February 28th 2000

3dfx Announces Voodoo4 and Voodoo5 Graphics Cards.

  3dfx showed off their new Voodoo4 and Voodoo5 graphics accelerator cards at the CeBit convention last week.  This means we should start to see them show up in stores sometime in mid to late Spring (April or May?).  The graphics chip that powers the new Voodoo cards is the VSA-100 (Voodoo scalable architecture).  This chip allows for parallel processing using multiple graphics processors. The most powerful new consumer level Voodoo card will be the Voodoo 5 6000 with file rates in excess of 1 Gigapixel per second.  This is a far greater fill rate than any existing graphics card, but will come at a price.  The estimated cost of the Voodoo 5 6000 will be between $500 and $600!!  Start saving your pennies folks.

  The only advanced video feature that the new Voodoo cards will not have is environmental bump mapping.  In addition, 3dfx will not be implementing on-card texture and lighting calculations like on the nVidia GeForce card.  But these cards will offer the highest fill rates available, along with full scene anti-alaising, T-buffer special effects including motion blur, large 32 bit textures, texture compression, and more.  In other words, these will be the premier 3D graphics cards for the year 2000.

  We've been waiting for this one for awhile. The VSA-100 chip will be the biggest boost in 3D graphics on the PC since the introduction of the first Voodoo card.  This is the big one folks, don't buy a new video card till these bad boys come out.  Watch a real video of the CeBit presentation here.

                               Dr. John

Time To Sell that Rambus Stock? Maybe not yet, but soon!

    Details are leaking out that Intel knows that SDRAM and Double Data Rate DRAM perform as well as, or better than Rambus memory, as reported in The Register today.  Despite this, Rambus' stock is not just going up up up, it's skyrocketing!  And it's possible that it will go even higher for a short time, if it is announced soon that the Sony PlayStation 2 will use Rambus memory.  But the loose-lips talk about the over-valuation that are starting to circulate will eventually turn the tide on Rambus' day in the sun.  And if the PlayStation 2 turns out to use a different kind of memory, then it's bye bye to the stock price's high.  

   Should be fun to watch for all those folks who don't own Rambus stock.  For those who do, decisions made now will determine whether they get that extra vacation in the Bahamas this year or not.  Choose wisely Grasshopper.

                                Dr. John

February 24th 2000

Faster Hard Drives On the Way.

  Just when you thought your new ATA-66 drive was the tops, the new ATA-100 standard is announced.  Expect to see these new drives show up on store shelves over the next few months.  Motherboard support may not be in place until this Summer.  On the SCSI front, Ultra 160 SCSI offers incredible performance with 10,000 rpm LVD (low voltage differential) drives.  Now Seagate has announced 15,000 rpm SCSI drives which should be available soon. The performance is phenomenal, with access times around 2ms. The price will be high at first, but the price on 10K rpm drives will likely drop after the 15K rpm drives debut.

                                Dr. John

Sunspots Scheduled to Zap your Computer Next Week.

   The so-called "Solar Maximum", the peak in the 11 year solar flare cycle, is scheduled to reach it's high point next week.  Large solar flares send millions of tons of highly charged "cosmic particles" hurtling toward Earth, where they can disrupt power plants, telecommunication equipment, and even computers.  Hopefully, he light show offered by the Northern Lights will make up for any electrical irritations.  

  So far, the sun has been relatively quiet leading up to the expected storms. You can check out all the fun here.

                                  Dr. John

February 23rd 2000

Rambus Stock On a Wild Ride.

 Defying all logic, the price of Rambus' stock continues to rise after the announcement that Intel's Willamette processor will use this type of RAM.  Here is a quote from The Register's Mike Magee.

"This particular share price is like a white-knuckle ride, and we can only think that some sort of collective madness has descended on Wall Street and punters for the price to fly so high".

  It is at least slightly amusing that despite Intel's claim that Willamette systems will use Rambus, it is much more likely they will have support for PC-133 SDRAM, and perhaps even Double Data Rate DRAM.  AMD and VIA are definitely going the Double Data Rate route for the Athlon and it's future chipsets, and Intel will be in a very bad position if they don't follow suit.  The Athlon already uses double data rate technology within the CPU itself, so it is a natural progression for them to extend this to the memory bus.

  It is unclear how long the misperceptions about the future of Rambus will go on, but reality will sink in soon.  For those of you who have been holding onto Rambus for a good time to sell, that time draws nigh.  For those who are adventurous, maybe it's time to "short" some Rambus stock.  You can read about "Stock shorting" here. 

