	Ground ZERO
	Latest Rumor
	Rumor Archive


	mesg. board

2015 © KickAss Gear

KickAss Gear News Archive: March 2000

March 31st 2000

Problems with VIA KX133 chipset?

The motherboard manufacturer Epox has issued a warning that using Win95a with the Epox 7KXA motherboard, sporting the new VIA KX133 chipset, causes the BIOS to become corrupted. The problem appears to be in the new chipset, rather than with the Athlon CPU. The obvious fix is to install a newer operating system. Win95a has many deficiencies, and as such should be updated anyway. This is just another reason to keep your OS and drivers up to date.

Stock Market Reality Check.

As the Nasdaq continues it's slow slip downward, it is dragging overpriced stocks like Rambus down with it. It's still hard to say what will happen with Rambus in the long run. A quick check of prices yesterday indicated that Rambus memory is coming down in price significantly. Dane Elec. quoted us a price of well under $600 for a 128MB stick of 800MHz RDRAM. That's almost half price compared with a few months ago, but still 5 times the price of PC-133 SDRAM. If Rambus can eliminate the low yield problems, the instability with more than two RIMMs, and they reduce their licensing fees substantially, and they drop the price to near that of PC-133 SDRAM and upcoming Double Data Rate DRAM, then they still have a fighting chance.

Intel was also down for the day, while AMD was up a few dollars a share.

Win2K Limits IP Addresses on Domain Controllers.

MSNBC has posted a story that says Win2K dies a horrible death by way of the new and forbidding "Active Directory" in the operating system when more than 51 IP addresses are installed on a single server. While this is not an issue for the average gamer, it's definitely an issue for web hosting companies and the like. But deeper than that has got to be the lurking suspicion that this is one of those 63,000 bugs that MS flatly denies exist in Win2K. It's a confidence shaker. Microsoft, in it's own defense, points out the following:

"... the issue only arises if the machine with multiple IP addresses is also a domain controller. Typically, any customer who deploys machines with lots of IP addresses would deploy domain controllers on separate machines."

This might lead one to wonder how many other Win2K bugs may have been known to MS at release, but were considered unlikely to be encountered by many customers, and therefore were relegated to "fix-it-at-SP1" status.

March 29th 2000

New Coppermine Celeron II "Released"

Intel has announced the debut of the new 0.18 micron, Coppermine-core Celerons with SIMD extensions. They should start to show up at distributors soon, but so far they are not even listed at Tech Data and other Intel vendors. Let's hope this is not another Intel "hype-ware" release, like when they "released" the Coppermine PIII last Fall. Coppermine-based PIIIs are just starting to arrive at distributors in significant numbers now, almost 5 months after Intel announced their supposed availability.

The most interesting marketing decision Intel made was to go with the 66MHz Celeron bus speed. This is far below what the Coppermine core is capable of, and brings up the question of bus overclockability. You can check out an early review of the 600MHz Celeron II at Hardware Central.

The bottom line is predictable, and shows that Intel has finally gotten the Overclocking thing figured out. In the past they have released Celerons (e.g., the 300A) that would easily overclock higher than, and outperform the existing high end processor that they produced. This cut into profits, and Intel does not want to make that mistake again. As such, the 66MHz bus on the Celeron 2 is a simple fix to Intel's problem. In order to get the processor running at 566MHz with a 66MHz front side bus, they needed to lock the CPU multiplier at 8.5X. Since most Coppermine-core processors crap-out between 750 and 825MHz (with normal cooling), it's clear that a 100MHz FSB, even with the 566MHz Celeron-2, is probably just outside the processors capabilities. This is not to say that you wont' be able to go to 100MHz on the bus frequency with core voltage adjustments, and extra cooling on the new Celery-2. It may be possible with some of them, but certainly not all. They will all run at 75MHz, and many at 83 MHz on the front side bus, but 100MHz will be an elusive prize.

Note: Remember, you need a Coppermine-type motherboard to run a Celery-2. Or, alternatively, you at least need a Flip-Chip adapter to run them on BX motherboards.

