mugurumakensei
Member
You're right. When I say AMD doesn't properly support DirectX 11 command lists, I mean they don't have support for it at all.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
AMD Launches Mantle in Beta Apparently its for Low End CPUs not GPUs!
- Thread starter viveks86
- Start date
You're right. When I say AMD doesn't properly support DirectX 11 command lists, I mean they don't have support for it at all.
Insane Metal
Gold Member
Star Swarm? :/
Test BF4 guys... also, questions, trying to help some friends: are the 78XX VGAs supported? And the A10 5700?
Test BF4 guys... also, questions, trying to help some friends: are the 78XX VGAs supported? And the A10 5700?
Skyfireblaze
Member
I don't even... whoever said that Mantle in its current form runs worse on 7xxx cards must have been wrong... I just installed the drivers on my 7970, enabled Mantle in BF4, restarted the game and played a bit of Shanghai 64 player and well I was blown away.
The first minute felt really stuttery as if the game was caching something but after that it became smoother and smoother, then for the first time I set all to Ultra and my fps pretty much stick glued to 60fps and the game feels quite smoother. It still drops at times when lots of stuff is happening but I never dropped below 52fps and even if it dropped it was never longer than 2 seconds, I said wow. Not that it ran bad before but it just feels... smoother, I guess that's the frametimes I feel
Well I felt them until the driver crashed but it's beta so I don't mind
My system:
Intel i5 3570k @3.4ghz (yeah I know I have to overclock that...)
Club3D 7970 Royal Queen @ 1100mhz GPU clock
8GB G.Skill 2133 DDR3 RAM in single-channel since my mobo is slightly borked
Windows 7 64bit
I will do more testing later for now I should sleep
And a side-effect, Firefox fell back to software-rendering o.o
The first minute felt really stuttery as if the game was caching something but after that it became smoother and smoother, then for the first time I set all to Ultra and my fps pretty much stick glued to 60fps and the game feels quite smoother. It still drops at times when lots of stuff is happening but I never dropped below 52fps and even if it dropped it was never longer than 2 seconds, I said wow. Not that it ran bad before but it just feels... smoother, I guess that's the frametimes I feel
Well I felt them until the driver crashed but it's beta so I don't mind
My system:
Intel i5 3570k @3.4ghz (yeah I know I have to overclock that...)
Club3D 7970 Royal Queen @ 1100mhz GPU clock
8GB G.Skill 2133 DDR3 RAM in single-channel since my mobo is slightly borked
Windows 7 64bit
I will do more testing later for now I should sleep
And a side-effect, Firefox fell back to software-rendering o.o
The Dear Leader
Banned
BF4 with Mantle works on like 3 cards right now.
RazorbackDB
Member
my gpu is stuck at 300MHz :/
Skyfireblaze
Member
Well my 7970 begs to differ, look at my post above!
As for my Firefox issue, it might have happened because I installed the drivers while Firefox was running so it might have simply needed a restart though setting gfx.direct2d.force-enabled to true in about:config and restarting the browser helped if anyone else has the same issue.
Skyfireblaze
Member
And another observation I made in BF4, ALT+TABing from and to the game feels instant now instead of taking one second, not that it matters
jediyoshi
Member
== Hardware Configuration =================================
GPU: AMD Radeon R9 200 Series
CPU: GenuineIntel
Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz
D3D
Average FPS: 14.43
Average Unit Count: 3459
Maximum Unit Count: 5676
Average Batches/MS: 504.54
Maximum Batches/MS: 734.92
Average Batch Count: 42127
Maximum Batch Count: 204199
Mantle
Average FPS: 34.12
Average Unit Count: 3963
Maximum Unit Count: 5504
Average Batches/MS: 659.88
Maximum Batches/MS: 1703.65
Average Batch Count: 23990
Maximum Batch Count: 109536
Welp, my old cpu is happy
GPU: AMD Radeon R9 200 Series
CPU: GenuineIntel
Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz
D3D
Average FPS: 14.43
Average Unit Count: 3459
Maximum Unit Count: 5676
Average Batches/MS: 504.54
Maximum Batches/MS: 734.92
Average Batch Count: 42127
Maximum Batch Count: 204199
Mantle
Average FPS: 34.12
Average Unit Count: 3963
Maximum Unit Count: 5504
Average Batches/MS: 659.88
Maximum Batches/MS: 1703.65
Average Batch Count: 23990
Maximum Batch Count: 109536
Welp, my old cpu is happy
Telephoto999
Member
Which AMD card do you own?