                               Dr. John

Intel Processor Shortage Continues.

  Intel processors continue to be very scarce, and according to our distributors, they will stay that way until sometime in March.  Currently, the Celeron 533, Pentium III 550 and certain Xeons are the only readily available CPUs from Intel.  Apparently, the change-over to the Coppermine Flip-Chips is the major holdup.  One more 0.18 micron fabrication plant is about to come on-line, so we should start to see more Flip-Chips in a few weeks.

                                 Dr. John

February 20th 2000

How is Intel like Apple?

 Intel continues to do the dumbest things they have ever done in their history, and they are being extremely stubborn about it. I have heard tell that at Intel, the marketing people run the show there, and the engineers are like underlings that must do marketing's bidding. The engineers complain, and say it won't work, but marketing says, "make it work!"

  The entire history of Intel has been geared toward making PC hardware as proprietary as Mac stuff. The so-called "P6" bus with the BX and now newer chipsets is all proprietary, and Intel wants to keep it that way. They charge about twice as much for their chipsets as VIA does (which is why VIA motherboards are so much cheaper), and in fact, they are suing VIA over the Apollo Pro 133 chipset saying it infringes on their proprietary bus architecture. 

  The i820 chipset has been a big disappointment for gamers with it's lack of PC-133 SDRAM support.  Now Intel has resurrected the sleeping  i815 chipset. The Solano2 chipset is designated #2, even though #1 never came out. Intel canned Solano1 because it had SDRAM support, and would have cut into profits on the i820 and Rambus. Now it's back, because Intel couldn't stop VIA from shipping a chipset with PC-133 SDRAM support. 

  Word is that the i815e, Solano2 will have native PC-133 SDRAM support.  This is a step in the right direction, but by the time it comes out will be too little too late.  Double Data rate DRAM (DDR-DRAM) is the way to go.  Intel will not support DDR in a desktop PC platform unless forced to do so by the competition. Don't look to Intel for new product innovation right now. They are still battling between the marketing and engineering departments.

  What I am really interested in seeing is a VIA second generation Apollo Pro 133 chipset. The current one needs some work, but it needs tweaking more than an overhaul.  For example, I would like to see ATA/66 support as robust as that offered by HighPoint's controllers.  If VIA can add DDR DRAM support to the next KX133 Athlon chipset before there is DDR support for the Pentium III, then AMD and VIA will take further market share from Intel.  The good news for Intel is that they don't have to worry about the Federal Trade Commission charging them with being a monopoly anymore.

                             Dr. John

February 19th 2000

Memory Prices Drop Like a Rock (Finally!).

    In just the last several weeks, memory prices have plummeted to their lowest levels in over 5 months.  The huge memory price spike is finally over.  Industry analysts (whoever they are) predict slowly rising prices for other key computer components in the near term.  We have not seen this, except in the case of rare items like Coppermine Pentium IIIs and DDR GeForce cards, who's prices have been creeping up due to supply and demand disparities.  But memory prices seem to keep falling.  As more manufacturers turn away from Rambus, they are no doubt dedicating more fabrication capacity to SDRAM production.

                               Dr. John

February 18th 2000

Microsoft denies Win 2K bug count.

    In a rebuttal letter, Microsoft Group Vice President, Jim Allchin, replied to reports on the Internet that there were thousands of "bugs" in the release version of Windows 2000.  Mr. Allchin said, "these reports are inaccurate".  While this seems to be a little short of "there are no bugs in Win 2K", it's good to know that the number of 60 some thousand is inaccurate.  Mr. Allchin goes on to ask, and subsequently answer the question "So, does Windows 2000 have 63,000 defects? The answer is a flat no." 

  To describe the situation more clearly, he states, "We also have a special advanced source code analyzer that we use. This tool generates a significant number of false positives (it "thinks" the code should be changed, but in fact it should not be). But, we track them all. The only way to be sure is to look at each hit and see if the issue is real or not. We love this tool. It helps us improve our code for readability and it can find bugs that our testing may not find. We also track our test code. There are over 10 million lines of test code—that can also have improvements, potential bugs, etc.—that we track in the same database.

  Finally, he queries, "Will customers run into bugs in Windows 2000? We worked harder than ever to ensure they would not. Windows 2000 is the highest quality product we have ever released--" 

  So there you have it, Jim Allchin says that Microsoft worked harder than ever to ensure there would not be bugs.  That means there are some, but it's a lot less than 63,000.  In fact, Windows 2000 is probably Microsoft's best software yet, but because of the fact that there are millions of lines of computer code, everyone knows there are still bugs in the first release.  Nonetheless, Windows NT was long overdue for an overhaul, and Microsoft delivered.

