November 30th, 2000

Dragon Down!

No I am not talking about a new 3-D role-playing game. I'm talking about Dragon Systems, a division of Lernout & Hauspie, a company that specializes in speech recognition software.

Speech recognition firm Lernout & Hauspie filed for bankruptcy protection in the U.S. yesterday. This company is best known for their Dragon Naturally Speaking software, which I happen to be using at this very moment to write this article. The software is excellent, but apparently their bookkeeping was not.

Reports began surfacing over the summer that indicated large profit shortfalls in the company's Korean sales figures. By the time this was thoroughly investigated it appeared that the sales numbers in Korea had been inflated nearly 10 times. The company says that it will reorganize and stay in business. This is good news to me because speech recognition is obviously the way we will interact with computers in the future. I want as much competition in the area of speech recognition as possible. Losing one of the larger players would not be good for competition.

                             Dr. John

November 29th, 2000

Abit VP6 Dual PIII Board Coming Soon.

For those of you who have been waiting for a really good, overclockable dual Pentium III motherboard, your wait is almost over. Abit is getting ready to release its new VP6 motherboard, with an estimated availability around mid-December. The price for the motherboard should the approximately $160. Here is what Abit has to say about the upcoming VP 6 motherboard.

ABIT is proud to announce the launch of the VP6.  This dual processor motherboard is a follow-up to the immensely popular BP6.  The VP6 is an ATX form-factor mainboard designed for use with all PIII Coppermine FC-PGA 370 Processors. It supports all normal industry standard functions, including Advanced Configuration and Power Management Interface (ACPI), AGP 1X/2X/4X and 100/133MHz SDRAM Memory Bus Settings. The VP6 features four 168-pin DIMM sockets supporting up to 2GB memory. 

Shhweet!

November 27th, 2000

P4 Awesome With SSE-2 Optimized Software.

The Register has a story on how well the P4 performs when the applications are optimized for the new SSE-2 extensions.  In fact the results are spectacular.  Most current applications use the underpowered floating point unit (FPU) in the P4, which hurts performance.  If those same applications were re-written to take advantage of SSE-2 instead, performance would jump dramatically.  But keep in mind that by the time software has SSE-2 support built in, newer, better, faster, cheaper Pentium 4s will be available.

                              Dr. John

November 25th, 2000

3dfx Comeback?

Reactor Critical has a speculative post about the upcoming VSA-200 "Rampage" chip from 3dfx.  It sounds nice, but without hardware assist, it may not be in the same ballpark as the NV-20 from NVidia.  

3D-Concept discusses the GP-1 chip that was made by Gigapixel, the company that merged with 3dfx earlier this year.  It will be interesting to see what 3dfx can do with this cool little chip.

November 24th, 2000

Rambus' Past Haunts Them.

The worm may be beginning to turn for Rambus Inc., but you can never tell with Rambus. First they're up, then they're down, then they're back up again. Well the roller coaster ride just goes on and on. When the initial benchmarks of the Pentium 4  were released, they indicated that Rambus really does offer a large memory bandwidth increase over previous memory types. This had to be excellent news for Rambus, whose memory design has been blasted repeatedly in earlier benchmark comparisons with existing memory types. But just when Rambus should be feeling their oats and riding high, their past questionable behavior comes back to haunt them.

An article at EBN details how Micron and Hyundai are working together to press forward with their antitrust case against Rambus filed with the Federal Trade Commission. The basis of this antitrust case, as we have discussed many times before, is the allegation that Rambus patented intellectual property which they misappropriated while a member of JEDEC, which is a memory standards committee. Rambus of course denies these claims, but the facts of the case seem to go against them. The memory architectures which Rambus has patents on were the very same architectures that were being standardized while Rambus was a member of JEDEC. Rambus quit the standards committee in 1996 and immediately filed the patents.

Well now they're going to have to answer to the Federal Trade Commission and explain how it is that they patented technologies that had been discussed with other memory manufacturers during the standards meetings. If the FTC rules against Rambus, their big cash cow (memory royalties) will head for that giant cash cow stable in the sky. We can only hope.

Reports this week from major memory manufacturers indicate that Rambus DRAM sales to computer makers have been very slow, and that currently they are producing more Rambus than is being bought.

