the X2 seems like an inch or two longer than the R600, so i guess it's fine with my casing.

After seeing the Green's new nuclear reactor which is under NDA. Probably yes.

Or maybe i'll just save my moolah and get my EOS5D with 580EX flash gun, Grid mesh, Battery Grip and 24-105L Walkaround special.

This post has been edited by empire23: Nov 22 2007, 07:28 PM

Really nice scores from the 3870. Looks like their driver optimizations are late, but coming.

Question is though, are those tests done with AA? Because IIRC the 2900 has somewhat broken AA units and does it via Shader ALU instead.

Anyways, Powercolor has allocated 30 3870s for Malaysia. 30. Three Zero. Not Thirty thousand. AMD lies about shitty yields and supply issues.

Atleast they're better than HIS. HIS only allocated 20

Yeah lol, the date is about 1 month plus back, so new drivers might have changed alot...or maybe a little.

I'm not too sure about the BIOS, but 7.12 is internal beta with new provisions for the Spider platform and few good improvement across the board, OGL boost reported too. Tested it that day, Overdrive is bugged.

Doing accounts for engineers assignment. I seriously need some fking sleep lah...stupid assignment.

This post has been edited by empire23: Nov 26 2007, 05:12 AM

Next Geforce high end part will not be G92 right now. It will use the G80 IIRC due to the need for high clockspeeds and the supply and clocking issues with the G92.

Hear it's internally called the G80 plus.

The idea isn't the larger amount of transistors or whatnot.

Highend parts have only 2 things in mind, yield, and clockspeed, and if the old process is unable to deliver on both, graphic card makers usually aim for a combinational process.

the G92 made in 90nm is a possibility, smaller processes don't really mean you'll get better clocks or usable chips per wafer. That's why the latest high enders of next gen cards usually use last gen processes because they're proven and easily optimized for.

The R520 used a larger process of 90nm while the RV570 and the resulting performance parts used the 80nm optical shrink.

The R600 used the 80nm TSMCHS while the RV630 used the 65nm and the RV670 the 55nm optical shrink while the 80nm is still being churned out.

The 7800 used the 110nm which was used on the midend of the 6 series.

The same 90nm on the 7900 series was brought forward to the 8800 series.


High end parts usually take a half step when it comes to processes. They prefer old methods because they're perfected and the yields are better, to Nvidia, heat isn't a major primary concern above money, they have a business to run, they need the yields to make sure they keep the green coming in. Just because you can build an engine with space age materials, and you get better performance, doesn't mean you should, especially when money is the core of your business. If 65nm were as great as you'd make it seem, why the shortage? An interim solution would be to use to all production capability available, optimize the existing process and keep pushing cards.

They will move, there's no doubt about that, but in my view, not now.

Just got a 3850 for mah mom's work rig lol.

Abit overkill....but i guess i use it too lol.