Coolbgdog12
Member
Your CPU has now graduated to the cinematic stage.
http://www.anandtech.com/show/7728/battlefield-4-mantle-preview
Limited data, but:
7-10% GPU improvement.
30%+ CPU improvement.
Limited data, but:
7-10% GPU improvement.
30%+ CPU improvement.
Thanks for sharing. I've got a modestly overclocked i5 760 paired with a 7950, so I was curious to see if I could expect decent gains.
Is mantle just for BF4 and Star Swarm right now? I don't have BF4 and SS is just a benchmark right?
This is looking promising though, I can't wait for a game I own or like enough to buy to come out with mantle support.
KojiKnight
Member
Woot, time to install and see if I get any noticable difference on my 7870!
mugurumakensei
Member
I don't need to own an AMD card to know they don't support Driver Command Lists. Any basic research will tell you they don't.
http://forums.anandtech.com/showthread.php?t=2346619
Telephoto999
Member
Ahhh okay. I kinda figured. Just making sure.
BTW, Mantle improved my average FPS on average about 25 on a 2500k @ 4.4 and a 290x @ 2560x1440
I was able to vsync 75hz just now on a 64 player seige of shanghai match, with a totall flat CPU graph in perfoverlay. Before I would use 60hz and vysync that and I will still get drops with Direct3d. All Ultra with 2x MSAA
Pretty amazing results. Can't wait to see what else comes from this.
KojiKnight
Member
There is a general 7-10% improvement on the GPU side of things with mantle. So assuming proper implementation in driver and game, there should be a noticable speed up regardless.
If you're running into a situation where your CPU is running behind, it will make more of a difference. Whether this means that developers will feel that more free PC resources means more they can use with high end CPUs, or whether that means they'll just ignore it and represent it as a speed boost on lower end machines.... who knows yet. Probably the ladder till it matures more.
rudieboy77
Member
DLing the new driver beta right now.
Skyfireblaze
Member
that makes no sense at all really, 7850K one of 3 cards AMD suggest will work properly....whats the rest of your specs
RazorbackDB
Member
OK fixed the 300MHz bug, had to uninstall MSI Afterburner.
FX 6300 OC
R7 260x OC
8GB Ram
FX 6300 OC
R7 260x OC
8GB Ram
Code:
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 7081
Average FPS: 19.67
Average Unit Count: 3967
Maximum Unit Count: 5562
Average Batches/MS: 397.36
Maximum Batches/MS: 884.96
Average Batch Count: 22366
Maximum Batch Count: 139110
Code:
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 10941
Average FPS: 30.39
Average Unit Count: 4101
Maximum Unit Count: 5619
Average Batches/MS: 712.55
Maximum Batches/MS: 1918.60
Average Batch Count: 26436
Maximum Batch Count: 187145
===========================================================Look at the batch counts and unit counts. They're much higher on the mantle benchmark
KojiKnight
Member
This... It was a mistake to call the Starswarm demo a benchmark. Benchmarks are 100% scripted and will play out the same way every time. One of the starswarm's appeals is that it's the exact opposite... It's a huge simulation with tons of individual non-scripted AIs engaging in a simulated battle. This means each time you run the demo it's going to be different.
To get MEANINGFUL results, you're going to have to run the bench mark 5 times or more on each setting then compare averages. Honestly, probably much more than 5 to get any good data if it's "close".
The better example for Mantle right now would be BF4 benchmarks until there is a true benchmark designed for it.
Coolbgdog12
Member
Bf4 is stuttery with mantle for me.
