HD4870X2 VS HD6970 VS HD6850 new benchmarks!
 

Hey guys! I want to share new benchmarks those i have made and presented in overclockers forums!

http://www.overclockers.com/forums/s...d.php?t=684245

Have a nice day!

Thanks! I love this kind of benchmarks.

Interesting benchmark results, thanks.

kinda pointless with a 1090T, benchmark would be far more interesting with an actual bulldozer chip, without the new chip this isnt really all that interesting.

kind of a pointless comment, as bulldozer isn't available...

That's a stupid comment. If my goal was to test processors, i would have used lowest resolution possible, far less games. Another thing is that bulldozer is not even out yet. Third thing is - if bulldozer had been out, i would not have the money to buy it. So what's the point of your notice - really stupid... AMD Pheonom II X6, though not as fast as Core i5 Sandy Bridge, is the best thing an AMD enthusiast can build on the new 990FX. You sounded like a troll with your post.

im an enthusiast if i wanted GPU tests on a 990fx with a Phenom II id look back at the 790fx and 890fx performance hasnt changed any i can say this because im running an old as dirt 790fx and in all those tests with a 965be my 6970 gets within 1-2fps of all your tests, thats why its pointless,

if your testing GPU performance on a new mobo thats fine an dandy, but it IS a 990FX it IS AMD's high end in a single gpu situation on an AMD cpu, the performance wont change enough to matter at this point. It may sound like im trolling but its the truth.

now actual Crossfire tests with a 6850 crossfire / 6970 crossfire 4870x2 quadfire would be interesting to see how the improved chipset handles that much data load on the PCIE lanes and memory subsystem

as for the Bulldozer comment well it is what it is, the memory bandwidth on Bulldozer is suppose to be far better then anything AMD has previously released where as the IMC on current CPU's by AMD are there biggest limiting factor in games its so bad in fact that in F1 2010 an AMD APU on low end FM1 board gets 20fps more then a 1090T do to the extra 3000mb/s the memory subsystem provides from a revamped IMC,

testing a board is great testing a high end board with single high end gpu on a CPU thats performance wont change anything no matter what board you use is a nice side project but not really anything to take note of.

what WOULD be of note is if the 990FX board offered better ram compatibility if it helped offer higher Ram speeds and tight timings, if the board allow extra overclocking headroom, these things are more pressing then,

Yes, you make sense now, but you forget that one of the key factors in this case also is the new Catalyst 11.7 suite. I made benchmarks before, and you can see how much of improvement do the new drivers add on HD6850..

true but i can tell you from experience new drivers help but not when a CPU is a bottleneck

6970 xfire on my 790fx is about 25-45% slower then any intel i5 based system do to memory bandwidth ;)

if the 990FX board allows a Phenom II to get near Intel memory performance with extra work more often then the older chipsets thats what id like to know

essentially does the 990fx allow mediocure Phenom II chips stretch the IMC further? if not then its of no relevence till tested with a Bulldozer chip

I think you're missing the point. The OP wanted to do benchmarks of this specific scenario. Just because your opinion is that it is "pointless" does not make it so. I found it interesting and useful, and I don't even use AMD platforms. Your post was in no way helpful or related.

It would be neat to see how the scaling compares with the cards you mentioned, but I also think that the OP was relevant, and very helpful to others. He took his time to create a well structured personal review. The more data end users submit, the more accurate the information is for potential buyers, etc.

thats your opinion, if i wanted general gpu performance id just read TPU GPU reviews etc,

mobo wont impact that nearly enough to matter, again if you start looking at memory bandwidth and what it can offer an AM3 chip then were getting somewhere THATS helpful,

because its not just a 990fx it applies to if the 900 series offers the ability to tweak a Phenom II further then older chipsets that a 990x and 970 for $100 would be a viable method to take a AM3 chip sitting around a bit further.

what the OP has is a good base to start from to explore how these things impact performance,

as of right now it looks more like a GPU review with an AMD based system.

I think you are reading into it a bit more than the OP intended ( I could be wrong ). It seems like you are turning his post into something more than it was supposed to be. I think he just wanted to post a nice constructive set of data based on his personal tests.

Another thing - if you don't like his review you don't have to read it - you can go to the "official" reviews on the main page. You're writing as if you were forced to read it, and it made you upset. Correct me if I'm wrong.

