend that this term has a different meaning in one patent due to representations made to the Patent and Trademark Office during the prosecution of that patent." Interesting, but perhaps the real crux of the case lies in the definitions of the words "multiplex" and "bus". Apparently, in all of their patent applications, Rambus defined their new memory interface as a multiplexed address, data, and control bus. For those of you who are technically inclined, Rambus Inc. has always maintained that the big advantage of Rambus DRAM was reducing the pin count on the memory sticks and sockets. They achieved this by multiplexing. Multiplexing simply refers to sending more than one type of data or control signal over the same circuit lines. On the other hand, SDRAM and DDR DRAM use distinct circuit lines to send control, data, and address information. If the distinction between standard and multiplexed memory buses holds, then Rambus will almost certainly lose this court battle. In the end then, it will be the wording of the patent documents themselves that may be the undoing of Rambus. The worst outcome for Rambus will be if the judge and jury determine that Rambus intentionally withheld information on their patent applications during their participation in a memory standards committee (JEDEC). If those allegations stick, Rambus could face racketeering charges under the RICO act. If you are deathly bored, you might want to read the judge's memorandum in the case here. There are links at the bottom of this article to the 77 page tome written by Judge Payne. He clearly has an excellent grasp of the technical details of this case. March 19th We Never Stop Litigating I thought it might be fun to head over to Rambus.com and check out what they had to say. I found their new slogan quite amusing. "We never stop innovating". Who are they kidding? Rambus has more lawyers in their employment than engineers! I suggest they change their slogan to "we never stop litigating". Mike Magee over at The Register suggests that the court case may end up hurting Intel. Intel has pretty much bet the farm on Rambus memory, and the current court case has undoubtedly sent collective shivers up Intel's spine. Not only is their technology partner concentrating more on litigation than innovation, they have pretty much infuriated the whole rest of the industry. Some of the ill will towards Rambus must eventually rub off on Intel. The court case was originally scheduled to begin tomorrow, but now it has been delayed until April 10th. It should be very interesting to hear the details that come out of the trial. So far the pretrial rulings have pretty much gone against Rambus. If the actual court case continues in this fashion, things aren't looking too good for the world's most infamous memory litigation company. March 16th Rambus Racketeers It seems a long way from being a company with the premier up-and-coming memory standard, to a company facing fraud and racketeering charges, but that's where Rambus Inc. finds itself today. While the stories yesterday that the judge in the case had already ruled against Rambus were premature, the case is definitely shaping up against Rambus. A ruling that does appear to have been made yesterday was that fraud and racketeering charges against Rambus could be heard by the court. I'm sure that's not how Rambus lawyers had envisioned this case going before opening arguments even began. Rambus share prices slid sharply yesterday, but may bounce back slightly over the next few days if no more rumors leak from the trial. But if it were my money, I'd think T-bills would be a safer bet! :) PS, If Rambus officials had simply learned how to play nice as children, this whole mess could have been avoided, and they might even still have some friends left. March 15th The Day The Rambus Died The ides of March seems a perfect day for this. Et tu, Rambus? Don't count them out just yet, but things are looking a tad scary for the patent wielding company. Judge Robert Payne delivered a painful ruling in pre-trial hearings yesterday, effectively hacking the Rambus defense off at the knees. Expert testimony went like this: “(The Markman) testimony was that the Rambus patent is a multiplex bus, and that we don’t use the multiplex bus in our SDRAM or DDR products. It means that because we don’t use the multiplex bus they use in their patents…we are not in violation of their patents.” The Judge agreed! This ruling basically undercuts the entire Rambus case against Infineon, Micron and Hyundai. The Judge will make additional rulings today, March 15th, but that may not be the end of it. Judge Payne delivered another payneful blow when he ruled that fraud charges could proceed against Rambus for their inappropriate use of the JEDEC standards committee. This seems almost too good to be true. “Attorney-client privilege has been forfeited under the crime-fraud exception as to certain topics, and therefore the Defendants may conduct depositions of Messrs Diepenbrock, Vincent, Crisp, Mitchell and Tate respecting the legal advice provided about disclosures of patents and patent applications to JEDEC by Rambus Inc, the disclosure policy of JEDEC and about the efforts by Rambus, Inc. to broaden its patents to cover matters pertaining to the JEDEC standards.” The end may be near for Rambus. Now they are not only facing the possibility of having to pay back all their ill-gotten royalties (which would break them), but of possibly facing criminal charges for their conduct as well! Oh how the fortunes of war can change. Links: Electronic News March 14th Mad Onion Makes
