Nuja, meer Test 1.09 maar het lijkt erop dat deze test versies meer beta zijn dat ID toegeeft.

Er zijn veeeel interessante dingen te melden. Ik stel voor dat je het gewoon helemaal zelf leest, maar goed...

Interface The interface in Quake 3 is much improved already. The server browser, while not quite as flexible as Unreal Tournament's is up to the job of getting you into games quickly and easily. You can filter by several settings, sort columns, and add servers to your favorites. [break]Vooral dat laatste maakt me blij. Ik had zelfs al eens een e-mail naar id gestuurd met het verzoek om zo'n feature.[/break] Music Just to get it out of the way first, I heard the Quake 3 music in quite a few of the levels I played. The music has a very moody, ambient techno flavor to it. It is reminiscent of Quake 1 music but better in many ways and changes style between levels more. It's the kind that gets you into a darker mood and more into the game. Definitely something that adds to the atmosphere. I didn't care for Quake 2's music all that much [break]Hmm. En 350MB aan hardeschijf ruimte kwijt zijn enkel aan wav bestandjes.. Ja, dat is fijn![/break] Levels The levels I played are for the most part much cooler than Q3Test's levels. They also have more detailed textures and special effects (that haven't been seen in screenshots). I played quite a few levels containing water too. Paul Jaquays seems to like putting water in his levels and there is a nice bubble effect when weapons are used underwater. ... Graeme Devine said that overall, there will be about 30 maps in the game. 6 tiers of 4 map sections along with 6 CTF maps. [break]BLIJ! Hoeven we zelf tenminste niet meteen aan de slag met het maken van levels (doen we toch wel, toch?) Genoeg informatie is er te vinden, maar eh... Tjek die zooi zelf, hier.

[Update] Ik vind steeds meer leuke dingetjes over Quake3. Nu een .plan update van Grim Deveane:[/break] Secondly, Graeme has word that Mr. Elusive, the guy behind the Omnicron bot for Quake1, and Gladiator bot for Quake2 has been contracted by id Software to work on the bot code for Quake3. A few astute gamers out there probably already spotted the mrelusive.cfg file in the Q3Test pk3 file and were wondering. Now you know!

Mr. Elusive (aka Jan Paul) is over here from The Netherlands helping finalize the game. He is of course the famed author of the Gladiator bot for Quake II. As some of you have been guessing he’s been working with John Cash on the bots in Q3A.
We’re very pleased to have Jan’s expertise on Q3A.

KIJK! Dat is het soort info waar we op zitten te wachten! Een beetje nederlandse input in Quake3!

Johny heeft zowel gister als vandaag een .plan update gedaan. Mocht ook wel eens, aangezien hij al sinds begin vorige maand meer niets van zich had laten horen.

My advice for display lists is to use them for just raw vertex/color/texcoord data, and keep your state changes done with direct commands. This allows you to still sort display lists to minimze state changes, and prevents drivers from ever having to check state internally. Some hardware architectures can nicely encapsulate all state changes in a single dma buffer, but register sharing among different fields sometimes requires the driver to do manual masks, negating much of the async display list benefits. Texture swapping also complicates state changes inside display lists. You want to make the display lists as big as practical, but there is a tradeoff betwen culling tightness and display list size. [break]Als jij weet waar dit over gaat, dan is het absoluut a must-read. Voor de gewone stervelingen onder ons is het tweede stukkie een beetje interessanter, ongeveer het midden van deze maand zouden wij verblijdt worden met een nieuwe testversie, maar die is dus niet gekomen:[/break] Ok, obviously we didn't get a release out in the middle of the month... We are still hashing out the single player game, so it still isn't immediately immenent. [break]Dus nog ff wachten. Verder nog wat technische veranderingen, waarvan hier een drietal opvallende:[/break] * reduced machinegun damage * sarge as default model * reduced com_maxfps to 85

Wat me opviel waren een boel items die een verbetering in snelheid zullen moeten brengen.

Je kan de rest o.a. hier lezen.

Ik lees bij 3DGPU.com dat het Engelse blad "PC Pro" wat benchmarks heeft gepubliceerd van de GeForce 256. Zie hieronder het tabelletje.

	GeForce	TNT2U	G400 Max
Quake II 1024x768x16	90.3	56.0	32.5
Quake II 1600x1200x16	54.6	26.6	26.7
3D Mark99, 1024x768x16	5,386	5,040	5,165

Let wel: de kaart gebruikt is een oude reference plank, en de drivers waren beta.

