Riva 3D heeft een aantal Dagoth Moor screenshots met opgeblazen polygon count op het net gegooid. Het detail van de bomen gaat behoorlijk diep. Neem als voorbeeld dit vredige tafereeltje (met linksboven een 1:1 uitsnijding):

PC Week heeft een verslag gepost van het hack-m'n-Linux-of-NT-doos contest dat zij enkele weken geleden uitschreven. Uiteindelijk werd de Linux bak als eerste plat gelegd, maar dat lag niet zozeer aan gebrekkige beveiliging in Linux als wel aan een exploit in een commercieel CGI script:

The hack that felled www.hackpcweek.com teaches a very important lesson: Security doesn't stop at the operating system. (See related story.)

PC Week Labs went to great lengths to take the same security measures on the Linux and Windows NT servers running the site that any IT manager worth his or her salt would implement. The successful hacker, known as Jfs, bypassed the firewall, the intrusion detection system and a locked-down server to exploit a hole in a CGI script on the Linux server, which was running Red Hat Linux 6.0.

The successful attack against our Linux server was a methodical assailment by a hacker with intimate knowledge of C, PERL and The Home Office-Online's PhotoAds classified-ad engine application. PhotoAds is publicly available (at www.hoffice.com), as is information on its known security holes and fixes. Companies that don't keep on top of application fixes will be at the mercy of hackers who do.

Owkay, hét grote raadsel (waarom prefereert Rosink AOpen planken) is ontrafeld. Nee, het is zeker niet de zogenaamd onovertroffen stabiliteit van deze borden...

A top US IT company has been unknowingly hosting a Danish pornographic Web site containing a clutch of sordid sex scenes.

The hardcore material was discovered on one of the servers belonging to AOpen Inc -- the world’s largest motherboard and PC components manufacturer and part of the $7 billion Acer group, of Taiwan.

Fourteen pictures depicting images of extreme depravity -- some involving teenage girls -- were discovered on one of the company's US FTP servers.

Although officials at the company knew nothing of the porn material it is apparent that its security has been breached somewhere along the line.

Aldus The Register.

Dell en Samsung hebben onlangs een deal van 8.5 miljard dollar afgesloten. Samsung zal nu de komende 5 jaar TFT LCD schermpies gaan leveren voor Dell PC's. Tevens kunnen de lui van Samsung nu een nieuwe LCD fabriek bouwen. Hieronder een gedeelte gejat van Yahoo News.

SEOUL (Reuters) - Samsung Electronics Co announced Tuesday a $8.5 billion deal to supply liquid crystal displays to Dell Computer Corp that would provide the financing needed to build a next-generation LCD plant.

``The deal with Dell Computer means that Samsung has acquired the necessary funding to build the fourth-generation TFT-LCD plant and to continue its research to improve the technology,'' the company said in a statement.

``The plant should help to solidify Samsung's position as a leader in this industry,'' it said.

Samsung Electronics, the flagship company of Samsung Group, said it would sign the deal with Dell Friday to provide the U.S.-based computer manufacturer with thin-film-transistor-LCD screens for the next five years.

As part of the deal, Dell would buy $200 million of convertible bonds to be issued Friday by Samsung Electronics. The price at which the bonds could be converted into Samsung shares was set at a 30 percent premium to last Friday's closing price of 200,000 won.

"Microsoft heeft voor het eerst in vijfentwintig jaar zijn visie en doelstellingen gewijzigd", zo zei onze lieve Bill vanochtend op de Telecom 99 beurs in Genève. Poehoee, dat klinkt revolutionair:

"Eerst wilden we de PC op elk bureau in elk huis krijgen. Nu voegen we hier mobiele apparatuur aan toe."

Gates geeft aan dat Microsoft allianties zal sluiten met telecombedrijven, en dat het software aan deze nieuwe partners verwacht te slijten. De topman voorspelt dat het gebruik van mobiele apparatuur voor Internet-toegang enorm zal toenemen. "En er is ingebedde software nodig om dit mogelijk te maken." Het softwarebedrijf maakt veel geld vrij voor onderzoek om de technologische ontwikkelingen op het gebied van mobiele communicatie te volgen. "Op dit moment investeren we jaarlijks meer dan zes miljard gulden in onderzoek en ontwikkeling."

Check Computable voor het complete artikel.