                              Dr. John

Next NVidia Card Preview!

Reactor Critical has a speculative preview of the upcoming NV20 chip from NVidia.  I think they may have overestimated the increase in performance slightly, but if "hidden surface removal" is fully implemented, a 2 to 3 fold increase in speed is not impossible.  Mixed with multi-resolution mesh technology built into upcoming games, we may be looking at games this Spring with up to 100 million polygons rendered per second (3 to 4 times the best card of today).

November 23rd, 2000

Happy Thanksgiving to All!

Dr. John and the KickAss Gear Crew.

November 22nd, 2000

Tom Takes a Swipe at P4.

Dr. Tom of Tom's Hardware took a good swing at the new Intel Pentium 4, and some of the benchmarks don't put the P4 in a very good light.  But Rambus does offer a fantastic memory bandwidth, which even DDR SDRAM may not be able to touch.  So Rambus might just help the P4 substantially when the clock speed is increased to 2GHz and beyond.

  The very best thing about the P4 is how fast is will go when Intel decides to crank it up.  The 0.18 micron version will go between 2GHz and 2.5 GHz, while the upcoming 0.13 micron version with copper interconnects will probably scale well past 3GHz!  Current Athlons won't go nearly that fast. But like I said, if you are shopping for a top-of-the-line rig, don't buy the first batch of P4s this year, a better version is slated for early next year. By then, AMD may have a new butt-kicker ready to go as well.

                              Dr. John

November 20th, 2000

Pentium 4: The Good, The Bad, and The Ugly.

The Intel Pentium 4 is now available for anyone who wants one.  That's right, you can buy a shiny new Pentium 4 today!  We can get one for you if you want it.  But before you dig out your credit card, consider the good, the bad, and the ugly sides of the Pentium 4.

The Good: the Pentium 4 offers a huge memory bandwidth increase over current systems thanks to Rambus DRAM.  More than double the bandwidth of the Athlon with DDR memory!  The Pentium 4 also has a longer "pipeline" which allows it to run at much higher clock speeds.  Along with this longer pipeline and higher clock speeds Intel has also greatly improved the branch prediction capabilities of the processor.  This means fewer mispredictions and improved performance.  The Pentium 4 also offers 144 new 3-D instructions built into Streaming SIMD Extentions 2 (SSE-2, note: SIMD = Single Instruction Multiple Data).  And finally, the Pentium 4 runs relatively cool compared with AMD Athlons.

The Bad: the Pentium 4's massive memory bandwidth does not help in any current applications.  So much for Rambus.  The Pentium 4 at 1.5 GHz is actually slower at most office applications that an Athlon running at 1.2 GHz.  The Pentium 4 has poor integer performance compared with a Pentium III or Athlon.  Also, the Pentium 4 has a very long latencies, which translates to a large performance hit when branch mispredictions occur in the longer pipeline.

  We also should not forget that the Pentium 4 based on the 0.18 micron fabrication process is a huge chip.  The production costs are far greater than for Pentium III.  So because of the high cost of producing the Pentium 4 processor, Intel will not be able to sell these new high-end processors at the kind of profit premium they are used to for their flagship chips.  On top of that, Intel is going to bundle two 64 MB Rambus DIMMs with each retail Pentium 4 that they sell.  They will need to keep the price down on the bundled Rambus, further cutting into their profits.  All in all I would say that the Pentium 4 is not going to make very much money for Intel in its current configuration.

  Finally we need to mention the fact that graphics cards become the limiting factor when the screen resolution in games is set at 1024 by 768 pixels or higher.  This means that your graphics card is the limiting factor and the Pentium 4 will not help.  And don't forget that until new games are written to take advantage of SSE2, the new instructions will not be used.