People Cry Those folks at MadOnion dot com have just released the newest version of their benchmark, 3D Mark 2001. And just like last time, the new version will make you cry, because you will quickly learn that your expensive new computer is not going to run the next round of 3D games above 15 frames per second. Older systems will display these benchmarks like a slide show. With the default settings (1024 x 768 x 32 bit), a Pentium III 633 machine with an overclocked Voodoo 5 card will give you a 3D Mark score of around 1200. Frame rates in high detail benchmarks dropped to as low as 1 frame per second with antialiasing turned off. That's not so good. A similar PIII with an overclocked GeForce 1 DDR card provided a default score of about 2000. Most of the high-detail benchmarks ran unacceptably slowly with this combination as well. Finally, testing a GeForce 2 DDR Pro card on an Athlon running at 1GHz gave us a default score of around 3000. Alas, many of the high-detail benchmarks still looked like a slide show. Very pretty... not many fps. I need to play around with settings and driver versions, but it's pretty apparent already that even a high-end rig will be brought to it's knees, and rendered useless, by 3D games with these requirements. If upcoming games like Unreal2 are as hardware-hungry as 3D Mark 2001, we are all in trouble. Maybe a 1.3GHz Athlon with a GeForce 3 card would help? Maybe not. Link: Madonion.com March 13th Is the 1.33GHz Athlon a Palomino? The Register suggests in a recent article that the upcoming AMD 1.33GHz Athlon may be based on the new "Palomino" CPU core. But based on everything I've been able to find out, the so-called "Palomino core" will not debut until this Summer, despite the announced 1st quarter release date. The confusion comes into play because the "Thunderbird-core" was not supposed to clock above 1.2GHz according to AMD. That would suggest that both the upcoming 1.3GHz, and 1.33GHz models should be Palomino-based (one might be a 100MHz model, and the other a 133MHz model). But if these CPUs were based on Palomino, AMD would have no reason for not disclosing the fact. There is one other possibility. We know from experience that AMD "cherry picks" their current 1.2GHz processors, probably from the center of the fabrication wafers. It is possible that some of them have been found to run well at 1.3GHz, and that's where the new chips are coming from. If so, expect availability of 1.3GHz Athlons to be poor initially. March 12th Register This Rambus! The Rambus court case is set to start March 20th. Financial experts expect a quick win for the dishonest memory design firm. No bad deed goes un-rewarded, right? But now we take a turn into La-La Land. It seems that a court filing by Rambus claims that no testimony is needed from ex-Rambus employee Richard Crisp, because he has not worked for Rambus since 1996. For those of you who do not recall, Richard Crisp was the Rambus employee who sent around numerous e-mails within the company which indicated that Rambus was planning on using submarine patent tactics while participating in the JEDEC memory standards committee. Rambus used information derived from the memory standards meetings to modify their existing patent applications to cover SDRAM and memory controllers. At the same time, they did not disclose the patents they were working on as required by JEDEC rules. They then quit the standards committee and finished filing their amended patent applications. The patents were not granted until 1999 and 2000, long after SDRAM had been on the market. With that background information we now move ahead to the present time. Rambus Inc. has filed a motion with the court to disregard any testimony from Richard Crisp, because his employment at Rambus ended in 1996. The Register now has a story about a meeting they had in 1999 with a certain Rambus employee named, you guessed it, Richard Crisp. According to The Register, Mr. Crisp was clearly still the employee of Rambus Inc. during that interview in 1999. So there you have it folks. The Register may even come into play in the upcoming trial. If Rambus continues to insist that Mr. Crisp has not been in their employment since 1996, the plaintiffs in the case may be forced to call Mike McGee of The Register to the stand to testify. How much weirder can it get? March 10th Buy PC Memory... Now! PC memory prices are at an all-time low. Indeed, it seems like memory manufacturers are in the middle of a massive inventory clearance sale. Part of the problem in memory land has to do with the general downturn in computer sales. But another problem is that memory manufacturers are at an uncertain crossroads. Do they emphasize SDRAM production, DDR DRAM production, or Rambus DRAM production? Neither they, nor the public, are certain which way to go. One other factor that led to the current situation of excess inventory was the memory shortage that occurred just over one year ago. This drove memory prices so absurdly high that memory manufacturers geared up production in an effort to cash in on the price hikes. Overproduction of SDRAM during the subsequent year was partially responsible for the fix that memory manufacturers now find themselves in. In any case, it's a great time to buy SDRAM. Prices can't go much lower, and you might just help keep some memory company from going out of business. March 8th The PC is Dead Jim.... Again! There have
been many stories about the premature death of the desktop PC over the
last 10 years. The reasons
given for the eventual demise of the PC range from a takeover by laptops
and handheld devices, to death by obsolescence, or even replacement by
Internet devices and smart home entertainment systems. The PlayStation 2
and Microsoft's upcoming Xbox are examples of inexpensive PC replacements,
and it is no coincidence that Sony and Microsoft are behind them.
