Ik lees op Tom's Hardware Technews dat Evergreen een PCI upgrade kaart heeft ontwikkeld waarmee je een oude Pentium doos kunt upgraden naar een (socket370) Celeron:

Using the AcceleraPCI, formerly called EclipsePCI, users can upgrade from an Intel Pentium processor-based PC to a Pentium II processor architecture without having to swap out the motherboard or most other PC components, such as the chassis. The AcceleraPCI does not require any software drivers or operating system modifications. Nor will the user need to reload OS or application software, which would be required after a motherboard replacement. The AcceleraPCI card fits into a PCI slot, and the upgrade is effective after the installation and a PC reboot.

The AcceleraPCI incorporates the Intel 440 ZX chipset and the latest Socket-370 CPU interface. Currently, users can upgrade to any 333MHz to 466MHz Intel Celeron processor, with 32MB to 256MB of high-speed PC66 and PC100 SDRAM memory. Evergreen is selling the AcceleraPCI, without the CPU and memory, for $199.

Best wel wazig hoe dat dan mag werken...

Dimension 128 heeft een review gepost van de Abit ZM6. Als je op zoek bent naar een geschikt socket370 moederbord dan is dit dé keuze:

I definately recommend the ABit ZM6 to any tweaker building a system on a budget. If you are looking for a platform on which to build a Celeron based system, and particularly if you intend to overclock, this is a very good choice.

Lees ook onze eigen review van de ZM6.

Next Generation Online heeft een preview gepost van the Need for Speed4: High Stakes. Nieuw in NFS4 is dat je nu eindelijk ook damage realistische schade kunt oplopen:

The game also now has fully polygonal damage and a particle system. This allows for some pretty spectacular effects. When your car slams into a tree on its left side, it will not only get dented on that side, glass from your headlight might fly out, leaving your car with only one light -- and as you change the camera to the outside of the car, you will see the appropriate single beam.

De release van de pc versie is overigens pas deze zomer.

Anand heeft in navolging van iXBT een review gepost van de Diamond Stealth III 540 (Savage4). Het artikel is weer super lang (23 pagina's) met veel benchmarks, waaruit toch wel duidelijk blijkt dat de Savage4 ver achter blijft bij de Voodoo3 en TNT2. Vooral het ontbreken van een dual pipeline blijkt een groot gemis. De performance drop bij 32-bit is weliswaar veel lager dan bij de TNT2, maar de framerate blijft een stuk lager. Hier een stuk uit Anand's conclusie:

For those of you expecting the Savage4 to come through as being the world's fastest 2D/3D accelerator, you're out of luck. However, the Savage4 does have its strengths as discussed above. The Savage4's ability to render at 32-bit color depths with a very small loss in performance will make the Savage4 a strong performer in Quake 3 Arena which is supposed to truly illustrate a difference between 16-bit and 32-bit rendering. If it does in fact illustrate such a difference, Savage4 owners will be pleased to know that the drop in performance they'll experience for the added image quality is next to nothing compared to the TNT2 and other competing solutions capable of 32-bit rendering.

Check 't zelf op AnandTech.com.

Live Like a Savage, een site over de S3 Savage3D/4, heeft een review gepost van de...Voodoo3!:

For anyone expecting the Voodoo3 2000 to run away with the lower-end of the 3D accelerator market, you are definitely out of luck, But it will be one of the top runners in the price/performance market. Even though it does not do 32bit rendering at all its Raw Speed will make up for this shortcoming , as evident in the Savage4 review at Anandtech which proved that the Voodoo3 2000 has better CPU scaling and better performance at higher resolutions than preview boards of its main market rival the Savage4.

The Register zei gisteren al benchmarks te hebben van de K7 en PIII maar door een foutje bleken alleen de specs van de twee systemen gepost te zijn, niet de benchmarks zelf. Benchmarks zijn er nu wel. Hier is de meat:

inBench99 Version 1.1 (5 times measure) CPUMark99 (CPU Integer Performance does not include MMX register)

- K7@600MHz 68.4 Mark

- Pentium III@600MHz 70.1 Mark

FPUWinMark (CPU Floating Point Performance does not include MMX,SSE,3DNow!,FSTORE Extension register)

- K7@600MHz 2,819 Mark

- Pentium III@600MHz 3,104 Mark

Interesting stuff...

Ja...interessant. De benchmarks zijn bdw afkomstig van een engineer van Acer. Of dat valt te vertrouwen? Zie The Register voor meer info.