The Ugly: The initial Pentium 4 will debut in a socket 423 format.  By early next year this will be replaced with a new motherboard/socket design, Intel's socket 478.  That means that people who buy the first round of Pentium 4 systems will not be able to upgrade to a faster processor later on.  This is going to be very much like the transition from the very expensive Pentium Pro to the faster less expensive Pentium 2.  Talk about planned obsolescence!  Also on the ugly side of the Pentium 4 equation is the fact that currently it is a Rambus-only platform.  And while Rambus seems to offer higher memory bandwidth, this does not translate into improved performance in virtually any application or game.  Finally, the Pentium 4 requires a new type of ATX case (ATX 2.03) with a new type of power supply.  The new Pentium 4 compatible power units provide more power to the CPU.  But the bottom line is, you can't just go out and buy a new processor and motherboard.

I think it is safe to say that the Pentium 4 as it currently stands is really only for people with money to burn.  It is outperformed in almost every benchmark by an Athlon running at 1.2 GHz. At lower resolutions, games run faster on it, but if you run your games at 1024 by 768 or higher, the Pentium 4 offers nothing but a high price premium.  The temporary nature of the socket 423 format borders on the criminal.  Why release the processor in such a temporary format that will be made obsolete within six months?  You got me!

Conclusion: The P4 is a forward-looking chip that is not ready for prime time. The next version will be out in 6 months, and will be much better.  Let someone else waste their money on the first batch, while you wisely wait for the socket 478 version, and see how it stacks up against AMD's future offerings.

Link: AnandTech.com  

More: Daily Rumor Real Audio

                              Dr. John

Intel Delays Almador Chipset.

An article at EBN discusses how Intel is delaying it's new Pentium III chipset, Almador. This chipset was designed specifically for the upcoming Tulatin CPU, also known as the 0.13 micron version of the PIII.  Now Intel will make changes to the i815e chipset to be able to handle the newer Pentium III.  This suggests that there are problems with the Almador design, and Intel does not want to repeat the i820 debacle.  As such, going with a proven design is more prudent.

                              Dr. John

Rambus = Borg.

A delightful article by Andrew Thomas at The Register equates Rambus with the Borg from Star Trek.  The comparison seems logical to me. But damit Jim.... I'm a Doctor, not an engineer!

November 17th, 2000

NVidia to "Crush" Competition.

NVidia is feeling their oats in the year 2000. Not only have they rolled out GeForce-2 and GeForce MX video cards, but now they are feeling so good that they are preparing to release a motherboard chipset! That's right, nVidia has been working on a low-cost, high-performance chipset with integrated GeForce-2MX graphics. The name of the new chipset is "Crush", perhaps with the intention of of crushing the competition.

The Crush chipset will support Athlon and Duron processors in the socket A format, and will also have support for DDR memory. Despite the fact that the chipset has a GeForce-2MX video chip built-in, boards based on the Crush chipset will also have a standard 4X AGP slot for upgrades.

The Southbridge of the chipset is the MCP-1, which is supposed to be part of Microsoft's upcoming X-box. It sports ATA 100 hard drive controllers and integrated 10/100 ethernet support.

The chipset is scheduled to be released late this month or early next month. Expect motherboards with the Crush chipset to be available early next year. The cost of super-fast Duron systems based on the Crush chipset should be very sweet indeed.

                              Dr. John

Duron Smacks Celeron!

Tom's Hardware has posted a Celeron 766 vs Duron 800 deathmatch.  Here's a quote:

"The 766MHz Celeron takes a beating at the hands of the 800MHz AMD Duron. The Duron doesn't merely smack the Celeron here and there, but pounds it into the dust in almost every benchmark. What makes this massacre even uglier is that the 800MHz Duron is available for under $100, while the 766MHz Celeron has been announced to sell at $170 in lots of 1000."

Nuff said!

November 16th, 2000

Rambus Looses Again!

An ITC (International Trade Commission) Administrative Law Judge, Sidney Harris, has blasted Rambus Inc for blatant "Judge Shopping" in their law suit against Hyundai. In addition, Judge Harris has imposed restrictions on Rambus' ability to file new law suits against other DRAM makers.  This is tantamount to trashing Rambus' claims with the international trade body.  

Rambus had filed suit against Hyundai for patent infringement, and was under the impression that ITC Judge Paul Luckern would be assigned to their case.  When the much tougher  Harris was assigned to the case instead, Rambus immediately withdrew their legal action against Hyundai.

Judge Harris stated: "The claim of 'privileged' litigation strategy is highly improbable, Rambus has engaged in impermissible judge shopping."