                                Dr. John

February 16th 2000

Gag me with Rambus.

    In an article in the Register today, Mike Magee reports from the Intel Developers Forum in Palm Springs, and he gives us some hints about upcoming goodies from Intel.  When can we expect Willamette, the next generation processor from Intel?  The current release date is in October, but that will probably slip.  Here are the tidbits that Intel tossed out for the hungry crowd.

1) Willamette's on-die L2 cache will be "slightly bigger" than the Coppermine L2. What is slightly bigger than 256Kb?  Originally they were talking about 1MB of L2! Hmmm.

2) It will not launch at 1.5 GHz.  Probably more like 1GHz to 1.2GHz.

3) It will be about 30% faster than comparable Coppermine chips.

4) It will have new SIMD-2 instructions

5) The math unit will run at twice the speed of the rest of the CPU.  Hey... Dual speed processors!

6) They will probably call it the Pentium IV.  Now this is just plain silly, they named their 5th generation chip the Pentium for trademark reasons. Now the name has become (5th generation-4).  Either give it a new name, or call it the Septium for crying out loud! (Sounds kind of like a sewage system, doesn't it?)

7) Now for the head kicker.  According to Dr. Albert Yu, Senior VP at Intel, the Willamette chipset (code named Tehama) will only support Rambus DRAM!  OUCH!  This needs further delving.

  According to Dr. Yu, Rambus will be the primary  memory subsystem for Willamette motherboards.  He said SDRAM would not be sufficient.  But then he goes on to say that Foster, the server version of the Willamette, will use a chipset that has Double Data Rate DRAM support!  This is absolutely ridiculous.  They are putting less expensive memory support into their server platform, and putting extremely expensive, unstable, and difficult-to-use memory support in their consumer-level platform? What's up with that?  

  Intel has been getting yelled at by every one of their major customers to axe the Rambus memory, and go for Double Data Rate DRAM.  Now we find out that they are still developing the next generation consumer-level chipset to work with Rambus!  What's going on over there?  Didn't they learn anything after the i820 disaster?  Apparently not.  If this does not change, Intel will again leave a planet-sized hole in their product offerings that VIA can drive all their trucks through.

  Rambus stock prices soared on the news, proving that stock traders don't know diddly about computer technology.

                                Dr. John

February 15th 2000

Say It Ain't So Bill!

    In a ZDnet article, it has been revealed that Microsoft has acknowledged thousands of potential bugs in the release version of Windows 2000.  While many gamers have been eagerly awaiting the release of Windows 2000 Pro, the version with Direct X support, we predict that this OS will fall flat in this arena due to high price and lack of driver support.  Also, due to excessive security code, etc, this will not be a streamlined gamers operating system.

   Like others, we recommend waiting until the first service pack comes out, and more driver support is available.  In addition, Windows 2000 is based upon NT code, not Win 98 code, so Windows 98 owners will not be able to use their existing software on the new operating system.  And at $400 to $500 a pop, it's not going to be bargain-basement software.

                           Dr. John

February 10th 2000

Does Intel Stifle AMD Athlon Distribution?

    We have not seen any mention of this on the web, but we sure get hit in the face with it every time we go to our major distributor's web sites to get parts.  No AMD Athlon processors.  That's correct.  Major hardware distributors such as Tech Data and Synnex, which sell Intel processors, do not carry any AMD products whatsoever.  Is it just that they don't think the Athlon is worth carrying?  No way. You can be just about certain that it's because they have been threatened by Intel with Pentium III shortages if they distribute AMD processors. 

  Wouldn't this be a great time for them to start?  Let's face it, both Tech Data and Synnex have nary an Intel processor to sell right now, so threatening them with processor shortages is little like threatening a condemned man with a jail sentence.  They have nothing to loose, and profits and market share to gain. 

  Currently we need to go through smaller distributors to get Athlon processors.  We have no idea what gives Intel such clout that it can dictate to distributors what they can and can not carry, but it's a disturbing aspect of this business.  Free markets should not have these types of restrictive distribution channels where one manufacturer can limit the supply of competitors products through third parties.  But I'm certain that Tech Data, Synnex, and other Intel-only distributors have gotten lots of complaints about the situation. Let's hope they get lots more.

                              Dr. John

Web Attacks Will Likely Get More Frequent.