DX11:
== Hardware Configuration =================================
GPU: AMD Radeon HD 7900 Series
CPU: GenuineIntel
Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
Physical Cores: 4
Logical Cores: 8
Physical Memory: 8553304064
Allocatable Memory: 140737488224256
===========================================================
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 10882
Average FPS: 30.22
Average Unit Count: 3851
Maximum Unit Count: 5350
Average Batches/MS: 652.43
Maximum Batches/MS: 1192.37
Average Batch Count: 22052
Maximum Batch Count: 120047
===========================================================
Mantle:
== Hardware Configuration =================================
GPU: AMD Radeon HD 7900 Series
CPU: GenuineIntel
Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
Physical Cores: 4
Logical Cores: 8
Physical Memory: 8553304064
Allocatable Memory: 140737488224256
===========================================================
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 15713
Average FPS: 43.65
Average Unit Count: 4321
Maximum Unit Count: 5584
Average Batches/MS: 1330.68
Maximum Batches/MS: 2608.51
Average Batch Count: 34100
Maximum Batch Count: 149030
43% Increase on my i7-3770 @ 3.4 and 7970 GHz.
Although I think what's more impressive than the performance increase is what this demo actually is. It seems very dynamic, and those sweeping shots of 2000+ ships on screen at once with full lighting is fucking mindblowing.
== Hardware Configuration =================================
GPU: AMD Radeon HD 7900 Series
CPU: GenuineIntel
Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
Physical Cores: 4
Logical Cores: 8
Physical Memory: 8553304064
Allocatable Memory: 140737488224256
===========================================================
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 10882
Average FPS: 30.22
Average Unit Count: 3851
Maximum Unit Count: 5350
Average Batches/MS: 652.43
Maximum Batches/MS: 1192.37
Average Batch Count: 22052
Maximum Batch Count: 120047
===========================================================
Mantle:
== Hardware Configuration =================================
GPU: AMD Radeon HD 7900 Series
CPU: GenuineIntel
Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
Physical Cores: 4
Logical Cores: 8
Physical Memory: 8553304064
Allocatable Memory: 140737488224256
===========================================================
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 15713
Average FPS: 43.65
Average Unit Count: 4321
Maximum Unit Count: 5584
Average Batches/MS: 1330.68
Maximum Batches/MS: 2608.51
Average Batch Count: 34100
Maximum Batch Count: 149030
43% Increase on my i7-3770 @ 3.4 and 7970 GHz.
Although I think what's more impressive than the performance increase is what this demo actually is. It seems very dynamic, and those sweeping shots of 2000+ ships on screen at once with full lighting is fucking mindblowing.
KojiKnight
Member
D3D on High
Mantle on High:
Very impressive and very noticable difference when viewing. I might try mantle on extreme now. Was basically unwatchable in D3D. Still not an "ideal" benchmark (hope a more proper one comes about) but it's still a very clear and noticable difference.
Mantle on High:
Very impressive and very noticable difference when viewing. I might try mantle on extreme now. Was basically unwatchable in D3D. Still not an "ideal" benchmark (hope a more proper one comes about) but it's still a very clear and noticable difference.
Here are my results on "Extreme":
DirectX:
Mantle:
My CPU is an i5 760 OC @ somewhere from 3.4-3.9 Ghz (Supposed to be 3.4, but CPU-z reports both ~3.4 and ~3.9)
My GPU is a 7950 (MSi Twin Frozr Edition)
You can see the avg. unit count was higher on the mantle benchmark, but the avg. batch count was higher on the Dx benchmark.
My min FPS on the Dx benchmark was about 5 FPS anytime the really action heavy scenes were on screen. With Mantle I was getting 30+ FPS in these action heavy scenes, and only once when the unit count was over 5,000 did the FPS dip down to about 15 FPS.
Results seemed very impressive to me, I'd love to see mantle support in some proper games soon (I don't own BF4).
DirectX:
Mantle:
My CPU is an i5 760 OC @ somewhere from 3.4-3.9 Ghz (Supposed to be 3.4, but CPU-z reports both ~3.4 and ~3.9)
My GPU is a 7950 (MSi Twin Frozr Edition)
You can see the avg. unit count was higher on the mantle benchmark, but the avg. batch count was higher on the Dx benchmark.
