Thief Patch Enables TrueAudio And Mantle: First Benchmarks

Toggle sidebar Toggle sidebar
Status
Not open for further replies.
The most interesting thing is the FX-8350 beats the i7-4770K with Mantle(minimum frame rates), and gets smoked on DirectX. Being 10% faster probably isn't within margin of error either, so it's interesting.The newest drivers (14.2) also are helping the Kaveri APUs pretty dramatically, at least with regards to the A10-7850K. The performance improvement fundamentally changes the limitations of the GPU, and makes it more consistent with what we expected when we looked at the specs. I'm a little surprised it hasn't been reviewed here, since the one site that did showed between 20% and 40% improvements in frame rates. A more thorough review would be interesting, to say the least.Either way, with hardware pushing against a wall, software is going to be an increasingly important driver (forgive the pun) in performance. It always was important, but now it's critical. Software companies can't be lazy anymore, and hope the next generation of hardware will carry the load to the next level of performance.It amazes me Microsoft had to wait for AMD to release Mantle before they realized they might want to improve DirectX. It wasn't always right there before their eyes? A hardware company had to figure it out, for them to copy? Intel always made better compilers than the trash Microsoft sells, but for AMD to show up Microsoft is something new, and should be embarrassing for them.
 
Since this is all about shifting the load from CPU to GPU, it seems to have no effect whatsoever when you've got a CPU powerful enough to prevent bottlenecking (e.g., the I7 in this graph, or probably any I5 post-sandy bridge). Really, this is only useful if you've mis-managed your build and have a GPU far more powerful than your CPU (e.g., an FX-4170 and an R9-270). It's a cool technology, but the applications seem limited.
 
Since this is all about shifting the load from CPU to GPU, it seems to have no effect whatsoever when you've got a CPU powerful enough to prevent bottlenecking (e.g., the I7 in this graph, or probably any I5 post-sandy bridge). Really, this is only useful if you've mis-managed your build and have a GPU far more powerful than your CPU (e.g., an FX-4170 and an R9-270). It's a cool technology, but the applications seem limited.
Click to expand...
Actually, if you take a look at the gpu help section on tom's own website there are a LOT of people that made such a choice.
 
Setting games to maximum details levels the playing field somewhat, thus hiding CPU inadequacies.
It amazes me Microsoft had to wait for AMD to release Mantle before they realized they might want to improve DirectX. It wasn't always right there before their eyes? A hardware company had to figure it out, for them to copy? Intel always made better compilers than the trash Microsoft sells, but for AMD to show up Microsoft is something new, and should be embarrassing for them.
Click to expand...
Well, some might say that it took a lot of wailing and gnashing from consumers about CrossFire frame pacing before AMD did something about that, but what forced them to do so was that their main competitor developed a way of measuring said issues...
 
I think we want a bit more in depth comparision than 2 graphs from different GPUs and different detail presets...
Click to expand...

I should hope so!

And that's why, as the article states, this is merely a tiny sample of the data we've collected. Full analysis coming in the near future.

Reading the article helps put the charts in context. :)
 
I think this also shows what happens when you tailor the software and code to your hardware.

Most software running on Intel gets a boost form it using Intel compilers etc. which doesn't work quite so smoothly for AMD kit.

If you tweak the software more to your platform than the other guys you will get the benefit.

As for 'unbalanced' systems re. CPU/GPU, well if I was a gamer and had just $300 to spend on a new CPU and GPU then I'd buy a $100 CPU and a $200 GPU every time. Certainly wouldn't buy an i5/i7 and just use the integrated or whatever $30 GPU I could get with the leftovers..
 
Don, just a little point that's worth keeping in mind, Tech Report had a lot of data suggesting that in BF4, AMD+Mantle could be the same as Nvidia + D3D, basically implying that AMD's DX drivers have greater CPU overhead than Nvidia's.Could you look into that as well, please?
 
Looks like mantel is just another useless gimmick that only works with low performance CPU. They should have spent the money making their CPUs better rather than waste developer's time with that useless backend.
 
u know the gains are limited now with 290X, but I can see how a i5 pair up future flagship GPU like Radeon 390X/490X in Crossfire and not bottleneck it! which means I dont have to dump my i5 and upgrade entire platform, especially Intel changing socket too often... I save huge bucks from here alone.
 
Looks like mantel is just another useless gimmick that only works with low performance CPU. They should have spent the money making their CPUs better rather than waste developer's time with that useless backend.
Click to expand...
Surely the fact that it helps somebody means that there was a point to it all, moreso considering Microsoft is taking it seriously.As regards overhead, I think I heard before that AMD cards seem to want more PCI-E bandwidth. Maybe that's related, I don't know. There's also the idea that the delta between AMD and Intel CPUs is reduced when using an NVIDIA card as opposed to an AMD card, but I think I'm going off-track here.
 
People should keep in mind that while Mantle is very useful for current low end cpu, it also means that your high end i7-4770 will last another year or 2 longer with Mantle support. So while mantle isn't immediate beneficial to those with 100+ fps with ultra settings, in another 2 years, you're going to want a mature version of Mantle.
 
The newest drivers (14.2) also are helping the Kaveri APUs pretty dramatically, at least with regards to the A10-7850K... showed between 20% and 40% improvements in frame rates.
Click to expand...

AMD A10-7850K Performance Optimized Catalyst 14.2 Driver
from Benchmark Reviews

- "Performance in Tomb Raider was increased by over 33%"
- "Bioshock only gains about 20% performance" ('only' 20% :lol:)
- "Battlefield 4 performance is increased by nearly 40%"

 
Looks like mantel is just another useless gimmick that only works with low performance CPU. They should have spent the money making their CPUs better rather than waste developer's time with that useless backend.
Click to expand...
How is that a gimmick? And how is it a waste for developers? Heck, it's something new that might create a handful of new jobs at developer level (how is that a bad thing....jobs) and AMD actually nudged Microsoft to have a hard look at what DirectX needs to be at (and will be at) so really, it's a win-win for all. I'm not an AMD fan due to too many experiences that were not too pleasant with their stuff, but still, Mantle is a good thing.
 
I'm not sure these people complaining have noticed that Mantle has increased FX4170's performance by 40% on average and minimum fps by 58%! He's right up top, close to Intel's 4770k (which has a platform way more expensive).

The only 2 things really missing today, considering the rush, is a benchmark with a lower-end Intel processor, and an older CPU.

If Mantle can alsoo boost Intel's Pentiums and maybe even those older Athlon X2 and E4500s, it would mean AMD could completely dominate the market for people upgrading older rigs.
 
What I'm really curious about is the performance difference in AMD's APUs. If these numbers are any indication, this could be a really big boost for those trying to build a cheap mini computer for gaming.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom