The best video card,IMHO
Home | Reviews and Features | Special Reports | Forums |

Page 1 of 3 123 LastLast
Results 1 to 15 of 35

Thread: The best video card,IMHO

  1. #1
    Deanril Guest

    The best video card,IMHO

    Today I read this article on the 3dfx v6 playing quake 3 at 2048X1536X16 ,kindof amazing if you think about it,but its not this card I love,its my GeForce DDR.

    I was wondering how my GeForce would do at my monitors max resolution(1600x1200)so I said WTF,usually I play at 1156x864.And I play high quality,32bit,and 32bit textures and geometry at full.

    So to my amazement ,it not only played,but it played extremely well at that resolution,when i would get into a fight,with 2 or more it would drop to 25 fps,and that was pretty much the bottom ,most of the time my FPS were around 40-50 fps=very playable.


    Now waht I am amazed about is,when I first bought it I tryed 1600x1200 which was a joke,but the drivers have matured SO much along with texture compression,that 1600x1200 is now very achievable.


    And my GOD does it ever look pretty,Im thinking if I had a monitor that would go up to 2048X1536X16 I could play aswell,im doing REAL well at 1600x1200x32 think about when I switch to 16bit as they did with the v6.


    have a look at 1600x1200 16 bit benches are more than a 50% difference.



    They say in the quick article that the GeForce 2 GTS could not do this,I think they are wrong,and havent even tried this.


    My Geforce is o/ced a bit which gives me the excellant performance,but this card has been good to me,and with the new drivers keeps getting better and better.

    My GeForce is at 150 core and 350 memory and is coupled with a (get this) a K63-460 and 192 meg .


    I wanted to share this so if there are some people out here withe a Geforce DDR you may wish to overclock a little and use the 5.14 drivers that I am using and start playing at 1600x1200 its beautiful!

  2. #2
    Join Date
    Aug 1999
    Location
    New York, NY
    Posts
    5,975
    Yeah, Deanril, I preordered my GF2 last week; hope to get it this week

    [EDIT: oops; forgot to add that I only have a tnt2u card now, so don't think I'm like upgrading from a geforce ddr )

    [This message has been edited by Huge (edited 05-16-2000).]
    Asus P9X79-Deluxe, Intel i7 3930k @ 4.2Ghz, 16gb DDR3-1600 Kingston HyperX Memory, Coolermaster Storm Trooper case, Corsair AX850 psu, Geforce 680 SLI, Corsair Model H100 w/c, Samsung 840 Pro 512gB SSD, Asus Blu-ray Writer 12x, Windows7 Pro x64

    MSI GS Series GS60 Ghost-003 Gaming Notebook 15.6", Intel Core i7-4700HQ 2.40GHz (3.2 GHz), 16GB Memory 1TB HDD 128GB SSD, NVIDIA GeForce GTX 860M 2GB, 4.36 lbs., Windows 8.1 Pro

  3. #3
    Deanril Guest
    Nobody has a comment on Quake 3 at 1600x1200??????

    Hmmmmm,i figure ther would be some doubters,hmmm

  4. #4
    Join Date
    Aug 1999
    Location
    New York, NY
    Posts
    5,975
    60fps baybee...well at least on a p3-900 or so; not gonna get that on this p2-450; waiting for new mobo to arrive.
    Asus P9X79-Deluxe, Intel i7 3930k @ 4.2Ghz, 16gb DDR3-1600 Kingston HyperX Memory, Coolermaster Storm Trooper case, Corsair AX850 psu, Geforce 680 SLI, Corsair Model H100 w/c, Samsung 840 Pro 512gB SSD, Asus Blu-ray Writer 12x, Windows7 Pro x64

    MSI GS Series GS60 Ghost-003 Gaming Notebook 15.6", Intel Core i7-4700HQ 2.40GHz (3.2 GHz), 16GB Memory 1TB HDD 128GB SSD, NVIDIA GeForce GTX 860M 2GB, 4.36 lbs., Windows 8.1 Pro

  5. #5
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936
    Let's do some maths here... 1600x1200 means 1,920,000 pixels. On the other hand, 2048x1536 means 3,145,728 pixels. The fact that it's 16 bit will help with memory bandwidth, no doubt, but it's still a good 64% more pixels to render, and 64% more z-buffer operations. It's also 70% more lines to process for the geometry part.