The Register heeft een nieuwsartikeltje gepost met meer details over de timing problemen die i820 moederborden praktisch onverkoopbaar maken:

Electronic Buyers' News in the US reports that the timing problem with the Rambus RIMMs stems from the fact that some of the motherboard traces are up to 10cm in length, causing timing-related signal integrity problems.

If this is indeed the cause of the problem, all existing Camino motherboards will have to be scrapped. It is not known if the trace lengths to the second memory slot are also long enough to cause problems.

One extremely high tech kludge to get around the problem is reported to be the fitting of a cap over the offending third memory slot. This would prevent OEMs and end users from fitting additional RIMMs, but, as reported here yesterday, this would mean a maximum of 512MB of RAM for all i820 based systems - not nearly enough for the kind of demanding applications the chipset was designed to handle. [break] Verder hebben ook onze Nederlandse vrienden van Computer!Totaal een interessant artikel gepost over de Camino chipset troubles: [/break] Aanwijzingen duiden echter op een probleem met Intels eigen i820-moederborden waardoor OEM's als Dell en Gateway geen feilloos werkende i820-systemen op de markt kunnen zetten, terwijl andere dat wel kunnen. Dit is natuurlijk slecht voor de concurrentiepositie van Dell en Gateway, waardoor ze Intel naar alle waarschijnlijkheid 'vriendelijk' hebben verzocht de introductie van de i820 uit te stellen. Deze introductie is voor onbepaalde tijd uitgesteld. Wij verwachten dat de introductie van de i820 zal plaatsvinden als de problemen met het moederbord zijn opgelost.

Het lijkt erop dat de problemen niet zozeer het gevolg zijn van een falende i820 chipset, maar worden veroorzaakt door het brakke ontwerp van de Intel's eigen i820 mobo's (de te lange trace length tussen chipset en RIMM banken). Check C'T voor meer info over deze Dell/Gateway conspiracy theorie.

Hardware Central heeft een preview in elkaar gedraaid van de i820 chipset. Geen benchmarks ofzo, en het meese is nu wel bekend, dus hier een hap uit de conclusie:

Intel has never really been put on the spot like this before, and their inexperience with dealing with it is showing. The delay of the i820 isn’t a disaster for Intel, but it doesn’t look wonderful.

Whether or not you ought to run out and buy an i820-based system remains to be seen. Certainly, it has its advantages, but it also has some issues that clearly need resolving. First off, the whole RDRAM/price issue is one that will have a large impact on the success of the i820, and Intel may be wise to offer some sort of PC133 SDRAM support.

In my view, the real potential of this chipset and the 133MHz FSB won't be clearly seen until Intel releases their Coppermine CPU; so, for now, perhaps the best thing you can do is to sit back and watch.

Het hele verhaaltje kun je hier vinden...

Transmeta heeft een nieuw patent aangevraagd dat verband lijkt te houden met de mysterieuze processor die ze daar volgens geruchten in ontwikkeling hebben. De beschrijving van het patent ligt een tipje van de sluier op. Hier een stukje copy/paste uit een nieuwsartikel van The Register:

Another patent granted to Transmeta seems to have established what the mysterious outfit is up to pretty clearly. Briefly, the company's processor is intended to be faster than anything built using current technology, and to be able to run any of the operating software for any existing processors - faster than the original.

This of course sounds like complete hokum, but Transmeta has the patent, US Patent Office number 5,958,061, and the application, which can be read here, explains in some detail how the company proposes to achieve this. Transmeta claims: "The present invention overcomes the problems of the prior art and provides a microprocessor which is faster than microprocessors of the prior art, is capable of running all of the software for all of the operating systems which may be run by a large number of families of prior art microprocessors, yet is less expensive than prior art microprocessors."

The mechanisms described in this patent tally closely with those outlined in a previous Transmeta patent (Transmeta reveals radical new chip design). Transmeta won't be using "a microprocessor with more complicated hardware to accelerate its operation," but instead will be combining a hardware processing portion it refers to as a "morph host" and an emulating software portion, "code morphing software." These two, the company claims, will work together as a microprocessor "with more capabilities than any known competitive microprocessor."

Met de huidige belachelijk hoge DRAM prijzen lijkt het voor bepaalde duistere personen ook weer interessant te worden om die onschuldige reepjes te ontvreemden. Dane-Elec, een of andere grote distributeur uit de UK, werd gisteren getroffen door een dergelijke 'RAM raid':

Dane-Elec today warned fellow memory distributors to keep DRAM under surveillance following a robbery at their warehouse. The Surrey-based distributor said it was broken into at 2am this morning. But the robbers were disturbed by a security guard and got away with less kit than they intended.