March 27th 2000

Rambus Tries to Ban Sega Dreamcast Imports (and Hitachi Counter-Sues Rambus)

In their ongoing battle to take over the memory industry, Rambus is going litigation-crazy. Their product doesn't sell very well by itself, but, (their marketing dept. may be reasoning) if they can ban imports of all other memory products, , then their lack-luster RDRAM might still become the new memory standard. So they are not only suing Hitachi on 8 counts of patent infringement for Hitachi's SDRAM, DDR and microprocessor products, but are also seeking to have the Sega Dreamcast gaming console banned from imports into the US, also for reasons of alleged patent infringement in their memory timing scheme. Since Hitachi does not use significantly different memory architecture than other SDRAM makers, it's as though Rambus thinks that most or all current SDRAM products infringe on their RDRAM patents.

Hitachi is not taking it lying down. It was reported in Electronic Buyers News and The Register today that Hitachi has counter-filed a motion with the International Trade Commission alleging Rambus is violating the Sherman Anti-trust Act. Hitachi's statement says it all.

"Rambus has used its lawsuits to assert that [synchronous] products compatible with the JEDEC standards infringe Rambus patents, making Rambus technology the dominant, if not sole standard in the industry."

Sounds about right to me. You can read about the original Rambus claims here. The argument boils down to the fact that Rambus gave SDRAM makers all the technical details of Rambus technology so that they could produce Rambus RIMMs. Instead, most memory makers decided that Rambus DRAM was too expensive, and not any faster than current SDRAM technology. According to Rambus, companies such as Hitachi then implemented memory timing schemes in SDRAM that Rambus Inc. had patented. If true, Rambus may have some merit to their case. But if Hitachi's statement that it uses JEDEC-adopted standards is true, then there is no case.

Intel's New Processor Packaging Formats.

We don't know much about it yet, but Intel is going to be using NASA-developed heat dissipating substrates for it's upcoming Flip-Chip products, including the Willamette (Pentium IV). It was reported in Electronic Buyers News that Intel will be using a new Flip-Chip Pin-Ball-Grid Array format for the Willamette and Foster microprocessors. These will be made with new, less expensive epoxy materials, will have NASA-developed heat dissipation materials, will have more substrate layers, and will use a new technology to form the interconnect lines in the substrate. All of these changes are intended to permit higher processor speeds while reducing production costs. The new processors will require new motherboards and chipsets.

If you are interested in much more detail on these new processor formats, there is still time to register for an Oxford University course entitled "Low-Cost Flip Chip Technologies for BGA, CSP and DCA Assemblies."

Intel's Flip-Chips Don't Do Multiprocessing?

The question is surfacing on the web. It was reported in the The Register today that Intel's Coppermine Flip-Chips do not support symmetric multiprocessing (SMP). According to Intel, this capability has been disabled in Flip-Chips! It has not been disabled in the SECC2 (Slot-1) versions. This sounds a heck of a lot like the rumors that Celerons would not work in SMP systems, but apparently, it's true. Don't dispair, newer Flip-Chips may not suffer from this affliction. Over at BP6.com, Tim Brinkley talks about how the new Flip-Chip stepping (cB0) has been enabled for SMP.

Don't Buy that Aureal-Based Sound Card Just Yet.

Apparently, Aureal is going out of business. The makers of the popular Vortex-2 sound chip, with "A3D" 3D sound is in serious financial trouble, and may wind down operations. The company makes their own cards, and supplies Vortex-2 chips to other sound card makers for their audio boards, including the Diamond MX line of sound cards. There was no mention of the problems on the Aureal web site, but at least one fan-based site, "Aureal 3D Central" has closed for good. If true, there will be virtually no competition left for overpriced Creative Live sound cards.

March 25th 2000

KickAss Gear is now running on a dedicated server, so things should be a little speedier. Thanks to Lisa and Steve for all the good work. There may be a few services that are down temporarily, but we should have everything running again shortly.

New Athlon Site: Interested in overclocking your Athlon, and want to know what kind of success others have had so far? Check out Kevin Morgan's redesigned web site in the UK, called K7 Core.

If you haven't heard already, the Unreal Tournament Bonus-Pack has been released! Get it here. It includes new maps, and new skins, and it's FREE! Look Ma, I'm a Nali War-Cow!

AMD Processor Lineup.

There's lots of buzz on the net about AMD's new lineup of CPUs for the year, and it looks pretty exciting. All the code names can get a little confusing, so I'll try to be concise.

