| 01.11 - T&L Investigated |
| Beyond made a very in-depth article about Hardware Transformation & Lightning. Besides of theoretical background they made an interview with 3dfx´s Scott Sellers and NVIDIA´s Nick Triantos who give some different answers on the same questions: T&L Investigated |
| 01.11 - OT: Gamers Gathering |
Off Topic: There´s a hughe
LAN-Party going on in Duisburg (Germany) in December:
The Power-Central in Duisburg is one of the greatest event-places in Europe. It is the location for one of the largest european LAN-Parties in this century. More than 5500qm give room for more than 1600 gamers. There a great tables, comfortable chairs and some rest places for the comfort of the players. A fully switched network allows to play on more than 25 High-Speed games servers..... Gamers Gathering, from 10.12.1999 to 12.12.1999 - Gibb you there! |
| 01.11 - November |
| Unbelievable... it´s November again....... *brrrrrrr* |
| 31.10 - Unreal Tournament Benchmarks |
There is rumor in the web that
Unreal Tournament runs faster with a Glide emulation than using D3D. Time to check that
out. I made some benchmarks with a Creative Graphics Blaster TNT2U using D3D and the
creative Glide wrapper Unified, a V3 3000 and a GeForce DDR (PIII 500, 128MB RAM):
So wake up! I was a bit surprised that the TNT2 is faster than a V3 using Glide. In Unreal it was way ahead of D3D cards. But as you can see on the results of the GeForce: graphic card performance is not the point here. It seems the CPU is the 'problem' (or still a bug in UT timedemo code...??) |
| 31.10 - Quake3 Tweak Guide |
| Mike of nV News made a very complete Quake3 (Q3Test) Tweak Guide: Tweaking Quake 3 on a GeForce (also of interrest for normal human with TNT(2) cards <g>) |
| 30.10 - GeForce comment |
| Raven Coder Jake Simpson made
some comments about GeForce after some coding experiments: Ok - time for some actual figures on the GEForce thingi now that I've had some time to play with it. I was doing some GL stuff with something I've been playing with at home, and since we have one of these lying around, I thought I'd *cough* slip my code on it *cough* and see what happened. So, in 1600x1200 mode - lets go worst case here - 32bit color, with a TNT2, and a 20,000 poly model, I get 9fps. Not too shabby when you think about it. With a GEForce256, same set up, I get 21fps. Some background for the programmers amongst us. Thats not doing anything special with the GL code. No clever extensions, except for glLockArray. I'm using vertex arrays for Normals, texture coordinates and the vertices them selves, locking the array and then using glDrawElements to wack them out. I AM using GL to do all the lighting, and there are two world lights out there. So we are seeing a 100% plus increase in speed here. BUT. And this is a big But. When I use a smaller poly count model, say 6000, and I don't get half the speed increase. Its goes from 30-31fps with a TNT2 to 35-38fps on the GEForce256. So what's going on here? Well it looks somewhat to me that the larger the vertex arrays are in the first place, the more the hardware T&L engine can do its thing efficiently. So the conclusion is that if you have lots of small models, the GEForce may not give you quite as much of a boost as you might want it to. Remember, this is not a game I'm testing here, it's just some results I got with what I was doing with GL. Your results may vary. .... Remember also that there is no specialized code in my tests. There are a ton of new extensions for GL detailed on the nVidia website, and I didn't use any of them. Using some of those may/should get you a ton more speed boosts from the card. I think the main point I'm making here is that the GEForce WILL give you some benefit on existing OpenGL games, just not as much as we might have all thought. However, with the right approach, both coding and design wise, the card can pay off in spades. .plan File - ThanX to Bluesnews! |