Alan Stanley, Dane-Elec general manager, said 128MB and 64MB modules were stolen. He did not disclose the value of memory taken, but said it was small, and less than a tenth of the company’s stock. Word on the street is that this could be anything between £1 million and £3 million.

[...] DRAM robberies -- burglaries, ram raids(where villains rammed vehicles through warehouse doors and hijacks were common in Britain in the heydays of high memory prices (1994-1996. Looks like they're back.

Check The Register voor meer info.

InterCosmos.com heeft een vergelijkings artikel online gezet over de zgn. 'software coolers'. De deelnemers zijn AmnHLT 2.1, CPUIdle 5.6, Waterfall Pro 2.99 en Rain 1.0. De laatstgenoemde wint met een neuslengte:

All software coolers really cool your processors so you should have no doubt about it. CPUIdle and Rain have tie in that aspect by cooling the most out of your CPU. It seems that all software coolers do not affect the CPU performance as shown in the benchmarks. Only CPUIdle and Waterfall Pro show a little dip in graphics performance while AmnHLT and Rain do not seems to have any impact at all. It is tough to pick the winner as each has its own strengths and weaknesses. Personally i would pick Rain 1.0 as the best software cooler due the fact that it is very simple and easy to use and also it is a freeware and it doesn't affect CPU performance at all. I would recommend you choose the programs that best suits you; if you go for feature packed programs then you may consider CPUIdle 5.6 or Waterfall Pro 2.99.

Quake kampioen Thresh heeft een editorial over Unreal Tournament in elkaar geschroefd. Hier wat plakwerk uit zijn relaas:

With its speedy gameplay and air-movement control, doesn't UT sound a bit like Quake 1? It does to me. Is that a bad thing? No way. At the height of its popularity, nothing could compare to the adrenaline rush of Q1, and for all the individualism and "innovation" in the FPS community (how many people out there are playing SiN multiplayer?), change for the sake of change statistically doesn't work out. There's nothing wrong with borrowing from something that works.

While we're not going to get into the whole game x vs. game y debate again, whether you end up liking UT or not, one thing is clear: Epic has NOT been holing themselves up in a secluded office coding away without seriously considering the input of the FPS community.

Volgens Omega Hardware heeft een anonieme bron binnen AMD het bestaan van de K6-2+ bevestigd. Die processor zal in 3 smaken gebakken worden, met 0,18micron koper-interconnect technologie:

Sources inside AMD (who will remain anonymous) have now confirmed the K6-2+. The K6-2+ is what AMD plans to market the .18(CS50)K6 (copper)line as. It will come in three flavors:

-No-cache(same as current K6-2)
-Full-cache(same as current K6-3)
-Half-cache(same as Intel Coppermine)

Currently, the K6-2+ is in the test phase, and has not yet started mass production.

Hardware One heeft een review van de Logitech Cordless Wheel Mouse online gezet. Hier een stukje over de staartloze werking van het beestje:

After the installation of the mouse & the software, the mouse worked like any other mouse (with a scroll wheel). The only difference is that when you have to move your mouse around the table, you do not have to deal with the wires involved. At 400 dpi resolution, the movement of the mouse for most applications is smooth. I suppose it will be a plus for the Quakers and FPS enthusiasts. The shape of the mouse provides for a comfortable grip, with its size just right for my palm.

However, the digital communications between the mouse and the receiver can be affected by the monitor and the placement of the receiver, in my case, my Viewsonic E771 monitor and the placement of the receiver on the top left corner of the monitor. At my normal desktop resolution of 1024*768*32 bits (under Matrox drivers), the mouse works fine. However, if the drivers get switched to “Standard VGA” at 640*480*16 colours (i.e., windows default driver & resolution), the mouse becomes a bit flaky in its movement. Moving the receiver away from the monitor helps or changing the communications channel can also help improve the mouse performance. I suppose placement of the receiver away from the monitor would solve most of these kinds of problems.