Low-end (ETA, Q2-Q3): "Spitfire" is the code-name for AMD's new Celeron-like offering. It will probably not carry the Athlon name, and it will be based on aluminum technology fabricated at 0.18 microns. It will come in AMD's new Socket-A format to reduce costs. This is made possible because the 128Kbs of cache is on-die, like with Celerons and Coppermines.

Mid-range (ETA Q3-Q4): "Thunderbird" is the code name for the new Athlons with 256Kb (perhaps 512Kb)of on-die L2 cache. They may be marketed under the brand name Athlon-Pro. They will be produced with aluminum at 0.18 microns. They will also come in the new socket format (Socket-A), and will require a new motherboard chipset called the AMD 760. This will support 200MHz Double Data Rate DRAM (I can see Intel engineers cringing in the corner) and AGP 4x. It is not clear if the new chipset will have better ATA-66 support than the AMD 750 chipset.

High-end: (ETA: Q1 2001): "Mustang" is the code name for the upcoming server platform from AMD. It will be based on Copper-interconnect technology, and will have up to 2MB of on-die L2 cache (note: yields will be lower, and costs substantially higher). This processor line will still use x86-32 bit code, and like the Thunderbird line, will require a new motherboard chipset, the AMD 770. This chipset will support 266MHz Double Data Rate DRAM, giving it a substantial edge over Rambus-based systems from Intel.

Outer Limits: (ETA???): "Sledgehammer" is AMD's code name for their foray into 64 bit territory. But unlike Intel's (faltering) Itanium platform, the AMD 64 bit code is backward compatible with IA-32 bit instructions, giving it a massive advantage relative to Intel's upcoming 64 bit CPUs.

It seems as if AMD has a more exciting lineup of new chips coming out, which is ahead of Intel's plans in several respects (Copper interconnect, Double Data Rate DRAM, etc). Intel has been talking about these things, but AMD has been acting on them.

Sony Making Illegal DVD Copies. (Bring On the Lawyers!)

The latest poop on the new Sony Playstation-2 is that it doesn't just let you play DVD's region free, but it also will let you make video tape copies of those DVDs! In other words, both the Region Code and Macromedia DVD encryption can by bypassed by pushing the right buttons on the game pad. Very cool, but it's the kind of thing that keeps corporate lawyers busy for years.

This seems to be a new trend that can make regular old DVD players into overnight success stories. If you are interested in buying and playing non-Region-1 DVD's, and making VHS dubs, then you'd better hurry because the DVD Gestapo are busy snuffing out the brush fires. But for at least a little while longer, you should be able able to go to your local electronics store and buy yourself an Apex 600A DVD player, which has a hidden remote control menu that lets you disable the Region-code check, as well as the Macromedia copy-protection. You can check out the details here.

March 24th 2000

Coppermine Celerons Debut Next Week.

Intel is ready to release their "Athlon-Killer" Celeron chips next week. The only way these chips will cut into AMD sales will be if they sell for under $150, and run at 600MHz. What they have going for them are SIMD instructions, and full-speed on-die L2 cache. But like older Celerons, they will be limited to 128KB of said cache. The other thing they will have going for them is bus frequency overclocking. Chances are these chips will be awesome overclockers (less cache means less problems when overclocked).

Currently, there is a shortage of motherboards that will run Coppermine CPU's, and even fewer that allow core-voltage adjustments which are necessary for high-speed (150MHz) overclocking. Add to this Intel's problems in producing enough chips to meet demand, and you are looking at a wait of a month or two before the Super-Celeries are available in quantity, with motherboards that allow for core-voltage adjustments. Unless the price difference between Super-Celeries and Coppermine PIIIs is close to $100, I predict they will not sell very well. If they are very inexpensive, Intel might have a winner on their hands.

Rambus Gets Desperate.

It seems as though Intel has been coaching Rambus on how to deal with their competition: sue them. In a move that imitates Intel's bid to the International Trade Commission (ITC) to stop imports of VIA's new motherboard chipsets, Rambus has filed suit with the same trade body to block Hitachi from importing SDRAM! According to an article in The Register today, Rambus claims that Hitachi (why just Hitachi?) uses a patented timing scheme in their SDRAM, DDR and microprocessor products. This move, taken directly from Intel's marketing repertoire, is sure to make Rambus even more popular with memory manufacturers than they already are. Me thinks that Samsung, a major Rambus backer, might have their finger in this jurisprudent pie as well. I can't imagine that Rambus' failures in their Engineering Department can be made up for by actions from their Legal Department.

