Related articles

• The most negative part of Windows XP seems to be the activation of the system, which can only be done on one computer per license. The corporate will probably not have a problem with this; he is buying it for or together with a complete system that probably never will need a reinstallation. The biggest problem is going to be for the enthusiast who might change his system such that a reinstallation will be necessary quite often. It might for him become a hassle to reactivate the license via Microsoft's support all the time. XP has a lot going for it. It's a little pointless to compare the performance of XP to previous versions of Windows. The reality is that every operating system from Microsoft uses more resources and takes up more of your hardware than the one preceding it. We wanted to look at it from the point of view of guys who have to sit in front of the screen 10, 12, sometimes 16 hours a day. So, a new interface is welcomed, if it is good. XP's new look and feel seems to be an improvement. The other really noticeable feature is that XP's remote assistance and access tools are going to directly target freebie applications like VNC, and things like pcAnywhere, or GoToMyPC. Microsoft is integrating features into the XP OS that will, in effect, deliver remote support and access technologies to anyone implementing an XP server. The remote assistance and access tools in XP worked very nicely, and are integrated with Microsoft's Instant Messaging so, they're pretty user friendly. A boon for help desk staff, and maybe a nice way for the average home user to lend a hand to grandma, or that sibling who never seems to be able to configure his PC. Maybe one strong argument for XP is that the sum of the parts is greater than the whole so, it's worth that upgrade price. On the other hand, we're using XP here in the office on a couple of computers, and heck, it's not bad. We'd buy into it. Barring the "anything-but-Microsoft" argument, there isn't a good reason not to take the plunge with XP. Once the full version is out, it'll be time to test out its gaming performance, and also, to look out how the more popular applications will take advantage of XP's features. So, it might be nice to be down on Microsoft because, they are evil capitalists with no soul (or, at least some people say), but that's not a good enough reason to shun XP. It might also be true that the Justice Department didn't do much of anything to thwart the barriers that Microsoft's monopoly puts in front of other PC software developers, but that doesn't really impact the fact that XP does a good job. Tough call.

• So what about that wide-area connectivity thing, Bob O'Donnell? "The problem there is, of course, we're not quite there yet. There's no reason why you couldn't put 3G wireless in this - those chipsets are available, and they're putting them in notebooks. Supposedly, second generation [UMPC] will have that. But the bigger problem is the service provider issue, which is, 'Great, I have the hardware, but you know what? Verizon's going to charge me 60 bucks a month.'" "Here's the challenge," offered Ross Rubin, "and here's why saying the device has potential is not apologizing for it: Let's think about the future of wireless connectivity and the speeds that will be available a year or two from now, and these broadband speeds that we're going to have access to - 2 Mbps, maybe more when you think about WiMax or 4G. What kinds of applications are you going to roll out on a 2 1/2" screen for a smart phone that can take advantage of that kind of thing?" In other words, maybe the next-generation handsets and connectivity devices may be the first to roll out WiMAX or other standards, but they'll also be the ones stuck with the miniature screen. So what services would consumers need wide-area broadband access to all the time? Rubin posed this question at a WiMax conference in 2005, and the answers he got all had to do with "on-demand" functionality - audio and video conferencing, and location-based services. It's at this point where the WiMAX attendees realized, said Rubin, that a smart phone has a little portability problem, where the two-inch UI no longer seems an asset. "UMPC is a compromise between this smart phone platform and the notebook platform," he continued, "with its family heritage being a lot closer to the notebook platform. But the intention is to get it down to a price where it is perhaps competitive with some of these higher-end smart phones, or that it could realistically be subsidized down the road by a carrier. That's why I think, of all the things that this product needs, probably the most important missing link right now is this high-speed wireless." "That would be great," agreed IDC's Bob O'Donnell, "but getting [carriers] to do that is going to be a challenge. My sense is, the service providers have different business models, and different ways of making this happen. Those guys have to wake up and [think of this as a] basic guns-and-butter economy [issue]. 