Read more.Quote:
Is the Mantle API all it's made out to be?
Printable View
Read more.Quote:
Is the Mantle API all it's made out to be?
I think that this is something that will increase in performance as we move forward. AMD has done a good job in getting it's APU's in modern consoles, and if it has a good takeup will be something positive
Not to mention the slowdowns that happen every 3 minutes for a second or so with Mantle running on the beta drivers (and the bug requiring you to disable the integrate intel graphics otherwise crashing on the options menu). Using the stable release Catalyst drivers 13.* and Direct X removes this issue. I know the latest ones are Beta and i'm not sure they should be used as a benchmark just yet as they are not finished.
I'm disappointed to see you didn't run the test with one of the high AMD FX cpu's :(
Mantle worked fine for me for a few hours and then took a mighty dump on me and despite removing, cleaning and reinstalling the betas 3 times no wits not stable on my rig what so ever and just crashes whenever I try to load a game and the drivers are now giving me direct x errors at random times so im going back to the old drivers as they where stable. Ill wait until a stable release comes out before giving them another go.
Tarinder..... you can borrow my sabertooth 990fx and amd 9590 for testing with am3+ If you want
On the contrary, the whole point of mantle is that it reduces the single/light thread botleneck imposed by the DX driver, and BF4 is heavily threaded so an FX 6300 or 8320 should see significant uplift. Testing on a quad core/quad thread CPU misses some of the benefit of the BF4 engine.
Mind you, all of this is a little moot given that *all* the systems under consideration pump out > 60fps @ 1080p ultra.
EDIT: also - a "roughly comparably" Intel CPU? Last time I checked the 4670k was £40 more than the 7850k ;)
I think a test with multiplayer would be far more relevant.
I think the difficulty is reproducing the same processor load between instances using multi-player.
I'm sure Mantle performance will improve with time, & it's understandable that AMD should focus on the lower performance CPU side of things, which is reflects their position in the CPU market. Still, a 5-8 frame improvement isn't great.
+1 on this, I'm surely not the only one that's got an ancient CPU paired with a lot more modern GPU - in my case six-way PhenomII with oc'd 7970?
That being the case, I'm not entirely in agreement with the statement in the article that you wouldn't have a "naff" cpu like an APU paired with some barn-burning GPU, simply because it's a whole lot less disruptive to replace a GPU than a CPU which, let's be honest, usually means having to replace the motherboard too. In which case you're talking about an OS rebuild. :( On the other hand switching graphics means a new driver at worst.
Interesting stuff on Mantle (as always). Don't think it'd make me want to run out and buy a 290 though.
Seeing as its still all new, its showing promise.
Phenom II X4 905e with a 7750 in my case. Given the low clock speed of my PII, it should actually see a decent boost in Mantle (but I'm not buying the game just to test the speed up from Mantle - anyone want to gift me a copy? ;) ). I also suspect the mobile chips will get quite a boost too, given the low clocks they run at (my laptop's an A10-4600M!)
I suspect as time goes on they will add the currently missing error detection & more features. Then GCN 2.0 will come out requiring greater abstraction, and at that point they will have at great expense have re-created OpenGL :D
Light weight systems tend to either die off or, like for example Linux started off as a lightweight Unix that could run from a 1.44MB floppy disc, they bulk up into something more robust and fully featured in line with the demands and uncertainly of life.
So more performance would be lovely, but I won't hold my breath.
The main use for mantle on higher-end system seems to be in multi-gpu configs. Can't see it being much use anywhere else (other then low-end systems) based on all the results we have seen so far.
I'd like to ask why the 780TI score is even shown here?, for an article thats about Mantle vs DX11, having the 780Ti score doesnt seem to help show the difference atall, it afterall is completely different hardware and drivers.