• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Cat 14.2 B (Mantle update)

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Has AMD stated that the HD 7000 cards are now fully optimized for Mantle?
I thought that the R9 280 & HD 7000 cards were still under-supported.
no, indeed they didn't state anything about this 😉 to me better than 14.2 for sure...but far from full playability...
D3D 11 with my rig is really fast in multigpu while Mantle suffers enough...i hope that i will see better results soon, give them time to improve and optimize...
 
We had a short run with a 7850 1g and it was very bad - like unusable. Will try different setups and settings tomorrow.
 
Btw its the first time since 15 year that i use beta drivers. I always thought people was crazy doing it - helping others doing their job but not getting payed. Lol. I just need this s..t for my bf4.

I am not to complain about a beta drivers because its called beta for a reason.
 
We had a short run with a 7850 1g and it was very bad - like unusable. Will try different setups and settings tomorrow.
Vram usage ATM in Mantle is different from D3D...until they fix it to avoid memory leak.
First try the lowest settings, apply... leave and restart the game, open console and write Render.DrawScreenInfo 1 (or add this line to a user.cfg file inside the main BF4 dir) and press enter... close the console and look at the Vram usage (which Dice added on the last BF4 update)... then if it's good try to increase some settings...anyway you will have to keep MSAA off
 
Last edited:
..helping others doing their job but not getting payed...
not properly, 'cause it's true without a doubt that hardware setups are so many and there can be so many minor bugs, really hard to see sometimes by internal testing,...that becomes almost natural to get the users feedbacks/reports and fix the issues...rather maybe they should begin to do this kind of job better than before...😉
 
Vram usage ATM in Mantle is different from D3D...until they fix it to avoid memory leak.
I think Dice fixed the memory leak, or at least "a" memory leak.
I am going to go out on a limb and say maybe this is a limitation/oversight/issue/problem with Mantle. I am a proponent of Mantle(just to make clear where I am coming from,) but this almost sounds as if they need a hybrid method of storage allocation, so they don't blow out cards
with only 1gig of memory.

Just guessing, but Repi has commented on this subject, and seems to acknowledge an issue and the need to cut back on settings(maybe not a solution that makes everyone happy)
 
I'm using 1,25GB with Mantle on 64P servers at 1080p with all low/off, cept mesh quality ultra, lighting quality high, texture filtering high.
 
Van what drivers were you using? did you uninstall your old drivers before instaling the betas *that seems to be one of the biggest in causing stuttering; not fully deleting old drivers*

Of course i did. I used display driver uninstaller. And the trouble persist. I use to have 13.12.
 
On my i5 3570k and 7850
I almost completely got rid of stuttering by adding this to my config file

thread.processorcount 3
thread.maxprocessorcount 3
thread.minfreeprocessorcount 0

brings my jobthread count to 2 on mantle, same as it is on directx. Seemed to give good results with the mantle cpu stutter, almost completely gone now, like one minor stutter throughout a round. More testing is needed to see if I wasn't just lucky
 
do you know if the combination of those commands was what made the experience better? Im still trying to figure out the first one you listed (thread.processorcount), because default value is -1 default.
 
do you know if the combination of those commands was what made the experience better? Im still trying to figure out the first one you listed (thread.processorcount), because default value is -1 default.

Yes, it sets the jobthread count

as I said I am not sure if it really helped or I was just lucky though.

Edit: depending on your cpu you may need different values. like 8 core you may way to try 5.
 
Last edited:
You guys with 7850's aren't seeing good performance because your card isn't powerful enough to see a cpu bottleneck. Especially if you aren't running low settings in game.

Mantle is ideal for people going for 120fps to match their refresh rate. If you have enough GPU horsepower to be cpu bottlenecked at your graphics settings mantle is going to scream.
 
You guys with 7850's aren't seeing good performance because your card isn't powerful enough to see a cpu bottleneck. Especially if you aren't running low settings in game.

Mantle is ideal for people going for 120fps to match their refresh rate. If you have enough GPU horsepower to be cpu bottlenecked at your graphics settings mantle is going to scream.
no, already explained to you...but you don't want to understand :biggrin: don't you? Maybe just one is not so powerful but the Mantle parallelism in rendering is different from D3D parallelism, so if you have 4-8 cores the performance must be better (definitely never worse) frametime etc.
then, two 7850's can cause more bottlenecks, instead, 'cause sometimes they are together more powerful than a single 7970...and less CPU overhead helps crossfire, already been documented...
but the main issue now with these cards is the Vram usage and lack of optimization...not the pure performance...and if a Pitcairn doesn't stutter and no weird fps drops in D3D at 1080p it has not to stutter in Mantle too...easy.
If it does, instead, there is clear problem about Vram, testing and optimization...
 
