Ik kwam bij Ace's Hardware dit interessant tabelletje tegen met single CPU server benchmarks van PC Professionell (duits ZD magazine). De K6-III blijkt het verassend goed te doen, voornamelijk vanwege z'n tri-level cache architectuur:

CPU	ZD serverbench 4.1	ZD Netbench 6.0
Athlon 600	41.9	64.18
Athlon 550	38.7	61.8
K6-III 500	35.1	54.1
XEON 500	32.5	50.1
PIII 550	32.3	43
Celeron 500	22.5	33.2

Het Mozilla.org team heeft Milestone 10 van Netscape 5.0 gereleased. De release notes lees je hier en de Unix, Mac en Win32 files kun van Mozilla's FTP wegplukken. Om deze posting wat op te fleuren hier een reaktie van Slashdot:

Less than 50 posts and I can already read how Mozilla is buggy, how it's late and how IE is better. Maybe some of you haven't been following the Mozilla development? I'd like to clarify a few things for you.

Mozilla is still very much alpha. It means that there's a lot of bugs. M10 is definitely not for daily use. These milestones are released so that we can try them, report bugs and take part in the development process. If you don't want to do it, then you're better off with an old Netscape or MSIE.

Those of us who do know programming understand really well why Mozilla is special. That's because programmers usually know to pay attention to basics. In Mozilla I see a really small browser which supports standards really well and has a really fast renderer. It's way more important at this point than having a beautiful screen or flawless scrolling.

I'd like to talk more about programming big projects. I've been working on a big project for a year now and customers are amazed as I can't show anything yet. That's because I've been concentrating on building tools - a compact database, fast and versatile search engine and lot's of controls. When these are working well, it's really fast to build the application and it will be fast, reliable and small. This is exactly what these guys at Netscape have been doing too.

Many say that the browser war is over. It's not as the internet is a moving target. We'll get new complicated technologies and browsers have to support them well. As the renderer in Mozilla is done well, it's easy to make Mozilla support these technologies. It wouldn't have been possible with Netscape4 or the code that the Mozilla team dumped when they decided to start from scratch. And as MSIE is a huge program, it probably also hides a lot of bad code which makes developing it worse.

Pittige posting bij JC, waar men de geruchten dat Intel fabrikante van Athlon planken onder druk zet door de levering van BX en ZX chipsets te stagneren, nieuw leven in blaast. Twee maand geleden zei onze vriend Tom (toch al niet bepaald pro-Intel) ook al zo iets:

A very disturbing usenet post by Dean Kent, a former reseller who kept his many contacts and moved on to create a "behind the scenes" hardware website. He has some very shocking stuff to say, and I fear he may have severely severed some or many of his contacts because of this. He asserts that Intel is not reimbursing anyone for the tens (maybe hundreds) of thousands of i820 boards created before Intel announced them as junk at the last minute before the Camino almost-introduction. He also says something along the lines of

"In August, there was a little pressure from Intel to keep manufacturers from making Athlon boards - but it has been steadily increasing ... Intel has begun using [BX/ZX shippment constraining] as leverage to punish those who make non-Intel based boards".

Dean goes on a bit and cites a couple examples. This is rather concerning. I mean, I'm all for Intel improving their microarchitectures (and making new ones -- jeez, I've been panting for Willamette for the past two years, come on already!) and I've no trouble if they market at more attractive price points. I'd even be generally okay if they started lying out their asses like Steve Jobs (face it, he's a smart guy, and I do like Apple's design strategy, but the term "truth in advertising" isn't in his vocabulary) and started overmarketing their products with the finesse of the "Got Milk" people (actually, the rocketship and moon commercial for the PIII is really sweet!). But if Intel is even slightly doing what Dean is suggesting, I'd totally condemn them. It's a very, very uncompetitive and rude thing to go to companies that support you and say something akin to "If you look the other way, we know where you live. We know where your wife gets her hair styled, and we know where your children like to play".

I remember that Tom Pabst, famous ranter, was ranting about this very thing and while I listen to him generally, I didn't take his words too seriously. I hope my shunning those rants wasn't a mistake those months ago.

JC heeft een posting op z'n site geplakt met info over de geruchten dat de FIC SD-11 problemen zou hebben met 650MHz Athlons:

From John Howland, "Actually: The FIC SD-11 is approved for the Athlon 500 thru 650 CPU - but only the rev. 1.6... the rev. 1.7 has not completed testing by AMD yet & is therefor not approved for any speed. That is direct from AMD". From Dean Kent, "Just spoke with FIC. They did *not* provide the information to that website - and in fact, did not even provide the motherboard. They have tested the 650MHz (and are now testing the 700MHz) with good results... They *did* however, confirm that they have found some boards do reboot - which is a QA issue. They are working on identifying them so no more will ship out like this. The majority of the boards shipped should not have this problem, according to FIC".

