April 29th 2000

 AMD Officially Announces the "Duron".

   The upcoming low-end AMD processor that has heretofore has been code-named "Spitfire" has now been given the official name "Duron". While it sounds like a brand of house paint to me, it is supposed to have the air of durability about it, while sounding just enough like "Celeron" to let customers know it's in the same processors class.  Based upon a new version of the Athlon core, it has on-die L2 cache which will offer a performance boost relative to the original Athlon.

Proposed Microsoft Remedies.

   The US government has proposed remedies in the Microsoft antitrust case.  They include breaking apart the operating system and applications developing sections into two distinct companies, full disclosure of their application programming interfaces, a prohibition on any special deals or restrictions for certain Microsoft customers, and a prohibition on "knowing interference of performance" with competitor's products on Microsoft operating systems.  

  Sounds about right to me.  It's too bad that Microsoft couldn't just behave themselves.  But now they get sent to the wood shed to contemplate their naughtiness.  If it ends Microsoft's rotten behavior, the taxpayers money will have been well spent. 

  Wait till Microsoft share prices drop more before buying.  But you can expect a company breakup to be essentially a de facto stock split, with both sibling company's stocks rising after the shakeup.

                               Dr. John

April 25th 2000

 Preview of ATI's New Chip, the Radeon 256.

   HotHardware has posted a preview of ATI's upcoming answer to the Voodoo VSA-100 and NVidia NV15 graphics chips.  If the actual cards based on the Radeon chips really have the capabilities that ATI shows on paper, then Voodoo 5 and GeForce-2 cards will have a real run for their money.  Competition is good.... VERY good!

                               Dr. John

April 24th 2000

 Prices Drop on Non-Existent Pentium III Processors.

  Intel has dropped prices (slightly) on it's line of (non-existent) Pentium III processors.  If you can find them, they should be a few bucks cheaper.  It's interesting that when SDRAM became scarce several months ago, the prices went way up.  The sad fact is, that Intel processors are much more scarce than SDRAM ever got, but the price is still going down, rather than up.

   The really unfortunate part about these scheduled price drops is that they do not let market forces work.  Supply and demand need to be able to slide relative to one another in order to keep producers and consumers from getting too far out of synch.  So when memory got scarce, it got expensive.  Regular folks put off memory purchases, and rich folk helped drive the price way up.  Demand ebbed, allowing supply to catch up again.  The higher price tempted more memory makers to get on the ball and turn out SDRAM.  This had the effect of bringing the supply levels back up and driving prices down again. The whole cycle only took 3 to 4 months.

  As I mentioned yesterday, many computer parts are in very short supply now, resulting in higher prices for the majority of the scarce parts.  But not CPUs!  The competition between AMD and Intel has gotten so fierce, and the prices have gone so low, that demand has greatly outstripped production capacity for both companies.  This will no doubt have an adverse affect on Thunderbird, Spitfire, Celeron-2 and Willamette supplies over the next year.

                               Dr. John

April 23rd 2000

Pentium III shortage getting weird.

  Due to CPU part supply problems, fabrication retooling, and high demand, Intel's processor shortage is reaching absurd levels.  On top of this, Dell computers acts as an Intel-only distributor, and Intel is obliged to feed this insatiable chip-eating monster until it hurts.  Only very limited quantities of a few select Coppermine Pentium III chips are reaching the rest of the supply chain. This will continue through most of the Spring, and that fact isn't helping Intel's rep with anyone but Dell.

   If this supply problem continues, it may affect Intel's ability to turn fabrication capacity over to Willamette (Pentium IV) production in the Fall, delaying the availability of the new processors until 2001. 

Athlon Supply Affected.

  The lack of Pentium III processors has led so many computer makers to switch to the Athlon processor, that shortages of the Athlon have become severe. AMD is bringing another fabrication plant on-line in the next month or two, so that may help alleviate the shortages by early Summer.  AMD is in the same pickle as Intel right now, they are having supply problems at the time they are getting ready to debut new chips, namely the "Thunderbird" and "Spitfire".  So the problem is... make more of the old chips (delaying the new ones), or just switch over to the new ones and forget about shortages?

Parts Shortages Lead to Price Increases.

   In addition to the processor shortage that has been plaguing computer makers for months, shortages of other computer parts have become severe.  High speed, 7200 rpm IDE hard drives from companies like IBM have been scarce for weeks.  DVD drives have likewise been few and far between. High-end, double data rate (DDR) video cards are also hard to find. The inevitable result of computer parts shortages is an increase in prices.  So far, prices have just started to rise, but if CPU and other parts continue to be in short supply, then it is likely that prices will rise further.  The usual response by computer parts makers is to overproduce parts in the wake of a shortage, so we can hope that prices drop again in the Summer.