Home of Lost Circuits en Sysopt.com hebben reviews gepost van resp. de Shuttle HOT 661P (Slot1 BX) en FIC VA503+ (Super socket 7).

iXBT Hardware heeft weer een primeur met de eerste review van de (niet-Ultra) Diamond Viper V770. Zoals we van iXBT gewend zijn een uitstekend artikel. Ze hebben de V770 over weten te klokken naar 140/200MHz, wat trouwens nauwelijks sneller is dan de standaard 140/150 (verhouding core/memory is helemaal mank bij 140/200).

Hier heb je een plak uit de review:

Summing up we would like to say the following. The overclocked V770 retained its leadership in 3D 32-bit color and provided beautiful graphics quality in 3D as well as in 2D at 1024x768 and less. However, the additional cooling was strongly recommended in this case. In 16-bit color Riva TNT2 (overclocked to 140/200MHz) failed to leave behind its competitor - Voodoo3 3000. This observation makes us think that even Riva TNT2 Ultra (175/200MHz) will fail to catch up with the overclocked Voodoo3 3000. So, we assume that it doesn't make much sense to pay the whole lot for 32MB video memory and a Riva TNT2 Ultra chipset since al the present games and those coming out in the near future do not require larger textures and 32-bit color depth support.

Unreal patch 224 is dan toch eindelijk gereleased! De patch kun je downloaden bij Epic of bij AGN 3D. Op de Unreal Tech Page vind je meer info over patch 224, waaronder deze posting over trage 224 servers i.v.m. massale drukte:

Unreal 224 servers are popping up very rapidly, and becoming filled to capacity rather quickly.  From the last couple hours of cruising the net and playing on various servers, we're seeing wildly varying performance.  As with all games, the quality of Unreal netplay is a function of how well the server is performing and how good your connection is (bandwidth, packet loss, and latency).

If you're having a bad experience, please try out a few other servers.  This can affect performance hugely.  In my game sessions here, I had some great performance with a 300 msec ping and 28.8Kbps bandwidth, and some astronomically bad performance with a 100 ping and 50Kbps bandwidth.  The bad performance fell into two categories.

Some servers have more users than bandwidth available, so they are dropping packets, experiencing 50%+ packet loss and escalating ping.  Some are just running very slowly, around 5 fps (I'm not sure why, perhaps other processes were running?).  This tends to happen whenever we release a patch, as people rush to get into gameplay, flooding the servers that have upgraded, users setting up servers on cable modems out of desperation to get into a game... It can get chaotic.  In the future, the new backwards-compatible network code should mean there are always lots of servers available when a patch is released.

So, bottom line:  Please don't bitch if netplay performance is erratic over the next couple days as servers get set up and stabilized.  If netplay performance sucks after a couple of days, THEN commence bitching. :-)

-Tim

Ik lees net bij onze vrienden van MS.IDG Nepwereld dat SURFnet het eerste Europese netwerk is dat aangesloten wordt op het supersnelle Internet2. IBM en Mikrosoft dragen respectievelijk 1,9 en 1 miljoen dollar bij:

Het snelle Internet2-netwerk ging in februari van dit jaar in de Verenigde Staten van start, na een voorbereiding van 30 maanden. In 1994 startte de overheid, universiteiten en het bedrijfsleven dit project dat een alternatief moest bieden voor het steeds trager wordende ‘gewone’ internet. De behoefte aan een snel netwerk was namelijk groot nu steeds meer mensen het net opgaan en steeds grotere bestanden worden verstuurd. Met behulp van de 1 miljard kostende Abilene-backbone kunnen snelheden worden bereikt van 2,4 Gb per seconde, zo’n 85.000 keer zo snel als een ouderwets 28k8-modem. Wetenschappelijke instellingen kunnen zo zeer snel omvangrijke videobestanden het netwerk overjagen. Het is de bedoeling dat het netwerk ook commerciële toepassingen krijgt.

Kijk, soms is het gewoon erg leuk om student te zijn :). Bijvoorbeeld tegen 500Kb/s Netscape downen van Tucows.