Voodoo Extreme heeft meer reakties toegevoegd aan hun Developers on T&L vs Fillrate artikel. De opvattingen van Michael Persson / Shiny Entertainment:

Since we are using scalable geometry for all characters in Messiah we are not as much fill rate bound as we are number of triangles bound. With upwards of 15,000 polygons per character on my K7 with a TNT2 we need a lot of transformations done in a hurry. I haven't implemented the T&L for the Geforce card yet, since it requires a fair amount of caching control as the geometry changes rapidly for the models and the Geforce pipeline is not really built for that. Imagine that. A next generation video card that likes static vertexbuffers? What was it videocard manufacturers have been preaching for years? "Use scalable geometry to get the most out of our card!". Oh, well. But the T&L when implemented properly should still present a significantly higher average number of polygons per character than any other current card, especially on processor bound systems.

The high fill rate in itself is not really interesting from our perspective, but when used to drive the T-buffer it makes a whole lot of sense. For my money the T-buffer is going to present the biggest change of how PC games look than any new rendering technique presented over the last few years. Antialiasing is suddenly enabled on all your old games, and for games programmed specifically for it, the depth of field and motion blur makes for very interesting effects in future games.

Since the T&L requires special programming efforts it will not immediately be available in a lot of games, but since the card pipeline is pretty much built for your run of the mill FPS it shouldn't take long before patches for the most popular games makes it out making the T&L a good decision for people that just can't take playing at less than 120 FPS.

David McNett (aka Nugget) van Distributed.net heeft een .plan update gepost over de problemen met de RC5 statsbox. Iedereen die getracht heeft om de afgelopen dagen z'n statistieken te bekijken zal gemerkt hebben dat de opperkoe op dit moment nogal kaputt is:

Note to self: SQL written while listening to Rage Against the Machine will need to be cleaned up at a later date. [break] ...en dat beschouw ik als bewijs dat Nugget een koele gozer is! RATM rulez! .

"Mooo, I won't do what you tell me" zei de Opperkoe: [/break] I've got a couple working theories on the stats slowdown, and some test code to run some benchmarks. Unfortunately I noticed that the monday- morning statsdb backup didn't happen monday as it should have, so I'm in a holding pattern as the tables bcp out to a safe place. As soon as this is wrapped up I'll get to run some time tests to see if today's changes fix the problem.

The "I searched for my stats and got nugget's" problem is fixed. Most of the php3 pages have defaults where they show id #1 or team #1 if no valid data is provided to them. I made a change to psearch.php3 last night that jumps straight to psummary.php3 if the search only yields one email address. It seems that psearch was confusing a lot of people who thought didn't realize that they could click on the email address and get to a full summary page. I screwed up when I made the change, but didn't notice when I tested it because (drum roll) I tested it with my email address. In reality, it wasn't passing the id properly to psummary, but I thought it was because the default was me. oops.

It's fixed now, and should be working just fine.

I'm pretty sure we're on the right track, that the recent slowdown in processing is triggered/exacerbated by the fact that as of last week, normal stats viewing now touches the "big" rc5_64_master table. Until we added phistory, no stats pages actually hit the main table. As explained in decibel's .plan, he disabled the phistory pages to see if that would help. As you know, it didnt', but that may very well be because we missed a spot. The "today this participant did foo% of their maximum" line also hits the rc5_64_master table, and we overlooked that. So, you guys have been hitting the rc5_64_master table this whole time, even with phistory disabled.

Assuming this is indeed the cause of the problem, I made changes to the php3 code today which may eliminate the conflict. If the changes do not help (and tonight's run also takes way too long), the next step will be to disable phistory and the "today compared to best day" psummary line during the update. (not all the time, just when the update runs).

If that doesn't help, well... we'll burn that bridge when we come to it.

Ik lees bij Slashdot dat Netscape Communicator 4.7 gereleast is. Geen idee wat er gefixt is of wat er beter aan is. Het is nieuw, dus het is goed .

De Win32, Unix en MacOS versies kun je hier van Netscape's FTP trekken. Bij de lokale Tucows is-ie op dit moment nog niet te vinden.

De chipbakkerijen in Hsinchu science park (waaronder TSMC, UMC en Winbond) hebben de produktie hervat, zo bericht EETimes. De schade aan de fabrieken blijkt mee te vallen:

Taiwan's fabs have resumed limited production after Taipower, Taiwan's national power utility, decided that providing power to the Hsinchu Science Based Industrial Park was a national priority.

Meanwhile, damage to Taiwan's fabs seems to be far less than originally feared. Last Saturday, Taiwan Semiconductor Manufacturing Co.announced that 20 percent of its fabrication equipment was intact. "We did not see any further losses after Friday's inspection," said Y.C. Huang, spokesman for TSMC. "All of our fabs have DI water, HAVC, gas, exhaust and power-supply systems now."