Last edited:
After playing BF4 a lot, I am still getting terrible stuttering. It may not be as often, but anytime it happens I am almost guaranteed to die.

But when it does not stutter, it runs way better than DX for me.
 
Stuka - I'm getting alittle also; some have been adding to the user config file for BF4; locking cores to like 3 if you have a 4 core cpu..

5 with 8 core......I'll be testing some of that tonight when I get home.
 
Turn off AA for the time being, Mantle doesn't seem to play nice with AA enabled in BF4 for now- or at least what I've gathered, I've had it bad on Zavod 311 with both 2x and 4x MSAA but with it turned off it's smooth as butter, but keep post-aliasing at high, it looks better even with no MSAA.
 
no, already explained to you...but you don't want to understand :biggrin: don't you? Maybe just one is not so powerful but the Mantle parallelism in rendering is different from D3D parallelism, so if you have 4-8 cores the performance must be better (definitely never worse) frametime etc.
then, two 7850's can cause more bottlenecks, instead, 'cause sometimes they are together more powerful than a single 7970...and less CPU overhead helps crossfire, already been documented...
but the main issue now with these cards is the Vram usage and lack of optimization...not the pure performance...and if a Pitcairn doesn't stutter and no weird fps drops in D3D at 1080p it has not to stutter in Mantle too...easy.
If it does, instead, there is clear problem about Vram, testing and optimization...

The point is that if crossfire is not working properly with mantle and 7000 series cards then you are basically either dealing with poor frame times, or a single 7850. There is not enough GPU horsepower with a single 7850 to really allow mantle to shine. As english is obviously not your first language I really do not think you fully understand what I am saying.
 
Turn off AA for the time being, Mantle doesn't seem to play nice with AA enabled in BF4 for now- or at least what I've gathered, I've had it bad on Zavod 311 with both 2x and 4x MSAA but with it turned off it's smooth as butter, but keep post-aliasing at high, it looks better even with no MSAA.

Hmm, I will give this a try tonight. I hate playing with no AA, but if it makes it smooth its worth a try until a fix is out.

If that doesn't work, will try the config change for the number of cores. Although I think I really need all 4, not sure 3 is enough with my CPU.
 
Hmm, I will give this a try tonight. I hate playing with no AA, but if it makes it smooth its worth a try until a fix is out.

If that doesn't work, will try the config change for the number of cores. Although I think I really need all 4, not sure 3 is enough with my CPU.

I actually think both msaa and post aa makes the look less sharp but I can also see why people use it.

I wouldn't fiddle a lot with the game setup. Just wait 2 to 3 more weeks.
 
Turn off AA for the time being, Mantle doesn't seem to play nice with AA enabled in BF4 for now- or at least what I've gathered, I've had it bad on Zavod 311 with both 2x and 4x MSAA but with it turned off it's smooth as butter, but keep post-aliasing at high, it looks better even with no MSAA.

Runs fine on my machine. 4xMSAA @ 2560x1440 uses up >3GB VRAM though.
 
Hmm, I will give this a try tonight. I hate playing with no AA, but if it makes it smooth its worth a try until a fix is out.

If that doesn't work, will try the config change for the number of cores. Although I think I really need all 4, not sure 3 is enough with my CPU.

AA puts the bottleneck back on the GPU making mantle basically the same as dx.
 
Just an update about my stuttering troubles on some DX games. After cleaning up my pc of annoying apps like teamviewer or adobe updater, it seems to work smoother. Still some stutters from time to time, but i see an overall improvement.
 
Just an update about my stuttering troubles on some DX games. After cleaning up my pc of annoying apps like teamviewer or adobe updater, it seems to work smoother. Still some stutters from time to time, but i see an overall improvement.

That makes sense, and also explains why reducing the number of cores in use helps also. I've noticed with Mantle using all cores on my PC, every other application becomes extremely unresponsive, especially Chrome, which spawns a thread per tab. It's really a throwback to gaming on a single core PC circa 2001, everything is fighting for resources and you have to turn off some background tasks to get a smooth experience.

DICE really should have left at least one core free by default, as the line between "gaming mode" and "normal mode" is blurrier for PC these days, people like to watch Netflix, streams, Skype and such while playing games.
 
Last edited:
Back
Top