'If I lower the price, I can get a lot more people, that makes up for the fact that I have to lower the price.' But that's not the UMPC guys' problem; that's the 3G network guys, and they have their own sets of issues." In other words, UMPC is going to have to appeal, somehow, to the carriers as a possible solution to a serious problem. But for them to realize this, it will be they - Verizon, Cingular, T-Mobile, and the others - to fall in love with this device. And as we've already figured out, falling in love with the UMPC is a lot more difficult today than many thought it would be last week. "The problem is, unfortunately, classic Microsoft," said O'Donnell. "They've got a good concept, but the first couple of iterations are not cuttin' it...Bottom line, it's a mixed bag. I don't think it's totally out to lunch. I think they're going to have a hard time with the first iterations at the price points they're talking about. Then the question becomes, by not having a great product out of the gate, did they blow their opportunity?" One could very easily answer, "Yes," explained O'Donnell, and make a case in the affirmative that would be very difficult to refute. "The reality is, no one is banging down the door saying, 'I need this product today.' To which I then [would have asked], 'Then why not have waited until you had it totally nailed at the price points you needed to hit, and then introduce it?'" "There's certainly a lot that needs to happen before Origami reaches its full potential, whatever that may be," added Ross Rubin. Even at the low end of the current UMPC price scale - $800 - a purchase will seem out of reach for many consumers, he argued. "Once we get down to $500, once the battery life gets a little better, once we see a few applications that really enable this platform to shine, instead of just repurposing today's notebook applications and, indeed, applications that are non-optimized for a device that lacks a keyboard, then maybe more of the device's potential will be realized." Maybe then, a year or so later down the road at least, we'll be able to answer the less-and-less-burning question, "What am I?" And the answer to that question may be more appealing than just, "I am a product category." "If you're going to launch a category, you need a hit product," stated Bob O'Donnell. "You can't just say, 'Here's a cool category...[now go] find the hit product.'" There are some who would argue that we should give this new product a fair chance in the market. The problem with that argument is, just how much of a product is this UMPC? How well thought-out is it, even as just a framework for a category for a platform for a market segment? Think about this simple question: Today's first-edition UMPCs, including the Samsung Q1, will run Windows XP Tablet Edition 2005; the full-blown Windows Vista that the UMPC framework promises will, of course, only be available this fall at the earliest. How will a customer upgrade? Vista will be supplied on CDs; where will they go? Does a consumer literally have to purchase one of the business premium editions, in order to create an administrator/client relationship and a remote PC uplink between her big PC and her little PC, just to be able to use the CD-ROM drive on the big PC? Will the connection be over a USB cable? How will that be secured? And will the customer need two Vista licenses; one for the big PC, the other for the UMPC? Or does he get any credit for having had to purchase Windows XP first? The fact that nobody with whom we've spoken - not our analysts, not our experts, not the people on the scene - have been able to glean the answer to these questions, is an indicator of a deeper dilemma: The UMPC is not a complete system - in many respects, not a "real PC." And so, a few weeks after that tantalizing question, "What am I?" was first asked, the only answers we can truly give at this point have to do with what the UMPC is not.