                               Dr. John

April 20th 2000

Rambus or Rambust?.

  For those of you with an insatiable desire to read MORE on the history and future of Rambus, check out this excellent (and lengthy) article on the ill-fated memory platform at 32bitsOnline.  We have discussed these issues before, but not nearly so comprehensively.  Here is a quote to whet your appetite. 

"By the summer of 1999, there were widespread rumors that the motherboard industry was under pressure by Intel to stop support of the VIA chipset. If you were not in Intel's corner, there might be some consequences for your chipset allocation, the bread and butter of these motherboard companies. AMD's Athlon also made its debut and got caught in the crossfire---to keep Intel's favor, went the thinking, maybe you should not support AMD either."

Napster Lives! For Now At Least.

   Everyone's favorite MP3 download and cataloging program is coming under increasing pressure of a class-action law suit by recording artists who are up in arms.  You can find out about Napster at their web site here.  Several large Universities were named in the original suit by the band Metallica, which prompted these edifices of higher education to ban Napster software.  Other recording artists are joining the suit, and will no doubt turn up the heat on little Napster.  We are rooting for the underdog here, mainly because MP3 music files do not have nearly the sound quality of CD's.  And anyway, as everyone knows, you can burn copies of music CD's on your computer's CD writer, which in our minds posses a far greater risk to the industry's profits in the future.  But with Napster, the industry has an easy, lawyer-poor target to shoot at.  The funny thing is, bands that go after Napster may find themselves loosing some of their most loyal fans!

                               Dr. John

April 19th 2000

Interview With 3dfx on Voodoo5.

  If you want to take a look at an interview with Alf Covey of 3dfx on the upcoming Voodoo5 card, you can check it out at HotHardware. Let's hope they keep to their schedule this time!

April 18th 2000

Intel In Another i820 Bind.

  Intel finds itself in a loose-loose situation with the i820 Cape Cod motherboard.  They have shipped lots of them to now highly unsatisfied customers.  On top of the RDRAM 3-RIMM problem, the Memory Translation Hub (MTH) performance problems, and the SDRAM Serial Presence Detect (SPD) SDRAM glitch, Intel now has a public relations problem.  

  It was reported yesterday in The Register that Intel was going to recall i820 Cape Cod motherboards because of the SPD SDRAM glitch in the memory hub (MTH).  In some cases, if the SDRAM is not properly detected by the MTH, the systems would not boot.  The folks who broke this story over at The Register often have their ePaper called the "National Enquirer" of Information Technology, mainly because IT professionals don't like the messages they receive from these messengers. In this case, Intel candidly answered an email query from the staff at The Register, saying that a recall was in progress.  A day later, Intel retracted the previous statement, saying no recall was planned or needed.

  Now Intel has a real public perception problem on their hands.  And of course, if they decide to do a full recall, they will have a major hit to their bottom-line for the quarter. So it looks like they have gotten themselves between a rock and a hard place again.  Best to bite the bullet, and recall the faulty product in order to keep their reputation semi-intact.

   On top of this, the Coppermine Pentium III shortage is now rumored to to be a long-term shortage (until June!).  New Flip-Chip PIIIs use different packaging materials, and Intel is having trouble getting the materials they need, in part because they bully their suppliers into dropping prices to the point of vanishing profits.  The only 100MHz Pentium IIIs that seem to be in fair supply now are the 550Es.  The good news is, the price is reasonable, they all seem to be the new stepping (with many bugs removed) and they overclock very well.

                               Dr. John

April 17th 2000

Tech Stock Madness.

   The enormous run-up on tech stock prices over the last 6 months was phenomenal, especially considering all the warnings from analysts that many of them were highly overvalued.  But big buyers kept on buying, driving prices up and up.  The big boys knew all too well that the time for gettin' was good, and profit taking turned into a plummet on the Nasdaq.  Expect this to continue until the big boys see a bunch of good deals on hammered tech stocks.  Then we will repeat the whole process over again, with the big boys taking the lions share of profits each time.  Rambus is one stock we have been warning folks about for two months.  Now that their share prices are down by over 60% in the last month, the big stock split they were talking about seems mighty far off.

                               Dr. John

April 14th 2000

Intel Piles On the Blunders.

   Intel's latest list of goofs makes one pause and wonder if blunders will never cease.  The newest list of problems strains credulity coming on the heals of so many similar problems.  Apparently the upcoming Solano2 chipset motherboards (i815) will only be supplied to large Intel customers throughout the Summer, and may not be made available to the regular distribution channels until this Fall.  That's way too little, and way too late. 

