Is this update for BF4 Mantle!?

[EarthDog]

I get it.. since i first posted in fact. You are looking into my post(s) too deep. I am just surprised that we are seeing polar opposite results using the same driver and, for all intents and purposes, the same card (7970). Apparently, that is somehow common. Usually with a driver update, you see, generally, similar performance increases off the same card not improvements and actual losses.

The issues with other drivers and BF4 in general are not relevant in the context to me... You and PB have a 7970 and used 14.1beta and the mantle API. There should not be (m)any differences considering the GPU is the same. I mean with 'normal' situations, if I change a driver, the performance increases are similar across different cards of the same type (say 7970). THAT is why I am surprised as I have never seen antithetic behavior off the same drivers and card.

 

[ninjacore]

Alright, I've got data points! Who wants to graph them!? Seriously, though. I don't have Excel on this PC, so this might have to wait til tomorrow, but I have about 10 mins of data on Mantle and then 10 on DX11. Same server, map, play style.

Annoying that Fraps isn't patched for Mantle. Anyone know an app that can record average FPS with the new API?

I attached the csv files if anyone wants to take a look.

View attachment mantleDXcompare.zip

PerfOverlay.FrameFileLogEnable 1/0 will write the log file to your Docs folder if anyone wants to do their own.

EDIT: I'm noticing some slowdowns playing with Mantle enabled. These drivers do not seem very polished...

 

[ninjacore]

I'll be honest, I have never used "frame time", so I don't really know how to interpret this. The "benching" tool they added into BF4 with the Mantle update logs "Frame Time", "CPU Frame", and "GPU Frame". Not sure what any of those are lol

I can't attach .xlsx.

Mantle Avg - 14.39987124
Mantle CPU Avg - 14.40034678
Mantle GPU Avg - 14.07741322
DX Avg - 14.18081768
DX CPU Avg - 14.183209
DX GPU Avg - 14.12847116



The frame time chart seems to go in Mantle's favor, but as I mentioned above, it feels pretty terrible playing with Mantle enabled.

This was also a relatively small sample. About 10 minutes on each API, Altai Range, same number of players, Ultra settings, 1080P, driving around on a dirtbike, blowing stuff up.

 

[EarthDog]

The lower the frametime the better, to summarize. According to the numbers, your FPS is about the same..(a frame comes out every 14.x milliseconds) with both with mantle showing a very, very small (negligible) lead.

Mantle = 71.03 FPS (1000/14.077)
DX11 = 70.78 (1000/14.128)

No changes there as that variance can happen between runs.

EDIT: You can see that even though the average frametime/FPS are the same, Mantle has some significant hitches in it for whatever reason. Your 'butt dyno' has told you right!

 

[ninjacore]

Cool. Thanks for the explanation

What's the difference between "Frame Time" and "GPU Frame"?

This was on a 280X. Did we ever confirm if that one is even supported yet?

 

[EarthDog]

I am not sure... perhaps that is how long it takes the GPU to make it versus to the CPU to display it... no clue.

As far as the 280x and any GCN (not GCN2) card, I was under the impression it wasn't included, but just the GCN2 the truly new Rx series cards (290x/290/260x) were the only ones showing a difference. However that same article shows a 7970 in there with gains, so I am confused. Another guy here (PBudder) says he is seeing improvements, while another is actually LOSING performance (WhiteR). So, the jury is still out...

 

[ninjacore]

Hmm. How are you guys measuring FPS in Mantle, White Runner and peanutbudder?

 

[EarthDog]

I think FRAPS they said?? It will still record/log, but it will not display the FPS on screen IIRC. You can see it by using the console command though (no qoutes): "PerfOverlay.drawfps 1"

Here is an article: http://www.guru3d.com/articles_pages/amd_mantle_preview,1.html




Quote from another article linked earlier in the thread:

Please note that only the following GCN cards will work for now: AMD Radeon R9 290X, R9 290, R7 260X and Kaveri APU.

In that same article, they reference a 7970...now you can see why I am confused!!

Then this:

Q: What hardware is supported with Mantle?
A: At this time, the following Graphics Core Next-based products are supported:

AMD Radeon™ R9 Series graphics
AMD Radeon™ R7 Series graphics
AMD Radeon™ HD 7000 Series graphics
AMD Radeon™ HD 8000 Series graphics
2014 AMD A-Series APUs (“Kaveri”)

 

[ninjacore]

Confusing indeed.

Oh well. I'll give it a few weeks. I was happy with my DirectX performance, tbh.

 

[White Runner]

I couldnt get Fraps/Afterburner to show in game FPS so I used the in game command...

