Discussion in 'AMD / ATI' started by jbunch07, Feb 26, 2008.
Found this interesting thought i might share and ask what you think?
quite uninspiring imo, then again so it would seem for nvidias next lineup, i wanna see something thats a generational leap like the 8800GTX was to the 7800/7900GTX
It's going to be uninspiring until DX10 becomes a requirement for games. Currently they want their cards to work with DX10, but are still pushing the performance in DX9. Each generation now is going to be nothing more than improvements on the last, there won't be any big jumps until DX10 matures.
my thoughts exactly, DX10 is a ways of being the norm.
I'm very disappointed in the 9800 GTX, specs wise and it doesn't seem to be a 'leap', not even a fart forward from the 8800. Let's hope the HD4000 get a 'leap'.
keep in mind it hasn't been benched yet, i doubt they would even release it if it was only like 10% better than a G92GTS, it should be decent.....
want one i do, r770 prefer i would, dx10 pointless is, real improvement over 9 it is not, hoping i am that opengl a comback makes.
Hope those prices are true though, <$300.
looks to me like ati is more focused on bang for that buck instead of being the most powerful...at least this is what the article said, im not going to make any speculations until i see some actual benching and real world gaming performance.
but as far as ati being out of the race....not a chance!
@ magnusegallo Yoda rules
Nvidia a joke, look hoow the 9800GTX is just a cheery picked 8800GTS with 3 way SLi.
what ATI knows about gddr5 is 5>4. People like bigger numbers so they'll pay bigger prices. I wonder how much of the" bang for the buck" has to do with keeping there cost down in chip production,can't spend money you don't have.
yeah they couldhave so easily clocked it up butloads at least and given us something awesome.
i reckon 800/2000/2500 easy, and with all the R&D resources they have;
1000/2500/2500 (512-bit) couldhave been possible....
i guess theyre saving it for a rainy day, that rainy day being when ATi steps up and takes them one head to head finally.
ah...its bloced...il check it out later
Power consumption too high for my liking.
What specs are you working off? I only ask because I see varying reports on the specs, for example, one report states that the 9800GTX will be different architecture to the G92, it will use the 55nm fabrication proces, it will have twice the shader processors of the current 8800GTS, it will have 32 ROP's (as opposed to 16), it will have 50% more TMU's, have a 512bit memory bus and have more than double the transistors.
Whether that happens or not, your guess is as good as mine, I suppose my point being, I am not convinced either way about this card and wont be until someone gets a pre release sample and put's it through it's paces but if it works out as some people think as a G92 8800GTS + 20% I too will be very disappointed, when I get home I will post the link in here to the predicted specs for the 9800GTX G100.
GPU-Z already reports its specs, take a look at this:
Here's the validation page: http://www.techpowerup.com/gpuz/mr6hu/
Where did that come from? I wont beleive it that until I see it, thats so bad, if it really is the case then someone needs to re-direct the worlds intercontinental ballistics missiles in the direction of NVidia's headquarters.
I know Tatty, it kinda spoilt my day. But GPU-Z is something worth trusting, the website maintains that they used a "special version" of GPU-Z, whatever that means. If it means that there's just a database entry and that 9800 GTX is based on G92 now and a later version could implement G100, that would be a good theory, just like 8800 GTS making a big leap from G80 to G92.
From what i've read, all those hd4000 specs are possibly fake. which is possible, they read like they just doubled up the rv6xx specs or something. I almost expect something totally new from ATI, probably something that simplifies their current VLIW archetichure.
That a real fake, much like the specs that were going around with 9800GTX with 1GB and 512bit bus and 256 shaders etc... the 4000 Series probably wont be launched until late this year if not the next year, i dont see a reason for ATI to release a new card, when they got the whole segment covered already from high-end to low end, with GPUs that compete directly with nvidias', and whats Nvidia's doing with the 9 series? it exactly what ATI did with their 3000 series, update the GPU and tack on a 256-bit bus instead of the G80's 320/384bit.
I find Nvidia 9 series highly disappointing(besides the 9600GT)i saw the benchmark 14k what a load of junk the "high-end" 9800GTX is my 8800GS scores 12k in 3dmark06...
This is an interesting find, but I have a question: According to the spec the ROP number is still 16 for the RV770. Why has AMD decided to stop increasing the number?
good question i was kinda wondering the same thing myself, of course theres not really anyway to tell the real numbers and specs of these cards until they actually come out.
A good question seeing as the amount of ROP's is the key for me, arguably more especially with ATI's current architecture, with so many shader processors only completeing a single calculation per pass. So many people look at performance purely relating to SP's when ROP's and TMU's are equally important.
i guess we will just have to wait and see what the testings shows us, but as of now i dont think ati has much to worry about especially when nvidia is just re releasing the g92 and calling it a 9800gtx, of course there are roomers or 9900 series... so if you ask me its just to early to tell who is going to win this war...but yes i would think that for that many shaders you would want more than just 16 rops.... but i guess thats how ati rolls
I've been wondering that ever since HD2k series.
Here Nvidia upped theirs to 24 and have been cruising with their performance 7800GTX - present.
Separate names with a comma.