From: GP <[EMAIL PROTECTED]> Buat yang males ngebrowse, ini artikel bagus diskusi dari www.gamepc.com GamePC Smackdown! : Nvidia vs. ATI 9/7/2000 After all the debates, reviews and rants, the dust seems to have settled on 3dfx and nVidia. The former "fill-rate is king" proclaimers were beat at their own game, relying on FSAA and the better image quality it produces to win the marketing war against nVidia. While nVidia seemingly winning the war with 3dfx, ATI, leaders of integrated video and OEM agreements, unveiled its Radeon product line and proved to be a worthy competitor to the GeForce2 GTS. ATI and nVidia are both at a crossroads in their companies. Both are supplying video technology to upcoming consoles, nVidia to the Microsoft X-box and Indremia and ATI, via its acquisition of ArtX, is part of the Nintendo GameCube (formerly Dolphin). However, both companies are coming at this battle from two sides. ATI is currently the undisputed king of economic video systems, consistently has their chips in most sub-$1000 computers and is the model for the "vertically structured" video card company. The fact that you can redily find video cards and systems based on the Rage II architecture is astounding, since it is over 3 years old. nVidia, on the other hand, is the current reigning consumer high-end king focussing almost totally on chip development while working with many third-party manufacturers to take their product to market. Both companies want a piece of the other's action. nVidia is working on getting the new GeForce2 MX series of chips into laptops and more economic systems. While, ATI's response with the Radeon is a very determined move by the company to get into the high-end consumer scene. GamePC once again lets their editors Todd and Chris go head-to-head to debate what may become one of the better rivalries. Todd : Looks like nVidia's got another serious competitor at the top. However, any fears that ATI's Radeon will topple nVidia's crown were crushed with the release of the Detonator3 drivers and the announcement of the GeForce2 Ultra. Speed kills and nVidia has proven that over and over. To their credit, ATI has put out a damn fine video card that compares well to the the GeForce2 GTS, but nVidia always something up their sleeve. It will be interesting to see what happens when their next chipset hits the market. There is already a significant performance lead over the former "fill-rate kings" (3dfx) and the Radeon measured up to the GeForce2 GTS until the Detonator3 drivers hit our Inbox. Chris : Nvidia's brute-strength approach to the video card industry can't last forever, as is very prevalent with their new GeForce2 Ultra card. Faster core, faster memory. Unfortunately, this memory is quite hard for a board maker to get their hands on, no matter how fast Nvidia's clock speeds are, they're always going to be limited by the top (and most expensive) memory technologies. ATI on the other hand, has taken a slightly different approach, adding a form of tile-based rendering to the mix that they call "HyperZ". HyperZ basically picks and chooses data that is going to be displayed, instead of drawing the entire screen at once (even drawing items you can't see because they're behind other things, etc). This doesn't necessarily eliminate the memory card bottleneck from ATI's Radeon line, but certainly puts a Band-Aid on the problem for now. Full tile-based rendering solutions are the way of the future, as the PowerVR "Kryo" card dictates. Even with slower memory and a slower clock, the Kryo can stand up with Nvidia's GeForce lines, simply by using efficient memory techniques. ATI's not at a full-tile based rendering stage yet, but they're already on the right track. Todd : Point taken, memory BW is a problem with nVidia's cards, one we've seen since the TNT2 Ultra. But you really have to tip your hat to nVidia. They showed that whoever they've hired to do their driver development is pretty darn good. The performance increase from their Detonator3 set really pushed their performance over the top. The GeForce2 Ultra was a minor disappointment, since we are waiting for NV20 (or whatever it's called), but its performance will keep nVidia on top for the next few months. Plus nVidia is so aggresive with their development, I have no doubt we will continue to see products that will keep us asking the question, "will they ever stop"? Chris : I've got to admit, Nvidia is kicking major gludius maximus in the driver department. Whether it's the employees they got from SGI or Matrox or something, Nvidia's drivers (especially OpenGL) have been getting released on a frenetic pace lately. ATI has never been strong in driver development, I'll give you that. Let's hope they can step up the pace for their Radeon-line of cards. As for the GeForce2 Ultra, sure it looks fast and what not, but there is really nothing innovative about this card. In fact, if you've got a quality-enough GeForce2, you can overclock it to these levels, and badabing badaboom, you've got a GeForce2 Ultra. ATI's followup to the Radeon looks to be the Radeon MAXX, which will effectively double the fill rate of the card with the addition of another Radeon graphics processor. While adding the "onboard SLI" features really hasn't worked for 3dfx and ATI in the past, when used with decent drivers, ATI's Rage Fury MAXX was quite a competitor. Hopefully ATI can get Windows 2000 drivers out and working right out of the box with the Radeon MAXX. If implemented correctly, the Radeon MAXX could be right up there with the GeForce2 Ultra and Voodoo5 6000 cards. Todd : That's a big if. We still have yet to hear of the actual shipping dates of the V5 6000 or if product outside the few demo boards exists, while a few other sites have posted numbers for the GF2 Ultra (yes, it is a speed upgrade, but it's still fast). Now add to the fact that ATI's Rage Fury MAXX has only Win98 support, and we have to question the way they impliment their dual chip cards already. Plus, ATI has said they want to release a product every 6 months. This is a major change