As I said a few weeks back, things have been turning against Rambus in just about every possible way.  This latest defeat almost ensures a continued downward slide for the Intellectual Property company.  

                               Dr. John

November 15th, 2000

3dfx Exits 3D-Board Business.

A report at CNNfn indicates that 3dfx is closing their Mexican plant, and getting out of the card-making business.  This ends two miserable years for 3dfx that witnessed their steep decline from the top of the 3D graphics chip business.  They are now going back to making the chips only, and selling them to video card makers such as Asus and Creative.  Unless they have developed some new super chip that they can bring to market quickly, it seems unlikely that they can regain their previous lofty position with such stiff competition from NVidia and ATI.  But if they concentrate on chip design, and forget about all that marketing and retail distribution stuff, they at least have a shot at it.

                               Dr. John

November 14th, 2000

Voodoo5 6000... D.O.A.

As I predicted a few weeks back, 3dfx has announced the cancellation of the Voodoo 5 6000 card! It seemed inevitable, but it's still comes as somewhat of a shock. Is this the beginning of the end for the company that began the 3-D accelerator revolution? We certainly hope not, but things are not looking up for the little graphics chip maker.

3dfx states that they will license the Voodoo 5 6000 technology to Quantum 3-D, a company that makes high-end graphics boards for 3-D simulations. Some of you may remember their ill-fated Obsidian line of graphics cards. They were some of the glitchiest video cards we have ever dealt with. Many of them just did not work correctly. My guess is, if Quantum 3-D tries to market the 6000 card to consumers, it will be another disaster. I have a sneaking suspicion that there were serious problems with the Voodoo 5 6000, ranging from power problems to heat dissipation to driver difficulties. It may have just reached the point where 3dfx was not confident that the product would be a plug and play video card for home computer users.

This puts 3dfx out of the running in the high-end video card market. Some flight sim fans and racing game aficionados may prefer the V5 5500 card, but for most gamer's, these cards just do not deliver enough performance for the price. By the time 3dfx comes out with their next generation video card it will probably be too late. By Spring, we will likely see GeForce 3 and dual Radeon video boards. Anything from 3dfx will most likely be too little, too late. Damn I'm depressed.

                                 Dr. John

November 13th, 2000

Direct X 8.0 Benchmarks.

For a quick look at Direct X 8.0, and whether you can expect a performance increase, check out our quick benchmark test here.

November 12th, 2000

Direct X 8.0 Released.

For those of you who have been asleep for the last 24 hours, or can't pull yourselves away from CNN's coverage of the never-ending election, we wanted to point out that Microsoft has released DX 8.0 for download here.  Many of the enhancements are just improvements to the "API" (application programming interface) which makes it easier for programmers to write DX8 games.  Other improvements include improved "scalability" which means that you can make bigger worlds with more characters. Check out the new features here. Performance tips for AMD systems are discussed here. Until games come out with specific support for DX8, don't expect to see big differences in performance or features.

                                 Dr. John

November 10th, 2000

AMD 2001 Plans Revealed.

EBN details AMD's plans for its processor lines in the next year. Perhaps the most interesting revelation is that AMD is targeting one version of its new 64-bit Sledgehammer processor at desktop PCs. The code name for this 64-bit CPU appears to be Clawhammer. This version of the Sledgehammer may not be available until the end of 2001. At that time it may be possible to put together relatively low-cost two-way Clawhammer servers for home use.

AMD's Athlon and Duron plans seem a little less aggressive.  AMD admits that the current Athlon will not be able to meet the clock speeds of the Pentium 4 over the next six to eight months.  But AMD officials conclude that a 1.2 GHz Athlon will still beat a 1.5 GHz Pentium 4.  However, Intel has indicated that the Pentium 4 may reach 2 GHz by mid next year.  AMD is planning on moving the Athlon to 0.13 microns. The code names for the new and improved versions of the Athlon and Duron core are "Palomino" and "Morgan".  But the question that everybody really wants answered is, "will the Athlon still be faster and cheaper than the Pentium 4 next year?" Only time will tell.

                                  Dr. John

November 9th, 2000

Memory Makers Dump Rambus.

