News Archive

Intel has not invited Rambus Inc to join it's upcoming memory standards forum. Other memory manufacturers, including Samsung and Micron, have invitations. The reason for the lack of a Rambus RSVP is that these hardware hounds are discussing ways of integrating PC-133 and Double Data Rate SDRAM support into future Intel chipsets. Expect Rambus' stock to take a hit.

This may be "too-little too-late" in terms of support for these memory standards, since the brand new line of chipsets from Intel have no support for these memory types. Unless Intel breaks all it's R&D records for development of a new chipset line, expect a long delay before the new chipsets with support for faster memory are actually available for purchase.

The holidays have left many hardware distributors with shelves as bare as Mother Hubbard's cupboards. We have noted shortages of almost all major computer components, including Pentium III CPUs, high-end video cards, hard drives, motherboards, and memory. These continuing shortages will almost certainly keep prices higher than they would have been if supplies were more plentiful.

Computer purchases have traditionally remained high just after the holidays, so the shortages may continue for several more weeks. But by the end of January, new parts which have been in production should enter the market, relieving some of the supply shortage. The earthquake in Taiwan a couple months ago occurred at a time when manufacturers were gearing up for the holiday rush. So we may still be feeling it's aftereffects on supplies.

December 28th

The recent space shuttle mission to service the Hubble Space Telescope was successful, and Commander Curtis Brown safely landed Discovery yesterday at the Kennedy Space Center in Florida.

The Hubble was fitted with new gyroscopes, a new protective outer layer, a new radio transmitter, a new digital solid state data recorder that replaces an old, tape-based unit, and the Hubble also got a new computer. NASA reports that the new computer is 20 times faster, with 6 times as much memory as the previous Hubble computer.

The new computer has been in development for at least 4 years, and is called the HST486. That stands for Hubble Space Telescope 486. That's right. The new "Advanced Computer System" upgrades the Hubble from a 386 computer to an Intel 486 computer! It's obvious that the folks at NASA won't be playing any Quake 3 Arena on the ol' Hubble. But the original Frogger might be an option.

December 24th

We want to wish everyone happy holidays from all of us here at KickAss Gear. Don't forget, this is just a practice session for the real turn of the millennium which occurs in 2001! But it ought to be a great dry run! Cheers!

December 19th

Many rumors were spread when the Athlon was first introduced that they could not be overclocked without special equipment. Like many Internet rumors, they were wrong. The early Athlons that were tested could not be overclocked because they were tested on motherboards that did not have overclock bus settings.

The Asus K7M has the Front Side Bus frequency settings in 1MHz increments between 100MHz and 125MHz, as well as having 133MHz and 150MHz settings. The Athlons we received recently were packaged in mid October, and will overclock very well to approximately 50MHz above their rated speed. Newer Athlons are reported to overclock substantially better. This may be due to the new Copper-interconnect technology that AMD is now using for fabrication.

We will have a full review of the Athlon's overclocking potential on the K7M soon.

December 16th

Many computer parts are in very short supply leading up to the last few shopping days before the holidays are over. Perhaps manufacturers were worried, and did not want to over-produce and over-stock this holiday shopping season.

You will be hard pressed to find Voodoo3 cards, IBM 7200 rpm drives, new DVD drives, 450MHz and 500MHz Pentium IIIs, and many other computer items. New items are in extremely short supply, including GeForce video cards, and high-speed Athlon and Pentium III processors. Some prices are rising slightly due to the increased demand for limited stocks of popular items.

December 12th

Many motherboard manufacturers have begun to turn out Slot-A Athlon motherboards, and some of them are very high quality! The Asus K7M is finally available in limited quantities, and looks like a very good mainboard. We will have Athlon CPUs, and the K7M for sale next week. Even companies like FIC seem to have fixed some of the problems with the early version of their Slot-A boards.

Unfortunately, the new "Coppermine" Pentium III is dead in the water, and sinking at the bow. Most of the problems appear to be a result of Intel getting too big for one section of this mega-corporation to know what the other sections are doing. Throw in a touch of greed, with their plan to get the entire computer industry to switch to the much more expensive Rambus DRAM, and you end up with a disaster. As we have said before, any plan that makes one essential part of a computer much more expensive, without drastic performance gains, is doomed to failure.

We wish we could offer both Coppermine and Athlon systems now, but due to a complete lack of acceptable motherboards for the new, 0.18 micron Pentium IIIs, this is not possible. Based upon rumors we have heard from very reliable sources, motherboards based on the i815 chipset, which will have PC-133 SDRAM support, may not be ready until June or July of 2000! This gives AMD plenty of breathing room to expand into the high-end market, where Intel has always been King.

December 10th

Intel keeps an updated specification sheet for the Pentium III Coppermine CPUs which describes the "bugs" (errata) in the design, and prescribes workarounds for them when possible. We downloaded the latest version from Intel, and were surprised to see the number of bugs is already up to 58!!! Errata number 56 is particularly interesting. Some of you may remember how much trouble the math bug in the original Pentium chip was for Intel. Called the FDIV bug, it caused Intel's first big-time processor recall.

Now a math bug (errata #56) has been found in the new Coppermine Pentium III CPU that can turn a positive number result into a negative number result! Oops! According to Intel, the system BIOS can be changed to eliminate this problem.

For those of you who know nothing of Scientology, it's the pseudo-religion created by L. Ron Hubbard in the 1950's which suggests all personal and emotional problems stem from bad experiences you had as a child. Called "Engrams", these deeply buried bad memories rule your life, always with negative consequences, whether you know it or not! For an immodest fee, they will fix your problems for you.

Who cares if these nuts want to believe such drivel? Businesses and consumers thinking of upgrading to Windows 2000 might. The disk defragmenting utility built into Windows 2000, "Diskeeper 5.0", was written by a group of programmers at ExecSoft, who are all self-professed Scientologists. You can check out the CEO's "About Myself" here. "Diskeeper" even has a semi-religious tone, doesn't it?

While this is probably no big deal to most Americans, Germans have been doing battle with Scientologists for years, and the European Catholic Church has even suggested it may boycott Windows 2000 if the situation is not addressed by Microsoft. What's next? Maybe there will be a new Win2K backup utility, perhaps called Jihad, which is written by terrorists ;).