  Also, it looks like another bug can be added to the list of problems with the current i820 Intel chipset.  Apparently, the i820 does not correctly read the SPD (serial presence detect) EPROM on new SDRAM modules.  This means that on top of the memory performance hit you will take by using SDRAM on an i820 board, your system may not work at all.

  And last, but not by any means the least important Intel tid-bit is that Intel has announced another (yes, another) shortage of Pentium III chips.  While the supply of low-end Coppermine PIIIs seems better than it has ever been, Intel says it will only have enough of the higher-speed chips to supply Dell and a few other Intel-only computer manufacturers.  So don't expect to see many 800MHz+ retail Coppermines at your local computer store until at least June.  This kind of nonsense is going to get Intel into some real hot water with smaller computer makers.  Even large companies like Gateway have turned away from Intel for obvious reasons.  You can't sell computers without processors.  It's as though AMD rubbed a magic lamp, and got at least three big wishes.

  And if you are looking for an older 550MHz or 600MHz Katmai PIII, you may have to do a lot of digging.  Intel has phased them out faster than anyone expected, and yet this has not improved the availability of the Coppermine parts to any significant degree. 

                               Dr. John

April 13th 2000

AMD Profitable, but Not Rambus.

   It's all over the Internet today.  AMD announced a far better than expected earnings report yesterday.  Expect the stock to generally rise as we approach the launch of the new Spitfire and Thunderbird processors in about 6 weeks.  This is great news not just for AMD and their shareholders, but for computer enthusiasts in general.  The more market share that AMD takes from Intel, the better off we will all be as consumers.  

  The only reason that CPUs are at 1GHz now, and that prices for processors are at an all-time low, is that Intel was unable to produce a viable product based on Rambus memory, leaving a huge gap in availability for high-end processors and motherboards that lasted for months.  AMD's Athlon would have caught on well based on merits alone, but the CPU/motherboard scarcity caused by the Rambus/i820 debacle boosted AMD sales significantly over where they would have been if Intel had released a functional i815/PC-133 SDRAM chipset, while providing Coppermine chips at a much faster rate initially. Competition had been sorely lacking in the computer CPU Dept. for years, but now that situation is thankfully over!

  Rambus' stock continues to slip, and only sales to game console makers have kept the company afloat.  If anything goes awry with the Rambus/Sony deal for Playstation2 memory, Rambus Inc. will be in big trouble.  It would be ironic, would it not, if Rambus ended up being the memory found almost exclusively in game consoles, and DDR memory ended up being the dominant memory in PCs.   Intel really gambled on Rambus, and it's looking more and more like they bet poorly.

                              Dr. John

Celeron II Overclocks Well, But Does not Match Copper PIII Performance.

   Based on the initial reports we've seen, and some discussion on the topic, it looks like changes in the Celeron II make it perform 20% to 40% slower in benchmarks than the Coppermine Pentium III at similar speeds.  The only two changes we are aware of are the reduced L2 cache, and the higher L2 cache latency reported at HardOCP.

  The review at AnandTech hypothesized that the difference in benchmark scores were entirely based on the difference in bus speed, rather than the L2 cache size or latency.  But it's clear that the Celeron II still does not match the Copper PIII's performance even when overclocked to 100MHz on the front side bus.

  The major question is why Intel set the L2 cache latency to 2 on the Celeron II.  The L2 latency on the off-die L2 cache on older Katmai-based Pentium IIIs is only 1, while on the new Celeron II the L2 latency it is set to 2. In Coppermine PIIIs it is set to 0.  Since the L2 cache on the Coppermine Celery IIs and PIIIs are both integrated on the die of the processor itself, there is no clear reason for such a long latency on the new Celery.  This may have been a deliberate hobbling of the L2 cache in order to insure the Celeron 2 could not outperform the Copper PIII at similar clock speeds.

  The bottom line is that the Celeron 2 is so much better than the Celeron I that there will be no reason to get the older type.  But if you can come up with about $50 more, you can afford a real Coppermine without a weakened L2 cache.  You can join a discussion of the topic at The Tech Report.

                              Dr. John

April 11th 2000

Get Ready for Spitfire and Thunderbird!

   According to Mike Magee at The Register, AMD will be releasing the Spitfire and Thunderbird processors within one or two months (May - June).  If new motherboard support is in place, this will be another big headache for Intel. Both products will include on-die L2 cache, and will outperform the Athlon significantly.  On-die cache means better overclockability, but with the AMD double-data rate technology in these processors, they will never overclock on the front side bus the way single-data rate Coppermines do.  No matter, they will be so inexpensive that serious overclocking is not necessary (that's been the whole purpose of overclocking, to not pay Intel top dollar for their chips).  AMD's prices and performance make overclocking their CPUs much less important.