According to The Register, 3 DRAM makers have decided to discontinue direct Rambus memory production. While these memory makers are relatively small, they may just be the first few dominoes to fall in a long line. It is interesting that this announcement comes just two days after an IT analyst commented that Rambus DRAM was a dead-end technology. It is thought by most analysts that double data rate SDRAM will become much more popular than Rambus DRAM over the next six months, and as such, we can expect most memory manufacturers to concentrate on double data rate SDRAM production.

                                 Dr. John

Constitutional Crisis

The Constitution has served the United States well for over 200 years, but occasionally it's provisions fail to account for situations that are out of the ordinary. When occurrences proceed beyond the bounds of Constitutional jurisdiction, a constitutional crisis ensues. We're currently in the midst of just such a crisis because the Constitution does not make provisions for a situation where so few contested votes in one place will determine who becomes the next president of the United States. So it remains up to state election officials in Florida to try to figure out what actually happened in the vote on Tuesday.

The Constitutional crisis becomes even more problematic due to the extremely poor design of the election ballots in Palm Beach Florida. At last count somewhere in the neighborhood of 19,000 votes had to be thrown in the garbage can because the people had voted accidentally for Pat Buchanan and Al Gore on the same card. The vast majority of these votes were obviously meant to be cast for Al Gore, and discarding them takes the voting rights away from those of 19,000 U.S. citizens.

At last count, the recount has almost erased George Bush's lead in the Florida vote, and there are still nearly half the precincts to report. The interesting thing as that no one is mentioning that this smacks of voter fraud. After only half of the recount has been done, far more "lost" votes have turned up for Al Gore than for George W. Statistically speaking, since the vote was so close, the recount should not have had a major discrepancy on either side unless something unfair had happened in the initial vote count. Statistically, this result is virtually impossible. If votes were missed accidentally, the recount should have given nearly equal numbers of new votes to both candidates. The fact that Bush's lead has been cut more than in half should raise some eyebrows.

Even though I do not think there are any such provisions in Florida law, I'm beginning to think a runoff election in Palm Beach is in order. I do not think there is any way this can be done, but considering the extremely large number of disenfranchised voters in Florida, and the slimness of the lead, something needs to be done to restore confidence in the electoral college system.

                                 Dr. John

November 8th, 2000

Analyst Trashes Rambus.

Hang onto your Rambus hats, it looks like they are in for a wild ride again.  The Register reports today that another IT analyst has declared Rambus memory to be a dead technology.  You can read the story here.

Intel Goes to 0.13 Microns.

According to another report at The Register, Intel has successfully moved their fabrication abilities to the much awaited 0.13 micron process, with copper interconnects.  When these babies become available, they will possibly be the most overclockable processors in history.

George Dubya Losses Popular Vote, Declares Victory.

The American political system, and many American news reporting organizations, were both shown to be in sore need of an upgrade last night.  

News organizations declared Al Gore the winner in Florida very early in the evening. This declaration was then retracted a little while later after more polling results came in. The media then declared George Dubya the winner by about 2 AM Eastern time, only to retract that statement a couple hours later. By the wee hours, only Florida and Oregon remained undecided, and it was clear that the election would be decided entirely on the outcome in Florida.  Finally the media decided it was too close to call and went to bed.

By morning George Dubya's staff declared that he had won, and that the younger Bush was now the President-elect of the United States. This announcement came despite the fact that less than 1800 votes separated the two candidates in the Florida's election, and that a recount was in progress, and the fact that many overseas absentee ballots had not even come in yet. In addition, a number of voters in one district in Florida claimed that the ballot's had confused them into accidentally voting for Pat Buchanan when they meant to vote for Al Gore. If you get a chance to see how the ballots were laid out you'll quickly realize how easy it would've been to make the mistake. This is really going to come back to haunt the Republicans.

On top of all of this, Al Gore easily won the popular vote throughout the United States. Prior to the election, it was the Republicans who were indignant about the possibility that Al Gore would win the electoral college, while George Bush would win the popular vote. Now that the opposite has happened, they say "that is how our system works." Nothing like sticking by your convictions.