December 8th

Intel quietly relegated the Pentium III 450 CPU to the scrap heap without any fanfare. The internal logic behind this retirement is that higher end CPUs are available, and the low end then becomes expendable. However, what this line of thought neglects is the simple, undeniable fact that none of the new Pentium III chips will run on most motherboards. So the PIII 600 CPU is the top end for motherboards based upon the BX chipset, and the PIII 500 is the low end. Not much wiggle-room in there.

Why does this matter? The BX is a dinosaur isn't it? While that is true in theory, in practice, the BX chipset outperforms i820 "Cape Cod" motherboards that support PC-100 SDRAM. This is because memory access is hobbled on the i820 when used in conjunction with PC-100 SDRAM. There is no "native" support for PC-100 SDRAM in the i820 chipset, so a memory translation hub is required. Systems take a 30% to 40% memory access performance hit when using the memory translation hub. Therefore, the BX chipset, which can be overclocked to above 133MHz with PC-133 SDRAM, outperforms the i820 with SDRAM.

What about Rambus on the i820? It will give you a few percent better memory performance than PC-133 SDRAM, but a 128MB stick of 800MHz Rambus DRAM costs approximately $1000. That's not an option. The bottom line is, if you want to get a new "Coppermine" Pentium III now, you will have to get a motherboard that either reduces performance, or one who's cost-of-ownership makes private aviation sound like a poor-man's hobby.

December 2nd

Didn't you know this was coming? We have been thinking for some time that the "errata" (nice techy-talk for "errors") in the new Coppermine chips were going to eventually grow in number, and end up including an actual noticeable "bug" that created problems for end users under some circumstances. The count of known "errata" in the Coppermine CPUs is up to something like 49 now, but software and hardware work-arounds can usually be devised to limit their effect on system operation.

But now that Coppermine systems are trickling out, it appears that some percentage of them have a bug that can't be fixed with work-arounds. It does not affect all Coppermines, but that means buying a Coppermine system now is a little like gambling. The bug is not a serious system-crashing problem, but it is certain to be irritating to anyone who gets an afflicted Coppermine system. The bug causes the computer to fail to turn on the first time the power button is pushed. A second button press should bring the system on-line.

This can't be good news to Intel engineers, who failed miserably with the i820 motherboard, and who have had problems with the next-generation IA-64 "Merced" CPU as well. Intel is perhaps growing too large to integrate it's various parts into a well-oiled machine for developing new hardware.

November 30th, 1999

The fact that a good number of folks have bought one version of Linux or another is one thing. But the additional fact that Red Hat's and Corel's stock prices have risen dramatically in the last few weeks means that corporate America is taking this operating system more seriously. This is great news. The two major problems that Linux faces are lack of available applications written to run on it, and some installation and drivers issues.

Yet now that Microsoft has been ruled a monopoly in court, and Linux is becoming more popular, it looks as if money may start pouring into Linux distributor's coffers. This will allow them to develop the operating system into a mainstream product, and viable alternative for some consumers. The increased ease of use that should result from increased development funding will help bring more applications developers on board, and Linux ports of existing Windows applications should become more plentiful.

Rumors have started around that Red Hat feels spunky enough to purchase Corel, the makers of Word Perfect, and a new, easier to install version of Linux named Corel Linux OS. It's starting to look like a Linux feeding frenzy is in the offing. We think the Corel version, which comes with Word Perfect, may help stimulate further interest in this open source operating system.

There is a whole lot of confusion out there about motherboard availability for Coppermine and Athlon CPUs. Athlon motherboards are available, but the quality is low. Wait for better quality motherboards before getting an Athlon system. Also, wait for motherboards with the VIA KX-133 chipset on them, they are better than boards with the AMD750 chipset.

Virtually no "Flip-Chip Coppermine" motherboards are available yet. The Flip-Chips don't work on existing Socket 370 motherboards. Even if you find a new motherboard with Coppermine support, don't get an i810E or i820 motherboard from Intel, they won't allow overclocking. Wait for companies like Asus, AOpen, and Abit to come out with their new motherboards with Coppermine support.

Very few good motherboards are available for Slot-1 Coppermine CPU's either. There is no reason to get an Intel Cape Cod motherboard, they also won't overclock the CPU. In addition, the motherboards with the i820 chipset that have SDRAM sockets use a memory translation hub that slows down memory access. This is not a good option. And Rambus is too expensive for anyone to consider seriously, so forget i820 boards with Rambus sockets. We are hoping that the i815 chipset makes it to market soon (don't hold your breath), with support for PC-133 SDRAM and Coppermine CPUs.

In general, we recommend that folks wait to get Athlon or Coppermine systems. You can find them out there, but you will not be happy with the quality or features on existing motherboards. So if you want a top-notch system, rather than just the first Coppermine or Athlon on your block, then wait. You will be glad you did. How long? It's hard to say. We estimate it may be one or two months more before good quality motherboards are readily available for either CPU.

Apparently, the European Union is going to take a look at the privacy implications of Intel's Pentium III "Personal Serial Number" (PSN). We thought this stuff had gone the way of Travelgate, Whitewater, and Filegate, but it looks like the European Union is worried that the FBI and National Security Agency of the United States may be making unlawful use of the PIII PSN to track the activities of US citizens on the net. Perhaps, the reasoning must go, this is just a short step away from the Feds deciding to go rifling through the dirty web laundry of Euro-surfers!

Is it just me, or do you have the feeling that spies and Feds have better ways to figure out who's doing what than sifting through terabytes of CyberSpace traffic for PSNs? How about a simple phone tap? And what if the bad guys are using Cyrix or AMD chips? And what if they switch between different Pentium III processors depending on whether they are doing legitimate or illegitimate activities? I'm still a little befuddled as to what exactly these minions of Uncle Sam are supposed to be doing with PSNs that they could not do more easily other ways. But then again, we are so dumb here, we are all planning on flying back and fourth across the International dateline in a 737 at Midnight on December 31st just to see what happens to the computers on board. Should be fun.

Intel is dropping prices on Pentium III CPUs. We can't say yet if the drops will be small or large, but if you are planning on getting one of the more expensive chips, you might do well to wait.

November 24th

Ever wonder what kind of stuff goes on behind closed doors in the computer industry? The details of Intel's secret deal with Gateway computers was reveled, sort of. They are not commenting on it, of course, but here's the scoop. Apparently, Intel offered Gateway as much as 100 million dollars to NOT use AMD parts. They offered Gateway a rebate in the neighborhood of $50 per Intel CPU purchased, as long as Gateway did not offer an AMD Athlon system for sale.