GA-Source meldt dat Westwood de komende weken details geeft over een Expansion Pack voor Command and Conquer: Tiberian Sun (spelen mensen die game nog?), genaamd Firestorm.

GA-Strategy is reporting that Westwood Studios will be releasing full details on the upcoming Tiberian Sun Expansion Pack, Firestorm in the coming weeks.

Intel en de American Medical Assocation zijn een samenwerkingsverband aangegaan voor de ontwikkeling van zachtwaar waarmee medische gegevens online toegankelijk gemaakt kunnen worden. Intel's aandeel in dit verhaaltje bestaat uit het beveiligen van de gegevens:

The alliance, as earlier reported by CNET News.com and announced today at Intel's Internet Health Day event in New York, will focus on creating so-called digital credentials that Intel and the AMA plan to eventually distribute to health Web sites, physicians, and consumers. Digital credentials are a piece of encrypted software that helps verify a person's identity to Web sites. Intel also announced that it has created a new Internet Authentication Services unit to develop these credential services.

Health care has the potential to be a huge online industry, and companies such as Healtheon and Medquist already have entered the space. But many privacy experts regard health information as being among the most sensitive information that can be distributed, requiring companies to safeguard that data. Digital credentials like those developed by Intel help ensure that only authorized physicians, insurers, and consumers can access a patient's medical transcripts or other health records.

Tja, je kunt nauwelijks meer nieuws noemen, maar er zit dus weer eens een beveiligings krater in Billies Internet verkenner. C|Net News.com brengt ons op de hoogte:

For security reasons, browsers typically restrict the kinds of things a Web site can do with scripts. But in this case, Microsoft's Internet Explorer 5.0 browser fails to restrict scripts when they are executed from within smaller windows within a Web site called frames. The command at issue here is the "document.execCommand," according to Microsoft.

In a security alert, Microsoft said it was working on a patch that would implement tighter security checks within frames. The patch is not yet available.

Gamers Depot heeft een review online geschopt van de Matrox G400 Max kaart. Voor de gebruikelijke benchmarks hebben ze de TurboGL driver gebruik, wat de scores erg omhoog pompt. Check de review hier.

It's very apparent after reading the benchmarks, that the G400 Max really kicks in on the higher resolutions. And if you're buying a card for more games that'll take advantage of 32bit color rendering, the G400 Max kills the TNT2, and leaves it there scratchin' itself wondering why it can't render 32bit that fast.

Although the overall OpenGL speed still isn't quite as fast as the TNT2 Ultra, I doubt you'll miss those few extra 2-8 frames once your eyes get spoiled to the best image quality for 3D-gaming bar none. I was overall impressed with the OpenGL scores now that the "Turbo Drivers" are out, and I'm sure it'll only get better as future drivers are released.

PC World heeft een roundup met 10 17-inchers neer geparkeerd. De Mitsubishi Diamond Plus 71 en Dell UltraScan P780 worden als Best Buys beoordeeld, met de Iiyama Vision Master Pro 410 op de derde plaats:

WHAT'S HOT: Crisp colors and sophisticated features such as BNC ports and a flat screen put the VisionMaster Pro 410 near the top of our list.

WHAT'S NOT: The $429 price is slightly higher than average, and you can't get technical support by phone on weekends. Also, you have to scroll through submenus to reach some of the controls.

IMAGE QUALITY: Strong on text and graphics, the VisionMaster Pro 410 approaches Best Buy quality. Colors look true but don't exactly leap off the screen, and the brightly colored text on the black background of our test Web page was softer and less readable than on our Best Buys.

BEST USE: A good, if somewhat pricey, general-use display, with above-average image quality and an extensive feature set.

Speedy 3D heeft een review van de Eye 3D Stereovision in elkaar gefrutseld. Het idee achter deze 3D glazen is vergelijkbaar met die van de Elsa Revelator en Metabyte EyeScream, alleen is de werking volgens Speedy 3D nogal matig:

I wasn't impressed with the Eye3D's at all basically sighting them as a good try but in being limited in likeability only to the Stereo enthusiast and those that have very fat heads and can stand to keep their head tilted back, better get that doctors appointment ready. The stereo itself was the only good point and the fact it supports almost all drivers is a plus, sadly the stereo isn't quite as clear cut as that with the 'Revelators' due to bad shutter design. You know it gets to a point where bigger stutters actually make the experience less interesting. So who should buy these? Only the Stereovision enthusiast that understands the settings + drivers + has a fat head and a big monitor with high refresh rate. Normal gamers would be better suited to 'Revelators', which are light, well designed and have native drivers. At around $60 USD, buy something else.

