The LAST R600 Rumours & Speculation Thread

O crap, you're right.

Why can't they make better fakes? Is it so hard?

 

I suspect that the x2800XTX2 will be GX2 with R580 and GDDR4 according to the spec on the table... also I agree with you that the table smell fishy or probably PR error.

 

Don't know where you get that from. People mentioned here that XTX2 would be probably developed to compete with G80 *refresh*, giving ATI time to get its own R600 refresh out.

 

Or this whole chart is just fake.

 

Were people this gullible in past launches? It seems every new R600 rumour is being immediately gobbled up as fact.

 

I think we're.. you know.. awaiting the return of our Saviour or something... it just HAS to be.

 

*Looks again* Ahh, yes oopsie

 

that should be 1.1Ghz GDDR4 if 7X0/2200 is to be believed.

GPU at 700Mhz then just below 800Mhz??

If the X2800XTX does indeed to 2000 more 3DMark points, then i'm very impressed.

Still, lots of varing results, cores, memory etc. so don't bank on things being real.

That fx57 site seems bull though, the FX55 actually has a higher score than the Core 2 Extreme QX6700? Seems wrong to me.

 

Here come, the INQ updated for the dragonhead2 of R600...
http://theinq.net/default.aspx?article=37588

The card would be about 9.5" and the picture in that post is not the photo

 

shows the same layout as the vr-zone pic.
so obviously that's not bogus.

 

This must be a joke. That picture shows very clearly 8-pin connector, ant that 240w power figure...

 

Wow..the chart is clearly showing some monster cards!

I think the 128 shaders is vec4..

And to have a GX-2 style card already? Yikes! AMD is going balls out.

 

Actully, I think those 128 shaders are scalar.

 

Again, maybe that 8 pin power connector is backwards with a 6 pin cord and that 8 pin connector is there for people that have 8 pin supported PSU's.

 

I hope those are vec 4 else it is not very impressive...

If they are scaler it is basically just twice as wide and half as fast as G80..

 

Which is why I believe in thread packing and clock domains.

 

Yes, that is the most likely scenario, but Inq reported "two 2*3 standard pci-connectors", which isn't true/accurate. They should have said someting like "one 2*3 and one 8-pin connector, which also works as a 6-pin connector".

 

What makes you think that part is fake? Because some source from the zone or Inq thinks these parts are 65nm? Just how reliable has the zone and Inq been lately?

I don't see anything wrong with them being on 80nm. I'm sure their will be 65nm refreshes down the line.

 

...and what would be 6th row from the left for?

 

Those would be the ALU's.