Does this sound familiar folks? As we have mentioned before, playing hard-ball is second nature to companies like Micro$oft and Intel. They would rather rely on shady business practices to do what their engineers and programmers can't: get people to prefer their product.

As computer manufacturers ourselves, we really have a bone to pick with these types of deals. Large computer makers already get special deals from companies like Intel and Micro$oft, in part to keep them beholden to these industry giants. And when the purpose of the "price break" is to coerce them into using Intel or Micro$oft products exclusively, consumers do not benefit.

Now Gateway has decided that the AMD Athlon is too good a processor to ignore, so they are going ahead with plans to produce an Athlon-based system by the end of the month. Intel has pulled the sweetheart deal, and now has little or no control over Gateway's product line. Let's hope that all the major computer makers decide that making products their customers want is more important than doing Intel's bidding. We are really sick and tired of hearing how the big players try to unfairly manipulate other companies at the consumers expense.

November 22nd

Unless you are a-hankerin' to put your shiny new 500E or 550E Coppermine CPU into a lowly i810E motherboard, you may have to wait to use that Flip-Chip. We are not saying the Flip is a Flop, just that it was released without any motherboard support. Asus has announced that the next version of several of it's Socket 370 motherboards will support the Flip Chip CPUs (like the MEW, rev 1.08) . But when will they be released? No word on that. A special socket 370 adapter card will allow Flip Chip CPUs to fit on new Slot-1 boards with the i820 chipset, but that will negate some of the benefits of the new Flip Chip design.

So for the most part, if you are interested in one of the new Flip Chips, your best bet is to wait for an i820-based Socket 370 board to be released. How long this will be is not certain, but companies are certainly working on it now.

Now that Comdex is over, you might be wondering what new hardware we can expect for the New Year? Well, unfortunately, the iMac has created such a feeding frenzy among computer makers, that about half the new stuff is just a bunch of ugly iMac clones, and attempts at making PCs look "better". I expect eventually, computers will look like an Art Deco paper weight.

Blue Tooth: No more wires (if we can believe them) piled up on the floor. Except for power cords, all the cables to peripheral devices will be eliminated if this technology actually works well. How much more you will have to pay for a Blue Tooth keyboard is unclear.

Voodoo VSA-100: The GeForce 256 is an impressive bit of technology, but what 3dfx has known for a long time is, a single chip can do only so much. The new VSA-100 chip by 3dfx has all the major features you could want in a new video card, like 32bit color, full-scene anti-aliasing, large texture support, etc, so speed becomes the main issue. However, rather than having to make a new video chip every year or two, 3dfx was smart enough to make a good basic chip that can be run in a multiprocessor configuration. Up to 32 VSA-100 chips can theoretically be integrated into a single system, with up to 2GBs of dedicated video memory. Now that's scalability! As memory prices drop next year, expect to see the 4-chip Voodoo5 6000 card with 128MB of RAM drop below $500. It will be substantially faster than the GeForce 256.

No More Oldies: Legacy devices are going for good, no more floppies, ISA slots or even CD drives. DVD and CD-R drives will probably replace standard CDs now that the price is so low.

Firewire: The standard that spun off USB, known as Firewire, will make it's debut in mainstream PCs in the year 2000. It's much faster than USB, and will first show up as an input jack for digital cameras and camcorders.

Ultra-3 SCSI is coming out. If you thought Ultra-2 SCSI was fast, how about doubling it? Sustained transfer rates will be 4 times that seen with ATA66 drives.

Games with new 3D technologies: Expect new games to have support for new 3D technologies that will greatly improve game speed and appearance. From "Level of Detail" (LoD) polygon count reduction which allows more detail in close objects, to "weighted vertex" programming, which allows materials like skin and cloth to behave more realistically, you can expect upcoming games to be much much cooler looking!

Athlons and Coppermines above a Gigahertz: Both of these CPUs should be scalable to at least a Gigahertz. Reports on overclocking of the new Coppermines have shown they will be overclockable to approximately 1.2GH in their present form.

Intel is pushing ahead with the lawsuit against VIA, the little chipmaker, and FIC and Everex, which are motherboard manufacturers that use VIA chipsets. Intel claims that VIA has violated patent agreements with Intel, but Intel will not provide any details. Of course, many larger companies use VIA chipsets on their motherboards, such as Compaq and IBM, but Intel did not name them in the suit.

Obviously, Intel is trying to intimidate motherboard manufacturers in the hopes of getting them to steer clear of VIA products, and stick with Intel chipsets instead. Of course, Intel can not make enough chipsets to meet the worlds demands, but apparently, that's not their concern. They just want to beat up on small companies in the hopes of putting the fear of litigation into larger companies.

Once again, Intel is not making the best product possible, and getting it to market before the competition. Rather, they are relying on their lawyers to do what their engineers can't: make sure everyone uses Intel chipsets.

Corel is releasing their new version of the Linux operating system (Corel Linux OS). The deluxe version will have Word Perfect Office bundled with it. They are planning on releasing Linux versions of Quattro Pro and Corel Draw soon as well. This is very good news, because it means that more software will be available for this operating system soon. The biggest shortcomings of Linux are buggy installations, and lack of software. Corel is helping change both of these problems with their new release.

November 15th

3dfx has produced a new graphics chip they are calling the VSA-100. The new chip has 32bit color support, full-scene anti-aliasing, T-buffer digital effects, large texture support, texture compression, and multiprocessor support for up to 32 VSA-100 chips on the same board! These boards will come in several flavors, starting with a 32MB Voodoo4 4000 card. Perhaps the most exciting card will be the Voodoo5 6000 card, with 128MB of onboard RAM, and 4 VSA-100 chips running in SLI mode. It will be able to deliver nearly 1.5 Gigapixels per second! That's some real pixel pushing power! Don't expect to see these cards available until sometime in the Spring.

It's the week we have been waiting for. All those non-disclosure agreements that manufacturers had to sign are history, and we get to hear what they have been working on.