    For the time being, I expect the web attacks that have hit sites like eBay and Buy.com will continue.  There will be lulls and spikes in activity, but the plain fact is, that security on the web is poor indeed.  The main reason why these attacks have happened is because they can be done fairly easily, with little chance of getting caught.  This will change as web sites slowly become more like little Cyber-Fortresses.  So in one sense, the folks who have all that free time on their hands to pull these stunts do in fact help beef up security by pointing out all the holes.  When money is lost due to these attacks, you can be certain that the problem will get lots of money thrown at it.  Eventually, the hackers will have to go after the smaller fish, because the big ones will be locked down.  

  The hackers are shutting down web sites, rather than robbing banks, because they don't have the same fear of getting caught.  After some of them get nailed, and jailed, you can expect the number of incidents to slowly drop.

                              Dr. John

Why Keep Making the Buggy Coppermine?

    Intel announced a couple weeks back that it would release the new stepping of the Coppermine Pentium III processor in early April.  But ever since then, we have not seen any Coppermine Pentium III CPUs show up at major Intel distributors.  This makes us wonder if Intel is still producing the original Coppermines at all, and if so, why?  It's obvious they do not have built-up inventories of the Coppermine CPUs that need to be liquidated before rolling out the new processors.  So why even make any of the old ones and ship them out the door?  Perhaps Intel is worried that the new stepping may have new problems, and wants to do more testing. But now that we know how buggy the original stepping is, I'm not sure if buying a Coppermine before April is a good idea.

                            Dr. John 

Intel Sics the ITC on VIA!

    Intel has managed to get the International Trade Commission (ITC) to open an unfair trade practices investigation into VIA Technologies, and others, for patent infringement violations it was reported on http://www.eetimes.com/ yesterday. This is after Intel filled a petition with the International Trade body in January charging VIA with patent infringement concerning the P6 front side bus architecture in it's new Apollo Pro 133 chipset.

  On February 8th, the ITC voted to conduct an investigation into the matter.  The first step is an evidentiary hearing to determine if the facts merit pursuing the case.

                            Dr. John 

February 7th 2000

Slot-Not is Not Happening.

    According to an article in The Register today, Intel has actually told computer manufacturers to use Flip-Chip Slot adapters for the new Flip Chip Pentium III processors.  What's so odd about that?  Intel released the Flip Chip Pentium III, which looks like a skinny Celeron, back in November 1999, but still has no motherboard support for this processor format.  Add to this the fact that they are phasing out the Slot-1 format much faster than they had originally intended to, and moving to the "Slot-Not" format ASAP.  

  This leaves consumers and computer makers in a bit of a pickle, since Slot-1 and Slot-Not Pentium IIIs are similarly scarce, and motherboards with native Flip-Chip Pin Grid Array (FCPGA) support just don't exist.  Therefore, even if you can find one of the rare Coppermine Flip-Chips, you need a special FCPGA-to-Slot1 adapter to make it work on existing motherboards.  This introduces another level of complexity to getting a system up and running with a Flip Chip processor.  And as you might imagine, since the FCPGA-to-Slot1 adapters are brand new, they are also difficult to obtain.

  Bottom line?  Intel debuted a new processor format without any hardware support in place, and now 3 months later, they still have not made such support available, despite the fact that they are converting their entire processor line over to the new form-factor.  In short, they really messed up at a time when AMD seems to be able to deliver oodles of Slot-A Athlon processors. A few more big blunders like this, and AMD will have as much market share as Intel.

                         Dr. John

February 4th 2000

Whither Willamette?

   One thing we have been wondering about with all the problems Intel has been having with processor production is what's going on with Intel's next generation 32 bit processor, code-named Willamette?  The Willamette is still basically a P6 at it's core, but it has additional decoders and execution units, a longer "pipeline" and larger buffers.  It will be based on the 0.18 micron fabrication process, and will have as much as 1MB of on-die L2 cache.  This processor will use a 200MHz front side bus, and will have a new SIMD instruction set.

  The Willamette will use a modified Socket 370 design with 423 pins.  So it's kind of like a Super-Coppermine, and will be introduced at speeds starting at a Gigahertz.  Don't expect them soon, they will not debut until late 2000, or early 2001.  Then you can expect that they will require new motherboard support.  The code-name for the Willamette-compatible chipset is Tehama.  Intel considers the changes in this processor to be significant enough to designate it P7, and will probably call this one the Pentium IV. 

  While this processor/chipset combination was originally conceived of as being paired with RAMbus DRAM, we wonder if Intel is instead working Double Data Rate DRAM support into the architecture instead.  It would be a much smarter move.

                           Dr. John