Last message on previous page:
According to THG's new ATI FireGL X1 updated drivers test, the FX2000 remains a significant leader. So yeah, when it comes to NV30 Pro, nVidia does not skimp at all, they DELIVER. Price is the only problem.


--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

I wont get a 9800. I think im done with new cards for at least a year at this point. 9700 pro will last me that long, I'll overclock if if I feel like im lagging behind

thanks again.

Nah, Shay, you missed what I was saying, I mean you can Driver /Software modify you new 9700 and turn it into a 9800-like card. However like I said earlier, wait until it's stable first, then soft mode it to 9800, and THEN overclock it. But for now play with the stable 9700 until you are feeling 'adventureous'. I didn't mean go out an buy another card, just, when you are ready start tweaking it.
Have Fun!

- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <font color=red>RED</font color=red> <font color=green>GREEN</font color=green> GA to SK

I have a stable Radeon 9700 pro. And am not interested in softmooding it. I'll wait for the 3.3 catalyst drivers for performance boost or will overclock my card. The softmod from what I've seen only gives like a 2-3% performance boost.

On another issue. I keep seeing Cebit reports of NV35 being so much faster than NV30. which I think is just more PR BS. its only a refresh of an existing product. I don't really see it being 30-40% faster than Nv30. I could be wrong but I just don't think it will be.

Yeah, you can wait, I was just saying Shay COULD softmod, should he so desire. Because you can softmod and THEN overclock as well. But just throwing that out into the mix.

As for the NV35, I can see how it would increase performance in some areas, it is expect to have 256bit DDRII running at 400-500(800-1000)mhz, vs the current 128bit DDRII, that should give it a significant boost in some areas, and is what alot of us feel the FX should have been in the first place. Anywhooo, we shall see if it's more PR than actual performance.

- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <font color=red>RED</font color=red> <font color=green>GREEN</font color=green> GA to SK

No matter what they do, they need to fix the cooling though.

<font color=red>GOD</font color=red> <font color=blue>BLESS</font color=blue> <font color=red>AMERICA</font color=red>

No doubt they will. So far the majority of the card makers are using their own cooling methods and most seem to be back down to a more silent treatment.

The problem lies not in here, it lies in the fact this is UNNECESSARY cooling.
The FX5600 Ultra, god's dried up fruit, runs at 46ºC, about 10ºC hotter than a FULL LOAD Radeon 9700PRO, according to the results tested on the OCA ultimate edition R9700PRO, when actively cooled by some 120mm fan.
Yet the chip runs at 0.15m.

Then nVidia, according to Ace's, preaches about GPUs using 120W as being inevitable. Forgive the language but what kind of utter bullcrap is this? ATi has their best chips around 60W, and the transition to .13m will ensure a drop to 40W easily. nVidia is only talking to themselves and they continue to shoot themselves in the foot. They have to stop this idiocy, they are only making fools off themselves like AMD is with their constant delays and blaming of the x86-64 technology on Microsoft.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

Some more info about the NV35 shows that it does indeed have 256bit DDR bus, but instead of DDRII, it shows DDRI @ 1GHz.. The info is <A HREF="http://www.nvnews.net/#1048095553" target="_new">here</A>.

One thing to make a point of is that both cards are at the moment downclocked to make the comparison fair, but as you can see, looks like the memory bus really helps out AA/AF performance at least.

...And all the King's horses and all the King's men couldn't put my computer back together again...

Ahhh...I knew the 128-bit memory bus was the bottleneck. It was obvious when you look at the high-quality benchmarks. Now this is very interesting, indeed.

<-----Insert witty sig line here.


<P ID="edit"><FONT SIZE=-1><EM>Edited by Twitch on 03/22/03 03:13 AM.</EM></FONT></P>

But despite all of this, a 500MHZ GPU core could've compensated for the poor AA performance, think about it.
Not only this, but look at the FX5600 Ultra memory bandwidth, which is awefully similar to the R9500PRO. So why does it abso-diddly-lutely suck as well?

The problem lies not only in memory bandwidth but the WHOLE core efficiency. It cannot process lots of raw triangles. And covering up the problem will not help. ATi is on their way for high bandwidth as well.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

there got infos out from nvidia tech dev, stating they realised the core is [-peep-] up. they had to disable with drivers stuff in the core that simply don't work as expected, etc. the core is like the first p4 core. problem is just, nvidia is not intel, they cannot simply spit out yet another core that easily (full redesign, that is).

nv30 is crap. and i personally don't believe in nv35. it looks like just another marketing issue.

"take a look around" - limp bizkit

www.google.com

It's obvious the FX is memory-starved. I don't know if this is a problem with the memory controller, or with drivers or with the core itself, but obviously DDR2 does not compensate for the 256-bit memory interface. That's one of the reasons why the FX always loses as the memory requirements are jacked upward. Higher quality settings require more memory bandwidth.

Note that I did NOT say that FX-256 is going to destroy Radeon. I also did not say that the 256-bit memory was going to fix all of the problems with FX. I said the benchmarks shown and nVidia's changes to FX were "interesting."

I predicted in one of my first posts on this forum that nVidia would be implementing a 256 memory bus on their next "Ultra" card. Probably pretty obvious considering FX's primary performance weakness is obviously memory bandwidth. So, of course, I probably find it more interesting than other people.



<-----Insert witty sig line here.

Oh my, I was not debating you or telling you anything wrongly. I totally agree with you man, however I noticed some oddities in that the core speed itself has problems, aside from the bandwidth (processing only 0.7 triangles per clock is not good).
Anyways I also hope that the NV35 is more than just "fixed". I keep feeling the architecture is there, so, AA and Aniso may still be bad... Who knows, nVidia better have learned from their lesson, as the FX5600 Ultra's release only makes trust worse.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

I noticed that the FX does not even begin to approach its theoretical fill-rate. While the Radeon acheives something like 97% of its theoretical fill, the FX was something like 80%. I don't think memory bandwidth is the cause of that. I initially thought improved drivers might go a long way towards fixing that, but I have yet to see it. I don't really know what the flaw is. If that's what you're referring to, then we do agree.


<-----Insert witty sig line here.

Something tells me that this 500MHZ clock is a fluke, but I could be wrong.
Even nVidia seems to realize something did not make sense with their clock speeds.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.