Abit announced two new motherboards based upon the i820 chipset (can you say "only two RIMM sockets?"). They are the Abit CX6 and the CH6. The CX6 is only for the rich, since it has two (yup, only two) Rambus memory sockets (RIMMs). Currently, Rambus is 4 to 5 times the price of SDRAM, and does not offer a significant performance boost. The CH6 is more interesting since it has PC-100 SDRAM support, and Softmenu II, which will allow overclocking of the front side bus. The question is, why did Abit stick with Softmenu II for it's i820 mainboards, when softmenu III is already available on their BE6-2 and BF6 motherboards?

Abit is also releasing a GeForce 256-based graphics card called the "SiluroT GT2".

Intel is announcing a bunch of motherboards that only Dell will buy. If you want an i820 motherboard with all sorts of junk integrated on it, Dell is the place to go. Intel has announced the following:

1) "Lockport" (D820LP), socket 370 with the i820 chipset, and integrated audio and networking,

4) "Templeton" socket 370 with the i810e chipset, with integrated video, network and audio!

5) "S820PN-2" Dual slot 1 with the i820 chipset, and integrated network and Ultra-2 SCSI controller.

None of these will be good for overclockers, but at least it shows how much Intel is pushing the new Flip-Chip design. All but one of these motherboards is for the new Flip-Chip Pentium III. You can read a Preview of the Flip-Chips here:

November 12th

We finally got our hands on a single, lonely 500E Pentium III. These CPU's come in the new "Flip-Chip" Pin Grid Array format (FC-PGA). They are not compatible with the pin configuration of S370 (Celeron) motherboards, and they use a different Voltage regulator module version. Therefore they won't work with current S370 motherboards. We will post a Preview of this chip tonight or tomorrow, with more details. Currently, only i810E motherboards will run the new Flip Chips.

Reports have been coming in with benchmarks that support what we suspected all along, that Double Data Rate SDRAM will beat the far more expensive and proprietary Rambus DRAM in performance. Just another nail in Rambus' coffin.

November 11th

Unlike any other debut of Intel processors we can remember, the roll-out of the Coppermine Pentium III's offers slim pickings for customers and computer makers alike. In a typical roll-out of a well-hyped product, Intel would have the distribution channels pre-charged with hundreds of thousands of pieces. This time, there were zero E-version Pentium III's for one week after the debut. Then on the next Monday, Tech Data, one of Intel's largest distributors, had only a few hundred 500E processors listed at one of their eight warehouses! The other E-version Pentium IIIs were still not available. Other Intel distributors did not seem to even know about the new processors. A few hundred of the 550E's showed up at Tech Data a couple days later.

This is certainly not your typical Intel processor debut. The interesting thing is that the only two processors which are being dribbled onto the market are the 500E and 550E, both of them in the new Flip-Chip Pin Grid Array (FC-PGA) package. These are bare chips, without the circuit board or the plastic housing. They will look similar to the current Celerons, and are pin-compatible with the socket 370 format. They are much quicker and easier to produce, so that may explain why they made it to market first. Therefore, if you are planning on getting a new 500E or 550E, remember, they go onto new i810E-based S370 motherboards, not Slot-1 motherboards. Expect 370 motherboards based on other chipsets that will take the new Coppermine CPUs soon.

To us, this trickle-debut is Intel's attempt to dampen enthusiasm for the Athlon 700 processor from their competitor, AMD, which actually is available.

November 8th

Perhaps Intel has been learning from Microsoft how not to run a company. We learned after Microsoft's defeat in court on Friday that Microsoft would often announce new products that were not ready, in order to blunt the release of products from competitors that were ready for market. Now Intel seems to be doing the same thing. Apparently Intel thinks their customers are not too bright, since the newly "released" "Coppermine" CPUs are still hype-ware, rather than hardware. They are nowhere to be found.

Is it a coincidence that Intel announced these unavailable parts just as AMD is about to release their Athlon 750MHz CPU? Most AMD Athlon parts are quite plentiful, but Intel has been using it's market muscle to make sure that motherboard makers don't produce Athlon motherboards. Asus recently admitted it would stop making Athlon Slot A motherboards after receiving pressure from Intel. They then retracted that statement, saying they will continue to support the Athlon CPU line, but we could not find a single mention of any Slot A motherboard on their web site today. An Asus official who did not want to be named was quoted as saying "Nobody can talk about the K7. It's a very sensitive topic, we don't want any employee to release any K7-related information to anybody." Sensitive topic? That's a polite way of saying "psst... there's a guy from Intel in the next room with a gun pointed at my head!"
ASUS MOBOs

Fortunately, a petition to motherboard makers that has been floating the web has received quite a bit of support, and may put pressure on Mobo makers to stop caving in to Intel's pressure. PETITION

Coppermine CPUs are supposed to start trickling in to distributors this week and next. But they are on so-called "allocation", which is reseller jargon for "very limited supply". Expect shortages of them until sometime in January. We also expect that the hype-ware release of the new Coppermine CPUs will have the desired effect of getting many customers to wait for these currently non-existent processors.

We will be announcing new systems based upon the AMD Athlon processor, and Intel's soon-to-be Coppermine processors, in the next few weeks.

Comdex, the computer trade show, is just a week away, and it looks like it's going to be a doozy. AMD will be showing off Athlons ranging in speed from 800MHz to 1 GHz. Who would have thought a year ago that AMD would beat Intel to the Gigahertz realm?

Abit wiil be releasing new motherboards, presumably based upon Intel's chipsets (i820 and i840). Abit also announced it would be entering the graphics card market! This is great news in our opinion, because with all the mergers among hardware makers and chip makers, there isn't much competition out there. We hope that Abit will make high-end video cards like GeForce cards with extra overclocking features. I'm psyched!

Expect lots of new input devices, and new generations of video cards, such as the Voodoo4 by 3dfx. We are also hoping that motherboard makers diversify, and offer new motherboards based upon VIA's Apollo Pro 133 chipset, and VIA's new KX133 Athlon chipset.

As Intel franticly cobbles together an i815 chipset with PC-133 support, and AGP 4x support, it seems like too little too late. No doubt, they will be able to use their market muscle to threaten Mobo makers into using it, instead of the competition's chipsets, which are already available.

But Rambus memory is getting trashed by all the reviews, as an expensive, non-scalable, less stable alternative to existing SDRAM. It clearly is not a technological breakthrough, nor is it a viable alternative due to it's high price. PC prices are still going down, and any attempt to make a key component much more expensive is doomed to failure.