Both of these giant corporations would love to eventually replace
the PC with these proprietary devices. Indeed,
that is the crux of the latest iteration of the PC-is-dead story.
Hale Landis
has been involved in the development of the ATA standard for personal
computers, and it was a recent commentary
by him that led to these new speculations about the impending death of the
desktop PC. He says that the
entertainment industry is putting all its corporate money and might into
killing the PC. Nobody can
make money on an open standard architecture, so many entertainment and
software corporations are working on a two-pronged attack on the PC.
One prong is to increase copy protection within devices that go
into modern PCs. This
includes strengthening DVD encryption methods, and even adding copy
protection schemes to upcoming hard drives.
The basic idea is to cripple the home PC's ability to copy and
store copyrighted digital material. The second prong of the entertainment industry's attack on the PC is to subsidize inexpensive, copy protected, proprietary PC replacement devices. Indeed it is estimated that Microsoft may lose as much as $125 on each Xbox they sell. It is assumed that Microsoft will match or beat the street price on the PlayStation 2, despite the fact that the Xbox will be more expensive to produce. But the logic behind this is inescapable. If the tides turn against the PC and in favor of Internet-capable consoles, Microsoft would lose big-time through declining sales of their operating systems. The Xbox would put them in a good position to come out on top during that transition. Some would
say that the recent spate of litigation against Napster, and against
authors of DVD encryption cracks, is the beginning of a third prong of the
entertainment industry's attack on the PC. And while
I buy all of the above arguments, I still say that the rumor of the PC's
death is premature. For one
thing, Intel, AMD, and others have a large and lucrative stake in keeping
PCs as they exist today alive, and many other smaller companies are making
a decent living at producing the parts in PCs.
The profit margins may not be as good as everyone would like, but
they are still making money. As
long as people buy PCs and PC parts, this will continue.
If on the other hand, 95% of the home computer-using population
moves over to Internet-ready game consoles, that will change.
But I
don't see that happening. I can't think of a single popular home
electronic device or system that has been completely replaced or fallen
from favor in the entire history of home electronics.
Standards have changed, but the technologies remain.
Hell, people still use tape recorders for crying out loud. Home
stereo systems are as popular as ever, and so are VCRs. The same will be
true for home PCs. They will
change and improve over time, but I expect a derivative of the home PC
will still be around in the distant future. Whether they are toaster-like devices that you can't even
open up, or plug and play, do-it-yourself computing devices with an open
architecture will depend upon the buying public. Considering the Mac G4 cube is not nearly as popular as
full-sized G4 systems suggests to me that even Mac fanatics would rather
have an upgradeable computer system, instead of a sealed home appliance. March 6th GeForce 3 Reviews... Finally! TweakMax has finally posted a review of the GeForce 3 with benchmarks included. The results are just what we expected. No improvement over the GeForce2 Ultra, and in some cases, slightly slower performance than the GF2 Ultra. This is because the drivers have not been optimized yet, and games have not been re-written to take advantage of the GF3's features. There is also a GF3 card review at Digit-Life. Keep in mind that games will not have built-in support for the GeForce-3 card for at least 6 months, and probably more like a year. NVidia may have bet wrong this time, since most folks won't pay $500 for a card that is not any faster than a GF2 card. But when Quake 4 and Unreal Tournament-2 come out late this year, or next year, you will definitely want a GeForce-3 card. Hopefully by then, they will be $300, rather than $500. Listen to the Daily Rumor for more. March 5th Pentium 4 vs. Athlon/DDR Scott Wasson at The Tech Report has posted his slugfest between Intel's Pentium 4 processor, and the AMD Athlon with 266MHz DDR memory support. It's the comparison we've all been waiting for! Scott is very kind to the Pentium 4, but on a bang-for-the-buck basis, (and in many benchmarks) the Athlon still wins. DDR memory definitely helped with the Athlon's performance, but not quite as much as everyone had hoped. When you're all done reading about the new DDR support for Athlon processors, you'll probably go out and buy an Athlon with PC133 SDRAM on a KT7A-RAID motherboard! :) March 3rd Intergraph to Become Next Rambus Inc.? Now that Intergraph has won its court battle against Intel, the royalties may begin to fly. It's still hard to determine how devastating this court loss will be for Intel. The nub of the situation is that several key components of the cache system in Pentium 1, Pentium II, and Pentium III computer chips were patented by Intergraph before Intel used them. Since it has been determined that these elements of the Pentium cache infringe on Intergraph patents, Intel will need to pay royalties, presumably for all previous processors sold in the Pentium line. Even at pennies per processor, this is going to add up big. But that's not the end of the story folks. It seems quite likely that all x86 processors, including the AMD line, use similar cache architectures. It is quite possible that Intergraph will go after the other chipmakers once it has finished