Why you should not buy AMD GPUs for DX11 games

[Monarch]

To 4GHz. If you were running your cpu at stock, that might explain it. Also 15.4 beta driver.

 

Actually mine is at 4.5 GHz. I ran the test a couple of times and I always get around 10 mil 900k. Did you run it multiple times and do you always get around 13 mil? 

[Kinda Bottlenecked]

Actually mine is at 4.5 GHz. I ran the test a couple of times and I always get around 10 mil 900k. Did you run it multiple times and do you always get around 13 mil? 

 

Ran the test for the 3rd time earlier. Got over 13 million in all 3 tests. 

Test 1

Test 2

Test 3

[Monarch]

Ran the test for the 3rd time earlier. Got over 13 million in all 3 tests. 

Test 1

Test 2

Test 3

 

Hmm. And again DX11 results are about the same as mine. Maybe it's cause I have single channel RAM. But still, both tests, with 290X and 970 were done in the same system so it's the diffference that matters, that's the point.

[Kinda Bottlenecked]

Hmm. And again DX11 results are about the same as mine. Maybe it's cause I have single channel RAM. But still, both tests, with 290X and 970 were done in the same system so it's the diffference that matters, that's the point.

 

True true, that definitely show the difference between the 290x and 970. But I'm definitely more intrigued by this now that it has happened.

What can this mean for future APIs I wonder? Might have been due to the architectural differences between the 290x and the 7970. 

[xAcid9]

Hmm. And again DX11 results are about the same as mine. Maybe it's cause I have single channel RAM. But still, both tests, with 290X and 970 were done in the same system so it's the diffference that matters, that's the point.

 

Yup, no doubt that AMD driver have more CPU overhead than Nvidia, that's why they created Mantle and want the industry to move forward A.S.A.P.  

 

3DMark API Overhead test love tight memory configuration, you don't need a 970 to get tonnes of drawcalls.