The worst part about the whole thing is that Intel is going to have to recoup their HUGE losses, and you all know whom is going to get stuck with that bill! Is is a coincidence that Intel just decided to delay their next price drop? NO!

November 5th

Microsoft got plastered in court today. But the loss of the antitrust case brought by the Justice Department against the maker of Windows was not nearly as interesting as the sordid details we got to hear after the decision was handed down by Judge Thomas Penfield Jackson.

We had heard rumors for years that Microsoft "did things" to their software to make it incompatible with the software produced by competitors. While many folks had more or less dismissed these allegations as unfounded rumors, it turns out that many allegations were indeed true, and then some! As one example, Microsoft programmers intentionally put phony error messages into Windows 3.1 installation, which would come up if the competition's version of DOS was detected, rather than Microsoft's own version. Hmmm. It's a real reassurance to know what the programmers at Microsoft are concentrating on.

This immediately brings into question some of Windows legendary instability. How much is bad programming, and how much is intentional sabotage of the competition's software packages? This is not just an idle question. Microsoft has obviously spent too much time worrying about how to get and maintain market share, and too little time thinking about how to make a rock solid and secure operating system. Microsoft got themselves into this situation by a fairly callous disregard for fair play. If they really felt their software was superior, they would not have been compelled to insert false error messages into Window's installation program.

This is not the end of the case, it's just the beginning of a long, and probably futile, appeals process. Based upon the information released today, Microsoft hasn't got a leg to stand on. But the incidents outlined in the legal actions against Microsoft are several years old. I want to know what the friendly folks at Microsoft are doing right now to Windows 2000 to make sure it doesn't work right with Netscape, or with Novell or with any other non-Microsoft product.

November 3rd

The price of Intel's CPUs, and the price of SDRAM have gone down in the last week, which is very good news. Intel released a bunch of new processors, and dropped prices on existing Pentium IIIs. SDRAM prices have fallen significantly, but we expect them to continue to drop. The rumors about memory chip shortages are still floating around, but we think they are just rumors.

Expect "E" version Coppermine Pentium IIIs to be slow to market. They are just coming out of warehouses, and supplies are very limited. Don't be fooled by the "B" version Pentium IIIs. They are not based upon the Coppermine core, even though they have a 133MHz front side bus. All "Coppermine" core CPUs have 256Kb of L2 cache, so if a Pentium III has 512Kb, it's based on the older, 0.25 micron core, not the new 0.18 micron core.

November 1st

Just for fun, we went over to Rambus.com to see what they had to say about "Caminogate", as the i820 disaster has come to be known. Rather than being down in the dumps, they seem to be having a jolly good time over there!

Not ones to be put off by a tumble in their stock price, nor the fact that memory manufacturers are jumping ship faster than rats on the Titanic, Rambus has shown it can have some fun in the face of impending disaster. Here's a sample of their humor.

Isn't that a gas? Platform stability! Ask Intel's engineers about Direct Rambus DRAMs legendary stability. And what is the "fun" they refer to? It's got to be that you'll pay 5 times as much money for Rambus as you would pay for PC-133 SDRAM. That's pretty funny.

"Rambus memory modules provide three times the memory performance as PC100 SDRAM-based modules."

That's fascinating, since it has been well documented on the web that Rambus offers only a few percent performance improvement over PC-100 SDRAM, especially when the PC-100 SDRAM is overclocked.

A quick look at their Quarterly statement revealed they only made 12 million last quarter, suggesting that Rambus DRAM is selling significantly worse than hot cakes.

"Rambus technology is an open standard, accessible to all semiconductor companies. Rambus provides licensees a full range of design, documentation and system-engineering services. In exchange, IC companies pay an up-front license fee and royalties"

An open standard? Oh sure, you just have to pay license fees and royalties! I liked the old kind of "open standard" better, you know, the ones without the license fees and royalties ;).

Alas, we had to bid Rambus.com a fond farewell, without finding so much as a mention of the delay in the i820 chipset from Intel. Why dwell on such minor inconveniences when you're having fun?

As the prognosis for the i820 chipset and Direct Rambus DRAM (DR DRAM) worsens, it is interesting to take a stroll down memory lane (is there a pun in there?) and look at how Intel and motherboard makers got themselves into such a fix in the first place. Check out this lengthy article on the Intel/Rambus saga here: Article (and excuse the imperfect grammar, they are Russian, yet their English is better than some Americans!).

We can only hope that the end result of this whole fiasco is that we get a good selection of PC-133 SDRAM-based, as well as Double-Data-Rate DRAM-based motherboards in the near future. We have no intention of paying Intel and Rambus 4 times as much money as Rambus is worth to put it in our new systems. If they want to drop the price to compete with other types of memory, DR DRAM has a slim chance of survival. If they keep up their arrogant, greedy ways, we expect DR DRAM to go extinct faster than the dinosaurs!

October 31st

Intel's Reaction to Mobo Makers using VIA's PC-133 chipset? Sue the Hell Out of 'Em!

Having had little success in getting their own PC-133 chipset to market, Intel has turned its attention to competitor VIA's PC133 chipset, and motherboard manufacturers that dare to use it. Previously Intel filed suit against VIA for patent infringement in the form of their new Apollo Pro 133 chipset. Now Intel has the audacity to file suit against motherboard makers that integrate the Apollo Pro 133 onto their motherboards.

FIC, known for making some of the first Athlon motherboards, is named in Intel's lawsuit, along with others.

More rumors are circulating that Intel has been directly and indirectly intimidating motherboard makers into supporting the i820 chipset, and into shunning the Apollo Pro 133. The direct method is to threaten litigation (or actually file suit), and the indirect method is to remind the motherboard makers where they get their BX chipsets from. Continued sham-shortages of the BX chipset, a proprietary Intel asset, will help make sure at least some motherboard makers continue to tow the Intel line.

All of this has hurt computer makers, and they are getting a little tired of Intel's bullying. The end result is more and more motherboard makers tossing caution to the wind, and getting on the Apollo Pro 133 bandwagon. Even our own personal favorite, Abit, appears to be working on a VIA-based motherboard. We can only feel a sense of Karma coming back to haunt Intel on this Halloween, as their strong-arm tactics begin to fall flat, and as VIA and AMD continue to keep the pressure on.