By yesterday, Taiwan's two major fab companies were well on their way to recovery. "TSMC's equipment recovery is now at about 70 percent," said Huang. "Our 6-inch fabs are now at 70 percent of production. Two of our 8-inch fabs are at 50 percent production. We are planning that by Sept. 30 our 8-inch fabs will be at 80 percent."

[...] The Sept. 21 quake will definitely cause PC shortages next month. "Usually Taiwan produces about 5 million core-logic chip sets per month," said Shen. "In September, because of the quake, it will be more like 2.8 million. The price of graphics chips will go up because they also will be in short [supply]."

[...] On a personal level, everyone here is very traumatized. "I keep feeling aftershocks that aren't there," said one marketing manager. "My head hurts, my stomach hurts, I feel disoriented. On [Sept. 22], they sent us home because of fears of aftershocks."

Indeed, Taiwan is not yet out of the woods. An earthquake measuring 6.8 hit the island on Sunday, immediately killing eight people. As of Monday the death toll was at 2,090. Virtually every company in Taiwan is donating money to relieve the misery of the survivors who are without homes.

EETimes bericht dat Micron, als gevolg van de enorme vraag naar (/tekort aan) DRAM chips, de prijs van z'n 64Mbit chippies met 20% verhoogd heeft:

Micron Technology on Monday (Sept. 27) raised OEM contract pricing on 64M DRAMs from $7.50 to over $9, effective immediately, and another hike may be right around the corner as the full impact of the Taiwan earthquake is felt, a company official said here.

[...] Spot market pricing is also showing no signs of slackening, with the latest reports indicating $19 to $20.50 as the range for 64M parts, Bedard said. "There's not a lot of volume going on there because the suppliers (were) woefully short, even prior to the earthquake in Taiwan, just meeting the OEM requirements for Q4." [break] Gelukkig lijkt er enige hoop te zijn voor de toekomst. Micron en Samsung verwachten de DRAM produktie verder op te kunnen voeren in het komende half jaar: [/break] There isn't enough increase expected in wafer outputs to take the pressure off DRAM pricing, he said. Samsung is planning to open a new fab next spring, and Micron expects to add another 5,000 wafers per week at its overseas fabs by spring, bringing that total to about 17,000, which would match its weekly output in Boise. But other than a joint venture fab NEC is opening in China, Bedard said he didn't know of any other capacity increases in the pipeline.

Micron expects that all of its Boise capacity will be shifted to 0.18-micron in about 4 months, and a small part of the Boise fabs are already being moved to 0.15-micron. Its overseas fabs are about a month away from being completely converted to 0.21-micron, with around 15 percent of those lines already at 0.18-micron.

Met de release van Windows 2000 in januari volgend jaar zal de vraag naar DRAM chippies echter ook weer omhoog gaan .

Storage Review vond het nodig om een review te posten van de 37Gig dikke IBM Deskstar 37GP UDMA/66 harddisk. De schijf draait op slechts 5400rpm en moet het in deze performance categorie wat betreft snelheid afleggen tegen de Western Digital 205AA:

An interesting comparison arises when we set the Deskstar 37GP against WD's Caviar WD205AA. Readers may recall that the specs and performance between the Deskstar 25GP and the Caviar AC420400 were quite similar. In this successive generation, however, we're seeing some differences. IBM has packed more data per platter on the 37GP than WD did with the WD205AA. On the flip side, WD states that its incorporated a variety of proprietary improvements in the firmware side that make the WD205AA a drive all their own.

A look at the transfer rates delivered by low-level WinBench 99 tests show that while the IBM packs more data per platter, the two drives sport virtually the same transfer rate on the outer tracks. When the test gets to the inner track, however, the WD proves to be slightly faster. It seems IBM's improved GB/platter arises from more densly packed tracks rather than squeezing in more data per track.

Glide Underground heeft een review van de Kaser (erg cheesy naampje) YO!MP3 portable MP3 player in elkaar geklust. Hier een stukje:

Last thing... sound quality. There's three levels of bass you can play your MP3's at: high, medium, and low. (duh) I prefer the high setting, as low and medium just don't cut it in my book. The sound is crystal clear coming from the provided set of headphones, that is, assuming the MP3 your playing was encoded at high quality. (128kbps preferably) The best part though, is when you shake the player when running, drop it on the ground, throw it against a wall, hit it with a hammer... well..... that's a bit too far.. but it won't skip! This is because there are no moving parts anywhere on the MP3 player (except maybe the battery case, or buttons) as on a CD player. Instead the audio is stored on a simply memory chip onboard the player, or on memory cards which can be purchased separately at 32MB a piece.