So with the election still somewhat in question, and the popular and electoral votes possibly going opposite directions, we can only expect one heck of a hullabaloo over the next few days. The news media are salivating like they have never drooled before.

November 6th, 2000

Abit KT7 RAID and Duron Overclocking.

Better late than never. I've posted a review of the Abit KT7-RAID motherboard, and how well it overclocks the AMD Duron 600 CPU.  You can check it out here.  The results are fairly impressive.

Many American's Depressed By Election.

Based on talking to people, and watching what little I could stand of the pitiful news coverage of the US elections tomorrow, I think I can say the majority of Americans are pretty depressed about the whole thing.  Sure, the hard core Republicans are very excited about getting back into the White House, but for about 60% of the public, this election seems like a circus side show.

Is it just me, or did we just spend the last two months talking about Al Gore's sighing behavior, and George Dubya's speech patterns? Personally, I would've liked to have had their feet held to the fire about issues rather than quirks, but then again I'm not in the news media. On top of it, the media seemed fully content to let dozens and dozens of the inaccurate statements spew out of both candidates mouths without so much as a "wait... hold on a second."

I have a distinct feeling that 50 to 60 percent of the American public haven't a clue as to the different policies that the two candidates will pursue. And how that could be after so much news coverage is beyond comprehension. The news media have been getting worse and worse since they changed themselves from a public service into a business, and I expect things to get more and more like the Ringling Brothers circus as time goes on.

Many people may disagree with me on this, but I think the root of the problem was turning both politics, and the news coverage of it, into moneymaking enterprises. You'll never get the straight story as long as money, big money, is involved. And the only way to remove money from the whole mess is to set up a system of public financing for all elections. It may sound drastic, but there's no other way to extricate the circus from our political system. Some portion of taxes needs to be set aside as a campaign bank, and the bank doles the money out to official candidates equally. Nobody gets the advantage of big corporate money. 

How in the hell we're going to get the corrupting influence of money out of the news industry is beyond me. But at least getting it out of politics would be a start.

                                 Dr. John

November 4th, 2000

DDR Pentium-4 Chipset?... Anyone?

  The Register reports today that Intel is talking to chipset maker VIA about licensing the design for a P-4 DDR chipset.  At the same time, Intel is supposedly talking with Rambus officials about dropping the ban that Rambus has imposed, barring Intel from making its own Pentium-4 DDR chipset until 2003.  This is basically the same story we have been hearing for months now.  The only thing that has changed is the urgency with which Intel is pursuing these goals.  

  With AMD's new DDR capable chipset nearing release, this puts additional pressure on Intel to come up with its own DDR solution, whether in-house, or by a third party.  It seems clear that VIA is in the process of producing a DDR Pentium-4 chipset, and will probably release it with or without a license from Intel.  Obviously, they would prefer that Intel license it to them at a reasonable rate.  

  The Pentium-4 is scheduled to begin shipping November 20th.  Based on what we have seen at distributors, this is probably the same approximate time frame that DDR systems from AMD will become available.  That means the DDR/Rambus debate is about to heat up big time. So hang on your hats folks, it's going to get very interesting.

                                  Dr. John

November 3rd, 2000

AMD 760 Chipset: Delayed or Not???

  The story continuing to develop around the new AMD 760 chipset is still far from clear. Reports at several web sites, including Tom's Hardware Guide, have continued to insist that there are no problems, and that the motherboards based on the chipset are shipping now. However, The Register reports again today that they have received reports of a problem, specifically with Gigabyte's new GA-7DX DDR Athlon motherboard. 

  The Register even points to the fact that just several days ago when Tom's Hardware Guide reviewed the Gigabyte motherboard, they admitted that it would not run at 133 MHz on the front side bus. Since this is one of the major features of the new AMD 760 chipset, it would seem that AMD may actually be having a problem with the chipset. It could just be a problem with Gigabyte's implementation of the chipset.  Only more time will tell.  Here is a quote from Tom's hardware review on Oct 30th:

  For those of you who check out the news here regularly, you may remember that yesterday I was complaining that this was beginning to sound like a short delay, that AMD would rather not call a delay. Will it looks like this same observation struck Andrew Thomas of The Register exactly the same way.  He reports today that earlier, AMD did indeed state that DDR-capable AMD systems would be available now.  In fact, they are not. His article is worth reading.