October 30th

Reports have circulated on the web that people are experiencing significant drops in frame rates in Direct 3D games after upgrading to Microsoft's Direct X 7. We have been testing Direct X 7 versus Direct X 6.1, and so far have found no reduction in frame rates. We finished testing a Voodoo3 3000 card with Direct X 6.1 and Direct X 7, and found benchmarks to be virtually identical. We will be testing TNT-2 cards next to see if there is a problem specific to them. We will post a full article with the benchmark results soon.

October 29th

Halloween marks the two year anniversary for KickAss Gear. In November we will be starting our third year of operation here in the Washington DC area. Thanks everyone, we look forward to many more years of providing you with high-end gaming systems and upgrades!

There has been a lot of talk on the web about Direct X 7 causing between 20% and 40% reduction in frame rates in Direct 3D games, after upgrading from version 6.1. We have noticed some slow down, and will be running some tests vs. Direct X 6.1 to see what the changes are. We will also test various ways to try to remove Direct X 7.0, and let you know if any of them work safely.

For Voodoo3 owners, 3dfx has released both Direct X 6.1 AND Direct X 7.0 compatible Voodoo3 2000 and 3000 drivers! This suggests that there are enough differences between these API's to require distinct drivers. We will also check out if these new drivers bring the frame rates back up to where they should be under Direct X 7. We will let you know what we find out. Get the new Voodoo3 drivers here:

3COM announced it will introduce a 56Kbps V.90 modem with tweaked firmware to reduce ping times for online gaming. 3COM will release the "Internet Gaming Modem" in late October, with an estimated street price of around $119. Ping times were 43% better with the Internet Gaming modem than with competitor's modems. So if ADSL or two-way cable is not in your foreseeable future, this may be the way to go!

The Pentium III now comes in a number of new flavors. The parts are listed below. Those with a 133MHz bus are "B" chips, those with the Coppermine core have an "E" designation. In some cases, like with the 533, 667 and 733 CPU's, Intel left off the letter designation since these Coppermine CPU's don't have a non-Coppermine counterpart. All CPU's with 256Kb's of L2 cache are Coppermine chips. This is the easiest way to tell if a CPU has the improved Coppermine core. The L2 cache speed is twice as fast in Coppermine CPUs than in Katmai chips.

	FSB	L2 Cache Size	L2 Cache Frequency
Pentium III 733	133	256	733
Pentium III 667	133	256	667
Pentium III 650	100	256	650
Pentium III 600EB	133	256	600
Pentium III 600E	100	256	600
Pentium III 600B	133	512	300
Pentium III 550E	100	256	550
Pentium III 550	100	512	275
Pentium III 533	133	512	267
Pentium III 500E	100	256	500
Pentium III 500	100	512	250
Pentium III 450	100	512	225

Names in RED are based on the 0.18-micron Coppermine core
Names in black denote CPUs with the old Katmai core
E - Denotes 0.18-micron Coppermine core
B - Denotes 133MHz FSB CPU
modified from AnandTech.com

The new 733 PIII, running on an i820 motherboard with 800 MHz Rambus DRAM equals or beats a similar Athlon in many benchmarks. The Athlon still wins in FPU intensive programs, due to it's better Floating Point math unit. But now we can say for certain that Intel and AMD are just about on par in a MHz for MHz slugout, and Intel's CPU's are overclockable, while the Athlon still needs major modification to overclock it.

As always, the lower speed Intel CPU's will overclock more than their higher speed counterparts. This is because the higher speed CPUs are already running closer to their maximum speed. Listen to the Daily Rumor for more.

The GeForce-based "Annihilator" video cards from Creative Labs have begun to ship, but since there are very large numbers of backorders, don't expect to see them available at stores anytime real soon. The Annihilator Pro, with faster memory, will not ship for at least a month.

The GeForce 256 does not offer significantly higher frame rates, but does give you more polygons per second if the game you are playing can take advantage of the hardware. Don't expect to see lots of games with this ability for months to come. The smart money is on waiting for the Pro version, and for prices to drop.

Memory prices have begun their long, inevitable drop. Prices are down significantly since the high point, about 3 to 4 weeks ago. Word from our distributor is that there is still a shortage of chips, but that it's not as bad as it was. Expect prices to continue to drop for the next few months, unless something goes wrong.

In a related story, it was reported in the Register today that technology analysts still say Rambus DRAM will become the predominant memory technology sometime next year. They must know something we don't. Not only is Rambus much more expensive, and less stable than PC-133 SDRAM, it's not significantly faster in any of it's current forms.

PC-133 SDRAM could be cheap enough soon to be able to put 1GB in your machine, whereas for the same money, you could have 128MB's of 800MHz Rambus DRAM, which worked a little bit faster. Considering the fact that for gamers, how fast your CPU and video card are, has much more effect on frame rates than the speed of your memory , Rambus is not destined to take the gaming computer market by storm.

October 25th

More details are emerging on the design problems with the i820 chipset, and they make Intel seem even more negligent in their pre-release testing. Apparently, the chipset itself is working properly, but the way it is wired to the other components on the motherboard is flawed. This means that the problem is on the motherboards made by Intel, but some third party motherboards are apparently designed differently, and may work properly.

Yet the root of the problem seems to boil down to resonance effects between traces on the motherboards. Resonance in electronic circuits is similar to what you get with resonant sounds with musical instruments. If you strike one string on a guitar, you will set up resonant vibrations in the other strings. Similarly, the pulses that travel through the circuit lines in the motherboard set up resonant currents in neighboring lines, which make their signals unreliable. To us, this sounds like an inherent problem with Rambus, since it is the high frequency range that it operates in that is causing all these resonance effects.

This resonance problem seems inherent to the three-Rambus-socket designed boards, but may even show up under some extreme circumstances in two Rambus socket boards, including the i840 with two pairs of interleaved Rambus sockets. Back to the drawing board!!

This engineering goof by Intel was found so late, that we have noticed that a number of computer companies actually have advertisements for new systems with the i820 in the October issues of several PC magazines! That was a bit optimistic, considering those companies could not have run all the reliability testing that should precede the release of new products. We recommend not buying a computer from a company that announces products before they have tested them.

This whole thing hurts the relationship between Intel and third-party motherboard makers, because the push toward Rambus has been Intel's pet project, and has not been popular with the other manufacturers from the beginning. If the motherboards already produced do not ever make it to market, the companies that made them on Intel's assurances will not only loose lots of money, but they will also loose confidence in Intel's recommendations.