Okee, iedereen ff meedoen: op Petition Petition (een of andere site die zich volledige dedicated bezig houdt met het draaien van petities) is een petitie begonnen die tot doel heeft om taiwanese moederplank fabrikanten ertoe te bewegen om meer Athlon borden te produceren:

As early as August 18th, reports began to surface in the PC industry's various online publications regarding the cause of this motherboard shortage, as well as the inexplicable lack of interest of motherboard manufacturers to design and produce boards for this hot new CPU. Many individuals (some with connections in the industry) have come forward to assert that a large corporation (who again will remain nameless in this context for legal purposes) has been pressuring these small companies to not design or produce motherboards for the Athlon. This larger corporations motivation for applying pressure is to protect long standing market share dominance from being comprimised by increased demand and sales of Athlon CPUs. They have threatened these Taiwanese motherboard manufacturers with phantom shortages of their own motherboard chipsets, which the motherboard makers need to produce boards for this larger corporations CPUs. Since the motherboards for this larger corporation's CPUs currently comprise the majority of most of the board maker's sales, most of the board manufacturers are very fearful of the threat of having this supply of chipsets suddenly cut off. This is an unethical and unfair business practice for this larger CPU producer to be engaged in, but they are doing it none the less.

De petitie kun je hier ondertekenen. Op dit moment doen er al meer dan 2500 mensen mee, dus het loopt aardig.

Ace's Hardware heeft een interessante follow-up gepost op de single CPU server benchmarks die zijn een aantal dagen geleden postten. Uit de benchmarks kwam de K6-III met 2Mb L3 cache erg goed naar voren, beter dan de véél duurdere Xeon. Ace's Hardware beredeneert wat het gebruik van Rambus DRAM voor gevolgen kan hebben voor server performance:

The L3-cache makes an incredible impact on the score of the K6-III. In other words, server and Internetserver applications are very sensible to the latency of the memory. The L3-cache has a low latency, but runs at the same speed as the SDRAMs. This could mean that RAMBUS PC800, with its higher latencies would be a pretty bad choice for a small server...

2CPU.com heeft een Symmetric Multiprocessing FAQ in elkaar geknutseld. Op zich wel handig om eens door te lezen, want er zijn nog steeds personen die denken dat SMP wonderen kan verrichten en het idee hebben dat 2x550=~1100MHz, wat in praktijksituaties zelden het geval zal zijn. Bij efficiënt SMP'en hangt alles af van multi-threading en er zijn maar weinig proggies die zelf multi-threaded zijn:

What is multithreading?

Programs break up the tasks they need to do into threads. By far, the majority of applications are written to be single-threaded. This means that the program executes in a linear fashion, and will only use one CPU. A multithreaded application is written in a way that lets it execute any tasks that it can in parallel across multiple processors. Writing a multithreaded program takes more time and skill than writing a single-threaded program because there are additional considerations such as thread scheduling that has to happen. Scheduling is what keeps the threads in synch, and makes sure that everything happens in order. That John Carmack was able to optimize Quake 3 Arena for multiple processors is a testimonial to his programming skills.

What programs that I use are multithreaded?

Well, at this point the only game that will use multiple processors is Quake 3 Arena. As more people buy multiprocessor systems, there may be more games available that use multithreading. On the high end of the scale, we have 3D rendering programs and CAD/CAM programs that are SMP-enabled. Adobe's fine suite of digital image and video manipulation tools use multiple processors, and use them well. L0phtcrack, the NT password cracker uses multiple processors. The RC5 challenge software is an excellent example of multithreaded software. In general, the majority of multithreaded programs will be high-end business software, especially in the design arena.

The Tech Zone heeft een review neer geknalt van de Swiftech MC2 cooler, waarmee je een broodje Celeron kunt smeren (vanzelfsprekend werkt dit grapje alleen op de inmiddels uitgestorven slot1 Celerons). The Tech Zone vergelijkt 'm met de Global Win VEK:

The performance of the MC² is very good. However it proved to be not much of a challenge for the modified Global Win which basically kicked its butt. Although the Global Win ran much cooler, because of its 50cfm + fans, neither cooler was able to run the Celeron 366 at 550Mhz. I consider a cooler superior if it can run the same CPU faster than any other cooler. In this case the Global Win VEK definitely performed better but was not superior because it was not able to push the Celeron 366 faster then a SwifTech MC².