| 30.10 - Voodoo3 Tweaker |
| A new version of this V3 Tweak
tools is available: - Added the Refresh Rate
tab. You can now change the refresh rate ! Download: Voodoo3 Tweaker |
| 30.10 - AGP Win2000 Problems |
| RIVA 3D reports of an not existing texture management in NVIDIA Win2000 drivers. NVIDIA explains the these drivers have not been optimized yet (for an not optimized OS <g>): Report |
| 30.10 - New Xentor drivers |
| Guillemot released new drivers for the
Maxi Gamer Xentor line based on NVIDIA Reference drivers v2.40: |
| 30.10 - ELSA Bios Part II |
| I saked ELSA for the cards
that can use the new released Erazor Bios. Here it
comes: ELSA ERAZOR III-16
(IMAGES\E31xxxxx.a2n) (Not all versions are available!) * Update 02.11.99 |
| 29.10 - Driver comparsion |
| Which driver is best for your
TNT2 card? Release 3 (v2.xx) or Release 4 (v3.xx) drivers? I ran some benchmarks with a Guillemot Maxi Gamer Xentor32 (175/183) on a P III 500 System (Tekram P6BX-A, 128MB)
The performance of v3.53 drivers is a bit slower than the v2.08 drivers. But the difference is only very small. Some applications have problems with v3.53 on TNT2. Shogo for example no longer shows fog with v3.53 on TNT2. This problem does not occur with GeForce on v3.53 drivers! In the following a 3DMark99MAX result table that shows v2.08 compared to v3.53 drivers:
|
| 29.10 - Guillemot buys Hercules |
| Guillemot bought the graphics card
pioneer Herkules which had some major financial problems in the last months for a price of
1,5Mio US$. Herkules was in financial troubles. The 98 sales were about 20Mio US$. Guillmot bought all rights on names, patents, domains, distribution, stocks and customer deals. The support of current Herkules products now is secured through the Guillemot takeover. Guillemot plans to include Hercules products in the current Guillemot product line but there are no details yet how this will look like. |
| 29.10 - SS7 Problems |
| During benchmark tests I´ve found some compatibilty problems of GeForce on Super Socket 7 mainboards. The system freezes after some time. A Guillemot Xentor32 with TNT2 Ultra at 175/183 runs pretty stable in this system using VIA MVP3 chipset. So SS7 owners that who to buy a GeForce should be carefully.... |
| 29.10 - Monster Shades |
| The new S3/Diamond 3D Shutter glasses are based on
Metabyte´s Wicked 3D: Launched in Q4, 1998, Wicked3D(TM) eyeSCREAM 3D stereoscopic eyewear was a hot selling product last Christmas. It provided immediate support for over one hundred and fifty games. Wicked3D now focuses on licensing its software technology. "Based on Metabyte's technology, Diamond Monster Shades is the ultimate gaming accessory for S3's Viper II 3D graphics accelerator and Monster Sound MX400 3D audio accelerator," said Michael Buchanan, Senior Director of Desktop Multimedia Products for S3 Incorporated. "Wicked3D's stereoscopic technology is currently the only solution for 3D gaming that supports hundreds of PC titles designed using the DirectX, OpenGL and Glide application programming interfaces (APIs) and is also compatible with the leading 3D game engines, said Vallal Jothilingam, Product and Marketing Manager for Wicked3D"... more S3 told me that it´s not sure now if they will ship the Monster Shades in Europe. |
| 29.10 - Erazor X² |
| ELSA announced a new GeForce cars using DDRate RAM: (I hate the name
DDR.... these letters stand for the old East-Germany communistic regime) Aachen, October 27, 1999—The GeForce 256 graphics chip of the ERAZOR X and ERAZOR X² features the innovative transform & lighting engine. This reduces the load on the computer's CPU and allows 3D games scenery even with over 50,000 polygons to be displayed in a quality not seen until now. The processor's QuadEngine with four parallel rendering pipelines provides lightning-fast on-screen imagery. The ERAZOR X² from ELSA reaches still higher frame rates with 32-bit color thanks to the Double Data Rate memory (DDR) that unleashes pace-setting performance from the GeForce 256...... more The Erazor X will cost about DM 599,- and the Erazor X² DM 799,- |
