Read your own graphs as well?
http://img.techpowerup.org/121125/Capture003145.jpg
It also shows the 6670 GDDR3 outperforming the 6670 GDDR5 when use on its own. Interesting maybe we should be a little more stringent on benchmarks.
Maybe a driver issue? Have you ever heard of bad drivers? It happens.
http://img.techpowerup.org/121125/Capture004931.jpg
Somehow adding the 6670GDDR3 hurts performance yet adding the GDDR5 card helps it. Again either someone is doing a piss poor job with their benchmarks or something is wrong in the setup.
Are you kidding me "it helps it", it gives 1 fps, this game is simply not well optimized for dual graphics. (Maybe AMD drivers fault)
http://img.techpowerup.org/121125/Capture005311.jpg
Oh wait here is one that takes advantage of the added bandwidth of the card. Interesting how this one game shows such a 10FPS improvement moving from the GDDR3 to GDDR5 card. Something no other benchmarking site shows.
Battlefield 3 needs a lot of memory bandwidth so I'm not surprised that the GDDR3 is bottlenecking the GPU.
http://img.techpowerup.org/121125/Capture006.jpg
Crossfire obviously isn't working in this one and somehow you get a 100% improvement with GDDR5.
Again, GDDR3 is bottlenecking the GPU and this game is simply not well optimized for dual graphics, probably AMD drivers fault.
http://img.techpowerup.org/121125/Capture007.jpg
Again crossfire isn't working and somehow almost 100% improvement with the GDDR5...
Again, GDDR3 is bottlenecking the GPU and this game is simply not well optimized for dual graphics, probably AMD drivers fault.
http://img.techpowerup.org/121125/Capture008.jpg
Again crossfire isn't working. I am starting to see a trend.
Again, GDDR3 is bottlenecking the GPU and this game is simply not well optimized for dual graphics, probably AMD drivers fault.
http://img.techpowerup.org/121125/Capture009.jpg
Interesting how all but one benchmark shows a negative improvement going to 6670 crossfire and this graph shows a positive one...Someone either lacks the ability to average or was looking at a different set of benchmarks than me.
44+43+42,4+18,9+33+17,3+38=236,6 = 100
35+35+38,8+31,4+31+15,2+63=249,4 = 105,4
I don't know where they took 14%, but it's 5% higher for the dual graphics GDDR3.
59+59+44,6+28,3+61+33,9+58,5=344,3 = 145,5
70+69+45,8+46,2+60+26,7+84,5=402,2 = 170
Again I don't know how they did their average... but they were near this time.
Yup your benchmarking site does such a wonderful job can't even get crossfire to work in all but what one benchmark. Yet they are who you base your purchase off of?
Dafuq? Crossfire was on, but hybrid crossfire does not always give positive results and you can see this in a lot of websites.
Example:
http://hexus.net/tech/reviews/cpu/46157-amd-a10-5800k-dual-graphics-evaluation/?page=2
I used this website since that is the only I found that compare dual graphics of the HD 6670 GDDR3 to the GDDR5.
Now I am not saying the GDDR5 card offers improvements simply that its not worth any extra money whatsoever. Overclock the GDDR3 card overclock the onboard DDR3 for the IGP both will offer more of an improvement than the GDDR5 card will on its own.
Can't you see that the HD 6670 GPU is bottlenecked by GDDR3? 25.6 GB/s is not enough for this card.
When dual graphics is supported, GDDR5 gives a real boost compared to GDDR3 in most games.