R520 informanina, part II (E3, etc)

The size of a chip is determined by its package and not by its DIE size.
With increasing pin counts on the package, it will become bigger and bigger, no matter how small the DIE size may get by technical progress.
And also you can't shrink the size of capacitors and coils.

 

Why would anyone assume that adding stuff (pins) makes boards smaller?

 

anyway AFAIK that is not a reference board, it is more likely a test/engineering board, so you would expect it to be on the large side.

NV35

 

Who says the chip's gonna have more pins in the first place? Do we know that for sure?

 

This link was posted in Shacknew's latest comments thread, it's shakey-cam of Prey on R520 I think:

http://dmode.datamachine.net/Preye32k5.wmv

 

Well, it's a possibility, but would depend on implementation going by recent ATI 512MB cards.

In the X850 XTPE Technology Preview the memory chips are distributed evenly on both sides of the PCB, while the more recent, but not as high-end X800 XL 512MB, as shown in Beyond3D's review here has all the memory chips on one side of the PCB only. Unless the R520 PCB also includes another discrete ASIC (Theatre 550?) on the same side as the GPU, then it could just be additional resistors, capacitators, etc and/or alternate GPU orientation to reduce memory trace lengths or similar resulting in the seeming PCB elongation.

Cheers,


BrynS

 

That was kinda cool! The portals and gravity effects were definitely quite interesting. While the gameplay doesn't depart very far from any other shooter, it looks like they're at least doing something interesting with environments.

Also, it seemed alot more like Half-Life than Doom 3, especially in the introductory clip. . .

 

Well, if we're three months out still, are we really confident that's final PCB?

If it is, that might tell us a thing or two about what the holdup is. Tho if they really are releasing new X8xx CrossFire variants at Computex, it could be just as simple as making their nut back on those. I still suspect that the extra time for driver polishing is at least appreciated, if not considered mandatory.

 

There's always the posibility that its some kind of engineering sample because ATI didn't yet want anyone to see a final card. Maybe not as deliberate as misinformation, just making sure that everthing is as locked down as it can be by not bringing the final product to a relatively open show.

 

lol, when I see a 9800pro, I say "this thing is SMALL!"
(and a X800XT is the same I think)

I'm used to voodoo2 (with or without SLI), voodoo5, I remember VLB video cards too.
Even a Trident 512K is bigger than a 9800pro or later ATI card

 

NO... it can't be! We have the same dream?!? What are the chances of that happening?!

I wish someone would make neon colored PCBs...

 

I´d like to comment on this, because i really am quite impressed how far 3DRealms managed to go with this game.

Early in the vid it just looks like another D3-title game, but after about 1 minute into the vid it really looks quite awesome. Not just the graphics but especially the possibilities and how they´ve implemented everything into their maps. 3DRealms always had some of the best ideas in the past (that´s also the reason why almost any shooter they brought out was excellent) and they seem not to have forgotten about this.

Those portals not only are there, because they look good, they also add great to the game´s atmosphere and playability - a design decision that is usually completely taken in other directions nowadays.

I´m REALLY looking forward to this game, i´m still quite excited about what i´ve just seen in this vid. It has some kind of "UNREAL" - feeling to it (first Unreal game) - and seems to let us explore a lot of different places and worlds.

Also, did you see how the map was changed in reverse-direction when that guy hit those "buttons" on the walls ? Frickin´cool


Great stuff !

 

Yes, I would think that the possibility of the card being an engineering sample quite likely, especially if launch is still a few months away, unless ATI really are delaying their schedule for the Crossfire launch and to better align their roadmap with Longhorn, in which case it might be the final PCB. If it is the final PCB, then the additional driver development time until launch in a few months would be beneficial, but the size of the card could be problematic and/or impractical for small-form factor/HTPC/shuttle-type users if it remains unchanged, at least from the impression of the card's size guestimated from the picture Dave posted.:?

