Overclock.net › Forums › Graphics Cards › AMD/ATI › AMD MANTLE Discussion Thread
New Posts  All Forums:Forum Nav:

AMD MANTLE Discussion Thread - Page 144

post #1431 of 2769
Quote:
Originally Posted by sugarhell View Post

With the new update

Test Duration: 360 Seconds
Total Frames: 19536

Average FPS: 54.27
Average Unit Count: 4400
Maximum Unit Count: 5743
Average Batches/MS: 831.62
Maximum Batches/MS: 3111.02
Average Batch Count: 17490
Maximum Batch Count: 116265

Huge imprevement i went from 30 fps dx to 54

Nice!

You still have lower avg batch and batch per second than Nvidia users with Deferred Contexts enabled under DX11. I wonder why that is?
Edited by MilesK - 2/4/14 at 1:22pm
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
post #1432 of 2769
I tried enabling it on bf4, at 5760x1080 i get 60fps solid with DX11, Mantle gives me 5 fps or causes the game to crash....

Mantle seems to work fine at 1080p though
Silent Dominator
(28 items)
 
Classy Lady
(26 items)
 
TUF Inside
(18 items)
 
  hide details  
Reply
Silent Dominator
(28 items)
 
Classy Lady
(26 items)
 
TUF Inside
(18 items)
 
  hide details  
Reply
post #1433 of 2769
Quote:
Originally Posted by MilesK View Post

Nice!

You still have lower avg batch and batch per second than Nvidia users with Deferred Contexts enabled under DX11. I wonder why that is?

Because with mantle you dont need to do a lot of batches
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
post #1434 of 2769
Quote:
Originally Posted by sugarhell View Post

Because with mantle you dont need to do a lot of batches

That doesn't sound right at all.

Doesn't Mantle allow you to do more batches, because you are less cpu bound?
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
post #1435 of 2769
Quote:
Originally Posted by MilesK View Post

That doesn't sound right at all.

Doesn't Mantle allow you to do more batches, because you are less cpu bound?

You have that backwards. You are less CPU bound *because* Mantle batches drawcalls differently than higher-level APIs. The CPU is feeding drawcalls to the GPU at a far higher multithreaded rate, therefore the end result is that you are less CPU bound than with DX/OGL.

EDIT: Point being, Mantle's drawcalls look very, very different from DX drawcalls. They are bundled and optimized and otherwise to-the-metal batched out, so it's not an apples-apples comparison between APIs.

.
Edited by infranoia - 2/4/14 at 2:01pm
Parasite
(18 items)
 
  
CPUMotherboardGraphicsGraphics
i7 4770K @ 4.7GHz Z87 MPOWER (MS-7818) Sapphire Radeon 290x @1100/1500 EVGA 1080Ti SC2 Hybrid 
RAMHard DriveHard DriveCooling
G.SKILL 2133 Samsung 850 Pro Caviar Black Corsair H100 
CoolingCoolingOSMonitor
Corsair HG10 Corsair H60 Windows 7 x64 Sony XBR65X850B 
KeyboardPowerCaseMouse
CMSTORM Quickfire XT Corsair AX1200i Antec P280 Logitec G700 
Mouse PadAudio
Black, came with my NeXTcube 25 years ago. Sound Blaster Recon 3D PCIe 
  hide details  
Reply
Parasite
(18 items)
 
  
CPUMotherboardGraphicsGraphics
i7 4770K @ 4.7GHz Z87 MPOWER (MS-7818) Sapphire Radeon 290x @1100/1500 EVGA 1080Ti SC2 Hybrid 
RAMHard DriveHard DriveCooling
G.SKILL 2133 Samsung 850 Pro Caviar Black Corsair H100 
CoolingCoolingOSMonitor
Corsair HG10 Corsair H60 Windows 7 x64 Sony XBR65X850B 
KeyboardPowerCaseMouse
CMSTORM Quickfire XT Corsair AX1200i Antec P280 Logitec G700 
Mouse PadAudio
Black, came with my NeXTcube 25 years ago. Sound Blaster Recon 3D PCIe 
  hide details  
Reply
post #1436 of 2769
Quote:
Originally Posted by infranoia View Post

You have that backwards. You are less CPU bound *because* Mantle batches drawcalls differently than higher-level APIs. The CPU is feeding drawcalls to the GPU at a far higher multithreaded rate, therefore the end result is that you are less CPU bound than with DX/OGL.

I'm not sure how that's backwards.

Mantle should provide higher drawcall performance.

Meaning more batches. Meaning more batches per second.

His results show lower batches/ms than DX11 /w Deferred Contexts.
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
post #1437 of 2769
Quote:
Originally Posted by MilesK View Post

I'm not sure how that's backwards.

Mantle should provide higher drawcall performance.

Meaning more batches. Meaning more batches per second.

His results show lower batches/ms than DX11 /w Deferred Contexts.

Draw call=/= batches
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
post #1438 of 2769
So just to make sure I understand this...

Star Swarm has fewer calls per batch in DX11 than in Mantle?

Whats the difference then?

You have similar unit count, the same game settings... Why is DX11 with Deferred enabled putting out similar FPS as Mantle?
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
post #1439 of 2769
Quote:
Originally Posted by MilesK View Post

So just to make sure I understand this...

Star Swarm has fewer calls per batch in DX11 than in Mantle?

Whats the difference then?

You have similar unit count, the same game settings... Why is DX11 with Deferred enabled putting out similar FPS as Mantle?

Similar? Where did you see that?
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
post #1440 of 2769
The web!
Code:
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:              Enabled
Temporal AA Duration:           16
Temporal AA Time Slice:         2
Detailed Frame Info:            Off
===========================================================


== Results ================================================
Test Duration:                  360 Seconds
Total Frames:                   18636

Average FPS:                    51.76
Average Unit Count:             4221
Maximum Unit Count:             5492
Average Batches/MS:             1292.31
Maximum Batches/MS:             5703.76
Average Batch Count:            23597
Maximum Batch Count:            108048
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
Hasfail
(8 items)
 
  
CPUMotherboardGraphicsRAM
i5-4670k @ 4.9GHz MSI GD-65 Gaming GTX 780 GHz Edition 16GB G-Skill @ 2400MHz 
Hard DriveHard DriveMonitorPower
WD Red 1TB HDD (Plex) Samsung 256 GB  2560x1080 LG AX860i 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › AMD MANTLE Discussion Thread