• When the new generation of graphic chips started shipping last year, it took quite a while until the first cards with Rendition's Verite 2x00 turned up on the market. 3DLabs and NVidia were a lot more successful in marketing their chips and opposite to these two companies, who got a lot of vendors using their chips, Rendition had first only one, Diamond, and only now the Thriller 3D of Hercules starts shipping. If you listen around large graphic card vendors, you wonder what the future of Rendition looks like, since I hardly came across any vendor that is planning future graphic cards with Rendition chips. This seems pretty sad to me, because the current Verite 2k chips are far from uninteresting and I think that Rendition certainly deserves a lot of respect in the 3D chip market still. Let's hope that they'll get a better reputation soon, because I think that the current products with Rendition chips are far from unattractive and I can take pride in saying that I was one of the first who published that. As mentioned above, there are currently only two mainstream graphic cards with Rendition's Verite 2x00 chips available, Diamond's Stealth II S220 and Hercules Thriller 3D. They don't use the same chip, the Stealth uses the cheaper V2100 with only 170 MHz RAMDAC and Hercules uses the more expensive V2200 with 220 MHz RAMDAC. This is not the only difference between the two cards, because whilst Diamond puts the Stealth II S220 into the low market segment and sells it for practically only $99, the Thriller 3D comes with a lot of fancy additions and is placed into the mid to higher market segment. I don't want to talk about the nice video in/out and the stereo glass feature of Thriller 3D, because I'll do that on the dedicated review page for this card. Rather than that is this article meant to discuss the 3D performance aspects of these two cards. First of all, what's the difference between V2100 and V2200? It's actually less than you might think. Due to the different RAMDAC speeds, the V2200 enables higher refresh rates. This can be a performance advantage in case you have a monitor that supports 160 Hz refresh rate at 640x480, because the 3D performance of the V2x00 is depending on the refresh rate due to the well known 'flip at vsync' issue. This 'flip at vsync' means that the graphic chip may only draw a new page into the screen buffer memory when a whole picture has been displayed on the screen. The lower the refresh rate, the more often the graphic chip has to wait until it can transfer a new picture into the frame buffer and hence the fps go down. This issue is well known from the 3Dfx Voodoo cards, where almost everyone has switched off this feature with the special Glide setting 'SET SST_SWAP_EN_WAIT_ON_VSYNC=0' in the autoexec.bat or in the batch that starts the game. Hercules also allows you to switch off this feature, but the game becomes unplayable by that, which does not happen in case of a Voodoo. The 'flip at vsync' is a specification in Direct3D which was modified by Matrox' Mystique for the first time last year, which upset a lot of people that said Matrox is cheating. Nowadays it's pretty common to switch off this rule as long as you can still play the game decently. Anyway, the performance of a game running under the Rendition V2x00 increases with higher refresh rates and hence the V2200 can have an advantage here, the V2100 is limited to a maximum of 120 Hz at 640x480. The 200 Hz setting on my Thriller 3D doesn't work, although my 21' monitor would be able to display that, 160 Hz works though. The other advantage of a faster RAMDAC is the ability to have a faster 2D engine, just because the memory is not used as long by the RAMDAC and hence free for the graphic chip. There are a little bit more advantages of the Thriller 3D however, its SGRAM can be clocked higher than that of the Stealth II. This is quite surprising, because both are using the same kind of SGRAM, only 10ns one. Since the memory usually runs at double the Verite's clock, it's very surprising that it survives almost 140 MHz. This should normally take 7 or even 6 ns SGRAM. Anyway, my Stealth II could only be clocked up to 62.6 MHz, the Thriller runs up to 69.8 MHz. By default the Stealth II is only running at 40 MHz, which is very little and that's why you can get a significant improvement in 3D performance by overclocking it to at least 60 MHz. The Thriller already starts off with 62.6 MHz and so the increase isn't quite as high as what you can reach with the Stealth II. The interesting thing is in any case, that the Stealth II at 62.6 MHz is just as fast as the Thriller 3D at standard settings. This is so impressive, because the Stealth is about half the price of the Thriller 3D. The most important things you have two know about both cards are that they produce a very good image quality. Also are cards with Rendition's Verite 2x00 chip the only ones that can do anti-aliasing currently. This is mainly a very fancy feature and hardly ever used in any game due to its performance impact, but it shows how seriously Rendition takes the implementation of as many 3D features as