The ATI R600 Rumours & Speculation Centrum

You call Direct3D10, 2x Radeon X1950 XTX DirectX 9 performance and superior image quality a lemon ?

I wonder why no pictures of R600's PCB have turned up yet...

 

Actually, that was a perfectly ambiguous statement as far as I could tell. He could claim he meant what you said. . . or he could claim he meant that R600 will give G80 owners a feeling that that they bought a lemon and should have waited. . .or he could've meant that R600 is a lemon compared to G80.

But any way you look at it, it's trolling, so warning given, and. . . move along, move along.

 

The deleted post was more interesting.

 

I was staring at the back of R600, and I noticed that they still have room for at least 100 more pins.


So can anyone take a shot at the number of ALU's and ROP's R600 might have?

I assume 96 ALU's, and 24 ROP's. Could be wrong though...

 

I get ya. Thanks.

But seriously isnt there atleast one review site that decided to take the heat spreaders off to take a GPU die shot of the G80? So atleast we could do a comparison between these leaked shots of the R600 with the G80.

 

It would seem we got our selves a XT and a gto.

 

You mean something like this?. So what would the pin count be on this?


Yes, it's a chop chop

 

I beat you R300.

 

Yeah, beat me with my own pic, damn! haha


p.s. What do you do Sound_Card, sleep on these forums 24/7?

 

lol, Im bored right now. I'm in off season training right now, and so I pretty much sit till I have to go to work at Pizza Hut.

 

Did ya guys actually count each individual dot, cause I'm getting 1700, I'm blocking off sections in photoshop with different colors, and x 4 each (quarters), and the orginal pic, the last supposed back side shot is a photochop .

 

I got 1805.

 

that sounds more reasonable, then 2100-2200.

 

Haven't borthered counting them all, but here's how I'd go about it. The center is 20x20 (400). The 4 sides off the center is 18x11 (792). A portion of the outer-ring is 45*3 (540). The corner of the outer-ring is 4x4 (64). Then from there you just have to deal with the odd corner sections of the inner bits.

There is at least 1796 pins (400+792+540+64) with a decent portion still uncounted.

 

Damn it -- they did it finaly.

 

Very Quick Estimate

20 x 20 = 400 -- middle square
11 x 20 x 4 = 880 -- middle outside square areas (not counting triangle edeges)
(10 + 9 + 8 + 7 + 4 + 3 + 2 + 1) x 4 = 176 -- (triangle edges)
5 x 4 = 20 -- extra triangle edge pins

= 1476 (not counting the outermost rows)

25 x 4 = 100 -- first outmost row
28 x 4 = 112 -- second outmost row
23 x 4 = 92 -- third outmost row
41 x 4 = 164 -- fourth and some fifth outside row
42 x 4 = 168 -- last outside row
4 x 5 = 20 -- extra pins

= 2132

some pins counted twice on the outer rows

= 8 x 2 = 16

2132 - 16 = 2116 total pins

Assuming 10% error margin, 1904 - 2328 pins total...

 

I got a pin count of 2140 (after careful recounting) on the original pin shot.

Anyone concur with 2140?

Here is my breakdown

20 x 20 = 400 -- middle square

11 x 18 x 4 = 792 -- middle outside square areas (not counting triangle edeges)

((10 + 9 + 7 + 6 + 4 + 3 + 2) x 8) = 328 -- (triangle edges)
4 x 4 = 16 (extra triangle edge pins, these are between the triangle edges)

16 x 4 = 64 (corners blocks of 16)

((12 x 11) + 3 extra dots) = 135 x 4 = 540 (outside pattern excluding corner blocks)

I get 2140 pins

 

my version



green = 20*20 = 400
blue = 18*11*4= 792
pink = ((41*2)+5)*4 = 348
orange = 47*3*4 = 564
corners = 9*4 = 36
TOTAL = 2140

edit: pink area error (42->41) fixed

 

That's good enough to be true