Meer info over deze kwestie lees je hier.

Ook bij Digital-Clips hebben ze al een exemplaar weten te bemachtigen van de GeForce 256 videokaart van Creative. De review wijst uit dat ze toch niet helemaal hebben gekregen wat ze wilden. De performance is wel hoog, maar toch enigzins tegenvallend. Het wachten eigenlijk ook op software of een game die alle mogelijkheden van de GeForce benut. Hieronder heb je een gedeelte uit de review, check de rest hier.

I'm sorry to say but the GeForce has, so far dissapointed us greatly. The anticipation, the hype has all come to nothing, a meagre growth in speed for running today's applications. Wait a minute, I think I just struck a chord. Lets think of the GeForce in a more POSITIVE light. Firstly, the GeForce was a completely new chipset designed from the ground up. It was supposed to be revolutionary, rather than evolutionary. Therefore, is this card ahead of its time? I'd like to think so. The major selling point of the GeForce is the ability to do Transform and Lighting on the GPU. No longer restrained as a mere video card, nVidia has decided its about time the video card had its own processing unit. Coupled with the advanced rendering algorithms such as Cube Environmental Mapping, this card is set for many months to come.

What today's benchmarks show is merely a fraction of what we should expect from the GeForce. Until a software product can be released to test the advanced features which make this card standout, we cannot really judge it. It would be like trying to make a person from the early 90's write a review about the TNT using primeval software that could not test its touted functions.

However, there still is a lingering problem in my mind. Why is the memory clocked so slow? 166 mhz does not seem to do justice to this card. Is the lack of memory speed the bottleneck for the GeForce? If so, why did Creative not manufacture a card which would deliver to nVidia's rated specs? We'll keep you informed as we look into this issue further. Watch out for the second part of the review, highlighted in Italics on the review navigation panel.

Ik hoop niet dat je al je geld in IBM aandelen hebt gepompt, want die hebben gisteren een nieuw dieptepunt bereikt. GA-Source brengt de info.

Shares of IBM fell to their lowest price since June Friday after an influential brokerage analyst cut revenue and earnings estimates and forecast flat hardware sales into mid-2000.

Separately, IBM said Friday it plans to trim 5 to 10 percent of the jobs in its personal computer group, or 500 to 1,000 jobs, as it folds its consumer line into its broader PC operation in a bid to restore the business to profitability.

Laura Conigliaro of Goldman Sachs said surveys of IBM's customer base suggest the growing impact from the so-called Y2K effect as some corporate customers delay spending on new computers as they focus on fixing older equipment.

Shares of IBM fell to just above $111 before recovering slightly to close at $113-1/2, down $2-3/4 on the day on the New York Stock Exchange Friday.

Over bij Voodoo Extreme staat een vrij techie artikel over de FXT1 Compression techniek van 3dfx. Hieronder heb je vast een klein gedeelte, lees de rest hier.

Texture compression is the process of removing redundant image information from the texture data. There are lossless texture compression algorithms, which when uncompressed yield the original compressed image (such as .GIF, .PCX, and .PNG) and lossy compression techniques, such as .JPG, which employ techniques which greatly compress images (textures) at the cost of image fidelity. The texture compression employed in FXT uses a lossy compression algorithm which generates textures with the same compression ratio every time. This is extremely efficient for bandwidth and storage concerns, because a constant compression ratio helps eliminate "hick ups" (temporary bandwidth bottlenecking, for example) which can occur if a certain image is not readily compressed, or not compressed as well as the others. (As is the case with GIF, PNG, and JPG)

Matrox heeft de nieuwe TubroGL driver gereleased. Met deze drivers kun je een vette performance boost verwachten, mits je een Pentium 3 of Athlon hebt... Klik hier om ze te downloaden.

Mirabilies heeft een nieuwe build van ICQ Surf gereleased. Betanews brengt het nieuws:

ICQ has posted another alpha of ICQ Surf to the FTP for download. ICQ Surf Alpha version is a remarkable Internet tool that offers an exciting way to browse Web pages on the Internet. Using ICQ Surf Alpha version, you may ramble the vast Internet highways with fellow Internet travelers while enjoying mutually fun and interesting Web browsing adventures. ICQ Surf Alpha version allows you to communicate through public chat and personal instant messages with other ICQ Surf users while simultaneously browsing the same Web page. ICQ Surf Alpha version is therefore, the perfect way to increase your social contacts and generally be among friends. ICQ Surf Alpha version is great for discovering others who share your interests or hobbies. It's the ideal way to share your knowledge and experiences, receive and transfer to other users' tips and ideas as you browse from Web page to Web page soaking up all that the Internet has to offer.