My min FPS on the Dx benchmark was about 5 FPS anytime the really action heavy scenes were on screen. With Mantle I was getting 30+ FPS in these action heavy scenes, and only once when the unit count was over 5,000 did the FPS dip down to about 15 FPS.
Results seemed very impressive to me, I'd love to see mantle support in some proper games soon (I don't own BF4).
Interesting that we're seeing very similar FPS performance across 7800 series, 7950's and 7970s even with wildly different CPU setups.
The biggest difference between these seems to be the Batches/MS, Average Batch Count and Maximum Batch Count.
Does anyone care to break down what a Batch is?
Your card is not yet supported for BF4, only R9 290/290X I believe.
I tested BF4 on my 7970 and got pretty much identical performance from DX to Mantle. So I'm assuming support isn't in yet.
The biggest difference between these seems to be the Batches/MS, Average Batch Count and Maximum Batch Count.
Does anyone care to break down what a Batch is?
Your card is not yet supported for BF4, only R9 290/290X I believe.
I tested BF4 on my 7970 and got pretty much identical performance from DX to Mantle. So I'm assuming support isn't in yet.
KojiKnight
Member
In extreme... (FX6300, 7870 both stock)
D3D:
Mantle:
Hooooly crap the difference on extreme was huge for my system... More games need to implement this so I can see if it equates to any tangible differences in the "real world"
D3D:
Mantle:
Hooooly crap the difference on extreme was huge for my system... More games need to implement this so I can see if it equates to any tangible differences in the "real world"
Difference is big in Star Swarm (Extreme) for me as well.
Gygabite Radeon 7950
Intel(R) Core(TM) i5-3570K CPU @ 4.2 Ghz
D3D :
Test Duration: 360 Seconds
Total Frames: 10865
Average FPS: 30.18
Average Unit Count: 4147
Maximum Unit Count: 5417
Average Batches/MS: 564.57
Maximum Batches/MS: 1496.24
Average Batch Count: 21599
Maximum Batch Count: 130191
Mantle :
Test Duration: 360 Seconds
Total Frames: 17263
Average FPS: 47.95
Average Unit Count: 4410
Maximum Unit Count: 5818
Average Batches/MS: 941.12
Maximum Batches/MS: 2848.17
Average Batch Count: 22333
Maximum Batch Count: 167869
Gygabite Radeon 7950
Intel(R) Core(TM) i5-3570K CPU @ 4.2 Ghz
D3D :
Test Duration: 360 Seconds
Total Frames: 10865
Average FPS: 30.18
Average Unit Count: 4147
Maximum Unit Count: 5417
Average Batches/MS: 564.57
Maximum Batches/MS: 1496.24
Average Batch Count: 21599
Maximum Batch Count: 130191
Mantle :
Test Duration: 360 Seconds
Total Frames: 17263
Average FPS: 47.95
Average Unit Count: 4410
Maximum Unit Count: 5818
Average Batches/MS: 941.12
Maximum Batches/MS: 2848.17
Average Batch Count: 22333
Maximum Batch Count: 167869
LiveSpartan235
Neo Member
Star Swarm on High
Gonna try extreme now
Gonna try extreme now
=====
longdi
Banned
Using 290X, still cannot play BF4 at 4xMSAA at 2560x1600, it drops below 60fps too often for my liking. 2xMSAA meanwhile seems to work better now with Mantle at maintaining 60fps!
Although the Mantle render is buggy, i encountered in 40mins of playtime, cannot click to spawn in vehicles, game hanged a couple times, cannot click to change graphics options.... mouse support feels broken...
Although the Mantle render is buggy, i encountered in 40mins of playtime, cannot click to spawn in vehicles, game hanged a couple times, cannot click to change graphics options.... mouse support feels broken...
Did you update your drivers ?
Yes. The option isn't even available without the drivers updated. Just tried BF4 as well and crashed a few seconds into the game, before I was even able to switch to Mantle, so it looks like a driver problem for me.