Voor personen met teveel geld (het apparaat kost 20.000 dollar) heeft men bij Pioneer een nieuw plasma schermpje met een 50" diagonaal in elkaar gebouwd:

Although plasma displays are expensive, their thinness and sharp images give them an edge over CRTs for large-format, direct-view displays. The first-generation units have been impressive, but new technology gives the 88-pound, 50-inch-diagonal Pioneer PDP-502MX ($19,995 list) eye-popping contrast with a wall-filling image--and it's less than 4 inches thick.

The wide-XGA display (1,280 by 768 pixels) arranges subpixels in vertical stripes--long U-shaped channels--as do other plasma designs. The key difference is that the Pioneer panel has horizontal dividers between each subpixel. Instead of a surface of long grooves, the panel is more like a huge waffle with rectangular depressions.

Nou, leuk om op te kweken . Check ZDNet voor de preview.

Nog een VIA Apollo Pro 133 review bij HardOCP, dit keer in de gedaante van een AOpen AX63 Pro. Wederom een jumperless plankje met support voor 32 busspeeds (van 66 tot 155MHz), Vcore tweaking, ATA66 en 512Meg DIMMen. Interessant bord voor overklokkers, want HardOCP is zwaar positief over de stabiliteit op hogere busspeeds:

I gotta say right out of the box here that this is one of the MOST STABLE mainboards we have ever tested at such high bus speeds here at the HardOCP.

One of the primo tests that I like to personally use to really see how stable a board is, is the HOCP RAM TEST. Stoopid? Probably. Do I think it means something? Bet your ass. We took sticks of AZZO PC133, Corsair PC133, and EMS PC133 and tested them all TOGETHER. Every stick we have here has been verified to independently run at 150MHz FSB. What we did for this test was fill all three DIMM slots with a different stick of ram (all 128Megs in capacity) and see if the machine would even boot. I have found in the past that many time different sticks of ram don't "Get Along" either because of the ram or the mainboard or a combination of them both.

We ran the machine in endless loop tests for three days using our multiple SDRam setup at 144MHz FSB. While we did experience some crashes in Winstone99 the system basically ran flawlessly at a 360MHz clock speed.

With just the AZZO SDRam PC133 stick in, we were able to achieve a stable 155Mhz FSB. This is the first time we have ever gotten to this plateau.

Sharky Extreme heeft een review van de Soyo SY-6VBA moederkoek met VIA Apollo Pro 133 chipset gepost. Dit plankje ondersteund in totaal 29 busspeed en kent dankzij de VIA chipset AGP dividers lager dan 2/3, wat 'm ideaal maakt voor PIII overclocking:

Offering a very complete set of frontside bus speeds, the Soyo board really is one of the best overclockers we have seen. Aside from having a 133 MHz FSB setting, the board also has the capacity for dividers that won't fry your components. Allowing your AGP slot to run at 66MHz and your PCI slots to run at 33MHz, while still running the FSB at 133MHz is what this board brings to the overclocking table.

While some of the more current BX boards allow a PCI divisor of 4, none can offer the AGP divisor of 2 as the 440BX chipset only supports a 2/3 divider. The one unfortunate part about this is that the AGP divisor must be set with manual jumpers and cannot be changed from the BIOS. Up 'til now, bus speeds near 133MHz could not even be tested without a PCI video card, as the AGP clock frequency would be far out of spec.

De AGP performance ligt volgens Sharky Extreme echter nog steeds iets onder die van de BX chipset.

The Register meldt dat de EU plannen om het aftappen van internet verkeer (waaronder realtime monitoring van emails) mogelijk te maken (voorlopig) heeft laten vallen:

According to senior sources close to the Finnish presidency of the EU, the European Council wants to redraft completely the text for the European recommendation on the Interception of Internet traffic, known as Enfopol.

Britain and the Netherlands, alongside industry bodies such as the European Internet Services Providers Association (EuroISPA), have been lobbying for some time to get the draft dropped.

As it stands the current proposal would have allowed law enforcement authorities extensive access to communications performed via the Internet and other "new technologies".

