R600 pictured

[D_o_S]

Pictures of AMD/ATI's R600 chip have started surfacing.

I will let the images do the talking:


View at TechPowerUp Main Site

 

[OnBoard]

GPU pr0n Socket?! Sexy sideways, wonder if it's just visual or actually some reasoning behind it. Atleast now they can't but it upside down anymore

 

[shoman24v]

Please whoop G80

 

[EastCoasthandle]

Trust me a few leaked benches are in your future

 

[tofu]

Nice man, the R600 core is almost as big as the IHS on the socket 775 CPU. Damn is this big.

G80's goin' down bro's. <- ATI party manifesto.

 

[Dippyskoodlez]

Funny.. we beat the INQ with posting this

 

[petepete]

Do you think now all the new cards will be as big as the 8800 ?? cause if its like half an inch smaller it would fit in my case perfect... i have an armor jr. and a 8800 barely fits lol..

 

[Steevo]

I can't wait for THIS card to come out!!!!!!!!

 

[HaZe303]

Sweeeeet... When this baby monster comes out, ill sell my G80 in a heartbeat, and buy one of these...


ATI FTW!!

 

[aximbigfan]

wounder how hot it will run...


chris

 

[XooM]

i thought i heard R600-based cards (x2800s i suppose) would be the same length as ATi's current top-of-the-line offerings.

 

[lemonadesoda]

I agree with Steevo. The R600 pictures look as fake as the Bitchin'fast!3D2000.

Clues?

1./ You don't change pin density 3 times on a chip... its way too difficult to design, manufacture and then locate accurately on the PCB.

2./ The whole concept of ball grid array FBGA or uBGA is a regular array of pins. Just like on any CPU.

3./ Over 2300 pins? Isn't this a bit high for accessing 256 or 512 bit memory?

4./ WolfLSI is a naughty graphics artist http://wolflsi.deviantart.com/

Who got fooled?

Hang your heads in shame!

 

[XooM]

i would love to see somebody good enough to fake that. you can see the damn camera's reflection! also, the layout is reminiscent of the... which was it... either the RSX or the R500. one of them had the chip at 45 degrees with the ram on the same package.

 

[ATIonion]

I agree with Steevo. The R600 pictures look as fake as the Bitchin'fast!3D2000.

Clues?

1./ You don't change pin density 3 times on a chip... its way too difficult to design, manufacture and then locate accurately on the PCB.

2./ There is no need for 45 degree resistors. Creates unnecessary location problems with the robot assembler.

3./ WolfLSI is a naughty graphics artist http://wolflsi.deviantart.com/

Who got fooled?

Hang your heads in shame!

on the last picture with the cpu next to the r600, the painted specs on the ati chip are dif. from the other pics shown...and some resistors are laid out dif. too...why is this?? im looking real hard to find proof that its fake or real....that dif. seems odd to me...

 

[HaZe303]

Its not fake... the pre-release version have been shipped to developers. Maybe pics are from one of them. Or some other version, who cares. But im pretty sure they are real.

And why the two pictures differ, is because all the pictures are not from same site/person/gpu. They are from 2 different places...

 

[lemonadesoda]

If those pictures are for real... then apologies all around. However, does anyone have the following info?

1./ Pin count on an R480 is 1345
2./ Pin count on an R520 is ???

However, pin count for the shown R600 > 2300 pins!!!

But a GDDR4 memory chip is only 136 pins. So 8 memory chips = 1088 pins. So what exactly are the other 1200 pins on the R600 doing?

So either the picture is FAKE, or the R600 is 1GB with 16 x GDDR4, running in a QUAD rather than DUAL set up, to double the memory bandwidth.

 

[Casheti]

LMFAO @ Steevo

 

[tkpenalty]

cool... i reckon ATI/AMD's latest apparation will succeed.... and geez thats small, i reckon they will make it in a real socket. According to the picture its LGA

 

[Dippyskoodlez]

on the last picture with the cpu next to the r600, the painted specs on the ati chip are dif. from the other pics shown...and some resistors are laid out dif. too...why is this?? im looking real hard to find proof that its fake or real....that dif. seems odd to me...

Its far from uncommon to get different resistor layouts in different revisions when its still under heavy development..

Hell, its common to find different layouts on released versions...

and who says a graphic artist can't have insider access to GPU's?

I have inside access to a few certain things(not like its gonna change the world.but still...)... but being what I am shouldn't change that.. credability Now if he made realistic 3d models of prototypes... thats another story..

Besides, you said he was crappy... so how'd he fake such a realistic picture of a chip thats never been produced...? You would need atleast something to model it after...

That, and way too many factors line up... such as rotated PCB for the 512bit memory... unless he's a chip architecht too, it wouldn't have been thought up by an artist... it actually has some engineeringh logic to it..

 

[Dippyskoodlez]

1./ You don't change pin density 3 times on a chip... its way too difficult to design, manufacture and then locate accurately on the PCB.

Too bad its BGA and its all soldered at once

Although PCB shouldnt be a problem... its 12 layer....

 

[ATIonion]

Besides, you said he was crappy...

i said what now??? i was just taking interest in the idea that the pics maybe fake..i never said they were...thus the questions i asked...

 

[lemonadesoda]

HAHAHAHAHA

Just take a look at

Fake shadows.

The chip is "straight on"

Look at the shadow of the top of the chip. It should be parallel with the chip, not sloping up (on the left) and sloping down (on the right).

There is no light source in the world that can do this, other than rendered or fake.

If you don't get it, then cut a postcard square shaped, and use a desk lamp (or two) to replicate those shadows. You'll be up all night!!!

 

[lemonadesoda]

@Xoom
@Haze
@Dippys

 

[ATIonion]

try this then.....if you think this is real....

these red lines should be the same length, but they are not....the text was added...

the text has been added...it is not in the same perspective as the chip...if the text is placed square to the chip then it should follow the same angle as the chip...it looks as if the text doesnt follow the rules of perspective...the first line of text is further back then the last line text, yet it all looks to be on a flat surface...and that pic shows the chip at an angle...


i think i explained that ok...i hope so ...

 

[ATIonion]

this shows what i am saying better...

the green line is straight but the yellow line is angled...all the other pics show the text being square to the chip...if that were true then the text would follow the yellow line instead it's straight.....could be a real chip but i think the text on it maybe fake...