possible. A special thing of all Verite chips is that it can be programmed with a special microcode. This enables the Verite to easily understand OpenGL when needed, you just have to load the correct OpenGL microcode. The Verite 2x00 is just as the V1000 using Rendition's RRedline graphic engine, which is supposed to be very powerful as well, however, there are hardly any new games that support this particular engine right now. Another advantage of the V2x00 is the ability to run with up to 8 MB of onboard memory, which enables 3D at a resolution of up to 1024x768 and more. The typical 3D performance of the V2x00 is pretty good in systems with lower CPU 3D performance, but it doesn't scale as well as e.g. the NVidia RIVA 128 chip. This means that it's not the greatest choice for high end systems, although it's definitely scaling better than the Voodoo Rush chip from 3Dfx. Socket 7 systems benefit greatly from it though and that's currently still the majority. Altogether can't you go too wrong getting a Stealth II or a Thriller 3D if you have got a Socket 7 system. You will get good 3D performance at very good image quality. However, the 2D performance of the V2x00 is a different story and I couldn't see any impact of the overclocking on 2D performance at all. Specification Of The Rendition Verite 2x00 This is the specification of the V2200: 55 million pixels per second fill rate (high quality pixels) 2 million triangles per second (edge walk) 550,000 triangles per second (fully setup, high quality triangles) Over 50M WMs 2D acceleration (WinBench'96) 100 MHz SGRAM / SDRAM memory (up to 16 MB) Exceeds Microsoft "Entertainment PC 97" requirements Exceeds Intel "Graphics Controller 97" requirements Complete on-chip triangle setup Asynchronous rendering with seperate setup, triangle and pixel engines Richest set of advanced 3D features True per-pixel perspective correction Sub-pixel / sub-texel positioning accuracy Flat and Gourad shading Z-buffering(optional) Anti-aliasing Comprehensive texture mapping support inc. video textures MIP-mapping Bi-linear and tri-linear filtering Support for texture animation, morphing and other special effects Complete set of blending functions including alpha blending, fog, chroma-keying Outstanding video features 60 fps full-screen MPEG-2 video playback Video input port Digital video output port "SofTV" for flicker-free TV output Complete 2D acceleration support including VESA2.0 and VGA AGP-66 / 33 MHz PCI bus interface Asynchronous DMA support Windows(R)95 (DirectDraw, Direct3D)/Windows NT 4.0 drivers If you want more in depth information about the V2x00 chip from Rendition, why don't you just read what Rendition has to say about their product . For all those of you who went crazy already, because you don't know how to overclock your V2x00 card, here the downloadable V2x00 overclocking program and I can also explain how you can do without it: You need to create a text file called 'verite.ini' and put it into the Windows directory. If you've got the Thriller 3D, it's already in there. It should have the following lines: [display] m=x n=2 SClkP=4 MClkP=2 x is the only variable you want to change, the rest is responsible for the settings of the ratio between chip and memory clock, which is 1:2 by default. x = Chipclock / 14.318 * 8, where 'Chipclock' is the clock frequency in MHz at which you want the chip to run at. Due to the odd 14.318 (typical clock quartz) you'll have to round x up and down a bit. This calculation stuff is all done by the above mentioned Win95 program, so you may as well get it. There is something I almost forgot, because I tend taking this for granted, which is certainly wrong. Overclocking the Verite is something you do at your own risk and there is definitely a variation amongst the cards with V2x00 chips regarding how far you can overclock them. This depends on the used SGRAM chips as well as on the V2x00 itself. My experience also showed that it differs from motherboard to motherboard, which is most likely due to variations in the PCI voltages. I don't think you take a big risk in damaging the card, but it can happen that your system simply freezes and you'll lose your data. Hence I'd recommend overclocking the V2x00 only for playing games, especially since the 2D engine does hardly benefit from the overclocking at all. It also makes sense putting a flat heat sink on the chip and especially using a 3Dfx Cool Fan , which works just as well as for Voodoo cards. Please be careful with the overclocking and increase new clock frequencies slowly step by step when testing it out. For the ones that haven't got the latest miniGL port for the V2k yet, download it here Summary After running all these game benchmarks on all these cards and this on two different systems I feel that I know well, which 3D accelerator I would recommend. Sure, in a few months the Voodoo 2 will rule the world, but what's the beef right now? In the first place you should be clear about your budget. If money isn't of importance, the Quantum3D Obsidian 100SB 4440 is the card you have got to get. This card offers Voodoo2 like performance now