AMD Mantle API [updating]

So looks like AMD made a mountain out of a molehill?

http://wccftech.com/amd-launches-mantle-beta-low-end-cpus-not-gpus/

A 290x with a top end Intel CPU sees BF4 performance increase a maximum of 10%?
With a 260x the increase is of just 1.4%!

The larger improvements are for AMD CPUs and even then only 17.3% at the highest settings.

And then they showcase Star Swarm Demo to make a point, hiding the fact that this engine was built from the ground up to do precisely that. Mantle only came after.

It looks like they are just launching this now, to save face. They even say its Beta version.

Did you read the BF4 patch link with Dice's own tests?

Multiplayer = 7970 + 8350 = 25% faster
Singleplayer = 2x 290X + i7-3970x = 55% faster
http://battlelog.battlefield.com/bf4/news/view/bf4-mantle-live/last/
 
No I hadn't, thanks for those.
The first example quoted is, again, using a crap AMD processor, without showing at all what a decent Intel CPU does on DX11.
The second example looks more interesting, but I'll save judgement about that kind of number after independent reviewing. After all, why would AMD not use it on their PR??

The 8350 performs very well in BF4 under DX. 8350+7970 is a very good example of an upper-midrange setup, which shows more the benefit of Mantle for the average gamer then the ultra-highend setups with a $1000+ I7-4960X CPU. If you can get 25% increase in a heavy multiplayer scenario (where it after all counts the most for BF4 gamers) by just changing API, thats great.

That benchmark is, in my opinion, the one who has most value for most people. :smile:
 
The 8350 performs very well in BF4 under DX. 8350+7970 is a very good example of an upper-midrange setup, which shows more the benefit of Mantle for the average gamer then the ultra-highend setups with a $1000+ I7-4960X CPU. If you can get 25% increase in a heavy multiplayer scenario (where it after all counts the most for BF4 gamers) by just changing API, thats great.

That benchmark is, in my opinion, the one who has most value for most people. :smile:

Thanks, that is why I included it. Wanted a good spread. But also don't have the resources or time to do super extensive testing and comparison. There will be plenty of hardware review sites that do their own extensive comparisons. Though you should ignore sites that test multiplayer on empty servers.
 
The biggest performance gains can be seen when the game is bottlenecked by the CPU which can be quite common even on high-end machines and this was main goal to improve on with Mantle.
Sounds like what some of us was saying turned out to be true, Mantle's main benefit is for the CPU. which doesn't seem to be bad at all by the way.

IMO, Dice's numbers sounds less reliable, comparing single scenes as opposed to a whole several minutes run (even testing an empty MP map) .. but I could be wrong, anyways The Mantle patch is now online,ready and downloading on my PC. shouldn't be hard verifying those results now.
 
"Also, multi-GPU support is patchy at best with Mantle today as AMD told me that there were some “show stopper” bugs with CrossFire that were causing crashes in BF4. Stuttering is also being reported with CrossFire…man do we not need more of that."

So those 58% increase on Crossfire should also be taken with a grain of salt. Who knows if the fixing of those bugs will reduce that number?

Mantle is definitely not ready yet by AMD's own words. Once again AMD's Marketing failed by promising things and dates they could not achieve.
 
What's up with the DICE claim that only "GCN 1.1" GPUs will be supported initially, yet they post benches on 7970 themselves?
 
Interesting comment on BF4 Mantle post:

The screenshots all appear to be 'shopped. In screenshot 2, FCAT analysis says max 33fps but the top corner says 39fps. Screenshot 5, FCAT analysis says max 23 but the top corner says 78fps. Also.. by the way the player appears to be holding the firearm it proves the player is idle, standing still, not moving.. therefore not really emulating any kind of real-world benchmark. No action in scene = benchmark fail. Are you benching wind or leaves rustling? Lastly, single-player results net no information whatsoever about Battlefield 4. It's a multi-player game by design. EA DICE if you need tips on how to benchmark properly, swing me an email. :)
 
What's up with the DICE claim that only "GCN 1.1" GPUs will be supported initially, yet they post benches on 7970 themselves?
A few of these known issues are worth mentioning here. For starters, AMD specifically points out that performance improvements for Mantle on the HD 7000 series, R9 280X and R9 270X “will be optimized for BF4 in future AMD Catalyst releases” leaving me to believe that only the R9 290 and R9 290X (as well as the 260X) are really being supported at launch. Only those parts based on the new Hawaii architecture with support for XDMA and TrueAudio seem to be the target.
http://www.pcper.com/reviews/Graphi...rings-Mantle-Support-Frame-Pacing-Phase-2-HSA
 
Interesting comment on BF4 Mantle post:
The screenshots all appear to be 'shopped. In screenshot 2, FCAT analysis says max 33fps but the top corner says 39fps. Screenshot 5, FCAT analysis says max 23 but the top corner says 78fps. Also.. by the way the player appears to be holding the firearm it proves the player is idle, standing still, not moving.. therefore not really emulating any kind of real-world benchmark. No action in scene = benchmark fail. Are you benching wind or leaves rustling? Lastly, single-player results net no information whatsoever about Battlefield 4. It's a multi-player game by design. EA DICE if you need tips on how to benchmark properly, swing me an email.