It is clear to us that PC-100 and PC-133 SDRAM will be the primary memory types in most Desktops in early 2000, with high-end computers packing Double Data Rate DRAM, rather than Rambus DRAM. If Rambus offered significantly better performance, it would slowly catch on, but it does not, despite the fact that it costs over 5 times as much as PC-133 SDRAM. Double Data Rate DRAM will almost certainly outperform Rambus DRAM, at a much better price.

The funny thing is, we have no idea how many i820 motherboards are already produced, and how much Rambus DRAM is sitting on shelves somewhere, waiting on Intel's "validation" process. We sure can't find any for sale! Based upon our brief communications with Abit, they are spending most of their time right now trying to get i820 motherboards ready for market, whenever that may be.

October 24th

Monday the 25th is the day folks. We should be seeing the new Coppermine (0.18 micron) Pentium III CPU's, and maybe, just maybe, some form of i820 motherboards. Not that we are excited about any of this, but it's a big deal for Intel.

Will they release the Cape Cod i820 motherboard with SDRAM support (and no Direct Rambus support)? Well, the big problem with these mobos is that they are marred by a memory architecture problem. They were designed for Rambus DRAM, and can not directly work with SDRAM. They need a Memory Translator Hub, which converts Rambus signals to SDRAM signals, slowing the system down. This is not a good option for power users.

Will they release Vancouver motherboards with only two usable Rambus sockets? Will Direct Rambus DRAM go on sale at memory distributors? It's kind of doubtful, but you never know with Intel.

The scuttlebutt surrounding this shaky debut is that neither i820 chipset motherboards, nor VIA Apollo Pro 133 chipset motherboards, can beat overclocked BX motherboards by any more than a few percent in most benchmarks. And AGP 4x is looking more and more unimportant as faster graphics cards come out with 64MB's of onboard RAM. For the full scoop, check out today's Daily Rumor.

October 22nd

Graphics chipset maker Real3D has died, and Intel is feeding on the corpse. Real3D was the company that gave us the under-whelming Intel i740 graphics chip, and the "Starfighter" graphics card, so it's no wonder they didn't do so well. Intel has purchased their patents, and hired some of the staff. Let's hope that they can do a better job on their next graphics chip for Intel.

Well, it's not too surprising, but it looks like companies are actually going to start selling Camino-chipset-based motherboards (including Intel's "Cape Cod", and "Vancouver" models). As we had mentioned earlier, warehouses were actually stocked full of the defective things for over a month now, and everyone is itching to unload them ASAP. If Intel rolls it's collective eyes toward the heavens, and gives the go-ahead, expect them to hit the market shortly.

These will be unmodified boards, and therefore, the ones with Direct Rambus sockets will only be able to accommodate two memory modules, while the third socket will only accept a "terminator".

But the i820 comes in two, equally unappealing flavors. The ones with the third, defective Rambus socket, and those with support only for PC-100 SDRAM (no PC-133 SDRAM support to go along with the 133MHz front side bus and new 133MHz Pentium III CPU's). So one flavor is defective, and the other is intentionally underpowered to tempt you to buy the defective one.

Maybe it's a good thing that the third "RIMM" socket for Rambus DRAM does not work. If it did, the poor souls that ante up for a Vancouver mobo might be tempted to fill up all the sockets, which for 128MB Direct Rambus DRAM modules would set our poor customer back in the neighborhood of $2400!!! And the memory would not be significantly faster than PC-133 SDRAM! Maybe Intel is just trying to protect us from our unbridled buying habits. :)

October 21st

AMD has been talking quite a bit recently about it's new, state-of-the-art 0.18 micron fabrication plant in Dresden Germany (Fab 30). This facility will not only produce Athlons using the 0.18 micron etching process, but will also be using the so-called "copper interconnect" technology that was pioneered by IBM. This first introduction of the initial stages of the changeover to copper-based CPU technology in a production CPU has already allowed AMD to boost the speed of the Athlon to 900MHz.

AMD says that they will have a working 1 GHz (1000 MHz) system up and running soon! Expect the newer, faster "copper" Athlons to be available early next year. Hey Intel, are you paying attention???

Intel is showing signs of being such a giant, unwieldy organization that the right hand has no idea that the left hand is about to chop it off. The problems with the i820 chipset are not present in the i840 chipset. Why? Because two different groups of engineers made the two chipsets! And there was so little communication between groups that one of them practically boned the entire project, while the other seems to have done a good job.

Well, the signs of corporate bloat are showing up in Intel's processor line as well. They are about to confuse the CPU situation so completely, that customers will have a hard time figuring out what CPU they just got in their new system. There will be 100 MHz and 133 MHz front side bus CPU's, there will be CPU's with 256Kb of on-die cache, and others with 512Kb of cache, and there will be "Coppermine" 0.18 micron chips selling along side older 0.25 micron chips, with little to tip off the customer which is which.

We posted the "codes" for these different CPU's a few weeks back, but it probably merits reiteration.

=> The Processors marked "E" will refer to Coppermine core CPU's, while those with an "EB" designation will be 133MHz Coppermine chips. Chips without such letter designations will be standard 100MHz 0.25 micron Pentium III's. We expect Intel will make it clear how much cache is on each CPU. So a designation like 600EB/512 would mean that it is a 600MHz Coppermine CPU with a front side bus of 133MHz and a 512Kb L2 cache, and if you do the division, the multiplier factor is set at 4.5X.

October 20th

Abit is working on several new motherboards. The interesting part is that they are putting out new BX motherboards, despite the chipset's advanced age. These may be a result of the problems with the i820 chipset, but they may also have been in the works from several months back. The BF6 and BE6-2 will be available in November.

Abit is also working on a VIA chipset motherboard, based upon the VIA KX133 chipset, to be called the VA6. This will probably not be available for a few months, but will offer advanced features not available with BX-based boards.

Abit is also one of the motherboard manufacturers that worked closely with Intel to get i820 chipset motherboards to market earlier this month. As such, they are still working on a way to ship the product that has already been produced, despite the memory slot problems. The interesting thing is that they are not currently working on an i840 (Carmel) based motherboard. This may change, since the i840 appears to be much nearer it's debut than the ill-fated i820. But again, the chipset is partly hobbled by the lack of support for PC-133 and double data rate DRAM.

