Mantle Performance Review (AnandTech & Legit Reviews

[Assassin]

Looks like Ananandtech and other press media got a pre release driver for testing. Below is their performance review:

 

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview

 

the data is quite interesting. 

 

Legit Reviews: http://www.legitreviews.com/amd-mantle-api-real-world-bf4-benchmark-performance-catalyst-141_134959/3

[Sharif]

interesting......

[Vitalius]

WOO

This means it's around 1 day away. According to what AMD told Linus, "You'll have about a day to test it before it releases." or something similar.

[<boss2939>]

hmm... I think mantel is a success especially in the cpu side.

[KamiKatze]

As long as it get's rid of the bugs, it can also degrade the performance, i don't care!

[Cookybiscuit]

What a waste of time, it doesn't do anything.

[ivhyn]

So it only tackles the CPU bottleneck... kinda worthless to me. 

[Fetzie]

So it only tackles the CPU bottleneck... kinda worthless to me. 

 

If you were expecting anything else, you didn't understand the presentations.

[Askew]

So, it's benefit appears to have been greatly exaggerated in discussions then?

 

If it negates the effect on gaming that a slower CPU has in gaming slightly, you are still gonna have a CPU that sucks at everything that isn't a Mantle enabled application.

What interests me is that it seems to benefit imbalanced systems, the advantage drops of vertically when you put for instance a 290X with an 8350 as most people would.

 

I didn't really go out of my way to educate myself on Mantle, but there has been a lot of very rabid anticipation of it on this forum and I think it's been way overplayed by some people.

[ACatWithThumbs]

Those benchmarks are stupid.
We need Multiplayer benchmarks not Singleplayer where the CPU get's not even really used.

[Blue fellow]

A bit disappointing, but still something I suppose. 

[TechFan@ic]

So it only tackles the CPU bottleneck... kinda worthless to me. 

It improved performance on an i7 4960X overclocked to 4.2Ghz, it can't possibly be useless if even the most powerful consumer CPU on the market showed healthy performance gains.

I was honestly not expecting any performance improvements in single-player since it's rarely CPU intensive.

[Humbug]

Looking good. Can't wait for the detailed info on frame latency, i'm expecting the frame delivery to be more consistent compared to direct x.

Also think this is going to be a big deal for people on highly populated multiplayer maps in BF4 which currently causes CPU bottlenecks even on good CPUs.

[TERAFLOP]

These are single-player benchmarks folks, in multi-player the performance almost doubled.


64 Players Ultra Details (Core i7-3770K, 16 GB DDR3-1333, Radeon R9 290X @ 1.000/2.500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)

http://www.golem.de/news/amds-mantle-api-im-test-der-prozessor-katalysator-1402-104261-3.html

Courtesy of @xAcid9

[Humbug]

I think the people who are dissapointed need to look at in perspective. This is a new immature API and yet it is providing more performance from the same silicon. Clearly it's allowing the hardware to be more efficiently utilized. And we are getting it for free.

[Humbug]

I was expecting decent performance improvement in multiplayer but this is insane. Almost looks to good to be true, will wait for more benchmarks but most review sites will not touch multiplayer. if this is true it gives people with AMD cards a competitive advantage.

[Guest]

Not bad AMD!!

[zeibis]

It would be nice to see how crossfire scales with two and three cards and at higher resolutions.

[Vitalius]

So, it's benefit appears to have been greatly exaggerated in discussions then?

 

If it negates the effect on gaming that a slower CPU has in gaming slightly, you are still gonna have a CPU that sucks at everything that isn't a Mantle enabled application.

What interests me is that it seems to benefit imbalanced systems, the advantage drops of vertically when you put for instance a 290X with an 8350 as most people would.

 

I didn't really go out of my way to educate myself on Mantle, but there has been a lot of very rabid anticipation of it on this forum and I think it's been way overplayed by some people.

Well, you also have to remember two things:

1. It isn't just about cpu bottlenecking. Part of it is enabling Developer's to make more expansive games. Which we won't see until new games are made (aka, a whiiiile).

2. Mantle is still premature as someone I quoted below said. There are more improvements coming. This just the "Alpha" (my name for it) Mantle update. AMD has emphasized more is to come in terms of performance boosting optimizations.

 

I think the people who are dissapointed need to look at in perspective. This is a new immature API and yet it is providing more performance from the same silicon. Clearly it's allowing the hardware to be more efficiently utilized. And we are getting it for free.

 

This. So much this.

It would be nice to see how crossfire scales with two and three cards and at higher resolutions.

Probably well. Mantle also brings HSA and frame latency fixing stuff with it too. Which should be great for frame stuttering on Crossfire setups.

[deathjester]

These are single-player benchmarks folks, in multi-player the performance almost doubled.

64 Players Ultra Details (Core i7-3770K, 16 GB DDR3-1333, Radeon R9 290X @ 1.000/2.500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)

http://www.golem.de/news/amds-mantle-api-im-test-der-prozessor-katalysator-1402-104261-3.html

Courtesy of @xAcid9

 

^ All that matters. Don't know if they are legit or not, but damnnnnnnnnnnnnn.

 

I don't see big spikes on my 4770k in that game, but 4x MSAA going from avg to perfectly playable is pretty sick. 

Siege Of Shanghai (1080p, 4x MSAA) 57.7 avg / 44 min 78.2 avg / 63 min

 

ALL CPU's can stuggle in many cpu bound games. Even my 4770k OC will dip to low 30's in Guild Wars 2. AMD's will dip to 15FPS in that game... 

 

If mantle can really do this in multiplayer on a 4770k? This is freakin awesome. 

[UnderADyingSun]

10 percentamitiles...fo free? I ain't gotsta buy nuthin? AMD, Lawdyyyyy lawddyyy I shall neva roam againnn!

[GeTecknics]

seems to atleast be a tangible increase in most cases it will probably get better with time though

[mymouthandi]

This is very interesting and I'm looking forward to seeing full reviews of Mantle performance and see how it develops when new games are released that support it. 

[LOLPhilips]

Those benchmarks are stupid.

We need Multiplayer benchmarks not Singleplayer where the CPU get's not even really used.

 Wouldn't running a multiplayer benchmark (every soldier is now controlled by a player instead of the cpu) use the cpu even less

[ACatWithThumbs]

 Wouldn't running a multiplayer benchmark (every soldier is now controlled by a player instead of the cpu) use the cpu even less

Nop, the Soldiers in BF4 are CPU bound and so are all vehicles and the netcode as well as the destruction and physics.

That's why you get a huge performance hit when you jump from a round with 32 players to 64 players it takes you up to 10-20FPS.

It calculates all players actions even when you can't see them.