ATI Radeon HD 4000 Series discussion

[zerazax]

The specifications of the HD4870 looks excellent except the memory interface which is 256bit. In addition, the price is good 349$

You do realize that 256-bit at 4Ghz memory (GDDR5) is the same as a 512-bit bus at 2GHz memory (GDDR3)? That's more than enough..

[adamsleath]

these info dribbles are pretty amusing.

[ownage]

I wonder if AMD will announce an USB version of the 4870.


link

[mascaras]

I wonder if AMD will announce an USB version of the 4870.


link

[Monkeywoman]

Source; http://www.hardware-infos.com/news.php?news=2075

translated page; http://translate.google.com/translat...F8&sl=de&tl=en

[LordEC911]

Source; http://www.hardware-infos.com/news.php?news=2075

translated page; http://translate.google.com/translat...F8&sl=de&tl=en

Hmmm... same rumored specs from over a month ago?

[Monkeywoman]

they might be the same rumored ones but atleast now there's some backbone to the source. These specs look real, what I'd like to know is what type of architecture they use. It can't be the exact same as the r600

[LordEC911]

they might be the same rumored ones but atleast now there's some backbone to the source. These specs look real, what I'd like to know is what type of architecture they use. It can't be the exact same as the r600

Separate shader domain?
Clockspeeds?
TDP?

Possible but not very confirmed IMO.

[zerazax]

Well the core clock speed matches the ES that w0mbat mentioned on B3d

TDP is possible - they're on the premium 55nm process from TSMC which is supposed to be better than the one used on the RV670 and should help TDP - and those clocks on the RV770PRO would certainly lead to lower TDP's

Separate clock domains aren't out of the question either. Although there isn't proof this is authentic, the math at least adds up, which is much better than what other sources have rumored so far

[FischOderAal]

Source; http://www.hardware-infos.com/news.php?news=2075

translated page; http://translate.google.com/translat...F8&sl=de&tl=en

"The RV770 Pro GDDR5 has been dissolved due to internal competition."

there was a HD-4850 with GDDR5?

[Origin_Unknown]

"The RV770 Pro GDDR5 has been dissolved due to internal competition."

there was a HD-4850 with GDDR5?

internal competition?

[Jowy Atreides]

internal competition?

yes, too many similar product lines are bad for a business.
One product will steal sales of another from the same comapany

[LordEC911]

internal competition?

HD4850 GDDR5 running as fast the HD4850 GDDR3 or the HD4870...

[madcho]

look's like the finnal spec.

[w0mbat]

look's like the finnal spec.

Isnt the final spec, cause its fake.

[Tim]

Isnt the final spec, cause its fake.

I know you know more, but this looks like a really good fake if true.

[Luka_Aveiro]

Isnt the final spec, cause its fake.

So, tell us the truth

[clonez]

prolly smth about the clocks not being final

lower clock speeds, but optimized shaders?

[Jack]

More unofficial benches OC Workbench

In 3DMARK06, Radeon HD 4870 will be 32% faster than 8800GT, 22% faster than 9800GTX and 12% faster than 4850.

LOL @ beautiful graph!

[SKYMTL]

Those aren't benches. And last I checked there was more than a 10% difference in performance between the 8800GT and 9800GTX.

[AliG]

Ugh for the last time guys, read or at least look through the thread before posting the same useless graphs/links. Besides, that's a performance ESTIMATION chart, not an actual result

[TG01]

Ugh for the last time guys, read or at least look through the thread before posting the same useless graphs/links. Besides, that's a performance ESTIMATION chart, not an actual result

Thx, and why should we read through 33 pages of posts when you can clarify this for us in a single post..

[perkam]

source

4850 625/2000
4870 725/3920

96v5 shaders = 480SP ??

We will see soon enough...

?!?!?!?!

What is with the 2X increase in Mem speeds between 4850 and 4870 0_0 !!!!

Perkam

[zerazax]

Unlike GDDR4, GDDR5 actually improves memory speed A LOT. As in, GDDR3's 2000 effective is GDDR5's base...

[Monkeywoman]

Not only the fight between the GT200 and the RV770 (we reported) is interesting. We could bring out some information about the coming RV730. What sounds very interesting is the the 256Bit memory interface. In the high end area nothing special, but for the middle class the newest sensation.

The 600 MHz and the 500 MHz core clock is for us a little to less, propably the values are from a engineer sample or AMD think they can win some performance with the higher memory bandwith.

We can not garantee that these are the final specifications!

source; http://www.ati-forum.de/general/news...pecs/?29c56f82