However, how much smaller/more compact will ATI be able to shrink the final card shown if it is an engineering sample, which perhaps has extra redundancy built-in, based on past products? Didn't Digitalwander have a 9700Pro engineering sample? Any chance of a pic or two to compare, Digi?

Alternatively, how likely is it that ATI would significantly "pad out" an engineering sample (if indeed the R520 shown was one) just to obsfucate the impressions of observers and competitors? Whereas one could understand Hercules using the much larger GeForce2 PCB for their KYRO2 product, even though a much smaller PCB could have been used, ATI probably have the engineering resources to create custom PCB's for multple products and further refine them as required, but even for ATI, to go through the process of creating an artificially increased PCB for a small-run early sample simply to mislead competitors, seems disproportionate.

Surely, it's more likely that the reason for the size of the PCB shown was to necessitate the current performance targets that ATI are gearing for on initial silicon and as the process matures and new revisions are available, that hopefully a smaller, less complex PCB can be used?

Cheers,


BrynS

 

Well, now that XB360 has "stolen" Ruby 3, the R520's launch demo(which must have been around for at least a month?), perhaps we're to read into this that R520 has been canned and R580 is being brought forward 2 or 3 months...

Are we sure that was R520, and not R580 spotted at E3?...

Jawed

 

Based on that thought, ATI could deliver both of ´em, without making any "losses". Could very well be a possible scenario, but highly speculative (don´t we love that here).

Until now it was believed, that R580 is "just" a more effective (ROPs, not pipes) refresh and they NEED more bandwith to USE that additional power.

Since R520 won´t be out before Q3/05 (end of July - September) they could´ve taken the opportunity that Samsung already has 800MHz GDDR3 (now in mass production) and brought forward R580, which already has taped out about the same time as R520.

They could delay R520 to use that chip for a possible R520 PRO and use R580 for their XT PE part. Maybe they also canned R520 altogether and just went with R580, because manufacturing capacity should be quite enough this round (TSMC has 3 FABs running with 90nm and 300mm wafer) to deliver the same chip to both market segments with disabled parts inside.

Just an extension of your thought, although highly speculative, but interesting none the less.

 

WHY CAN'T THEY JUST SHOW US THE DAMN CARD A DAY AFTER IT'S HINTED IT WILL SOMEDAY EXIST?!?!??!?!??!?!?


....sorry just had to get that out of my system....

 

To build up the anticipation and the drama?

I've been noticing a LOT of negative ATi "rumors" coming out lately, until I hear from ATi I ain't putting a lot of stock in them.

 

Interesting thing about so-called "negative rumors" concerning a company is that we can be fairly certain the company being criticized is not the source of such rumors... (I think that follows logically.)

Therefore, it is likely safe to guess that the source of negative rumor surrounding company X is likely the result of efforts made by certain corporate competitors of company X (since they would be the only logical beneficiaries of the fruit of such rumor.) So then, it is also likely that the competitors of company X may fear the upcoming products of company X, because if they did not then there'd be no reason to start such negative rumor-mongering in the first place, would there?...

The reality of the situation then may be described as the inverse of the rumor, and therefore the presence of such rumor likely proves that the upcoming products of company X will actually be very good. That's the way I've always seen this sort of thing play out, anyway...

 

It is just as likely that the source of negative rumors is the result of Company X trying to correct too much incorrect hype. After all, people have completely unrealistic expectations about what the next-gen parts should be and are therefore woefully disappointed when reality hits.

For example, Dave himself has been the source of some "realistic" info concerning ATI's next parts that some people read as "negative" data. Do you think he is spreading Nvidia-fed info?

 

Well duh! Everyone knows what an nVidia fanboy Wavey is.

























I think it's sort of silly to take ANY info about EITHER company too seriously before we see the actual hardware, and bickering about rumors and who started them seems very non-productive.

Haven't we been down this road enough times, haven't we learned yet?