in their philosophy, but this means we won't see a Radeon MAXX for another couple of months. By then, nVidia's NEXT chip will be well on the way. nVidia has always said they are happy with their chips if one of theirs can best two of the other companies in terms of performance. I see no reason to change this philosophy. ATI's HyperZ does benefit 32-bit performance at high resolutions, but again the last set of Detonator drivers pushed a standard GeForce2 above the Radeon in terms of performance in all areas. I think it will be more interesting to see how a tile-like rendering will improve performance on an nVidia card. One thing to keep in mind, other than the HyperZ, ATI's feature "checklist" looks exactly like 3dfx and nVidia's, which means they have mostly been catching up at this point. S3 took a chance on S3TC (texture compression). Matrox took a chance with environmental mapped bump mapping. nVidia took a chance on T&L. 3dfx took a chance on FSAA. There really isn't much risk taking with the HyperZ, it is a must to get the most out of your memory BW. When ATI starts to take chances on features that no one else has, contributing to the innovation in the video card field, they can command a leadership standpoint in the market. Until then, they show they can produce quality video cards, but they are always looking up at someone else. Chris : Hold on here, you think EMBM is a bigger chance in the industry than HyperZ? Sure, they both have stupid names, but HyperZ actually helps any game out there without game-specific coding. ATI is the first major video card manufacturer throwing tile-based rendering into the mix, which is definitely a big step. In my opinion, tile-based architectures are going to be the future of the video card industry. 3dfx is working on one with Gigapixel, Nvidia supposedly has one in the pipeline, and god only knows what Matrox is working on (something that supposedly eliminates the memory bottleneck of video cards altogether), can't you see a trend here? ATI's taken the first baby steps, and it's turned out to be a success. Personally, seeing how close the Geforce2 and Radeon scored, it looks like dual Radeon chips on a MAXX board could theoretically outperform a GeForce2 Ultra. But hey, that's just speculation. Mathematically, the original Radeon is capable of 1.6 gigatexels per second, whereas the GeForce2 Ultra just crashes through the 2 gigatexel per second mark. Two Radeon chips could easily surpass the Ultra, even though there is a small bit of performance overhead due to splitting up data over two chips. Todd : Yes, it was a bigger chance because it REQUIRED specific-coding (until everyone else adopts it). That's why T&L is a chance and why 3dfx's T-buffer is a chance. HyperZ is not much of a chance at all because it doesn't require any coding. I agree tile-based architectures are the future based on the numbers, but no one has been able to do a tile-based renderer and keep on the cutting edge of everything else. Until someone pledges to a full-fledged tile-based architecture then it isn't a chance. HyperZ can always fall back to the standard Z-buffer style of operation. Let me state it again: once ATI comes out with features that push the envelope for game developers and other manufacturers, then they will be a viable top dog in the high-end of consumer video cards. Until then, they are simply a company that puts out competitive products, but doesn't set how high the bar is. They are always chasing, never leading. I've tried to avoid the whole thoughput thingie, but since you brought it up... We've already seen that almost all the latest video cards are CPU limited. The only true benefit we are getting from faster/more video processors/GPUs is faster speeds at high res (like 1280x1024 and higher, which 99% of people don't use) and faster FSAA (which may be appealing depending on how good the implementation is and the amount of compatibility). So for the majority of computer users, a faster CPU will probably help just as much as a faster video card. (shut up Chris, I know you have a gigachip and your OC'd GF2 MX is almost as fast as my GF2 64 on a 750.) When nVidia can hit with something like driver optimizations that increase performance and add features time and time again, they show they are (in both hardware and software) top notch. Chris : *Chris asks himself "Why did I choose do defend ATI, WHY?". Ok, I'll lay it out on the table. Nvidia definitely has the upper hand at this stage of the game, but the video card industry can change in an incredibly short amount of time (i.e 3dfx's fall from grace), and ATI seems to be getting on the right track. 6 month product cycles, high-end based cards, and new technology are paving a way for them to be accepted back into the gaming community. Nvidia's brute force approach can't last forever, and if they don't come out with some new technology (no, NOT a 3rd generation T&L engine), ATI will be right there to pick up the pieces and sell their cards. Now, if ATI would only stop pissing off Steve Jobs, maybe they could keep the Mac market under wraps for the next few years. Todd I totally ignored that *other* segment of the computer market. I guess ATI really is the OEM master. Point taken, but do you really believe that nVidia will keep with the brute force approach? We've heard rumors to the contrary. I guess we'll find out in a few months. 3dfx is definitely the dark horse, depending on what they do with the Gigapixel technology. Unfortunately, ATI and nVidia both have HUGE contracts thanks to their console graphics technologies. Plus, nVidia is working toward motherboard chipsets now. Maybe nVidia has to stop looking at other video companies and take a look up at their next challenge, Intel... -- Garind P =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= "oc ur mobo not urself or anybody else" -- Compu-Mania MailingList, provided by PT Centrin Utama Unsubscribe: [EMAIL PROTECTED], body: unsubscribe Compu-Mania Archive: http://www.mail-archive.com/[email protected]/ Info: [EMAIL PROTECTED], body: help