| 29.10 - ELSA Bios |
| ELSA released new Bios versions for ELSA Erazor III (maybe also for Pro??) and Synergy II: Erazor III Bios / Synergy II Bios |
| 29.10 - Voodoo3 drivers |
| 3Dfx released new drivers for
Voodoo3 2000 and 3000: What's new in
1.03.00: |
| 28.10 - 3Dfx Napalm |
| Avault
reports, that 3Dfx will give out some details of the Voodoo3 successor Napalm at
Comdex on 15.11: 3dfx is going to drop some Napalm at Comdex 99. The company has sent out invitations to its Comdex event in Las Vegas that indicate it will be revealing details about its next generation 3D graphics card at the computer trade show. The event is Monday, Nov. 15 at 8:00 a.m. at Madame Toussaud's Wax Museum at The Venetian. Expect more details at that time on the successor to the Voodoo3 line, anticipated at this time to appear around March 2000. |
| 28.10 - T-Buffer in Detail |
| Reverend The Pulpit got
3Dfx´s Gary Tarolli (Chief Scientist and Chief Technology Officer of 3dfx) to explain the
T-Buffer technology easily: The visual distractions caused by aliasing artifacts not only look bad, but they ruin your overall game experience. When you are immersed in a game you are in a state of willing suspension of disbelief. You know the images on the screen are not real, yet you have temporarily suspended that disbelief, and are immersed in the virtual reality. Distractions, such as a phone ringing, break you out of the suspension of disbelief. Graphics artifacts, such as aliasing in any form also jar you from this state of mind. The T-Buffer is designed to eliminate these artifacts and preserve the suspension of disbelief. It is just the first of a series of new graphics innovations that will be coming from 3dfx in the coming years. The article can be found here. |
| 28.10 - Savage2000 News |
| Thresh´s FiringSquad ran some
first benchmarks of S3´s Savage 2000: The Savage2000 numbers are a bit slower than the SDR GeForce 256 numbers, but keep in mind that the Savage card is underclocked and running on beta drivers. To quote one of the S3 guys, "we get new drivers every day." If the 125MHz card can put up these kinds of numbers, we have to wonder how the finished card is going to perform. Savage2000.com got some technical details of the Savage 2000 Chip: We have confirmed with several S3 employees that the FINAL clock speed of Savage2000 will be 125MHz core and 155MHz memory. This configuration will result in fill rate of 500 MT/s. The original specs stated 700 MT/s. We also have preview benchmarks results on hand from P3 600 (133) and K7 600. Stay tuned for more information and side by side comparison to GeForce 256 DDR. |
| 28.10 - GeForce compared to TNT2 |
| Sharyextreme made an in depth
article of the GeForce compared to TNT2: Unquestionably, GeForce is the current fillrate 'king'. GeForce boards equipped with single data rate SDRAM are showing high resolution 16-bit fillrates that are 1.65 to 1.75 times, and 32-bit fillrates of 1.3 to 1.5 times that of the TNT2 at the same bandwidth. Quake III Arena Test ver 1.08, which uses hardware transformation, shows the greatest benefit with a 16-bit fillrate that is 1.75 times that of TNT2. At high resolution and 32-bit color depth, the T & L advantage evaporates because of bandwidth constraints. At 183 MHz memory clock or 2.9 gigabytes of bandwidth, GeForce enables 'reasonable' frame rates (30 plus to 60 frames per second) in current games up to a resolution of 1600x1200x16 and 1024x768x32. GeForce should not be judged purely from a performance standpoint. Its implementation of mip-mapping and trilinear filter is superior to TNT2. Read the whole stuff here: NVIDIA GeForce in depth |
News archives - more October news
Copyright: RIVA Station 1999 - Lars Weinand
No Copy without Permission!