Het volk van Got Apex? heeft een paar AGP4x vs AGP2x benchmarks gedraaid op het i820 plankje dat zij daar hebben liggen. Uit de tests blijkt dat het snelheidsvoordeel bij gebruik van AGP4x transfers in de meeste gevallen niet hoger dan 2 frames per sec is. Poehee...

Impressive? Not at all. A step in the right direction? Definitely. What we see here is what I think we'll be seeing in alot of the newer pieces of hardware coming out today. Headroom for growth. There was a time that PCI didn't yield that much more performance over ISA.

Benchmark grafiekjes en andere leuke info vind je hier.

Western Digital roept 400.000 harddisks terug. Het gaat om enkele 5400rpm modelletjes die tussen 27 augustus en afgelopen vrijdag het levenslicht zagen. Hier wat info uit een nieuwsartikel van EETimes:

Western Digital Corp. is recalling nearly half a million hard disk drives because a faulty chip might cause failures as the drives age. WD did not name the chip supplier.

The unusual move came after WD's testing process revealed that after six to 12 months, the drives might not power up due to problems with a chip from an outside supplier. The roughly 400,000 drives were made between Aug. 27 and Sept. 24. The recall is only for recently-introduced 6.8 G-bytes/platter versions of the firm's Caviar line, which is used extensively in PCs. No other WD drives are involved in the recall, the firm said. The Caviar line is sold to OEMs, resellers and retailers.

In a conference call, WD executives did not name the chip supplier. However, Chief Operating Officer Russell Stern explained that "This is a motor driver for the spindle and the voice coil positioner. We are single-sourced for this part, but there are multiple fab locations. The problems are tied to one fab location." [break] Western Digital heeft een pagina online gegooid met meer info over de falende drives: [/break] During extensive quality testing procedures, Western Digital has identified a hardware issue that affects a very specific group of WD Caviar 5400 RPM EIDE hard drives. This issue could cause the hard drives to fail to power-up after six to 12 months of full-time use. The affected drives were manufactured between August 27, 1999 and September 24, 1999 and range in capacity from 6.4 GB to 20.5 GB. None of Western Digital's WD Expert, WD Enterprise, and WD Performer hard drives are affected.

As part of our ongoing commitment to quality, we are providing a drive utility that identifies the affected WD Caviar hard drives. Even though this issue affects a small number of WD Caviar drives, we highly recommend that you test your drive using this utility.

Affected Model Numbers:

WD64AA
WD68AA
WD84AA
WD100AA
WD102AA
WD135AA
WD136AA
WD170AA
WD172AA
WD200AA
WD205AA

There is no need to test any model other than those listed above.

Het testproggie kun je hier gaan halen.

Sharky Extreme heeft wat lopen babbelen met Scott Sellars van 3dfx. Deze vraag over de busbreedte van de GeForce 256 en Savage2000 is op zich wel interessant (beter dan de standaard marketing propaganda die in dergelijke interviews wordt verspreidt):

SE: Quad Pipeline: NVIDIA's GeForce has support for 2 pixels x 2 textures x 2 128-bit pipelines, while S3's Savage4 has 2 x 2 x 2 OR 1pixel x 4 textures x 1pipeline, indicating a true 256-bit pipeline. The Matrox G400 architecture is also called 256-bit. Describe Voodoo3's pipeline in these same terms and (if possible) indicate if there are any changes on this score for Napalm?

Scott Sellers: I don't think it's accurate to compare the number of pixels and/or textures processed in a 3D engine to its "bitness". In classic terms, the "bitness" of a graphics accelerator is how many data bits are output on the memory bus. So, GeForce should be called a 128-bit accelerator when it uses single data rate memories, and a 256-bit accelerator when it uses double data rate memories. The Savage 4 only supports up to 128-bits of single data rate memories, so it should accurately be called a "128-bit" accelerator.