But opponents of the controversial text criticised it for being too vague and lacking in safeguards.

In a statement today, Jean Christophe Le Toquin, president of EuroISPA, said: "The biggest flaw of the Enfopol proposal was that many of the points have not been clearly defined.

"The current text calls for 'full time, real time access' to 'new' forms of communication, without even attempting to explain what exactly is or is not covered and would pay for the surveillance."

Ars Technica heeft een review gepost van de Logitech Wingman Formula Force. Hier een stukkie over de force feedback performance van dit stuurwiel:

Compared to the other wheels I tried (from Act Labs and Microsoft) the Formula Force was beautiful, and truly shined in games with FF enabled. I'll talk about Need For Speed: High Stakes because it is the reference game for perfect FF use, and because I've played it much more than I have Midtown Madness. As far as I'm concerned, all developers who are implementing FF in their games need to play Need For Speed: High Stakes with a good FF wheel. Icy roads in the game feel like icy roads in real life, you even get that sickening feeling as the front end of the car starts to let go, right before you start doing donuts at 85 mph down a mountain road. Wet roads feel like wet roads, and make your wheel handle sluggishly. Even the bumpy cobblestone roads in some of the tracks are accurately modeled. Good force feedback is difficult to quantify with words. It just feels right.

Bij de weg: tijdens mijn dagelijkse wandeltocht naar school keek ik gisteren per ongeluk naar binnen bij de Dynabijt, en zag daar tot mijn grote schrik een Wingman FF kloon van Trust (vergelijkbaar rood gekleurd, alleen dan van twijfelachtige Trust kwaliteit).

Ik kreeg van Ruben Frijns een mailtje dat BSE in de nieuwe editie van hun BSE in Beeld nieuwsbrief (die ik zelf ook ontvang maar nooit lees ) wat info heeft gepost over de nieuwe P3C Camino moederplanken van Asus. Op de BSE site gaat het archief niet verder dan september '98, daarom voel ik mij genoodzaakt het hele ding te rippen :

Het rommelt in de Computerbusiness. Intel brengt binnenkort officieel de 820 Chipset uit. Ondertussen is men gestopt met het fabriceren van de BX-chipset, de chipset waar de meeste bestaande Pentium III systemen mee werken. Helaas krijgt Intel de 820 chipset niet foutloos. Daarom heeft men de introductie van de 820 chipset uitgesteld.

Asus heeft al nieuwe moederborden klaarliggen gebaseerd op de nieuwe Intel chipset. Op de website van Asus is er (nog) niets over te vinden.

Er staan 4 modellen klaar. De standaard features: Intel 820 Chipset, UltraDMA/66 IDE, AGP 4x Bus, 133 MHz Front Side Bus, 1,6 GB Rambus Bandbreedte, 3 RIMM Sockets, Asus JumperFree Bios.

De Asus P3C heeft alle bovenstaan de mogelijkheden. Met als uitzondering 2 RIMM sockets en 3 DIMM (PC100) sockets.

De Asus P3C-E heeft alle bovenstaande mogelijkheden. Ook zal er een versie komen met Audio on-board.

De Asus P3C-D is de Dual uitvoering. Hierin kunt u twee Pentium III processoren plaatsen.

De Asus P3C-L is de uitvoering met de Intel 82559 10/100Mb LAN controller.

Uitvoeringen met SCSI controller on-board (P3C-S en P3C-DS) volgen later.

Met de introductie van de 820 chipset gaan we ook gebruik maken van een nieuw soort geheugen. Na Simm en Dimm gaan we over op RIMM. Rambus geheugen is nog erg duur en moeilijk te produceren. Asus springt hierop in door speciale 'DIMM Risers' te leveren. Deze converters maken het mogelijk om het DIMM (PC100) geheugen te gebruiken in de RIMM sockets.

Prijzen en levertijden van de nieuwe reeks moederborden zijn op dit moment nog niet bekend. Zeker nu Intel de introductie van de 820 Chipset heeft uitgesteld. Een feit is dat er nu wereldwijd een tekort aan BX-moederborden aan het ontstaan is. Dit zou de uitgelezen kans zijn voor VIA, de concurrent van Intel. Maar hun fabrieken staan in Taiwan dat getroffen is door een aardbeving. Het tekort zorgt er nu voor dat de prijzen van de moederborden stijgen.