Intel's so-called "Coppermine" Processors are coming out now, made with the 0.18 micron etching process, and based upon the original P6 core with added SSE instructions. They are rated to run at 133MHz on the front side bus, but Intel did not adopt the PC-133 memory standard, so there is no support for the faster SDRAM on motherboards produced by Intel. (They want you to buy overpriced Direct Rambus DRAM). Yet other motherboard manufacturers make mainboards with overclock settings for the front side bus, which will allow end users to run the new Coppermine chips at 133MHz or higher with standard PC-133 SDRAM. These Intel systems will not outperform Athlon-based systems.

Intel is also pushing the release date for their next generation 32 bit processor, code named Willamette, up to January or February 2000. Despite the fact that it is still a 32 bit processor, this one should be substantially faster than the Coppermine-based CPU's. Willamette seems on track for it's accelerated release in early 2000. If this CPU can be coupled with Double Data Rate DRAM, it has a chance of beating AMD Athlon systems.

Still looking hazy, and just beyond the horizon, is the IA-64 processor, code-named "Merced", now called "Itanium", which seems to be going the way of the i820 chipset. Several delays have put this CPU into the "will it ever be ready?" category. This is Intel's first foray into 64 bit architechture, and represents a big step forward. Aimed at the server market, this chip still could provide an awesome gaming platform if the price is right, and if they will run an Operating System that you can play games on (Win 2K Pro?).

October 19th

To go along with their new Coppermine based CPU's, Intel is getting ready to release i840-based motherboards with AGP Pro 50, ATA/66, and Rambus support. The first motherboard from Intel based upon this new chipset will be called the OutRigger 840 (OR840). These motherboards will have 4 Rambus sockets, for a total memory capacity 2GB of Rambus DRAM. Based upon the current cost of Rambus, only Bill Gates will be able to afford one with 2GB's of Direct Rambus DRAM.

This is at the heart of the problem with Intel as a company, and their new product, the i840 chipset. Intel is banking (litterally) on the hopes that corporations and power-users with money to burn, will go out and buy i840/Coppermine/Rambus systems, despite the high cost and marginal performance increase.

AGP Pro 50 runs at data transfer rates of 266MHz, and Intel claims it will achieve data throughput rates of 1Gbps. AGP Pro 50 sockets are backwards compatible to AGP 4x. The 50 refers to 50 watts maximum power consumption, and it has the same data transfer rate as AGP4x.

The main drawback to the i840 is it does not support PC-133 SDRAM. That means you will have to be wealthy to get one of these rigs populated with Rambus DRAM. We know we can't afford one! The big question is whether Abit and other board makers will ignore Intel's specs, and add overclocking goodness to their i840 boards. If we can get an i840 board from Abit with Softmenu III technology, we should be sitting pretty. We just don't know yet if it can be done.

Rumors are spreading that Intel plans on winning the chips wars. Word is that they are preparing to release their new Willamette-core CPU's as much as nine months ahead of schedule. These are next-generation chips, and the first Intel has produced that are not based upon the original P6 die architecture that debuted with the Pentium Pro. The Willamette will be Intel's first truly new 32 bit processor in 6 years.

The thinking behind the accelerated release of the Willamette is that Intel knows the Coppermine core (basically, the Pentium Pro core with SSE in a 0.18 micron format) will not beat the Athlon. See what competition does for us folks? We would have had to wait a year for the Willamette if it were not for AMD's Athlon. But what this also means is that folks who go out and buy Coppermine-based systems now will be kicking themselves in January, because the newer faster systems will be on the way.

The shortage of good-quality motherboards for the new Athlon processor from AMD is causing problems for system integrators and customers alike. Not only are Athlon (Slot A) motherboards hard to find, but the ones available are giving folks trouble. We expect this to change as more and more motherboard manufacturers turn out Athlon boards, but for the time being, Athlon systems will be hard to get. You can go here to sign a petition to motherboard makers to get on the AMD bandwagon.

October 17th

After a number of embarrassing goofs, Intel is getting ready to roll out the big guns on October 25th: the i840 (Carmel) chipset, and new "Coppermine" CPU's based upon the new (code-named "Cascades") Pentium III core. According to Intel insiders, who could never be biased on this subject, the higher-end Cascades CPU's running with Direct Rambus DRAM on an i840-based motherboard will beat the highest speed Athlon processor. This we have got to see to believe!

The i840 chipset was apparently designed and tested by a different group of Intel engineers than the folks working on the embarrassing i820 Camino chipset. Let's hope the i840 guys were at least smart enough to fill up all the Rambus sockets during testing!!

Intel is being cautiously tight-lipped about details, but we may actually see Cascades CPU's on i840 motherboards sometime in November. Intel may also release "Coppermine" core Celerons on November 7th as well (with a 100MHz front side bus??), but this is not a certain date.

The first round of Cascades-bases CPU's will be Xeon's (which translates to "way overpriced" in Intel jargon). These CPU's are based upon the new 0.18 micron fabrication process, and fit the new Slot2 socket on i840 motherboards. Pentium III and Xeon chips based upon the Coppermine core will have 256Kb of on-die cache, with an additional 512MB to 1MB of L2 cache.

Intel is targeting the Server market primarily, but also expects high-end gamers to go for the powerful new chips and motherboards.

So if you are in the market for a new computer, you might want to wait and see what Intel has up it's sleeve for November. As with all new technology, the first product may have unforeseen bugs, will be very hard to find initially, and will be priced very high.

October 16th

Not waiting anymore for the Intel i820 chipset, Abit is continuing to push the capabilities of the BX chipset beyond it's specs. Abit will be releasing the BE6-2 and BF6 mainboards in the month of November, with ATA/66 support, and the new Softmenu III BIOS setup utility. Softmenu III offers overclocking bus speed options in 1 MHz increments from 66MHz to 200MHz! This is what overclockers have been requesting for years, because it will allow the end user to overclock their systems to the absolute maximum speed they are capable of. You'll be able to squeeze that last couple of MHz out of your rig. The major difference between the BE6-2 and the BF6 is that the BF6 will have 6 PCI slots, and one ISA slot.

Abit has also said they are working on a VIA chipset motherboard, called the VA6 (you heard it here first folks!), which should be quite awesome. We have no word on the exact specs yet, but expect it to have official PC-133 SDRAM support, AGP 4X, and built-in ATA/66 controllers, all with Softmenu III technology. They probably will not be ready until early next year.

December 29th