In early February, Rambus stock was trading at less than $50 a share. By mid March, it was above $450 a share. On Wednesday of this week, Rambus was down to $250 a share, but it had recovered to $350 a share on Thursday. Expect more price instability as the Spin-Meisters spin their technology tales.

Microsoft Gets Office 2000 Updates Just Right.

Microsoft has released a Service Pack for Office 2000. The buzz on the web is not good news. First off, registration of the update is mandatory. You need to contact Microsoft during the installation to get an installation code. This is as convenient as driving to pick up your pizza delivery yourself. On top of this insult to their paying customers, early reports are that there are serious new bugs in this bug-fixing update. One bug makes it so you can't use any of your Office 2000 applications after installing the update. Sweet.

March 21st 2000

Death Of the Modem?

3COM/US Robotics has announced that it is getting out of the dial-up modem business, as part of a corporate restructuring. Not only are profits meager, but it is likely that 3COM realizes that "broadband" Internet access devices will completely replace dial-up modems over the next two to three years. Slow deployment of cable and DSL service in many parts of the world will ensure that companies like Creative will continue to sell dial-up modems for some time to come.

VIA and AMD May Have Something Up Their Sleeves.

Rumors continue to circulate on the web that AMD and VIA may combine forces to combat Intel. They are already doing so in practice, even if it's not an official merger. But an official merger would give the two companies much more clout in the industry, especially since VIA does not have any distribution channels in place. As a combined company, they could coordinate their chipset and processor releases and provide help for motherboard manufacturers in designing new motherboards based on their products. This would almost certainly improve quality and reliability.

Rambus' Stock Finally Caves.

The share price for Rambus stock finally took a big hit yesterday, and is now down by almost $150 from it's high just a week ago. This is about as surprising as the sun rise to us, but it's not entirely clear why the stock went so high despite bad reports on the memory technology circulating the web for over 6 months. Investors seem to point to the Tom's Hardware report on the problems with Rambus last week as the major culprit. If true, then it's clear that many tech-stock investors are clueless about the technologies they buy into. Our guess is that profit-taking was inevitable once Rambus had gone from less than $50 to nearly $500 per share over the last year. The Nasdaq has been getting hit by profit-taking over the last week in general, and Rambus was an obvious target for investors looking for profits.

March 18th 2000

Sony Pirates DVD's By the Million (But they haven't been arrested yet)

Now that Sony has sold over one million PlayStation 2 game consoles, it has been reported at the Gaming Intelligence Network that the new units can play DVD disks from the US, even though they are supposed to only play Asian DVDs. Most DVD disks are encrypted in such a way that they will only play on DVD players made for a certain region of the Earth. So region 1 (US) DVD disks only play if the player is a region 1 player. But Sony has, to the delight of movie makers, sent out over 1 million PlayStation 2 units to Asian customers that will play DVD movies from the US. A few flicks on certain game pad buttons, and your DVDs become region free! This is the kind of thing that gets small fry throw in jail and fined huge sums of money. So far, Sony hasn't announced a recall.

This comes on the heels of the announcement that some memory cards for the PlayStation 2 have problems, and can prevent the units from playing any DVDs. Apparently in some cases, saving a game to the memory card would cause the system to stop playing the game, as well as DVD movies.

On top of this, many new home DVD players have begun surfacing in stores which have simple menu systems that allow the user to make their players region free. We have seen them at work. A simple, undocumented menu can be pulled up, and the region codes and Macromedia encryption can be turned off. All of this has got to make movie makers sweat. You can check out the details here.

March 17th 2000

Coppermine Celerons Coming! (But Intel Bungles Again)

According to a snippet in The Register today, Intel is planning on releasing "Coppermine-core" Celerons soon. This move is a response to the very aggressive pricing that AMD has instituted for it's Athlon line of processors. But as usual, Intel did not learn from it's previous mistakes. Their original release of the Celeron flopped because it had no L2 cache, and performance suffered. Now they will be releasing the new Celeron (Celeron Pro? Celeron III?) with half of it's L2 cache disabled! That's right, they will be regular Coppermine processors with 128Kbs of L2 cache turned off, and just sitting there.