Voodoo Extreme heeft een review van de Voyetra / Turtle Beach Montego II Quadzilla neer geknalt. De Quadzilla is gelijk aan de gewone (dual speaker) Montego II, alleen wordt er een extra 'breakout' kaart meegeleverd waarop Turtle Beach een tweede speaker uitgang en een S/PDIF uitgang geprakt heeft (saampjes eten ze dus 2 slots op ). Deze exotische constructie heeft als bijkomend gevolg dat de Quadzilla niet vergelijkbaar is met het standaard Aureal SuperQuad reference design:

The Quadzilla does an admirable job of mating the Vortex2 with quadrophonic technology. Everything sounds great, and works just as it should. Unlike other dual-output Vortex2 cards, the Quadzilla does not use the Aureal reference drivers. This is both positive and negative. On one hand, the revised drivers add some handy features, especially the rear volume control in the Windows Mixer panel. On the other hand, though, this means that the drivers are not updated as often as the drivers for every other sound card out there. Aureal’s latest Vortex2 drivers significantly decrease CPU utilization (though, it’s so low that in reality, the difference is imperceptible), but the Quadzilla drivers do not yet take advantage of these improvements. But, in the end, it doesn’t really make that much of a difference...

Michael "Saxs" Persson van Shiny Entertainment heeft op het Messiah forum van de Developer's Corner een reaktie geprikt op een vraag of T&L acceleratie zal worden ondersteund in Messiah. Ja dus:

Tony put in support for T&L for the characters last weekend. Because of the way the vertexbuffers have to be constructed the characters only take advantage of the clipping and lighting, not the transformation. The background will support the full T&L in the very near future.

Saxs

The Register verwacht dat Intel binnenkort de prijzen van z'n PIII Xeon server processors omlaag gaatl brengen, waardoor deze dichter bij de desktop CPU's zullen komen te liggen:

Price lists seen by The Register show that there is not a great deal of price difference between a Coppermine Xeon and a Coppermine Pentium III.

The price for ten boxed Coppermine Xeons at 733MHz is $865 each, while the price for ten boxed Coppermine 733MHz/133 is $805. Meanwhile, the price for ten boxed 667MHz Coppermine Xeons is $688, while the 667MHz PIII/100 is $632. A 600MHz Xeon is $533 when bought in quantities of ten boxes. These Xeons have 256K L2 cache, while the Pentium IIIs have 256K ECC cache and active cooling -- fans to you and me.

[...] There could be several factors behind the move. Intel is now largely pushing its Merced-Itanium technology as the choice for high end servers. At the same time, boffins at the company are feverishly working away engineering its Willamette IA-32 processors, now expected next year. Given these future introductions, it will be important for Intel to introduce further differentiations in its processor line.

Ja, dat is wat Forbes meldt. Volgens hun heeft Intel het Rambus probleem opgelost en zouden er dit jaar nog i820 systemen moeten worden gelanceerd. De productie zou vrijdag al kunnen worden hervat. Hieronder een gedeelte uit het bericht.

Now the word is that Intel has fixed these problems and may announce the production schedule as early as this Friday. Intel and Rambus have not publicly disclosed what the problems are, but industry newsletter The Microprocessor Report says that there are memory-reliability issues on systems that have three memory slots.

"The problem was with some of the wiring specifications, and it seems they have fixed those issues," says a person familiar with Intel and Rambus. "The big question is can Intel convince the computer makers (OEMs) to re-launch systems in the fourth quarter."

Beyond the Camino 820 chipset, Intel is said to be facing no problems when it comes to Rambus-related memory. On Friday, Intel may also announce that the Carmel 840 multiprocessing chipset, which uses Rambus-technology, has no quality issues and will probably go into production in the first week of December 1999, dispelling any rumors that future chipsets would also be plagued with problems.

Target PC heeft een interview met een ventje van Kryotech gepost. Naast wat standaard geneuzel ook info over toekomstige Kryotech produkten:

Target PC : Any upcoming news for our readers ?

Kryotech : We are scheduled to release a dual Athlon KryoCooled system in late Q1 or early Q2 2000. Each processor will run at least 1GHz.

Dual Athlon op 2x1GHz...jammie!