    So basically it would be like me looking at your 1600x1200x16 numbers and saying "yeah, but even a Viper II can do more fps in 1024x768x32". See what I mean? Unless you're gonna compare at the same resolution, it doesn't mean jack squat.

    We also don't know what level they ran. Demo001 is good and fine, but I can easily think of more taxing levels.

    And finally, they were talking about the game being "very very playable" at that resolution. With a non-overclocked card. Looking at your numbers for the non-overclocked ge-forces in 32 bit, I wouldn't say 21 to 24 fps _average_ qualifies as too playable. With a 21 fps average, you'll get low values down to 15 fps or less in a real frag fest. You can play, no doubt at that value, but you'll be frag bait.

    ------------------
    Moraelin -- the proud member of the Idiots' Guild
    Moraelin -- the proud member of the Idiots' Guild

  6. #6
    Deanril Guest
    Where do you get 21-24 average,mine is a GeForce DDR (the first version)Im thinking mine would do 16 bit,2048x1536 but im KNOWING that the GeForce 2 WILL DO IT.And they just didnt do the comparison.


    Mine at 1600x1200x32 is very playable,it dips to 25 during firefights ect,thats the bottom,and the top of the spectrum is 40-50 is what im seeing during normal 1 on1 type fights.
    1
    600x1200x32 in quake3 with geometry on FULL and 32bit color depth and 32 bit textures ,textures at 3/4 is no-small feat my friend,the drivers along with the s3tc have made this thing like getting a brand new higher performance video card,only I didnt have to fork out $300+ boloneys on it,this simply amazes me!!!I bought it way back in DEC when creative shipped a few out,and I snatched one up cause i had to have it,and 6 months later the thing performes atleast a 1/3 to 1/2 better then release.


  7. #7
    Join Date
    Aug 1999
    Posts
    3,425
    First of all Deanril, 1600x1200@32bit?!!! KICK ***!

    The reason I'm looking forward to the Voodoo5 6000 is to see if it can do 1024x768@32 4X FSAA. If it can do that at nice FPS then put me down for one. The Voodoo5 5500 is fine but I'm not impressed with its performance when you turn on 4X FSAA or even 2X FSAA. Hopefully the Voodoo5 6000 will be able to run 4X FSAA like it is supposed to run, PLAYABLE!

  8. #8
    Deanril Guest
    Loache at 1600x1200 there is no need for FSAA,the jaggies are so small you never see or notice them.In addition everything looks so crisp,think about the rocket trails ect..Very crisp,not to be Biased against 3DFX,but I will take higher resolution over lower res with FSAA anyday of the week,it just looks better to me.......


    And thats exactly what the argumant is going to be here,buy a voodoo5 ?? and play at 800x600 4x FSAA or 1024x768 2x FSAA or buy a GTS 2 and play at 1600x1200 to me I already know the winner,whoops one last option buy a voodoo 5 6000 at $600...............

    [This message has been edited by Deanril (edited 05-18-2000).]

  9. #9
    Deanril Guest
    Hey Moraelin go here think you will see what you seek,it will do it,and thats what they have been demoing the V 5 6000 at NORMAL and 2048x1536..........

  10. #10
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936
    Dunno, Deanril, the numbers were taken from your own cute diagram. If you can't be bothered to scroll up, let me refresh your memory:

    Asus 6800/64 MB ... 24.3
    Suma Platinum 64 MB ... 23.1
    Leadtek Winfast DDR Rev B ... 21.9

    And, no, the GeForce 2 won't get much higher, either, since it just does not have the bandwidth. That giga-texel crap exists only in marketing, not in the actual product sold. Most actual improvement in an actual game I've seen is like 30% in a couple of games, but in the rest it tends to stay somewhere between 5 and 10% actual improvement.

    But let's return to those numbers. Now these are the _average_ numbers, since that's the only thing Q3's demo displays. If you'll remember what average means, it means sometimes you'll get more, but sometimes you'll get less. It's also for a _demo_ _playback_, where the computer doesn't have to bother with AI, path finding, collision checking, and lots of other details. I.e., if you were to do the exact same sequence in an actual game, you'd likely get less.

    So, sure, maybe in _some_ levels you'll get more than that, but in a lot of them you'll get a lot less. And, oh, 1-on-1 would hardly be the normal multi-player situation. Try with 16 people shooting rockets at each other, and then tell me how it feels :P A wild guess is that it will dip a helluva lot lower than 25 fps. (And, no, the maximum FPS doesn't really matter. Only the minimum one.)