The only way this will fly is if Intel sells them for under $100, but you can bet they will be almost $200 at release, with starting speeds at around 550 MHz. Since the 500E Pentium III is only a little more than $200, it is not clear why anyone would want to buy the same CPU with half of it's L2 cache disabled. Even at $150, most people would just assume pay another $60 in order to have all the L2 cache working.

Rambus gets Railed (But the Stock is Still Doing Great!)

According to a review at Tom's Hardware, Rambus DRAM performs poorly compared with PC-133 SDRAM. Rambus DRAM has higher latency, a narrower bus, runs much hotter, and cost many times more when compared with PC-133 SDRAM. I did find it amusing that Tom's ad-laden web site was advertising i820 motherboards with Rambus support right next to the very article that blasts the new memory.

Rambus Inc. was still trading above $400 per share this morning. Profit taking anyone?

March 16th 2000

Coppermines Start to Arrive (5 months after their "release").

Over the last week, Intel seems to have finally started shipping some Coppermine CPUs. Supplies are meager though. Everyone was surprised two weeks back when Slot-1 versions of the 550E Pentium III started showing up at distributors (they were supposed to be released in Flip-Chip format only). We have a small number of them in stock, and we have begun testing them. The biggest problem with Coppermine CPUs right now is the scarcity of good, overclockable Coppermine-capable motherboards. We are working with several motherboards, and will post our results shortly.

March 14th 2000

VIA and AMD bugs.

We spend a lot of time pointing out Intel's foibles, so it's only fair to bring up the fact that the AMD/VIA team have their share of hardware woes as well. To our knowledge, AMD does not post it's bug list the way Intel does, but it's obvious that there are bugs in the Athlon CPU. In addition, both the AMD 751, and VIA KX-133 chipsets have some bugs, not to mention poor ATA/66 support. We have run into video bugs with Athlon systems, and it has been difficult to determine if the problem lies with the CPU, chipset, or drivers. Many Athlon system owners have reported video problems, only some of which are fixed by driver updates.

In any case, Athlon systems do not appear to be any more problem prone than Pentium III systems while running Windows 98. But the rush to market that occurs with most new hardware ensures that bugs will be present.

March 13th 2000

AMD's Price Drops Make Overclocking Unprofitable!

AMD has taken an axe to it's processor prices in what has got to be the stuff of Intel's nightmares. The price of the AMD Athlon 700MHz processor has dropped so low, that the "Golden Fingers" overclocking cards are almost too expensive to be worth the price. But not to worry! As the new Athlons become available, based on the 0.18 micron fabrication process, the Golden Fingers overclocking cards should take them from 800MHz up to a Gigahertz or higher. We have not been able to get our hands on any 0.18 micron Athlons yet, but hope springs eternal.

The newer Athlons will overclock higher for the same reason that the 0.18 micron "Coppermine" Pentium III processors do. The smaller trace lines that make up the circuit elements require less power and produce less heat, permitting higher speeds. But remember, there will be little or no difference between the 800MHz Athlon, and the 1 GHz Athlon, except for the multiplier factor that AMD sets at the factory. So if there is a large price differential between the 800MHz and 1 GHz models, overclocking the Athlon will be very cost effective again.

March 8th 2000

Intel: What's Going to Go Wrong Next?

Intel has been having oodles of troubles lately, and there does not seem to be any good explanation as to what the underlying problems are. The latest reports are that there are serious flaws in the i840 chipset, not unlike the problems with the ill-fated i820 chipset from Intel. The end result is that the chipsets don't work with the higher speed Xeon processors that use a 133MHz front side bus, and can even fail completely during use.

This does not just mean that Intel can't sell the i820 or i840 based motherboards, but now they can't sell the high-speed, PC-133 Xeon processors due to a lack of motherboard support. And if it were not for third party manufacturers producing Flip-Chip adapters for Slot-1 motherboards, there would be no way to use Intel's new Flip-Chip Pentium III processors, due to a lack of motherboard support for them. As it stands now, Intel can not get the bugs worked out of their 2 new Flagship chipsets, they have failed to promote and support Rambus as the new memory standard, they can't produce nearly enough processors to meet current demand, thus loosing major market share to rival AMD, their upcoming 64-bit processor is massively behind schedule and in trouble, and on top of it all, these problems will have to start cutting into profits soon, if they haven't already.

