First retail 6870 photo; now includes full Chinese leaks and Guru3D review discussion

Page 44 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Red Storm

Lifer
Oct 2, 2005
14,233
234
106
Trying to figure out why your 6870 >= 5870 predictions didn't materialize?

I don't know about you, but to me the 6870 is clearly > 5870. ~$100 cheaper, less power draw, better CF scaling, better tessellation performance (I personally don't think this particular figure is all that important, but it's there so whatever)... :)
 
Last edited:
Feb 19, 2009
10,457
10
76
yeah that explains the low #, 10.11 gives the 6K series a huge boost. i'm surprised AMD did not made these drivers public for launch.

edit: AIBs had time to sample the cards with 10.11 and it was much faster than 10.10, to clarify.
 
Last edited:

busydude

Diamond Member
Feb 5, 2010
8,793
5
76
yeah that explains the low #, 10.11 gives the 6K series a huge boost. i'm surprised AMD did not made these drivers public for launch.

Come on man, get a grip!. 10.11 seriously? AMD makes them public in, lets see, November. Why do you have to make it so obvious for the world to NOT take you seriously.

My edit in reply to yours: Oh, I see. So, you are saying that 10.11 drivers will make 6870>5870 in terms of performance?
 
Last edited:

arredondo

Senior member
Sep 17, 2004
825
37
91
Two questions....

- Any updated info on whether or not the MLAA feature is exclusive to the 6xxx series?

- Would my rig get a significant FPS boost at 1080p with a 6870, or is my Q6600 too much of a bottle neck to make much of a difference? Keep in mind that I can't overclock my HP-installed CPU.
 

tijag

Member
Apr 7, 2005
83
1
71
yeah that explains the low #, 10.11 gives the 6K series a huge boost. i'm surprised AMD did not made these drivers public for launch.

edit: AIBs had time to sample the cards with 10.11 and it was much faster than 10.10, to clarify.

This will be very easy to prove/disprove.

if 10.11 drivers come out and a big performance bump for the 6xxx series cards isn't included, we can completely ignore Silverforce11.
 
Feb 19, 2009
10,457
10
76
This will be very easy to prove/disprove.

if 10.11 drivers come out and a big performance bump for the 6xxx series cards isn't included, we can completely ignore Silverforce11.


NP.

I'm just curious why they didn't release their best drivers for a launch.. i'll ask around.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
NP.

I'm just curious why they didn't release their best drivers for a launch.. i'll ask around.

The only reason I can imagine is they only need to beat the gtx-460. They might still want to clear some 5870 and don't want to devalue them anymore than they can get away with.

This is assuming they do have better drivers they are holding back, of course. ;)
 

v8envy

Platinum Member
Sep 7, 2002
2,720
0
0
It's easy to buy into future drivers bringing performance improvements. This is, after all, a changed architecture. The 5 series was very similar to the 4 series architecturally, and that in itself was similar to the 3 and 2 series. And yet, ATI still managed to eke out impressive performance improvements from the drivers over the years.

I guess AMD believes performance with current drivers is enough to paint these in a positive light. Just think how impressive the upcoming high end parts will look, relatively speaking, with fixed drivers AND more hardware. The 6870 will be competitive with the GTX480 by then, but few will realize this.

Looking forward to some re-benching done with drivers 2-3 months from now. And I may just have to bid Linux goodbye on my desktop machine until NV is competitive again. This generation from AMD may just be that good.
 

busydude

Diamond Member
Feb 5, 2010
8,793
5
76
This pic made me LOL
lol.png
 

BladeVenom

Lifer
Jun 2, 2005
13,365
16
0
guru3d 6850 and 6870 crossfire review Mirror from HardOCP.

http://www.zi0x.net/6800cfx/

I was just checking to see if anyone had posted that.
Although it is loading slow. If you just want to see some of the benchmarks, they are in this thread. http://hardforum.com/showthread.php?t=1555744&page=2

http://img181.imageshack.us/img181/6977/cod1.png
http://img522.imageshack.us/img522/5329/cod2y.png
http://img828.imageshack.us/img828/3276/bfbc2c.png
http://img109.imageshack.us/img109/663/warhead.png
 
Last edited:

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
A Chinese review site also broke the NDA:

http://translate.googleusercontent....le.com&usg=ALkJrhhfwcYxueYIWzEL5BbF7PUP1BzM2Q

985 core overclock here (which is capped; CCC typically caps overclocks at a low number; the oc'er also didn't overvolt the card, either). That isn't impressive but it's capped by CCC so who knows what MSI Afterburner can do. Also note that in order to heavily oc a GTX 460 to get to stock 6870-caliber speeds, you would drive up your power usage by a LOT, something like 100 watts, in additioning to shortening the life of the GF104 chip inside of it. I'm wondering if other reviewers will get better oc's.

I did hear something funny on another board, though: HD6870 = GTS 450 die size, GTX 460 power consumption, and GTX 470 performance.
 
Last edited:

Grooveriding

Diamond Member
Dec 25, 2008
9,110
1,260
126
A Chinese review site also broke the NDA:

http://translate.googleusercontent....le.com&usg=ALkJrhhfwcYxueYIWzEL5BbF7PUP1BzM2Q

985 core overclock here (which is capped; CCC typically caps overclocks at a low number; the oc'er also didn't overvolt the card, either). That isn't impressive but it's capped by CCC so who knows what MSI Afterburner can do. Also note that in order to heavily oc a GTX 460 to get to stock 6870-caliber speeds, you would drive up your power usage by a LOT, something like 100 watts, in additioning to shortening the life of the GF104 chip inside of it. I'm wondering if other reviewers will get better oc's.

I did hear something funny on another board, though: HD6870 = GTS 450 die size, GTX 460 power consumption, and GTX 470 performance.


Heaven on extreme tess., STALKER & Lost Planet 2

Looks like they took care of tessellation and DX11 performance.

heaven.png

cop.4aaa.png


lp2.4aaa.png
 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
Who cares what the HD6870 can overclock to.
The HD6750 is clocked much lower. Be much more interesting to see how high it can go.
 

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
These are some great results coming in, keep it up guys :thumbsup:. The 6870 performance is very impressive, approaching the 5870, even in DX9!, and especially considering the die-size and power consumption. Looks like most of the leaks we've had were spot on. This also leads me to believe that Cayman could be an absolute monster :D.
 

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
power.png

Looking at power usage it looks like it might be voltage limited. I'll bet it'll O/C better with some extra volts?
 

notty22

Diamond Member
Jan 1, 2010
3,375
0
0
Who cares what the HD6870 can overclock to.
The HD6750 is clocked much lower. Be much more interesting to see how high it can go.
Guru3d got it to 897
the 6870 to 950. IMO, the single 6 pin maybe held it back a little.

Who is he kidding with this 'november' driver ?
Desktop_2010_10_21_17_26_42_096.png
 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
power.png

Looking at power usage it looks like it might be voltage limited. I'll bet it'll O/C better with some extra volts?

Those power numbers look dodgy as hell. The HD5850 has never used that much power that I've seen. According to AT, it uses less than the GTX460.
Equally the GTX460 on that only uses 15w more than the GTS450, which disagrees with AT figures, and the 5770 and GTS450 are much closer than that shows.

I wouldn't put any stock in what those numbers 'show'.