Voor wie de Unreal Tourney demo nog wil downloaden heeft Voodoo Extreme een lijstje met mirrors bijeen geraapt:

- File Planet - UT Demo Full Version (228) - 53 Mb
- File Planet - UT Demo Linux Server (228) - 49.6 Mb
- File Planet - UT Demo Patch From 3dfx Version (228) - 9.4 Mb
- AVault
- 3Dfiles
- 3DAGA
- Unreal Universe (full version)
- Unreal Universe (patch)

AnandTech heeft een review gepost van de erg brute SuperMicro SC-750A server kast (volgende week hoop ik m'n SC-760A binnen te krijgen ):

When dealing with server cases, noise isn't that big of an issue. Generally speaking, you don't really care whether or not your server is loud since most of the time you will be tossing in some very noisy components into your server or high end workstation that will create quite a bit of noise even if your case doesn't (10,000 RPM hard drives aren't quiet). Instead of making noise a primary consideration, cooling is an even bigger factor with server cases than it is with any other type of case simply because of the type of hardware that will generally find its way in these cases.

On either side of the external 5.25" bays are mounts for 80 mm cooling fans - two per side. The fans simply snap into the including brackets and that assembly then snaps into the case. Three additional 80 mm cooling fan mounts are located at the front of the case. Let's not forget the previously mentioned 92 mm and 40 mm exhaust fans mounted above the power supply and behind the CPU, respectively. Preinstalled in our SC750-A sample were the 92 mm exhaust fan above the PSU, one 80 mm intake fan at the front of the case, and one 80 mm intake fan adjacent to the drive bays.

Wow, errug heftig nieuws bij Omega Hardware. Volgens een deep throat bij AMD hebben zij daar al Athlon's op 1,1GHz lopen en is het 800MHz model reeds in produktie, maar wordt er gewacht op de release van snellere PIII's (eind oktober?). Eat this:

According to sources inside AMD, there are major developments happening, all of which benefit the end user.

The Athlon (k7) is in testing at 900 and 1000mhz, both versions seem ready for production using standard cooling methods! The Athlon 800mhz version is already in production, and is sitting ready to ship as soon as Intel produces a faster chip than is currently available. AMD engineers also overclocked an Athlon 700mhz to 950mhz using standard cooling, and in lab testing commonly hit 1000mhz. Athlon 800mhz cores are overclocking to 1100mhz with a much better than expected success rate.

AMD is also producing copper k6-3 600mhz parts at it's new Dresden Fab, this is higher than copper k6-3 were originally going to go, and with a litle tweaking, AMD expects the copper k6-3 core to hit 800mhz. According to source "The copper Athlon (k7) core will easily make it to 1500mhz or higher using standard cooling methods"

Impressive, I wonder what Intel has up it's sleeve?

There are unconfirmed reports of an "AMD K6-2+". (source would not comment)At this time, it is unknown what this chip is, but it suspected of being a copper K6-2, meant to compete against Intel's Celeron, in the sub $800 market.

Source will remain anonymous, but is an engineer in AMD's core design and testing facility.

Tim Sweeney heeft een paar postings op de Unreal Technology page geplakt, o.a. over problemen met de Unreal Tourney demo:

Direct3D

- You need DirectX7 to use Direct3D. Get it on Microsoft's Download Page.
- Some users of Windows 98 + Sound Blaster live cards + NVidia video cards are finding the demo locks up on exit, sometimes with the music looping. Restarting the computer or killing the TournamentDemo.exe application gets you free.
- In 16-bit color, weapons may interpenetrate walls when you get real close.
- If you have any problems, make sure you have the latest Direct3D drivers for your card.
- I've had a report of very ugly up fonts in the game on a TNT, looking like this. The solution is to go to your D3D options inside the TNT display control panel, and change the value of the "Texel Alignment". A value of 3 apparently fixes it, although you should try other values if you're having similar texture weirdness.

Patch to Upgrade 3dfx Version

- Due to an oversight, the patch does not add the OpenGL or S3 Metal rendering drivers. If you have patched the 3dfx demo and really want to try these drivers, you will need to download the full 55MB demo. You are still best off using Glide for your 3dfx card, or Direct3D for all other cards except S3s.

Networking

- Occasionally when the server switches levels, users with high-ping connections may see the "Connecting..." screen appear twice, and reconnect to the server twice.
- When picking up a weapon, sometimes the animations flicker.
- In 3dfx Glide, if you alt-tab away from Unreal while connected to a server, you will time out and be disconnected. If running a listen server, all other clients will time out.
- We have many reports that internet gameplay performance sucks with all games for users with WebTV installed (particularly the version bundled with the Voodoo3 3500). Uninstall it.