Gamefan heeft een twintigtal screenshots van Force Commander op het net gemikt. Echt geweldig vind ik 't er nog niet uit zien:

Tenslotte hebben we ook nog een GeForce 256 review van de opper tweaker himself. Ook bij Tom weer véél benchmarks, waaronder handige (maar teleurstellende) CPU scaling resultaten:

You're probably just as disappointed as I was when I ran the benchmarks. GeForce scales almost identical to TNT2, only in Descent 3 you can find a noticeable difference. The only real exception is the TreeMark. I was impressed to see that Celeron 400 scored the identical result as Pentium III 550 with GeForce, while with TNT2 the score was only 75%. After talking to NVIDIA about this issue I was told that most of the CPU-power is currently lost inside the rather young GeForce-drivers and that the CPU-scaling will get less once the drivers have matured. [break] Tom heeft trouwens zowel de SDRAM als DDR versie getest en de beperkingen van de lage SDRAM bandwidth komen in zijn benchmarks duidelijk naar voren. In z'n conclusie is Tom, in het verleden regelmatig beticht van "pro-nVidiaïsme", slechts matig optimistisch over de GeForce: [/break] Let's have a look at the good sides first. Geforce256 performs well and is ahead of the competition in the vast majority of the benchmarks, especially at high resolutions. Those benchmarks look as if they are from the last century, if you realize GeForce's new features and its ability to display highly detailed scenes with huge amounts of polygons. Buyers of GeForce will indeed get the fastest and most advanced 3D-chip out there and the pleasure to run games on GeForce may grow as new and demanding games show up on the scene.

However, first of all I expected GeForce to be further ahead of the competition even in today's benchmarks. I am definitely disappointed with the memory interface, which deeply depends on the availability of double data rate memory. This means that the cards with single data rate SDRAM, which are shipping right now, are not as fast as they could be, simply due to the memory issue. I also expected to see some kind of advantage of the transfer and lighting engine in current applications. The CPU-scaling benchmark shows that GeForce scales about the same as TNT2 right now though

Ook onze vriend Anand heeft een review van de GeForce 256 op z'n site geparkeerd. Zoals gewoonlijk weer véél benchmarks, waarbij de kaart ook werd getest op een Athlon 700 systeem. Het door Anand geteste GeForce model beschikt over normaal SDRAM en in de meeste tests blijkt de framerate dan ook nauwelijks hoger te zijn als die van de TNT2 Ultra. Memory bandwidth in combinatie met slecht ontwikkelde drivers vormen het grootste probleem:

The drivers are the downfall to the GeForce 256's Direct3D performance. In some cases the performance difference between the GeForce 256 and the TNT2 Ultra was absolutely nothing and in other cases the GeForce 256 was actually slower than the TNT2 Ultra. These issues are entirely driver related and will be resolved within the upcoming weeks. For this reason we omitted the Celeron scores which proved nothing other than the fact that the TNT2 Ultra was 0.5 fps faster than the GeForce 256 due to the lack of good GeForce 256 drivers. NVIDIA has had their OpenGL ICD for a while now which makes the OpenGL part of the GeForce's drivers not as bad of a situation as the state of their Direct3D performance. Once again, there is no reason to get worried about these issues as they are completely driver related.

Looking at the Athlon 700 and Pentium III 600B scores we truly see the potential of the GeForce 256. Expendable is a very CPU dependent benchmark which unfortunately means that the reader must do a little more work in order to extract meaning from the numbers. A difference of 2 fps in the Expendable timedemo is a considerable performance advantage and shouldn't be dismissed as a "mere" 2 fps increase. This 2 fps increase could result in a much larger performance increase overall during actual gameplay.

The memory bandwidth issue from the OpenGL testing is present in the Direct3D benchmarks as well, where the Matrox G400MAX with its 200MHz SGRAM outpaces the GeForce 256 in 32-bit color at higher resolutions.

George Broussard heeft op het 3D Realms message board een posting geplaatst over de complexiteit van hedendaagse (single player) 3D Shoot'em ups en waarom het zo lang duurt om die games te ontwikkelen:

“Today’s games are MUCH more complex. In Duke 3D all the characters were sprites. Today they are polygonal models, have to be skinned, and then animated or motion captured. A VERY long, tedious and complex process. Also games today are "deeper" and require more than key/door stuff. That means more time to develop. Note all the so-so fps games that didn't sell well in the last 2 years and you will see that they were simple compared to more complex efforts like Half-Life. A game like DNF is probably 5-10 times more complex that Duke 3D was.”