Mantle performance for the AMD Radeon™ HD 7000/HD 8000 Series GPUs and AMD Radeon™ R9 280X and R9 270X GPUs will be optimized for BattleField 4™ in future AMD Catalyst™ releases. These products will see limited gains in BattleField 4™ and AMD is currently investigating optimizations for them.

I posted the link back on the 2nd page, and after doing more research, it seems not all 7970's/280x's are seeing improvements. Back on 13.2 (Latest non-beta) drivers, I haven't had any issues during my 3 hour session last night (DX11)... I'll give it more time and try again down the road.

 

[EarthDog]

Just a quick note... I am still on the first page. You can set the amount of posts per page. With that in mind, I always note post #'s since that will not change per user. For example, my post #8 I mentioned it as well.

EDIT: I just noticed WhiteR has TWO 7970's... with them saying CFx is not optimized, I am not surprised now at his result. I, for some reason, was only thinking one card until I read the signature, oops!

 

[White Runner]

I actually have 4 Gigabyte 7970's (Sig is kind of out of date... gonna fix that now)... but currently, only 1 is in my gaming rig... and you're right, post # is way better than page number.

On a side note, I have come across many more 7970's/280x's that have performance gains, prolly 5-1 over those that have issues; if UPS shows up with my PSU before the office closes due to snow, then I'll take home another 7970 and test Mantle/14.1.

 

[EarthDog]

Ok, though you may want to wait for them to say CFx results improved as they specifically note they don't..............................but.........................then again, they said that with the 7970/280x as well...

hopefully your stuff comes in... looking forward to your testing.

AS for me, I have a 290 on the test bench. As soon as I get a coupe more numbers, I will do a comparison too, as well as use our repeatable passage in the game for benchmarking.

 

[White Runner]

Ok, though you may want to wait for them to say CFx results improved as they specifically note they don't..............................but.........................then again, they said that with the 7970/280x as well...

hopefully your stuff comes in... looking forward to your testing.

AS for me, I have a 290 on the test bench. As soon as I get a coupe more numbers, I will do a comparison too, as well as use our repeatable passage in the game for benchmarking.

Ever since beta, crossfire in bf4 (for me) resulted in many crashes, to the point where I went down to 1 card, and haven't gone back since. Even @ 1440p, one 7970 has been enough for adequate fps, though there are times (levelution or whatever) when events happen on maps that drop FPS like whoa... might be time for some xfire testing too

 

[iLoki]

In the patch notes for the mantle driver, the focused cards are the R9 290 and 290X, R9 280 (not X) and a few others.

The R9 280X and 7970 are specifically listed as having not been optimized in this release, so that is the likely explanation for the inconsistency in performance improvements/losses.

 

[dyckah]

they make a 280 that isnt an x?
weird, ive never seen one of them before!

 

[shadowdr]

I haven't actually done any testing in the way of frame rates but Mantle seems different, not really better, just different. There are obvious hiccups where it feels like lag but other players don't seem to experience it. The colors on BF4 are darker since the last patch so I am still getting used to that and the sound bug. Driver has stopped responding and recovered is new, it still crashes the game and you must ctrl alt delete to close the window and rejoin. It seems to crash more often at least every thirty minutes, some games I have to rejoin up to three times before it ends. Overall it seems a little smother but due to the sheer amount of aimbots and hacks, game play is about the same but some effects seem more clear like blowing snow and sparks.

To sum it up, it looks a little better and smother but there are obvious stutters in gameplay that were not there before. Crashes are more often, but FPS must be higher because I can see that hacks are not actually even facing at me when I get killed. I think it will be a better experience when the game is actually finished.

 

[EarthDog]

they make a 280 that isnt an x?
weird, ive never seen one of them before!

Not that I know of...

In the release notes, they only mention an R9 280 in CFx frame pacing improvements, not mantle, but it looks like they are missing the X or it is an unreleased card.

 

[SF101]

280 would prob be a 7950 if it was released. probably arguably more popular buy then the 7970 and 7870 not sure why they didnt come out with a 280 "7950" it was such a good priced card @ 250$ perhaps the manufacturing cost was to high to sustain that price as its nearly identical physically as the 7970..

But miners do love it as its slightly more energy efficient per kh's then all the rest.

And gamers loved them because they oc'ed well and had similar to 7970 performance for alot less $.

 

[wetlayed]

Yea all this is confusing, downloaded the latest and greatest beta drivers and i'm not seeing the the ability to enable to mantle. When i try to run the 32 bit version, the game freezes when i click on options. Think i have a problem here lol