Anand has the details about r520,rv530,rv515

[Geo]

PE 540x16 = 8640 | Ultra 400x16 = 6400 --> 35% to ATi
XT 600x16 = 9600 | GTX 430x24 = 10320 --> 7.5% to Nvidia

Quite a leap IMHO.

Hence Rys, above, I 'spect. Hope he's right. I certainly don't believe that there will only be a 50mhz difference on the core between XL and XT. . .but then this latest XL leak shows it 50mhz higher than last time; so I can't foreclose the possibility that the greater gap will be caused by XL going down to 500. . . (as previously reported).

650x16 = 10400

But I really do think there's going to significantly more to the calculation this time, and from more than one angle. Interesting days are just ahead!

 

[digitalwanderer]

I think if there is only a 50Mhz difference on the core 'tween the two that 50Mhz will be a bigger difference than it used to.

 

[Geo]

A little sideways, but related. . .ATI's next quarterly conference call is Oct. 6. So are we feeling pretty good (I am) that R520 will be announced/reviewed before that date (even if only one day)?

 

[trinibwoy]

Hence Rys, above, I 'spect. Hope he's right. I certainly don't believe that there will only be a 50mhz difference on the core between XL and XT. . .but then this latest XL leak shows it 50mhz higher than last time; so I can't foreclose the possibility that the greater gap will be caused by XL going down to 500. . . (as previously reported).

650x16 = 10400

But I really do think there's going to significantly more to the calculation this time, and from more than one angle. Interesting days are just ahead!

Even with 650 on the core, it doesnt change my conclusion much. That would give it parity to the GTX in terms of numbers, which means that performance/clock would need to have improved considerably over the R420.

 

[digitalwanderer]

I think the R520 pretty much has to have much better performance-per-clock, the $64,000 question I think is "how much".

 

[Jawed]

I think the R520 pretty much has to have much better performance-per-clock, the $64,000 question I think is "how much".

50% I reckon.

Jawed

 

[Slappi]

Wow I guess you can't be weary of an ATI release of a new card around here.

They paper launched the hell out of everything in the past and now since nVidia actually puts out a card fast I guess ATI is forgiven of all past paper launches.

Going by their past record of launches, ATI should have their highend cards out in volume about 2-3 months after their official launch.

That would put it around the new year.

You bash me for no reason. I am not bashing anyone here. I don't agree with most of you about ATI's upcoming releases but I don't bash you. You call me fan boy, troll and other crap just becasue I don't think ATI will pull it off on a timely basis?!?

That's messed up.

 

[trinibwoy]

You bash me for no reason. I am not bashing anyone here. I don't agree with most of you about ATI's upcoming releases but I don't bash you. You call me fan boy, troll and other crap just becasue I don't think ATI will pull it off on a timely basis?!?

Well that might fly at other forums, but since you've been spewing a lot of nonsense based on practically nothing then you are deserving of all the backlash. Given Nvidia's handling of the GTX launch, ATi would be foolish to not follow suit. That was one of the reasons for the delay, no?

 

[Mark]

A little sideways, but related. . .ATI's next quarterly conference call is Oct. 6. So are we feeling pretty good (I am) that R520 will be announced/reviewed before that date (even if only one day)?

Nah they chose the 6th because it's the day after the official start of the 2005/2006 NHL Hockey season. You know how hockey crazy us canucks are!

 

[digitalwanderer]

50% I reckon.

That would be great, but where you coming up with that number from?

 

[Jawed]

That would be great, but where you coming up with that number from?

ATI's estimates of current GPU utilisation, versus what a more advanced scheduler can do.

Although the comparison is blurred somewhat by the fact it's based on the efficiency of Xenos - where unification of the pipelines also improves utilisation.

So I'm just being optimistically hopeful...

Jawed

 

[digitalwanderer]

Thanks Jawed, that makes sense...I was expecting you to say, "Well, I just pulled it out of me arse.".

Even a 40% gain would be more than enough.

 

[Big Bertha EA]

Wow I guess you can't be weary of an ATI release of a new card around here.

You are getting the same reaction at every ATI fan site. What do you expect, especially without a shred of PROVEN information to base your biased claims?

They paper launched the hell out of everything in the past and now since nVidia actually puts out a card fast I guess ATI is forgiven of all past paper launches.

Huh? nVidia never did a paper launch either? Get your facts straight or stop posting. Very simple...

Going by their past record of launches, ATI should have their highend cards out in volume about 2-3 months after their official launch.

I guess it never occurred to you that possibly one of the reasons they delayed THIS launch is because of past "paper" launches? Nah, that would be too much of a logical and objective viewpoint for you....

That would put it around the new year.

Yawn...

You bash me for no reason. I am not bashing anyone here. I don't agree with most of you about ATI's upcoming releases but I don't bash you. You call me fan boy, troll and other crap just becasue I don't think ATI will pull it off on a timely basis?!?

Bash you for no reason? You REALLY feel like you have done nothing to deserve the response you have received? Nobody here is bashing you as a person. They are bashing your incredibly biased and tainted comments as it doesn't take a genius to figure out what your agenda has been in this thread and other threads across the internet regarding this topic. I actually don't mind negative commentary. However, when that commentary is based upon NO FACTS and you have gone out of your way to spew your bias on most major forums discussing ATI products, then I do mind that and you can expect similar responses from myself and others if you continue. If you don't have anything good to say and nothing to back up your claims, then shut your piehole.