Ik heb een build van vorige week geprobeerd en kon niet echt zeggen dat ik het handig vond werken. Het idee is wel leuk, maar de uitvoering is IMO nog crappy. Als je zelf wilt proberen dan kun je 't proggie (1,27Mb) hier downloaden.

Betanews heeft een handige posting met info over de status van Windows 2000 neer geprikt. Over een aantal dagen kunnen we de eerste pre-RC3 builds verwachten, daarna zal Win2k klaargestoomd worden voor produktie midden november:

Over the next few days, Microsoft will be looking hard for any last-minute Windows 2000 bugs, working to get Release Candidate 3 ready to ship to ISVs and top beta testers. RC 3 will not be as widespread as RC2, due in part because it is merely a pre-RTM checkup build. Starting Monday, Microsoft will only be acknowledging major flaws in Windows 2000 and those which do not live up to the standard will be ignored. Microsoft will have two weeks to stomp out any bugs from the point code development will cease in what is dubbed an 'Escrow Build' and RC3 will go out the door. Here is the current timeframe for Windows 2000 release to manufacturing:

• 10/10/99 - Last Windows 2000 Issues Accepted
• 10/11/99 - Final Windows 2000 Bug Prioritization
• 10/12/99 - Windows 2000 RC3 Escrow Build
• 10/25/99 - Windows 2000 RC3 and RTM Calldowns. This is when the final build sign-off occurs.
• 10/26/99 - Windows 2000 RC3 Release
• 11/17/99 - Windows 2000 Hits RTM during COMDEX
  
  These dates are subject to change.

RC3 is likely to contain new drivers but no other major changes from RC2.

EETimes bericht dat naast Samsung en NEC ook Huyndai z'n Rambus produktielijnen zal overshiften naar de fabricage van SDRAMs:

South Korea's three top memory makers are contemplating shifting production lines geared for manufacturing delayed Rambus DRAMs to other memory products as demand for the latter increases.

Samsung Electronics, Hyundai Electronics and Hyundai Micro Electronics had arranged their production schedules to begin making direct Rambus DRAMs in late September. But Intel Corp.'s decision last month to delay release of its 820 chip set, code-named Camino, has forced them to consider switching to DRAM production as prices and demand for memory chips rise.

[...] The delay is seen postponing shipments of Rambus-enabled PCs until after the Christmas shopping season.

Alan Cox heeft in een interview met ZDNet UK uitgefloept dat de release van Linux 2.4 voor november of december staat gepland:

Linux developer extrordinaire, Alan Cox, revealed exclusively to ZDNet UK Wednesday that the 2.4 kernel of Linux could be available for developers by November. Casually asserting himself as a true Linux guru, Cox couldn't resist telling ZDNet: "Linus and me are hoping to have 2.4 out by November or December." The revelation will, no doubt, delight Linux disciples worldwide, but there's more: Cox made encouraging noises concerning hardware developments and suggested the most significant forthcoming events for the Linux community are in this area. "The biggest development is going to be scaling," he told ZDNet UK News at the Linux Expo in London's Olympia. "There are a lot of multi-processor machines on the way from vendors and SGI who are really interested in 256 way machines."

Volgens The Register heeft Intel de uitgestelde release van de i820 Camino neer geprikt voor begin 2000. Da's een behoorlijk heftige vertraging...

Coppermine based Pentium IIIs will hit speeds of almost 800MHz sometime in Q2 next year, according to the latest leaks from Chipzilla’s roadmap.

[...] The troublesome Camino i820 chipset is now targeted for early Q1 2000 with the 810e entry-level chipset being pushed upmarket to support midrange systems as well as cheap ‘n’ cheerful Celeron boxes. We confidently predict that the 820 to fail to appear in this timescale – prove us wrong, Intel..

Verder zegt men bij The Register dat we in het tweede kwartaal van 2000 een Celeron op 566MHz kunnen verwachten (de Celeron gaat in kloksnelheid dus een flink eind achterop lopen op de duurdere PIII modellen).

Positief nieuws bij The Register: de DRAM prijzen zijn sinds het begin van deze maand met 20% gedaald. Natuurlijk zijn de prijzen nog steeds belachelijk hoog, maar op z'n minst is er weer hoop voor de nabije toekomst :

DRAM prices have fallen over 20 per cent since last month's high in the aftermath of the Taiwan earthquake.

