-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
***THE BF4 BENCHMARKS THREAD***
- Thread starter pgi947
- Start date
I don't post here often. My specs are:
I74930k at 4.4ghz
32 gigs of trident x ddr3 @ 2400mhz
3 r9 290x's at stock
The 27" IPS monitor which shall not be named at 120hz
Using cat WHQL 14.4
My experience has been this in bf4:
I run ultra with 2x'msaa, post aa off, and fov of 90. Frame pacing method 1
In dx 11 the frame times are definitely more choppy than in mantle. I also lose about 10-15% performance.
In mantle there will be a periodic very brief frame drop to single digits once about every 20-30 minutes last maybe a fraction of a second Otherwise it is pretty flat. The frame times are super flat if I cap fps at 125, if I uncap frames it is no where near as flat and I don't really gain anything as far as average frame rate but my max is higher.
Seeing others results with 780ti's I believe nvidia is still smoother than mantle. Although I do believe mantle is flat enough to where it is a negligible difference.
My results on caspian 64 player are quite a bit less than some here post and some are using slower processors and 1 less card and more aa. Not sure why this is, maybe it's because the map they are on is not as taxing? Lower fov? I don't believe it is my pc as it is pulling 14k in fire strike extreme (with the 290x's overclocked). Anyway I will keep fiddling with it to try and see if maybe it is an issue on my end.
I74930k at 4.4ghz
32 gigs of trident x ddr3 @ 2400mhz
3 r9 290x's at stock
The 27" IPS monitor which shall not be named at 120hz
Using cat WHQL 14.4
My experience has been this in bf4:
I run ultra with 2x'msaa, post aa off, and fov of 90. Frame pacing method 1
In dx 11 the frame times are definitely more choppy than in mantle. I also lose about 10-15% performance.
In mantle there will be a periodic very brief frame drop to single digits once about every 20-30 minutes last maybe a fraction of a second Otherwise it is pretty flat. The frame times are super flat if I cap fps at 125, if I uncap frames it is no where near as flat and I don't really gain anything as far as average frame rate but my max is higher.
Seeing others results with 780ti's I believe nvidia is still smoother than mantle. Although I do believe mantle is flat enough to where it is a negligible difference.
My results on caspian 64 player are quite a bit less than some here post and some are using slower processors and 1 less card and more aa. Not sure why this is, maybe it's because the map they are on is not as taxing? Lower fov? I don't believe it is my pc as it is pulling 14k in fire strike extreme (with the 290x's overclocked). Anyway I will keep fiddling with it to try and see if maybe it is an issue on my end.
The 290X's only have enough vram for 150% resolution scaling (4k res) at 1440P ultra preset. 3.9gb vram usage. Any higher and you'll likely exceed the vram limit and get stuttering. Tested it myself recently. This was using Mantle btw, vram usage may be lower in DX.
Last edited:
3.9gb.
Mine only use 3.2gb at those settings.
5120x3200 uses 5.2gb using you know what cards.
3.9gb.
Mine only use 3.2gb at those settings.
5120x3200 uses 5.2gb using you know what cards.
I tested it over the course of a long round when i did my benchmarks here. Usage starts off low but creeps up the longer you play. I'm also playing on Metro which is not one of the worst maps for vram usage either. Look at the screenshots.
1440P Ultra Preset +x4 MSAA - Mantle API
140% Resolution Scaling = 3584x2016
1440P Ultra Preset +x2 MSAA - Mantle API
150% Resolution Scaling = 3840x2160
Sorry for the rather lacklustre screenshot. Was pinned in base after i finished benching and this was the best i could get whilst under fire.
Next up, x4MSAA @4K.
1440P Ultra Preset +x4 MSAA - Mantle API
150% Resolution Scaling = 3840x2160
Screenshots please. Include the UI info i have on mine. Try Operation Firestorm after 30 minutes of gameplay.
Caspian Border for high VRAM use
Good shout that lad. Same expansion pack as well innit?
Yea Second Assault DLC.. I play a Second Assault only server quite a lot and i notice VRAM will go up more on that server than if i play most other maps. To be honest i think Shanghai is starting to use less VRAM lately so if Dice have been working on something there i dont know.
Good VRAM test is that, having Caspian along with Firestorm and Gulf of Oman
Screenshots please. Include the UI info i have on mine. Try Operation Firestorm after 30 minutes of gameplay.
Just played for 20mins in a 64man game and the memory usage was exactly the same at the end as it was when I started.
That was using my GTX 690s @1600p maxed.
We're talking about Mantle and 290X's. Not DX and AMD/Nvidia cards.
Gosh and I though this was the BF4 thread.
Check you inbox matty boy I have just sent you the GPUZ stats of my 690s maxing BF4 @1600p. As you can see they have the same VRAM peaks and dips all the way through the run.
3.9gb.
Mine only use 3.2gb at those settings.
Still waiting on your Mantle 290X vram results. That's what we're talking about here Kaap. 690 DX results are meaningless as we're talking about Mantle specifically.
Just had a go with Mantle and DX11 on the 290s @4K settings
Mantle used 4.4gb of VRAM per card.
DX11 used 3.2gb of VRAM per card.
DX11 was very smooth and playable with good minimums and and very high average.
Mantle was all over the place with fps huge spikes on the CPU and the average or minimum fps was no where near as good as DX11.
I don't think multi GPU setups, 4K and Mantle are a good option as it uses way too much VRAM and that is what is causing the spikes on the CPU. At the moment DX11 seems to be the best option for extreme resolutions.
Just had a go with Mantle and DX11 on the 290s @4K settings
Mantle used 4.4gb of VRAM per card.
DX11 used 3.2gb of VRAM per card.
DX11 was very smooth and playable with good minimums and and very high average.
Mantle was all over the place with fps huge spikes on the CPU and the average or minimum fps was no where near as good as DX11.
I don't think multi GPU setups, 4K and Mantle are a good option as it uses way too much VRAM and that is what is causing the spikes on the CPU. At the moment DX11 seems to be the best option for extreme resolutions.
So i was right.
I have just enough vram at 1440P Ultra preset with 150% resolution scaling (4K). Its a smooth and playable experience. However its right on the limit so its possible after a map or two it would exceed that limit and begin to stutter. Possibly on a bigger map, like Caspian it would not be enough and would stutter from the outset. As you're at 1600P, your usage will be even higher. Its how i knew straight away when you said '3.2GB usage and stays there' that was not correct regarding Mantle.
I expect resolution scaling uses more vram than running 'proper' 4k as well, as it has to store the image whilst it downsamples it to the resolution of your monitor.
So i was right.
I have just enough vram at 1440P Ultra preset with 150% resolution scaling (4K). Its a smooth and playable experience. However its right on the limit so its possible after a map or two it would exceed that limit and begin to stutter. Possibly on a bigger map, like Caspian it would not be enough and would stutter from the outset. As you're at 1600P, your usage will be even higher. Its how i knew straight away when you said '3.2GB usage and stays there' that was not correct regarding Mantle.
I expect resolution scaling uses more vram than running 'proper' 4k as well, as it has to store the image whilst it downsamples it to the resolution of your monitor.
So you would agree that DX11 is better for anyone considering 4K.