16 pipes for R420 too?

I'm not so sure. 150W seems absurdly high. A 3.4 GHz Prescott has a TDP of 103W (but it has only 125 million transistors, is manufactured in 0.09 and runs at 1.40V vcore).

Any idea how much power you can draw over a 12V cable safely?

 

That number is awfully acurate, the picutures being show around the net does not have the same cooler on it. Two molex connectors should be a obvious clue.

 

Good to know, thanks.

 

Dunno guys where these numbers come from but IMHO there's the WS AGP Pro110 w/ 110W AFAIK.

 

I am not supporting "my favorite", I am calling for people to be more open minded. I'm still waiting for my last apology from you, or you finding anything negative I've said about ATI. I have not "flipped" my argument at all.

 

Actually...
Or you can add/subtract 1,2,3 to/from exponent.

 

The figures (originally from my memory) can be found in the AGP specification documents. For example, for AGP 2.0, it states the following maximum current draws for each voltage supply:

VCC3.3 = 6.0A = 19.8W
VCC5 = 2.0A = 10W
VCC12 = 1.0A = 12W

Total = 41.8W - which doesn't include current draw for the IO system and buffers. AGP Pro connectors use additional voltage pins, of course, to offer that 100+W.

 

That's a somewhat more complex operation than simple shifting, and requires different logic.

 

Different logic, yes. More complex, hardly. If you are going to support all the multiply/divide-by-power-of-2 modes of PS1.4, a 7-bit adder on the exponent of an FP24 number is going to take a lot less added complexity than the barrel-shifter you need for the fixed-point version.

 

Yes. My thought was 'According to who????'

 

new r420 clockspeed rumor?

http://www.darkcrow.co.kr/

XT@600/600
Pro@500/500

 

Incorrect.

 

always destroying my dreams

 

500/500 and especially 600/600....would Seriously be unbalanced in the memory bandwidth department.

I suspect though, that the memory clocks themselves are close to being right...but I doubt the core clocks are that high. I have no idea myself, but my guess is more along the lines of:

Pro (12 pipe): 450/500
XT (16 pipe): 450/600

 

prove it

 

I'd be skeptical of those core speed with a 16 pipe layout. Highly doubtful.

 

600 is probably pushing it, but I think 500 or maybe even 550 is reasonable on a 16pipe .13u low-k chip considering what ATI has done in the past wrt to the 8pipe .15u R3x0.

 

Is it me or darkcrow is down?

 

down

 

Down...