    ------------------
    Moraelin -- the proud member of the Idiots' Guild

    [This message has been edited by Moraelin (edited 05-19-2000).]
    Moraelin -- the proud member of the Idiots' Guild

  11. #11
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936
    Also, as a quick reminder: jagged edges are the least of your worries in real life, IMHO. The thing that annoys ME the most are the shimmering moire patterns. And, no, those won't disappear by just increasing the resolution. Plus increasing the resolution has the side effect of making everything look even more polygonal than it already does.

    Me, I'll take a 640x480 with good FSAA any day over an aliased 1280x1024. (Since as we can both see, 1600x1200 isn't really an option with those frame rates.) Whether or not YOU prefer otherwise, though, is of course your own decision.

    ------------------
    Moraelin -- the proud member of the Idiots' Guild
    Moraelin -- the proud member of the Idiots' Guild

  12. #12
    Join Date
    Nov 1999
    Location
    Oregon
    Posts
    1,990
    I want to make only one point :
    The gigatexel fillrate is not totally marketing hype. On this page, Anandtech ran a 3DMark2000 test where the GF2 achieves 1.2 GTS at 1024x768. This is 25% below their touted 1.6 GTS but is still pretty damn fast.

  13. #13
    Join Date
    Aug 1999
    Location
    New York, NY
    Posts
    5,975
    Wow, you guys are really going at it!

    While the v5-6000 is nice, 2 points bug me

    - pricetag: $600 (ouch)
    - who knows when they'll be available.

    I'm picking up a GF2 to replace my old tnt2u. If the 6000 comes out and blows away the world, then I'll consider getting one. I want real world tests/benchmarks to go through first; not this preview crap.
    Asus P9X79-Deluxe, Intel i7 3930k @ 4.2Ghz, 16gb DDR3-1600 Kingston HyperX Memory, Coolermaster Storm Trooper case, Corsair AX850 psu, Geforce 680 SLI, Corsair Model H100 w/c, Samsung 840 Pro 512gB SSD, Asus Blu-ray Writer 12x, Windows7 Pro x64

    MSI GS Series GS60 Ghost-003 Gaming Notebook 15.6", Intel Core i7-4700HQ 2.40GHz (3.2 GHz), 16GB Memory 1TB HDD 128GB SSD, NVIDIA GeForce GTX 860M 2GB, 4.36 lbs., Windows 8.1 Pro

  14. #14
    Join Date
    Jun 1999
    Location
    sf,ca, usa
    Posts
    2,109
    The Matrox G450 might just be half that price.
    Intel Duron 866 mhz
    Abit KT7 Raid
    TNT Ultra 64 mb
    64 mb Ram
    WD 4.3 gb HD
    Logitech Explorer
    19" Sony TFT display
    56K Cable Modem
    Abit Floppy
    Iomega Keyboard
    [b] Meet the ladies
    12X CDROM
    14.4 DVD

    Windows 95 ME

  15. #15
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936
    Heh... lemme get this straight... The same 3DMark2000 that nVidia *****ed and moaned to high heaven about? You know, when it showed that the CPU can do T&L a helluva lot faster than their "GPU", in the high polygon count test?

    Either way, if you understand how that multi-texturing test works, you'll also understand why the GeForce 2 doesn't perform like that in Real Life games.

    That is a "synthetic test". The gaming tests were the first two ones: the helicopter and the adventure. The multi-texturing test is a different beast: it's a test that's supposed to stress the chip's capabilities, without stressing the memory bandwidth too much.

    So, yes, I have no doubt that the NV15 chip itself could attain 1.5 giga-texel if it was coupled with a memory that has zero latency and infinite bandwidth. That is good and fine, but that is NOT what it has on the board in practice. And not surprisingly, that is NOT what happens in a real game.

    But let's go back to that test, and see why I say it doesn't stress the memory too much. That test has a very small number of textures. Four, to be precise. Stretched across only 2 triangles. (Or 8 triangles in the single texturing mode.) Also quoting from the readme noone reads: "The Fill Rate tests use Z-buffer, no mipmaps, bilinear filtering." The readme doesn't say anything about the size of those bitmaps but they look a bit stretched to me, too. I.e., at a wild guess, one texel is stretched across at least two pixels wide and high. So, guess what? The texture memory used is very low, and the accesses to texture memory are a helluva lot lower than in a real game.

    ------------------
    Moraelin -- the proud member of the Idiots' Guild
    Moraelin -- the proud member of the Idiots' Guild

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •