• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

More 6870 Benchmarks...

If these are true then only 1 decision remains...

To buy or not to buy? Or in my case, to sell or not to sell lol.


But like i've said before, is it really worth upgrading from a top end 5800 series card to the new 6 series or is it just more viable to sit this one out and wait for the 7 series :L.

Just can't decide!
 
If this is good as it claims to be then I might consider shifting from my current rig. Then again by then I'd probably need to go onto sandybridge etc anyways :s
 
That crysis result is interesting, I set my 5870 back to stock to compare.

At the same settings I get min 22.65 and average 32.4, 6870 got 43.55 avg.

So the 6870,if this is true, is just over 30% faster than a 5870 which is pretty impressive on the same 40nm process.


Techpowerup tells me the 480 is 12% faster than a 5870 so I guess there is little chance of the 6870 not beating it hands down IF the above numbers turn out to be true.


The vantage xtreme gpu score above is also 17% more than my card clocked at 950/1275
 
Last edited:
straxusii said:
That crysis result is interesting, I set my 5870 back to stock to compare.

At the same settings I get min 22.65 and average 32.4, 6870 got 43.55 avg.

So the 6870,if this is true, is just over 30% faster than a 5870 which is pretty impressive on the same 40nm process.


Techpowerup tells me the 480 is 12% faster than a 5870 so I guess there is little chance of the 6870 not beating it hands down IF the above numbers turn out to be true.


The vantage xtreme gpu score above is also 17% more than my card clocked at 950/1275
Click to expand...

If the numbers are right then that is 32% on very bad drivers. It would be like running your 5870 with a driver from three months before the release of the card.

Its a certanty we will see much more performance from this card by the time it comes to release.
 
Last edited:
How do you know they are very bad drivers, ATI's had a year to develop and works on drivers for their next gen, I very much doubt the drivers are in a bad state, and considering it's basically the same tech as the 5*** cards they should be **** hot on the driver front. If it's real benchmarks at all.
 
jigger said:
If the numbers are right then that is 32% on very bad drivers. It would be like running your 5870 with a driver from three months before the release of the card.

Its a certanty we will see much more performance from this card by the time it comes to release.
Click to expand...

Well call me anal but I have a record of my vantage GPU score with Oct 09 drivers, I got X8004 at stock.

So a direct comparison with performance at 5870 launch says the above is almost 50% faster, I doubt we'll see that though as I guess a lot of the improvements they made to vantage would also apply to the above 6870
 
This benchmark will ( if true) most likly be the start of the performance part of the driver development. The card ( going by rumours ) is quite a departchure from the 5870 core. It's a hybirb of two cores Raven.
 
Comparing heaven 2.1 on my stock 5870 same settings I got 16.3 fps so obviously the new architecture has a totally revamped tesselation implementation (if the above are true)
 
RavenXXX2 said:
How do you know they are very bad drivers, ATI's had a year to develop and works on drivers for their next gen, I very much doubt the drivers are in a bad state, and considering it's basically the same tech as the 5*** cards they should be **** hot on the driver front. If it's real benchmarks at all.
Click to expand...

The fact is we don't know, but based on every other graphics card release, drivers mature very quickly between "preview" drivers and "review/release" drivers.
 
I couldnt find an equivalent 480 heaven 2.1 test, closest I found had no AA and scored identically to the '6870' above so clearly the above is faster but don't know by how much.
 
For comparison, here are a couple of benchmarks I just ran with the same settings, using my GTX480 and a mild overclock (800/1000):

I'm using a core2 duo (e8400) at 4Ghz [unigine reports 4.5Ghz because I'm using a lower multiplier...], and 4Gb of dual channel DDR2, so CPu speed and memory bandwidth are significantly reduced. I can re-run on an i7-930 system in a few days, when I get the memory that ocuk forgot to send out with my order last week :(


Unigine:




Crysis:



I don't have vantage, but I can grab it if people are interested in the results.

If these are genuine, then the 6800 looks to be quite impressive. Although, I have to say, the fact that these come out so close to the GTX480 makes me a little suspicious. The crysis result could well be expected, but unigine makes such heavy use of tessellation (with tessellation set to extreme), and for two such radically different architectures to produce virtually identical results... Well, it's puzzling anyway. We will see if/when other benchmarks come in from multiple sources.

One thing is for sure, if these results ARE genuine, then the 6800 is certainly not just a minor revision of evergreen. In order to get the tessellation performance indicated above, AMD must have made massive revisions to their pipeline procedure, in order to maximise geometry throughput (as nvidia did with their "polymorph engine" design).
 
Last edited:
straxusii said:
^you need to do them at stock to compare properly dude.
Click to expand...

Just take ~10% off. I'm not running them again...


... I'll re-run on the my i7-930 system when I have it, using the same setup. As it stands I have a much slower CPU and memory anyway, and I'm running on a different OS, so stock runs won't really tell you any more than the shots I already posted.
 
Last edited:
GPU-Z detects all things memory related except the size, starting to look fake to me but we will see soon enough. If it is real then they have improved tess performance greatly, some would argue that it wasn't very difficult going by the 5*** performance in tess.
 
RavenXXX2 said:
GPU-Z detects all thing memory related except the size, starting to look fake to me but we will see soon enough. If it is real then they have improved tess performance greatly, some would argue that it wasn't very difficult going by the 5*** performance in tess.
Click to expand...

A massive amount of redesign work went into the GTX480, in order to produce the "polymorph engine" (really hate that term...), and remove the geometry bottleneck that is otherwise present in the pipeline. Unless AMD have come up with some other radical method to perform a similar task, then it will be anything but "not very difficult".

Unfortunately it isn't just a case of adding a larger tessellator unit. Hence my comment that the 6x00-series must be a radical redesign of the architecture. Either that, or these are overclocked GTX480 benchmarks in disguise :p
 
You must log in or register to reply here.
Back
Top Bottom