  I'm not one to split hairs, but I was also under the impression that DDR AMD systems would be shipping now, and that motherboards and memory modules would be available to system integrators. They are not. On top of this, we have seen no sign of the new 133 MHz front side bus Athlons. The PC manufacturers that The Register was able to contact said that the earliest systems with this configuration would be available was the end of November. When asked about this situation, AMD reportedly replied: "If that's the case, then the information we were given to write the press release must have been incorrect."

  So you can decide for yourself, is there a problem, or isn't there? Is there a delay? Sounds to me like they are having stability problems with the new, higher bus speed. And how can you ship motherboards if they are not stable at their rated speed?  Why was Tom's Hardware Guide given such an early version of the board to test just 5 days ago if new, improved boards are available?  Stay tuned for more updates.

                                  Dr. John

November 2nd, 2000

AMD 760 Chipset Not Delayed?

  Several reports surfaced in the last twenty-four hours that claim the initial story at The Register about the delay in the new AMD 760 chipset was incorrect (see story below). We will be contacting reporters at The Register today to see if we can learn more about the situation. But a post on Tom's hardware site from late last night states that Gigabyte has denied the initial story in The Register that they would be recalling motherboards with the AMD 760 chipset. Let's hope that it turns out that The Register did get this story wrong.

Motherboard companies including Gigabyte state that AMD 760 chipset motherboards are still scheduled to ship sometime this month. However, to me that seems like a delay that they would rather not call a delay. We had been led to believe initially that DDR based motherboards with the 760 chipset would be shipping now. For companies to say they are still on schedule for shipping this month suggests to me that there is at least a small delay. Let's hope it stays small.

We have spoken to several memory manufacturers about the availability of DDR memory modules. Crucial technologies, a subdivision of Micron, does have DDR memory modules for sale. However, other large manufacturers such as Dane Elec. have stated that they are not gearing up DDR SDRAM production yet because they are not certain how much demand there will be. I think they are being overly cautious, since DDR memory will obviously become very popular over the next six months. Pent up demand is growing, so I would say it's a no-brainer to gear up DDR production.

So the question is, when will AMD 760 chipset motherboards actually hit the shelves? If they begin to arrive before Thanksgiving, AMD will be in a good position for the holiday shopping season. If the delay slips past that time, there may not be enough time left in the shopping season for AMD to take advantage of.

                                   Dr. John

November 1st, 2000

Jeremy is Back!

  Jeremy Allford, the mastermind behind Allgames Net 3D (AGN3D), is back online with his own website.  It's great to see Jeremy back again, and you can check out his new website here.

AMD 760 Chipset Recall?!

  At this point the details are too sketchy to make any definite statements, but initial reports have come out suggesting that the just released AMD 760 chipset with DDR memory support has been recalled. A report at The Register states that Gigabyte has confirmed there is a problem with the new AMD 760 chipset, which is preventing them from shipping motherboards built with it. The multiprocessing version of the new AMD chipset has been reported delayed until the middle of next year.  If these reports turn out to be true, this is a major blow for AMD.   If the problem the the 760 cannot be pinpointed and rectified quickly, AMD may not be able to have enough product in place for the holiday shopping season. This will give Intel some breathing room with the Pentium 4 release.

So if you were getting all ready to run out and buy a new Athlon system with the AMD 760 chipset, it looks like you're going to have to wait. You can hold off on purchasing that DDR memory too. As I have said many times before, it's not always the best idea to try to be the first one on your block to get brand new hardware. The initial release of any new type of hardware tends to be bug prone and overpriced.  So fire up Unreal Tournament or Quake III Arena, and enjoy the system you have for a little while longer.

                                    Dr. John

Rambus Snags Samsung.

  According to a report at SBN, Samsung has agreed to pay Rambus Inc royalties for all the SDRAM, DDR DRAM, and memory controllers that Samsung manufactures.  This leaves Micron, Infineon and Hyundai as the only memory makers still fighting the lawsuits leveled by Rambus.  For more of my ramblings on Rambus, check out my Rambus R.I.P. editorial from yesterday. 

                                    Dr. John