and blows away anything else on the 3D game accelerator market. The cheaper version 100SB 4400 does it for people that have some concerns about how much money they spend. It runs Direct3D games exactly as fast as the 4440 and Glide games that don't support the 2 TMUs of the 4440 wil also not be faster on the 4440. Only GLQuake and Quake II can take advantage of the 4440 right now. The architecture behind the Obsidian 100SB is like 2 Monster 3D with 4 MB texture and 4 MB frame buffer combined on one card, doing the now famous SLI. Add another two Texelfx chips with 4 MB each and you've got a 4440. If you've calculated correct you came to 16 MB RAM, 2 PixelFX and 2 TexelFX chips on the 4400 and 24 MB RAM with 2 PixelFX and 4 TexelFX chips on the 4440. Quite impressive, ain't it? A Monster 3D has got one PixelFX, one TexelFX and 4 MB RAM. If you should think that money is certainly an issue, but you still want to get the best possible performance and the best quality as well, then get yourself a add-on card with 3Dfx Voodoo chip. The best bet is currently the Canopus Pure3D or in Europe the Miro Highscore 3D, which are identical. These cards offer 6 MB RAM, 4 MB texture memory and a beautiful video out for NTSC as well as for PAL. Since the above cards are hard to get, you can still go for a Diamond Monster 3D or for one of the other 3Dfx Voodoo clones. Just be particular about the pass-through cable, it should be of high quality, especially if you've got a big monitor and should do some normal work on that monitor as well. Now people that still think that a Voodoo is too expensive or who prefer to have only one card have got the option to decide between NVidia's RIVA 128 or Rendition's Verite 2x00 chip. If you want to play GLQuake or Quake 2 the decision is currently pretty easy, go for a Rendition Verite 2x00 card. If you have faith in NVidia's ability to come up with a really decent OpenGL ICD soon, and if you want the best Direct3D performance currently available, then go for a RIVA card. The RIVA card that I recommend is Canopus' Total3D 128V because of it's nice video in/out feature and the good video capture software that comes with it. However, there's still no AGP version available, which may be an issue to you. So if you really want an AGP card with the RIVA, I'd recommend the Diamond Viper V330, just because I've got more faith into Diamond's drivers than in drivers of STB, Elsa or Asus. If your choice should be the Rendition chip, it really depends on what exactly you expect and how much you want to pay. The difference between the Verite 2100 and the V2200 is only the internal RAM DAC. The V2100 has only got a 170 MHz RAM DAC, the V2200 has got a 220 MHz RAMDAC, which enables higher refresh rates. The Diamond Stealth II S220 comes with the V2100 and it's originally clocked with only 40 MHz. My Stealth II can be overclocked to 62.6 MHz and shows a frame rate increase of over 25% in the tested games. Clocked at this speed it's exactly as fast as the Thriller3D from Hercules, which clocks its V2200 at 62.6 MHz by default. The advantage of the Thriller3D is the better overclockability, mine can be clocked up to 69.8 MHz, better refresh rates and video in/out as well as 3D shutter glasses output. That makes the Thriller3D much fancier. I personally like the 3D glasses output a lot, if you've ever tried these cool H3D glasses with a game that supports it, you won't forget it. However the Thriller3D will be double the price of the Stealth II as it seems. If you are still not sure what you want to buy, you're probably someone who works on his computer professionally as well. For this case you may want to consider cards with the Permedia 2 from 3DLabs. This 3D chip wants some good CPU performance if you plan on playing games with it, but its professional OpenGL support and the excellent 2D performance are something that might be more important to you. As long as you don't care for video in/out as offered by Elsa's Winner/Office 2000, you should really go for the Diamond fire GL 1000 Pro. This card was giving the least trouble of all three Permedia 2 cards that I've tested. It comes with OpenGL support for 95 as well as NT, whilst the Elsa only supports OpenGL under NT and the Fire GL's drivers are offering better performance and image quality. The Hercules Dynamite 3D/GL is mainly good for trouble right now. Repeated driver problems as well as image quality issues wouldn't let me recommend this card at all. So what's with the ATI Rage Pro chip? I can't help it, but this chip lacks in too many cases to be worth a recommendation. It does not support GLQuake's or Quake II's OpenGL engine, it has obvious problems with Direct3D, it has got only very weak support of professional OpenGL under NT, so that it doesn't leave much else than it's excellent 2D performance in combination with its video in/out features. However, if I want the best 2d performer with the best picture quality and the best RAM DAC I rather go for a Matrox Millennium II. The Rage Pro is neither fish nor meat, it's no gamer's card, but it's also not really a professional card either. So what is it?