Everything written on this forum is opinion basically. I just don't think it was very nice how you imposed yours onto his nice thread. However, it is his thread, not mine. My apologies OP!

no not upset, there just isnt really much point to AMD performance testing right now. unless it comes to getting that extra magic ram bandwidth that helps the newer gpus stretch there legs. again ill reference the AMD APU having 3000mb/s more bandwidth then a 1090T and both at stock the APU with the same gpu as a 1090T put up 20fps more in one of the gaming tests,

Im of the point now having played with Phenom II's since they were am2+ only that the boards and performance hasnt changed,

it hasnt changed to the point

2x 5850s on a 790gx with a 940BE with DDR2 up to a 790gx 965BE to a 790fx DDR3 965BE at 4ghz the gpus performance remained the same across all those platforms, not enough has changed,

basically all the OP has shown me is that from 790gx up to 990FX DDR2- DDR3 nothing has changed, from Jan 2009 - to today, the motherboards havent changed anything in favor of helping AMD's chips its interesting in a sad kind of way that after 3 motherboard revisions and a jump to DDR3 the performance difference is pretty much null. for AMD's gpus, and the fact that looking at Intels forced socket change, with each transition performance with those same gpus tested by the OP goes up. they perform better, its a sobering test to the fact AMD needs to release bulldozer and have it not suck, It also drives home the point AMD has some very nice motherboards and nothing really worthwhile to stick in them performance wise to really put them to use.

And im not reading to far into it the OP put up numbers numbers that have been added to a database of information ive kept, its not like hes the only one to test shiny gpus on an AMD board, ive multiple forum reviews as well on 5850 xfire, 6970 xfire, 4870x2 performance, etc,

maybe its because your on an Intel platform your having a hard time grasping where im coming from looking at this info and info across the web

and it is the internet, if you dont really have thick skin to take information provided for what it is, then i dont know what to tell you.

I have not read read such hilarious comments in a long time! :toast:

crazyeyesreaper, i understand your comments; you misunderstood me a bit, that's ok.

I suppose to say its not the OPs fault AMDs hardware isnt really moving forward.

I look it like those vonage commercials where they throw there bill on the pile, your information while interesting for what it is, is kinda just like that tossing the same info on that big old pile just this time its AMD's latest and greatest platform testing there old and current gen GPUs and discovering oh wow nothings changed. nothing against the OP really, it just becomes a pain in the ass to always see the same results,

the FM1 socket APUs on the other hand turned my head upside down with that 20fps boost the APU had for a 1090T system, that new platform had a bit of a surprise right there something different from the norm. something to serious raise your eyebrow at and think wow that cant be right.

just again i suppose thanks to the OP in taking the time to post the findings, but shame on AMD for changing absolutely nothing on the chipset side of things.

Sounds like you need some INTEL in your diet :D

i have a 2500k + ECS P67 and 1866mhz Mushkin 2x4gb kit here just im having some parts sandblasted and painted, and having a vinyl decal done up, meaning its not operational currently.

I bet you can't wait. It will truly show you just how obsolete your AMD platform is.

I would rather see this
 

All the benchmarks should be run on the best platform AMD has to offer versus the best that Intel has or at least an attempt should be made to show whether or not crossfire or SLI is worth it and how well it works on different platforms. I get that can be an expensive test but without testing on a variety of platforms I feel that the results are less reliable. You just get a clearer picture of performance scaling with different setups being tested. It doesn't have to be just multi GPU performance. It is important to assess overall performance in general.:)

im well aware how behind my 965 is a 2500k + single 6970 gets the same frame rate as a 965BE + 2x 6970s lol

I dont see anywhere stating that you installed the latest CAPS? but if its already in the driver then fair enough, as i used the latest CAPS for Crysis 2 and my FPS went up ALOT.

Otherwise good benchmarks, interesting to see how the old 4870X2 went against newer cards, im surprised at some of the results, but i must admit i haven't played all of those games apart from 3 of them.

I know that 2 4870X2's in quad-fire will out perform 2X6950's in Heaven benchmarks ;)

Do you have actual results of two 4870X2's in quad-fire beating 2 6950 2GB's in heaven? What resolution? It wouldn't even be right unless it was in DX11 and the 4870 is not capable. So, that's a pretty weak comparison.

I don't see what would be wrong with the comparison as long as both were using DirectX10 in the benchmark.

Example: 2x 6850 vs 2x 4870x2

There's nothing wrong with it, but it's a little pointless as DX10 is no longer the golden standard. As a throwback test it would be neat to see how they match up, yes.

You do understand that having DirectX10.1 hardware doesn't actually prevent you from playing a DirectX11 game, even using the DirectX11 codepath, right?

Pretty much the only thing of any worth that you miss out on is Tesselation. Feel free to make your own conclusions as far as how valuable you feel that feature is worth but no need to make a mountain out of a molehill; a 4870 (or 2 or 4 of them) plays DirectX11 games just fine.

Tesselation:
http://gotnorice.no-ip.org/tesson.png

No Tesselation:
http://gotnorice.no-ip.org/tessoff.png