That's messed up.

No, your overtly biased comments are though.....

 

[Geo]

Waaay back in March/April, when we were InfoMania-ing it, a big topic of conversation was how close R520 is to R420 vs R600. The conventional wisdom (I was a little more aggressive) was that it was a relatively minor refinement of R420.

Now we have the ring bus confirmed. So, how's the CW feeling these days about this question? Are we seeing any evidence of other C1/R600-ish concepts/pieces in R520? Does the CW still feel that it would be pointless to even go further down that path without DX10 in place?

 

[Slappi]

You are getting the same reaction at every ATI fan site. What do you expect, especially without a shred of PROVEN information to base your biased claims?



Huh? nVidia never did a paper launch either? Get your facts straight or stop posting. Very simple...



I guess it never occurred to you that possibly one of the reasons they delayed THIS launch is because of past "paper" launches? Nah, that would be too much of a logical and objective viewpoint for you....



Yawn...



Bash you for no reason? You REALLY feel like you have done nothing to deserve the response you have received? Nobody here is bashing you as a person. They are bashing your incredibly biased and tainted comments as it doesn't take a genius to figure out what your agenda has been in this thread and other threads across the internet regarding this topic. I actually don't mind negative commentary. However, when that commentary is based upon NO FACTS and you have gone out of your way to spew your bias on most major forums discussing ATI products, then I do mind that and you can expect similar responses from myself and others if you continue. If you don't have anything good to say and nothing to back up your claims, then shut your piehole.



No, your overtly biased comments are though.....

Fact is that ATI have dropped the ball and they are late. I believe they will paper lauch the XT and XL and deliver on the PRO. ATI isn't going to come out and say it is a paper launch we will just have to realize it over the coming weeks and months as we most likely will see slim availability.


Call me bias all you want I could careless. I have been an ATI customer and I backed them in the past. I used to be an investor in nVidia and they burned me bad so I don't have much love for them at all.

I am pissed at ATI because I want them to produce something decent in comparison to NVDA and it doesn't look to me as that will happen.

 

[demonic]

To be honest, if it was exactly the same fps as the 7800GTX, I wouldnt be complaining at all..

(if HDR+Adaptive AA was enabled)



but i can wish i guess.

I am pissed at ATI because I want them to produce something decent in comparison to NVDA and it doesn't look to me as that will happen.

Quite frankly, you dont know what they have done. Why not give them the benefit of the doubt. Otherwise come launch, and you will look like an nvidiot.

 

[digitalwanderer]

Hey BB! Long time no see, have you gotten to play with an R520 yet? And what other sites has Slappi been popular at lately...

 

[Kaotik]

We all gave up.

I think it's:

• pipes per array
• TMUs per array pipe
• arrays
• ROPs per array pipe

which translates as:

• 4 pipes per array
• 4 TMUs
• 3 arrays (therefore 12 pipes total)
• 8 ROPs

Jawed

So R580, being 16-1-3-1, would be 48 piped solution? Doesn't sound feasible to me

 

[Big Bertha EA]

Hey BB! Long time no see, have you gotten to play with an R520 yet? And what other sites has Slappi been popular at lately...

Hiya Dig....wish I had an r520, but if I did, I would likely have to upgrade to PCI-E and I am not quite ready to do so. That being said, I am beginning to shop around for parts to become PCI-E capable soon.

As far as Slappi, I have seen him at Rage3D posting the same doom and gloom BS. It still kills me that people can make decisions about products which have no confirmed or official information or that have not been released to the public. What an insane mindset....

 

[Jawed]

Waaay back in March/April, when we were InfoMania-ing it, a big topic of conversation was how close R520 is to R420 vs R600. The conventional wisdom (I was a little more aggressive) was that it was a relatively minor refinement of R420.

Now we have the ring bus confirmed. So, how's the CW feeling these days about this question? Are we seeing any evidence of other C1/R600-ish concepts/pieces in R520? Does the CW still feel that it would be pointless to even go further down that path without DX10 in place?

I don't think there's any direct evidence. The best evidence, so far, is that R520 is a 16-pipe card running 60MHz faster than its predecessor and Dave Orton saying it's "competitive" with GTX (which is 20-50%, depending on benchmark, faster than X850XTPE in DX games). It can only do that with significantly enhanced shader efficiency.

Obviously I have my colours nailed to the mast, as per my signature.

I reckon that the scheduler, while not as full-scale as Xenos's (since it isn't scheduling vertex shader batches) is the foundation of the scheduler we'll see in R600. In a sense I see this as ATI testing this kind of scheduler in the PC space before going the whole hog with an entirely unified scheduler in R600.

A comment was made recently suggesting that R520's vertex shader pipelines would gain their ability to do vertex texturing by utilising the fragment shader TMUs. In other words, that the TMUs are shared both by the vertex and fragment shader engines.

If that's the case then it sorta makes R520 semi-unified - and actually makes the scheduler even more complex and arguably barely a whisker away from the scheduler in Xenos.

Jawed