[...] Richard Gordon, a Dataquest analyst, thought the worst was over regarding yo-yoing prices.

"All the froth of the last couple of weeks over the earthquake is over and people seem to have seen sense," he said.

Gordon said contract prices had started to firm up – at about $170 for a 128MB module – and prices in Q4 were back to where they had been in Q1.

"Barring acts of God like the Taiwanese earthquake, we will see some softening of DRAM pricing right through to December.

Oja, onze eigen Pricewatch reageert ook inmiddelijk: de 128Mb prijzen zijn voor het eerst sinds de start van de Price tracker - 3 week geleden - omlaag gegaan.

Onze Bill heeft een stel anti-abortus/pro-life fundamentalisten op z'n dak gekregen omdat de Gates Foundation steun geeft aan geboortecontrole en gezinsplannings projecten, iets waar die pro-lifers het niet helemaal mee eens zijn. De grap van dit alles is dat het erop lijkt dat Bill niet blindelings achter z'n PR mannetjes aan lijkt te wandelen:

It seems that our article about Microsoft's and Gates' charitable activities was a hot button for many people, and there's more to report on the saga.

A focus of the Gates Foundation philanthropy has been population control and planned parenthood, and not surprisingly this has resulted in howls of anger from pro-lifers. It is not our intention to enter this debate, but we would make the point that, surprisingly, this does show that Gates and his wife are following their own inclinations, since Microsoft PR would clearly have liked to steer him away from controversial donations.

The Planned Parenthood Federation of America is getting $5 million for its planned parenthood global partners program. This didn't go down very well with Pro-Life Infonet, who noted the paradox that Rory John Gates was born just two days before the gift to help the world cope with fast population growth.

Check The Register voor het complete nieuwsartikel.

Het kwam ook al enigzins naar voren in het AMD conference call transcript bij JC: AMD heeft bekend gemaakt dat zij pas in het derde kwartaal van volgend jaar een 1GHz Athlon verwachten. Da's denk ik toch redelijk teleurstellend nieuws. Hier de info van Overclockers' Workbench:

At this week's microprocessor forum, AMD announced that they will be shifting their operations to 0.18micron. They also announced that AMD will release 1 Ghz processor in 3Q of 2000. At the forum, they also displayed the 800Mhz version of Athlon and hopes to release it next yr.

AMD's vice president Fred Weber says that AMD is planning on it's Athlon Ultra series. AMD will also announce the EV6 which would bring the bus from 200Mhz to 266Mhz and will support 266mhz DDRRAM. L2 cache will increase from 1mb to 2mb. It will also support 4x AGP.

Weber said that other than increasing the bus speed., AMD will introduce a new technology known as Lightning Data Transfer. AMD will also be working with API and Hotrail to research on new multiprocessor technologies.

In navolging van Creative en Guillemot springt ook Leadtek op het Double Data Rate treintje. Net zoals bij de kaarten Annihilator en 3D Prophet is de DDR versie fors duurder: 100 dollar extra, maar dan heb je wel een verdubbelde fillrate. Hier de press release van FPS3D.com:

LEADTEK Research Inc. today announced its new version of the WinFast GeForce256 graphics accelerator with additional DDR technology. While keeping the same set of advanced features the WinFast GeForce256 offers, the WinFast GeForce256 DDR has significantly widened the memory bandwidth to reach workstation-class graphics performance. In addition to delivering superior 3D performance for high-end OpenGL based authoring and CAD applications for Windows NT users, the WinFast GeForce256 DDR also further enhances performance for demanding multimedia applications in the consumer market such as gaming, DVD playback and game authoring under Windows 95, 98 and 2000.

DDR (Double Data Rate), the latest memory technology, is capable of doubling the frequency of data fetching to dramatically increase data transfer throughput. Traditionally, each memory cycle is triggered by either the positive or negative edge in the clock cycle. DDR technology enables both positive and negative edges to initiate the memory cycle, effectively doubling the memory data bandwidth.

By utilizing DDR memory on the already powerful WinFast GeForce256, the speed of buffering and texturing will be greatly increased, thus releasing the bottleneck between the graphics processor and the memory. In an advanced graphics subsystem, instead of using the PC's central processor, massive graphics computation takes place at the on-board GeForce256 processor and its memory buffer; freeing this bottleneck gets particularly important in order to achieve higher graphics performance.

The WinFast GeForce256 DDR will be available at the end of November through Leadtek's distribution channels, and the estimated price will be $100 higher than the non-DDR version.