  As with the case of the original Intel Celeron 300A outperforming the Pentium II, it looks like the low-cost Spitfire chip from AMD is outperforming existing Athlons running at the same clock speed. That on-die L2 cache really helps!  So the scuttlebutt is that AMD will slightly delay the Spitfire debut in order to release both the Spitfire and Thunderbird at the same time.  That way, little brother won't be outperforming big brother.  We can't wait to see how these new chips perform.  Let's hope there is not a huge shortage of the new "Socket-A" motherboards at the time of their release!!!  Are you listening mobo makers???

                                  Dr. John

Double Data Rate Memory Coming Too!

   Memory manufacturers are gearing up for production volume on double data rate memory (DDR).  DDR RAM is a variation on SDRAM which manages to send two hunks of data during each clock cycle, rather than one like SDRAM.  But the underlying architectures for the two memories are very similar.  This means that, unlike Rambugs memory, existing fabrication plants can quickly convert to DDR RAM production.  Rambugs requires a complete overhaul, and tons of new expensive equipment.  Plus you need to pay big royalties to Rambugs Inc.

   AMD's upcoming Spitfire (low-cost), and Thunderbird (performance) CPUs use AMD's double data rate technology, as in the case of the Athlon.  So adding support for DDR memory in the AMD/VIA chipsets is the next logical step.  It will convert the entire CPU-memory bus into a double-data rate bus, with notably higher bandwidth than a single data-rate CPU-memory bus.

  This is starting to look like a brilliant marketing campaign by AMD/VIA (have they merged?).  They seem to be getting everything in place for a major release of new product this Spring.  If they avoid Intel's i820/Rambugs mistakes (releasing stuff that isn't nearly ready, and forcing expensive, unproven technology down our throats), they will gain even more market share from Intel.  While Intel keeps mucking around with Rambugs and i815, AMD/VIA will be releasing proven new technologies that will boost performance without raising prices.  At least some companies still care about what consumers want.

                                   Dr. John

April 8th 2000

GeForce-2 On the Way?

   NVidia has announced it will be "releasing" the NV-15/GeForce-2 video chipset at the end of April.  If this is anything like their last release, then don't expect to find more than a handful of video card with them available for the next few months.  It is becoming common practice in the computer industry to "announce" product availability long before the product is actually ready in production volume.  This is typically done to blunt the press-release of a competitors product, which may also nearing the initial "release" date.  Then a small handful of products are sent out to CompUSA and Electronic Boutique to get the buzz going on the discussion forums.  "I got mine!" manages to get everyone else hot under the collar to get the new piece of equipment.  Then, for most folks, the waiting begins.  I find it very amusing that it is still nearly impossible to get a DDR GeForce card (except maybe at CompUSA for full retail price), and now NVidia is about to announce the next generation video card.  Computer product life cycles are so short now, that they are over before the product even reaches the stores!  Who's gonna want a GeForce when NVidia is dangling the GeForce-2 in front of everyone?

                              Dr. John

April 5th 2000

The Bell Tolls for Apex.

   If you didn't get your Apex AD-600A home DVD player yet, it may be too late to  get one with the "hidden features".  Apex announced today that the secret menu has been disabled in the current shipments of the player.  You can read about the special-ness of this DVD player here. 

April 2nd 2000

Athlon/GeForce Schism.

   In the recent issue of PC Gamer magazine, technical editor Greg Vederman tested 4 Athlon-based motherboards with the 3D Prophet SDR GeForce video card to assess stability.  He reached the conclusion that GeForce cards are not as compatible with Athlons as they are with Pentium IIIs.  It is true that GeForce cards often act up on Athlon systems, but we would like to set the record straight.  

  The odd behavior that Greg noticed with Athlon systems and the GeForce card also occurs with Pentium III systems, just a little less often.  One symptom was the system dropping back to the Windows desktop from 3D Mark 2000.  We have observed the exact same behavior using GeForce cards with Coppermine Pentium III systems, especially when the Coppermine is overclocked on a VIA chipset motherboard.  Our tests so far have pointed more to AMD and VIA chipsets on current Slot-A motherboards as the probable culprit.  Further, these problems are not seen when other types of video cards are used in Athlon systems.  Our best guess is that these types of problems have to do with how the motherboard chipset addresses the AGP slot.  I expect these glitches will be fixable with updated drivers for both the GeForce cards, and the AGP miniport drivers for Slot-A motherboard chipsets.