Audio

- A3D support is back. For A3D support to work properly, you need the latest drivers from Aureal.
- EAX support is back.

Cosmetic

- Ping doesn't always align right in the scoreboard.
- Ping doesn't show up in scoreboard if resolution X size is less than 640.

Beta1 is nog maar net gereleased en Bill breekt al weer beloftes over een compleet nieuwe user interface in Windows Millennium. Lijkt erop dat Millennium niet meer dan een veredeld service pack gaat worden. Betanews heeft de info:

Microsoft has sent word that Windows Millennium will no longer include the planned user interface updates as previously announced. Slated to include HTML based shells containing Activity Centers and Desktop version 2, the UI changes will instead only ship with Neptune in 2001. This is a fairly recent decision from the computing giant, despite what other sites have claimed. Millennium was set to include many of the UI enhancements Neptune will contain.

"As Beta 1 approached, the Microsoft Millennium team decided to focus all of their efforts behind delivering a timely release in 2000 with advancements primarily in the areas of 'It Just Works' and Home Networking, but also in Digital Media and Entertainment and improvements in the online experience. All will be delivered within the familiar UI," noted Consumer Windows Division team member David Fuhriman.

The lack of UI enhancements does not mean however, that Microsoft has scrapped the idea of a better out of box experienece. "There will be some incremental improvements in the UI, but nothing major. Millennium will include a variety of enhancements such as a streamlined

start-up process, increased hardware and software support, support for the Easy PC initiative, and System File Protection (SFP) that will provide users with a more reliable operating environment. I understand that these are not specifically UI related but I think you would agree that all contribute to an enhanced OOBE," says Fuhriman. To the standard user, Millennium at this point looks and feels like Windows 2000 and will probably stay that way. It is even possible Millennium may merely act as a service release update for new PCs as Windows 98 SE did, not having a widespread retail launch. For a detailed look at Millennium, read below.

Marco van Zuylen van Beyond3D heeft een demonstratie van S3 in Amsterdam bijgewoond. Hier een hap uit zijn interview met een mannetje van S3:

The first cards will be the Savage2000, based on a 166Mhz core and 183Mhz memory. In the new year, there will be the Savage2000+. With a minimum core speed of 200Mhz. In fact our 3d core can be clocked higher, but we need better memory for it. When we can put 250Mhz memory on the Savage2000+, the core will be clocked at 250Mhz. All this must be in our policy of price/performance. So if the memory prices will drop in the new year (ed note: doesn't look like this is going to happen after the recent increase in memory prices) we even will be making boards with 64MB ram onboard. But the first boards (166/183Mhz) will be with 32MB.

Hmmm, de S2K+ op 250MHz zou een fillrate van 1000Mtexels/sec op kunnen leveren. Ook interessant is Dot Product 3 Bump Mapping support waarover in het interview gesproken wordt. Check 't hier.

Ik lees bij AGN 3D dat Nullsoft een nieuwe versie van WinAMP heeft gereleased. De fixes en nieuwe features in v2.25d:

- Mjuice.com's secure MP3 format now supported
- Winamp is now freeware! Thanks to all who have previously registered, your support is really appreciated!
- Skin selection menu in main menu, browser in preferences
- Builtin visualization settings are now in preferences
- Added streamed-file saving to in_mp3.dll (no shoutcast saving, sorry)
- Added extended-M3U/PLS support (files include title and length info)
- Better multi-byte character set support
- Restructured some internals of Winamp for bitrate reporting for speed
- Added recent stream list to open location box
- Automatic IE integration for SHOUTcast
- Updated AudioSoft plug-in
- Adds option for old-school playlist editor buttons

Je kunt het proggie (1,8Mb) bij AGN Files naar beneden halen.

CGO heeft een review van de gisteren gereleasde Unreal Tournament demo gepost:

The demo version of Unreal Tournament does more than just show the promise of a game that should be here soon—it's quite a bit of fun in it's own right. It's plain that it can't be compared to the Quake 3 Test, as this truly is a polished demo, representative of a final product that should be quite exciting, while Q3Test is merely a technology test and not a slice of the cooked pie, so to speak. The Unreal Tournament demo may surprise even those nay-sayers who aren't interested in "yet another deathmatch game" or don't think a single-player game against bots can be fun. It's proof that, if done right, computer AI opponents and teammates in a shooter can be as compelling as they are in sports games, race games, flight sims, and virtually every other genre. The download is a bit hefty, but anyone the least bit curious owes it to himself or herself to give it a try.