So what's next? Maybe a large asteroid will impact Intel's newest 0.18 micron fabrication plant, or an Alien Armada will enter Earth orbit and claim that Intel has infringed on thousands of Intergalactic patents. Could be!

March 6th 2000

Memory Prices Rise Again.

I am beginning to wonder if hardware manufacturers have figured out that shortages mean higher prices. A steady supply means slumping prices. After the phenomenal rise in memory prices last Fall, it seems that "shortages" of many parts, from BX chipset motherboards, to Pentium IIIs, to GeForce video cards, has led to less dramatic, but noticeable price rises. So, not unlike the OPEC oil cartel, putting the breaks on production seems to be the way hardware makers are halting the erosion in PC prices. Memory producers have apparently reduced output to prevent SDRAM prices from slipping further. It appears to be working, as memory prices edge up again, after a very short, but welcome dip.

March 2nd 2000

Flip-Flop on Flip-Chips?

A quick check on the availability of Intel processors this week revealed something interesting. No! There still aren't any Coppermines available, but a new Coppermine showed up on the list of Intel processors, namely, the 550E Coppermine in the SECC2 format! Originally, we were led to believe that the low end Coppermines would never be released in a Slot-1 format. Now we see that indeed they are going to be released that way.

Initial reports on the web indicate these overclock to at least 150MHz on the front side bus, meaning they are great overclockers. But the question is, what made Intel change their mind about the Slot-1 format for low-end Coppermines? Maybe it's the complete lack of motherboard support for the Flip-Chip format. Months after their release, we still have not seen a single overclockable motherboard with a Flip-Chip compatible socket. Even the current Tyan Trinity 400, which has a Coppermine Slot-1 socket, also has a regular Celeron s370 socket as well. The next version of the board will have the s370 socket replaced with a Flip-Chip socket. But right now, you need a Slot-1 to Flip-Chip adapter card to run Coppermine Flip-Chips on existing motherboards.

Maybe the appearance of the SECC2 550E Pentium III is an indication that we are going to have to wait even longer for Flip-Chip compatible motherboards. But the Coppermine famine may be over soon. Intel is scheduled to release large numbers of Coppermine chips this month. Be aware that Intel has announced they will be distributing both the old (A2) and the new (B0) stepping Coppermines at the same time. They have asked distributors and resellers to sell the old and new parts "interchangeably". They obviously hope that most customers are not concerned with the reported bugs in the earlier stepping.

March 1st 2000

Rambus Stock On a Roll.

Yesterday, Rambus Inc.'s stock climbed to over $300 a share in response to the ongoing Rambus conference in Japan. The stock was off a few dollars at last check ($290), but may skyrocket again with the launch of the Sony Playstation2 in a few days, which is supposed to use Rambus memory cards. Sony says it will ship 2 million PlayStation2s in the first few days after the launch. The PlayStation2 will bring console gaming much closer to PC gaming, and will no doubt be a massive hit with the public.

In contrast to Rambus, AMD's stock price has been languishing below $40 a share. Considering the opposite business positions these two hardware makers have gotten themselves into recently, technology watchers have got to be wondering what investors know that they don't. AMD has been grabbing processor market share from chip giant Intel hand-over-fist, while Rambus has not even made it to market in numbers, in part due to Intel's blunders with the supporting motherboard chipsets. AMD is making a big name for itself with consumers, while Rambus isn't even on the radar screen. So my question is, when will investors notice that AMD is actually making money, and that it's shares cost about 8 times less than Rambus shares?

New Celerons?

Word is leaking out that Intel is getting ready to release a new type of Celeron processor at the end of this month. These will apparently be based upon the "Coppermine" core, in the 0.18 micron fabrication format, and will have SIMD extensions. Conflicting reports are out on the web that they say will work at 66MHz or 100MHz. Regardless, they will probably need the updated voltage regulator module for Coppermine CPUs, and therefore will not work on BX chipset motherboards. You can view Intel's processor roadmap here.

2000 © KickAss Gear