C|Net News.com heeft een interview gepost met Linus Torvalds, de geestelijke pappa van Linux. Volgens Linus zal de huidige maatschappij langzaam veranderen in een entertainment society. Een prettig vooruitzicht:

Torvalds, who's become successful because of his technological prowess, said he believes people drive technology, not vice versa. Money isn't necessarily all that motivates a person these days, he said, but it is a critical part of the picture. "I actually think money is maybe the most fundamental idea that humanity" has come up with, he said. It's a "universal bartering tool" that allows a person to concentrate on saving up for the future. "People say money doesn't buy you happiness, but those people are usually people who don't have enough of it," Torvalds quipped. In the future, people will be motivated increasingly by entertainment, he predicted, noting that entertainment encompasses not only frivolous activities like playing video games, but also serious endeavors like going to the moon. "It may not be in 5 or 50 years, but in 150 years, we will be living in an entertainment society" where the most highly sought-after activity is something that keeps the boredom away, he said.

...alleen jammer dat wij het niet mee gaan maken :(.

Mike Magee van The Register claimt benchmarks te hebben van een PIII en K7 systeem. De benchmarks zouden afkomstig zijn van een ventje bij AOpen. De info vind je in dit nieuwsbericht. De grap is alleen dat er helemaal geen benchmarks staan, wel de specs van twee onwaarschijnlijk vette PIII en K7 systemen (Windoos 2000, Matrox G400 enz.). Dit is wat ze er bij AMD Zone over denken:

Mike Magee claims to have gotten some K7 Vs PIII benchmarks from the same Acer guy that said the marks for the K7 barely beat the PIII.  Thing is that Mike's teaser for the article says things are looking good for AMD, yet there are no marks, only the specs for the systems, which for some reason is using the Windows 2000 beta release.  I've been using that sucker for a couple of weeks and I most definitely wouldn't use if for benchmarking, unless of course it contained optimizations for the KNI instructions, and wanted to make any comparison as biased as possible towards Intel.  Plus it uses a G400 chipset from Matrox.  Another unreleased product, no telling if the drivers include 3DNow! or KNI optimizations, and I seriously doubt these are final.  It also seems funny to me that this Acer guy hasn't been shut down by AMD yet from releasing marks, unless of course he is not from Acer at all.  I have no idea if there are any further optimizations for either KNI or 3DNow!, so maybe I'm wrong.  But I must say, trust no one.  In fact, until I get benchmarks from a trusted source with the K7, I will be trusting no marks.  And believe me, I can easily count the sources I trust one hand. ;)

Ik lees op AMD Zone dat AOpen een nieuw socket7 moederbord gereleased heeft. Hier heb je de info:

AOpen introduces its new super socket 7 motherboard – MX59pro – to public today. Utilizing the VIA MVP4 chipset, MX59pro is fully optimized for AMD K6-2, K6-III with write allocation support and Cyrix M2 CPU for linear burst support from 66Mhz to 100Mhz of system bus frequency. The Integrated 2D/3D Graphics Engine and Video Accelerator with advanced DVD video capability incorporates UMA architecture and utilizes 2 to 8MB of system memory as frame buffer. It provides a flexible and high performance solution for graphics and video playback acceleration under various color depth and resolution modes. The AC’97 link with onboard CODEC enables full-duplex multi-media experience for sound under Windows 9X and NT environment. Onboard 512KB piplined burst SRAM L2 cache enhances the overall system performance for both read and write transactions. Designed with microATX form factor, MX59pro provides the expendability with 1 ISA and 4 PCI v2.2 compliant slots, 2 DIMM sockets for maximum of 512MB SDRAM. Dual Ultra DMA 66 Bus Master IDE channels and two USB ports are also provided for mass storage media and easy-of-use peripherals.

Ik lees op CPU Central dat de nieuwe MSI MS-6905 slot1 riser cards met dual Celeron jumper eindelijk leverbaar zijn:

Looks like MSI has taken our hint, and they have begun selling their new MS-6905's. The one source I have found for them so far is Census Computers at www.censuscomputer.com. I'll keep you posted as I find more sources.

EBNews heeft scans van de Intel Pentium III price roadmap zoals die verstrekt worden aan OEMs. Helaas zijn zijn de scans nogal klein en daardoor praktisch onleesbaar. Hier heb je de bijbehorende info:

Intel Corp. has released its April pricing roadmap to OEMs, which gives the firstspecifications of the new Coppermine or Pentium III microprocessor with integrated cache.

As Intel moves to its 0.18-micron manufacturing process, the company will integrate greater amounts of on-chip cache on its mainstream microprocessors, as it already does on its Celeron chips for low-cost PCs. The smaller die size of the processor core will allow Intel to integrate a total of 256 kilobytes of cache directly onto the Pentium III and Pentium III Xeon processors designed for desktop PCs, as well as workstations and servers.