What was interesting about it? There was no FCAT analysis done in those benchmarks. The poster obviously doesn't understand the difference between frame time (bottom left) and frame rate (top right) from the picture. The rest was an observation about how the benchmarker stopped to take a screenshot.

I don't see your point? :???:
 
Interesting comment on BF4 Mantle post:


What was interesting about it? There was no FCAT analysis done in those benchmarks. The poster obviously doesn't understand the difference between frame time (bottom left) and frame rate (top right) from the picture. The rest was an observation about how the benchmarker stopped to take a screenshot.

I don't see your point? :???:

They are using a "momentousness" FPS to show the difference, rather than a full benchmark run?
 
They are using average FPS to show the difference. The method used is described there too. :???:

Average over how many seconds/minutes? You can benchmark for 3 seconds and have an average as well. Sorry, but the description of the method is not nearly good enough. Furthermore, they simulated multi-player with bots, not in a real world scenario.

And it still stands that AMD themselves didn't show numbers as good as this ones. Why?
 
Average over how many seconds/minutes? You can benchmark for 3 seconds and have an average as well. Sorry, but the description of the method is not nearly good enough. Furthermore, they simulated multi-player with bots, not in a real world scenario.

And it still stands that AMD themselves didn't show numbers as good as this ones. Why?

Ask Repi for how long, he's in this thread. Point is, that the numbers presented is average, meaning its more then a single frame. You are moving the goal posts now. The poster you quoted didn't have a clue, so I don't see how that post was interesting in any way.

There will be enough independent benchmarks soon. If you don't like this one, that's ok with me. I found it interesting, since its the first benchmark we've seen. :smile:
 
Thanks, that is why I included it. Wanted a good spread. But also don't have the resources or time to do super extensive testing and comparison. There will be plenty of hardware review sites that do their own extensive comparisons. Though you should ignore sites that test multiplayer on empty servers.

I'm looking forward to see more benchmarks. :)

Any clue when 7000 series and R270/280 will be supported?
 
There are no bots in BF4 whatsoever .. the screen taken from MP clearly shows an empty map.

I was looking for the methodology but I couldn't find it? where is it exactly?
 
There are no bots in BF4 whatsoever .. the screen taken from MP clearly shows an empty map.

I was looking for the methodology but I couldn't find it? where is it exactly?

Its here:
http://battlelog.battlefield.com/bf4/news/view/2955063418790832031/7/

Test case 2: Standard 64-player multiplayer
CPU: AMD FX-8350, 8 cores @ 4 GHz
GPU: AMD Radeon 7970 3 GB (AMD will add support for the AMD Radeon™ HD 7970 in a later stage of Mantle’s release schedule, learn more)
Settings: 1080p ULTRA 1x MSAA
OS: Windows 8 64-bit
Level: Siege of Shanghai
Level was tested with 64 “pseudo players” that we have for our own internal testing that simulates heavy game workload that we have in multiplayer in order to get more deterministic results compared to full real multiplayer. 64 players on the large Battlefield levels is really demanding of the CPU so this test case is primarily CPU-bound.
Result: 18.87 ms/f -> 15.08 ms/f = 25.1% faster

To simplify measuring performance in the game we’ve added a new tool to the in-game console to record frame times for later analysis. Simply run “PerfOverlay.FrameFileLogEnable 1″ to start saving frame times and “PerfOverlay.FrameFileLogEnable 0″ to stop. The resulting .csv file will be located in Documents/Battlefield 4 which can be opened & graphed by Excel or other applications for viewing.Another in-game tool that is useful to use is “Render.DrawScreenInfo 1″ that will now show additional on-screen information about your CPU & GPU config, resolution and as well as if Mantle or DirectX 11 is used for rendering.
 
There are no bots in BF4 whatsoever .. the screen taken from MP clearly shows an empty map.

I was looking for the methodology but I couldn't find it? where is it exactly?

Well they say this:

Level was tested with 64 “pseudo players” that we have for our own internal testing that simulates heavy game workload that we have in multiplayer in order to get more deterministic results compared to full real multiplayer.

Although, yes, that does not need to mean bots. Which would actually make it worse I guess...

All this looks too rushed and amateurish.. if they wanted to make splash with mantle, they needed to make sure everything presented was crystal clear and objective. Which the numbers presented aren't.

EDIT - They just removed the last screenshot (which by sheer coincidence (?) was the only one showing very different FPS than what the averages show) after someone commented it was from Single Player and not Multi Player. Really??? How professional...
 
Last edited by a moderator:
I get a feeling some people here can't be pleased no matter what.

I on my side am exited to get free performance uplift no matter how big or small.
People saying the benefit is too small on top of the range Intel 6 core CPU paired with cheap Radeon R7 need to have a reality check. Maybe you can do better?

There is one aspect of Mantle I'm interested in. How big of a difference will it make to computer power efficiency. Imagine 2x290X in CF capped to 60FPS constant. Using DX11 requires quite a substancial CPU to feed GPUs which burns power. What effect will Mantle and it's more effiecient nature have on total system power? I'm going to find out later on today :)
 
Back
Top