with Intel's carcass. If so, you can expect the extra costs to be passed on to each company's customers. So is Intergraph positioning itself to become the Rambus Inc. of CPUs? The parallels are frightening. Neither Intergraph nor Rambus make the products that they hold patents on. Both companies are relatively very small compared with the corporate giants that they will be collecting royalties from. And both Intergraph and Rambus could drive up PC prices by charging large royalties for products they have no hand in producing. Also, in both cases, the patents were issued many years before the patent holders decided to try and collect royalties. Whether Intergraph will turn its attention to AMD is still an open question. Intergraph has had a long running fight with Intel, but not with AMD. Indeed, in 1999 Intergraph exited the PC business claiming it had been forced out by Intel's unscrupulous actions. So Intergraph has a much bigger bone to pick with Intel than with AMD. March 2nd Rambus May Start to Lose Royalties Several months back NEC and Hitachi formed a joint DRAM venture that they named Elpida (I know, it's a dumb name, but that's another story). In order to avoid litigation from Rambus Inc., Elpida agreed to pay Rambus royalties for SDRAM products that Elpida produced. But now the president of Elpida has announced that they will renegotiate their terms with Rambus, because several other large memory manufacturers have refused to pay Rambus royalties. Hyundai, Infineon and Micron have opted instead to take Rambus Inc. to court for antitrust violations. The president of Elpida has said this puts his company at a disadvantage relative to the companies that do not pay Rambus royalties. I do not expect this announcement worries Rambus very much, but it does indicate a large reservoir of ill-will in the memory industry towards Rambus. The entire industry is awaiting the outcome of the antitrust court case against Rambus. And while many suspected that the case would end quickly in Rambus' favor, it now appears that it may drag on for some time. As the case drags on, other memory manufacturers who agreed to pay royalties to Rambus may reconsider their positions as well. In my mind, it would be wonderful if Rambus started to run short on cash to pay their army of lawyers. Everyone knows, when the money goes, so do the lawyers. But I'm not holding my breath just yet. Intel Loses Lengthy Court Battle with Intergraph A very long and complicated court case between Intel and Intergraph was finally decided in Intergraph's favor. The U.S. Federal Circuit Court of Appeals ruled in favor of Intergraph concerning patents that Intergraph acquired. These patents pertain to so-called Clipper technology. Many of these patents involve accessing cache memory, and the technology is used in previous and current Pentium processors. The court ruling means that Intel will now be forced to pay Intergraph royalties, and these costs may have to be passed on to consumers. If so, Intel may have even more trouble competing with AMD on processor pricing. Indeed, this court ruling puts Intel in a sticky situation, where certain aspects of their core technology have been patented by an angry rival corporation. There has been much bad blood in recent years between Intel and Intergraph, and now it is probably time for some payback. Maybe now Intel knows a little bit better how memory manufacturers must feel about paying royalties to Rambus Inc. March 1st Microsoft... Innocent, Judge Jackson... Guilty! Microsoft's supporters have been blasting Judge Thomas Penfield Jackson's rulings in the Microsoft antitrust case for many months now. Judge Jackson ordered the breakup of the software giant into operating systems and applications subdivisions. Now the case has reached the appeals court. The appeals court case, which began on Monday, is clearly leaning in the direction of Microsoft. The seven appellate court justices seem intent on ignoring all the technical issues, the new legal territory, and all of the wrongdoing done by Microsoft, and instead are focusing on taking apart Judge Jackson's rulings piece by piece. The appeals court justices have apparently already decided in Microsoft's favor, and are just trying to justify the decision. They clearly do not understand the technical aspects of the case, and don't care to. They do not even seem to understand simple logic in some instances. For example, concerning the Netscape vs. Explorer issue, the justices state: "By definition, they [the public] must already have heard about it [Netscape], or they wouldn't want it. If you can get it it's not foreclosed." So there you have folks, because most Windows users never heard of Netscape, they were not being denied anything, due to their ignorance. And those who knew about it, and did want it, could just go out and buy it. Therefore no problem whatsoever existed! Hmmm. Paraphrasing Mel Brooks: "It's good to be the Judge." Perhaps the worst outcome of letting Microsoft off Scott free is that they will feel vindicated despite their despicable corporate behavior. Indeed I expect Bill Gates will probably ratchet up more of the same type of tactics that got him in hot water in the first place. Why stop being an unethical jerk if the courts say it's OK? This will also open the floodgates for other large corporations to feel they have a free hand in dealing with competitors anyway they like, regardless of ethics, and in a vacuum of law. So rather than doing the tough thing, namely drawing a legal line in the sand that will stand the test of time, the appeals court seems headed for a quick and easy resolution in Microsoft's favor. I can't imagine a more unjust outcome. Copyright 2001, KickAss Gear |