At the same time, Intel will begin to add the Pentium III instructions to its mobile processor lineup. The company will apparently also fulfill its earlier pledge to increase the speed of its mobile microprocessors to match those designed for desktop PCs.

Tim Sweeney heeft de Unreal Tech Page weer geupate met de laatste stand van zaken rond Unreal patch 224, die binnenkort gereleased zal worden:

224 Progress Continued

The team found about 10 bugs in Unreal 224t that are significant and merit fixing before release, so I'm just wrapping up fixes for them now, and we'll have another release candidate (224u) in internal testing this afternoon. Thanks for the patience...we're almost there.

Blues News posted a huge 3DS Max rendering of a bunch of Unreal Tournament characters.  These are the actual meshes (same poly count as the game), but textures of this resolution will only be available on cards that support S3TC texture compression such as the Savage4 (though other cards will have support soon).  Current 3D cards will have to make due with four 256x256 textures per player.  Our new skinning method enables player meshes to be mapped with four textures, so we can have separate faces and armor.  This Ultima Online inspired approach lets players customize their look more than past 3D action games.

-Tim Sweeney

Gurutech heeft een korte preview gepost van de K7. Hier een stukje over de 128Kb grote L1 cache van de K7:

So what have we learned from these statistics? Well, the more L1 cache a CPU has, the faster it's going to be, because L1 cache is considerably faster than L2. But then, the more L2 cache a CPU has, the faster IT's going to be, because L2 cache is considerably faster than RAM. While the K7 has the same amount of L2 cache on it as the PIII, it has four times the L1 cache, which means faster performance.

GA-Source heeft een Q&A gepost met Joe Lau van Hercules, met info over de nieuwe TNT2 kaarten van Hercules. Hier alvast een stukje uit het interview:

What do you believe is Hercules' greatest strength in the video board market?

Hercules provides the best available graphics card based on any chosen chipset. We begin by providing enhanced Hercules video utilities that allow users to really tweak their video to provide maximum performance out of their specific systems. Our newest enhancements are our HercuMeter and our MoniTUNE utilities. The Hercules HercuMeter is a slider that allows adjustment of both the engine and memory clocks on the board. This really allows users to squeeze every ounce of performance out of the board. The Hercules MoniTUNE utility allows users to custom configure their monitor limitations to provide the most video options for their given system.

Hercules was also one of the first board manufacturers to incorporate active cooling fans onto our high performance video cards. This assures stability of the hardware even under adverse conditions. This also allows Hercules to deliver the fastest boards straight out of the box. Users can be assured of top notch performance even without touching any tweaks.

De Hercules Dynamite TNT2 zal trouwens gebaseerd zijn op de hoger geklokte UltraTNT2 chip. In het interview van GA-Source staat ook nog wat info de Terminater BEAST 99, die gebouwd is rond de S3 Savage4.

CGR en Bleusnews hebben interviews gepost over Unreal patch 224. Hier een stuk uit het interview van CGR met Tim Sweeney:

How's the progress going for the Unreal 224 patch?

Tim: We're almost there. We've gone through tons of internal testing. We had a release candidate yesterday that almost cleared testing, but a few bugs were found and we're trying again. The programmers are focusing 100% on the patch 'till it's released.

What cool features will the patch sport?

Tim: Improved Voodoo3 support, improved OpenGL support, mesh level-of-detail which significantly speeds up bot and multiplayer games where lots of people are visible, in-game Internet server browser (similar to GameSpy), improved stability. There's lots more, see http://unreal.epicgames.com/readme224.htm.

[break]En hier een stuk info uit het interview van Blue met Mark Rein:[/break]

Blue: You guys have been accused of falling prey to feature creep, both on Unreal's original release, and on these delayed patches. Do you agree?

Mark: Guilty as charged! What's happened to us is that by the time we realize that a patch is relatively stable, with only one or two things to fix, we are no longer working with the same version of the source code meaning we would have to undertake the same renewed set of testing to get another build out. This makes you say to yourself "hey, if we're going to have to do all that testing over again, we should wait until this one other great feature we're working on is finished," and that's how you get caught in a feature creep cycle. This leads us to today where we're testing a whole bunch of new features in 224 like the mod installer, improved OpenGL, continuous LOD on the models, texture LOD, the integrated server browser and others.In retrospect we should have released more than one patch to accommodate these improvements and at some point removed the word "beta".