                                Dr. John

Rambus, Rambus, Rambugs.

   Just in case you weren't sick and tired of hearing about "Rambugs" memory, (as we like to call it after all the technical problems were announced), we've checked the stock price for you today.  Yesterday's Black Monday for the Nasdaq took it's toll on Rambugs, seeing the stock drop by over 10% of it's value in a single day.  Share prices are now down by almost 50% from their high just 3 weeks ago.  This was no doubt a bit of profit taking by those who bought the stock several months ago when it was selling for less than $50 a share.  Just a couple weeks ago it was approaching $500 a share.  So many investors are probably thinking the Rambugs boat may be getting ready to sail, and they want to make sure they don't get left on the docks holding a bunch of paper.  

                                  Dr. John

April 1st 2000

Bill Goes for Broke.

   As you probably have heard, negotiations between Micro$oft and the government have broken down, and now we are waiting for Judge Jackson's ruling on the Antitrust case against Micro$oft, probably tomorrow.  Why should anyone care?  It's my opinion that the world of computers would be a better place if Bill Gates was not such a greedy, self-important person.  And I don't mean that as a personal attack, it's just that Bill's personal behavior, unlike the personal behavior of average folks, has direct and indirect effects on the computer operating system in most of the World's computers.  The fact that Windows 98SE is a graphical interface running on top of a 16 bit operating system, with an integrated Internet browser that slows performance down, has everything to do with Bill Gates and his desire to crush his competitors, and nothing to do with logical or practical system design.  But Bill continues to insist that everything was done, "for the consumer", in the spirit of "the freedom to innovate".

   I've talked to lots of people about this case, and to my surprise, most people in the computer business side with Bill.  They shrug off any unfair, unsavory or illegal behavior on Micro$oft's  part as "business-as-usual in the computer industry."  I guess they have given up on fair play, and even the rule of law when it comes to IT (information technology).  You can bet your bippee that Bill had the final say on whether or not to accept the government's offer in the case.  And for all I know, the government smelled blood, and asked for too much.  Micro$oft's case is very poor, so that may be a distinct possibility.  But it's also my guess that Bill is counting on taking it to the Supreme Court, where 5 very conservative Justices sit, just enough to overturn Judge Jackson's ruling.  That may be Bill's plan, and it could work.  This is the same court that was unable to distinguish between free speech, and money, and lumped them together as a single concept.

  Before I go further, I want to say that Windows98SE and Windows 2000 are modern miracles of programming.  Part of the miracle is that systems with so much baggage could be made to work at all, but work they do, and they have more features than you can shake a chip at.  Now if Bill would just bite the bullet, and give in on a few key issues, the case could be over, and we would actually have a better operating system in the long run.  He has got to remove Internet Explorer, there's no getting around that.  It's not part of the operating system, and shouldn't be treated as such.  Performance would be boosted by as much as 20% without the integrated browser. Next, Bill would have to sign some papers that said he promises to be a good boy, and stop mucking up other peoples businesses unfairly with his massive monopoly.  No more phony error messages when competitors products are detected, and no sweetheart deals to companies that don't sell competitor's products, and no more free bundling of products with his operating system just for the purpose of squashing a company that Bill wanted to buy, when they wouldn't sell.  If Micro$oft hadn't been such a bully in using their monopoly position unfairly, they wouldn't be in this situation.

  What I don't think Micro$oft should be forced to give away is the source code for their current products.  It's their property.  But they must sign papers to the effect that they will stop the practice of hiding instructions from software makers to give Micro$oft products an edge.  Why?  Because that is monopolistic, and would be a sign that the company needs to be broken up.  Let's suppose, for example, that Intel made the major operating system on the planet, and hid key codes from Micro$oft so that the Intel OS worked much better on Pentium CPUs than the Micro$oft product.  Clearly that would be an unfair monopoly in the Wintel world, eventually leading to a solely Intel world.

  Finally, should Micro$oft be broken up?  It's my opinion that it would actually benefit Micro$oft.  Each division would be leaner and more integrated, and able to adjust to changing conditions faster.  It would be like a de-facto stock split, and each baby-Bill stock would grow in value, providing larger dividends in the long run for investors.  

  So next time you find yourself slipping, and starting to think Bill is the slickest entrepreneur since P.T. Barnum, think back to the last driver problem, Windows crash, lost document or non-functional peripheral you had to suffer though, and remember that your operating system was designed more by Bill  and the marketing team, than it was by the software engineers at Micro$oft.

                               Dr. John