LiveSpartan235
Neo Member
Ok now on Extreme
michaelius
Banned
Star Swarm is comically bad benchmark
It can have 20 fps variance beetween runs on same hardware:
All above are on 290x gtx780 is error in description.
It can have 20 fps variance beetween runs on same hardware:
All above are on 290x gtx780 is error in description.
iamshadowlark
Banned
Star Swarm is comically bad benchmark
It can have 20 fps variance beetween runs on same hardware:
All above are on 290x gtx780 is error in description.
Its procedurally generated, you never get the same demo twice so of course there will be different results...
Figured it out. If anyone else is having crashing problems, disable your integrated graphics card in your device manager.
Seeing huge gains on 7970, 2500k @ 1440p. Went from 23.06 fps average on D3D to 46.84 using Mantle on the StarSwarm benchmark.
rudieboy77
Member
I got 20 fps jump from 29 to 49 fps on that Swarm test, with a 4770k/7970.
Star Swarm Stress Test
Its purpose is different to Benchmark and it even doesn't say benchmark in name. Steam called in benchmark in shop but it isn't as name suggest.
-Benchmark is created as linear as possible to get every bench as close as possible.
-Stress test is created to get out of your PC last juices increasing amount of things on screen and so on.
It is good to compare total power of PC with amount of ships on screen, batches and so forth. FPS is secondary in this situation. And if you want to use FPS to compare you need to stick it with other numbers from demo.
Its purpose is different to Benchmark and it even doesn't say benchmark in name. Steam called in benchmark in shop but it isn't as name suggest.
-Benchmark is created as linear as possible to get every bench as close as possible.
-Stress test is created to get out of your PC last juices increasing amount of things on screen and so on.
It is good to compare total power of PC with amount of ships on screen, batches and so forth. FPS is secondary in this situation. And if you want to use FPS to compare you need to stick it with other numbers from demo.
My starswarm stats on extreme
== Hardware Configuration =================================
GPU: AMD Radeon HD 7900 Series
CPU: AuthenticAMD
AMD Phenom(tm) II X4 955 Processor
Physical Cores: 4
Logical Cores: 4
Physical Memory: 8548290560
Allocatable Memory: 8796092891136
===========================================================
== Configuration ==========================================
API: DirectX
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================
DirectX
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 4434
Average FPS: 12.32
Average Unit Count: 3790
Maximum Unit Count: 5357
Average Batches/MS: 289.63
Maximum Batches/MS: 506.81
Average Batch Count: 26778
Maximum Batch Count: 123878
===========================================================
Mantle
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 13241
Average FPS: 36.78
Average Unit Count: 4244
Maximum Unit Count: 5389
Average Batches/MS: 673.74
Maximum Batches/MS: 1487.32
Average Batch Count: 21641
Maximum Batch Count: 112660
===========================================================
Even with variance from this kind of test, that's one hell of a gain. It will be interesting to see how much better this can get when Mantle matures. Hell, if I read the release notes right, this wasn't even using my Crossfire setup.
== Hardware Configuration =================================
GPU: AMD Radeon HD 7900 Series
CPU: AuthenticAMD
AMD Phenom(tm) II X4 955 Processor
Physical Cores: 4
Logical Cores: 4
Physical Memory: 8548290560
Allocatable Memory: 8796092891136
===========================================================
== Configuration ==========================================
API: DirectX
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================
DirectX
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 4434
Average FPS: 12.32
Average Unit Count: 3790
Maximum Unit Count: 5357
Average Batches/MS: 289.63
Maximum Batches/MS: 506.81
Average Batch Count: 26778
Maximum Batch Count: 123878
===========================================================
Mantle
== Results ================================================
Test Duration: 360 Seconds
Total Frames: 13241
Average FPS: 36.78
Average Unit Count: 4244
Maximum Unit Count: 5389
Average Batches/MS: 673.74
Maximum Batches/MS: 1487.32
Average Batch Count: 21641
Maximum Batch Count: 112660
===========================================================
Even with variance from this kind of test, that's one hell of a gain. It will be interesting to see how much better this can get when Mantle matures. Hell, if I read the release notes right, this wasn't even using my Crossfire setup.