OH I just think that trinity was only a try.. to prepare the card for 3-way with different GPU's
prepare yourself for Asus HD4850X3 Trinity ;)
2 times faster than 280 with almost same price (or little less)
Printable View
http://www.tweaktown.com/popImg.php?..._4850erly2.jpg
According to GPU-Z, this can never be HD4870:
800mhz clock (GPU-Z displays 750)
memory clock should be 1800mhz (3600mhz effective, afaik, it's named GDDR5 and not GQDR5)
Bandwitdth is 57,6GB, so it is not GDDR5 for sure...
It's an HD4850 overclocked to 750mhz (maybe bios edited, :shrug:)
This is, in fact, very good news, HD4870 can only be better than this, which is great.
Are you going to be able to use crossfire X and say put the 4850 and the 4870 x2 together?
Nice but we are going to be limited to only using the 4000 line we can't do like a 3870x2 and a 4870x2 in crossfire X right? Also how much better is the 4850 going to be over the 4870 I can't decide if I want to get a 4850 and then a 4870x2 later on and just crossfire them.
http://www.tweaktown.com/news/9691/e...and/index.html
I don't understand this. They say this is suppose to be HD4850 but GPU-Z shows some really weird data. Clock is much faster and it shows GDDR5 yet bandwidth is out of whack. On the other hand score is much higher than default HD4850.
I thought HD4850 will only support GDDR3. Unless of course they really meant to say this is HD4870.
Smells like fake to me. Or at least very strange.
Actually I believe it IS the 4870
The 750MHz core cant be reached by CCC right now so unless they've already done BIOS editing, it has to be the 4870 since thats the core clock for it.
Also, 4 x 900 = 3600 MHz which is the stock mem clock of the 4870 as well. I know its not supposed to be QDR, but read the GDDR5 whitepaper by Qimonda: http://www.qimonda-news.com/download...whitepaper.pdf
GDDR5 has a lot of differences so it might not be properly detected by GPU-Z anyways. That might also be why Vantage crapped out and couldn't give a proper score since it might not be detecting the card properly.
However those test results are ridiculous for the card... if it truly performs at that level, then the 4870 might be closer to the GTX260 and 9800GX2 than we thought.
Actually, the only differences between HD4850 and HD4870 are GPU clocks (625mhz, 800mhz respectively, IIRC) and memory type (GDDR3, GDDR5 respectively)
It also sounds very strange to me too, I don't know what to think, but my guess is its fake... :shrug:
This is the other side of my thinkings, guess I'll need more facts to convince me :shrug:
BTW, wasn't HD4870 supposed to be 800mhz core clock???
I haven't used Vantage at all, but something is confusing me in that pic. Shouldn't there be an X or P or something before the score (maybe the glitch that caused the 0 made the letter not show up?) How do we know that the score reflects an apples-to-apples comparison vs. the 4850?
This site has a nice intro and a video of the micro stuttering problem. I bet someone has already posted this link but there is a vr-zone forum post that claims that the 4870x2 will not have any micro stuttering due to a very high bandwidth link between the GPUs and/or maybe other rendering methods like SFR tiling which require that sort of high bandwidth link. Although I was sold by the R600 pre-release hype this 4870x2 sounds very interesting indeed.
Yeah I agree. How do we know this is Xtreme at all?!
I suspect this might actually be HD4870 but not on Xtreme. I think it was done on Performance. That would prolly explain why some test would be lower and some higher than Performance test on HD4850 we have seen so far. Those more CPU dependent are lower and the rest are higher. After all this was done on Phenom at 2.6GHz.
I just don't believe HD4870 would be that much faster. The performance gap would be too big.
Of course I would love to be wrong but I call this BS.
dont think so
http://img207.imageshack.us/img207/8527/46701ws3od9.png
regards
its 750 - i don't have the link to the slides anymore, but if it is 1.2 TFp then 750 x 800 x 2 = 1.2 TFp exactly.
Also, GDDR5 does work differently:
The rest of it you can read here:Quote:
Bandwidth first: A system using GDDR3 memory on a 256-bit memory bus running at 1800MHz (effective DDR speed) would deliver 57.6 GB per second. Think of a GeForce 9600GT, for example. The same speed GDDR5 on the same bus would deliver 115.2 GB per second, or twice that amount. Take any GDDR3 bandwidth on a given clock rate and bus width and double it, and you get GDDR5's bandwidth. Of course, the marketing guys love big numbers and would undoubtedly not call it 1800MHz, just as 1800MHz GDDR3 is really running at 900MHz. Expect the marketing guys to call memory at that speed 3200MHz.
http://www.extremetech.com/article2/...2309888,00.asp
From the sound of it, it IS indeed seen as 900 MHz and for GDDR3 it would just be 1800 MHz effective, but GDDR5 doubles that to 3600 MHz effective.
Hey look at that. If the 4670 can run at 800Mhz then the 4850 should be able too.
They had the Cinema 2.0 announcement + the FireGL card anouncemnt on the same day as the GTX280 release, and all these sudden leaks this week tells me that AMD is indeed trying to crap on the GT200 release