Overclock.net › Forums › Industry News › Software News › [Anand] Ashes of the Singularity Revisited: A Beta Look at DirectX 12 & Asynchronous Shading
New Posts  All Forums:Forum Nav:

[Anand] Ashes of the Singularity Revisited: A Beta Look at DirectX 12 & Asynchronous Shading - Page 9

post #81 of 1175
Quote:
Originally Posted by Mahigan View Post

AMD do support multi-threaded command listing as well as deferred rendering. That's not really the issue, the issue is the command buffer size. It is tailored to match the queue size in the Command Processor Q0. That means 64 threads wide. So basically, GCN is constantly streaming commands from the DirectX Command Buffer, hitting thread0 of the CPU hard.

GCN is very parallel and thus needs a lot of work items being scheduled in order to get full utilization. So if the CPU is instructed to run some complex simulation, you get a GPU hardware stall. The GPU pretty much waits there for more commands. This of course hits AMDs draw call rate hard under GCN and DX11.

With DX12, the DirectX driver is spread amongst many cores. This helps ensure that if one CPU core is busy with other work, another will keep streaming commands to the Command Processor.

This is a mix of an API overhead and hardware issue (Queue size is too small on the Command Processor).

NVIDIA have an edge here because they've been using a Static Scheduler ever since Kepler (GTX 680). This means that a large segment of NVIDIAs scheduler is in the driver (software). Their scheduler is multi-threaded (what AMD call a hidden CPU thread). Ontop of that, NVIDIA's Gigathread Engine can hold many more threads than AMDs command processor.

So if CPU thread 0 is stuck doing other work, CPU thread 1 will be used by NVIDIA's scheduler. So NVIDIA's Kepler and newer architectures never skip a beat.

Evidently, if you move to DX12, NVIDIA won't see much in the way of a performance boost over DX11 but AMD will.

The other side of things is that under DX12 it is now NVIDIA who incur a larger CPU overhead as their Scheduler is in software and taking up cycles from the CPU that AMD GCN aren't.

Like a roll reversal but not nearly as bad as what AMD suffered under DX11 in draw call intensive titles.

OIC. So can I say this is due to hardware design rather than poor drivers (everyone seems to say its poor drivers).
post #82 of 1175
There are a some hardware/scheduler limitations, but the big issue is that AMD's poor DX11 drivers cannot keep the card fed with instructions. DX12 can keep the card fed.
post #83 of 1175
I wouldn't put too much stock in Async Compute going very far. Just like Mantle, it will most likley be short lived. Plus Ashes is an AMD backed game which is a HUGE bias. Just like the Nvidia backed games have a huge bias over AMD.

AMD vs Nvidia is ALWAYS back and forth. Some games AMD has been better at, other games Nvidia will be better at. Its just how the cookie crumbles.
post #84 of 1175
Quote:
Originally Posted by escksu View Post

OIC. So can I say this is due to hardware design rather than poor drivers (everyone seems to say its poor drivers).

Yeah, people say that AMD lack Multi-threaded command listing as well as Deferred rendering and I thought the same until a little birdie explained it to me.

There's a Microsoft Deferred rendering and multi-threaded command listing test and it works really well on AMD GCN hardware (meaning the drivers support the features).

What AMD can try and do is tune their hardware to fetch commands right before a complex simulation begins, but that varies on the title. Some titles are getting to be very complex. This is why it takes AMD some time to "optimize" a game. Take Rise of the Tomb Raider, that's a draw call intensive title. AMD are making gains but they'll likely never be able to reach NVIDIA's performance levels under DX11. Ashes of the Singularity is even more insane to try and optimize for GCN under DX11. So they left it alone.

Now if a DX11 title is relatively lower on draw calls and heavier on compute we see GCN reaching parity or near parity on the Top Cards and passing or tying nVIDIA on the Hawaii/Grenada cards. Lucky for AMD this has been a trend. That's why Kepler is falling back and the 290x passing the GTX 780 Ti. It is not due to NVIDIA abandoning Kepler. It is due to games being more compute bound. (don't be fooled by Kepler's theoretical compute numbers, it takes Kepler 9-10ns to complete a MADD instruction vs 4ns on GCN2&3 and 5-6 on Maxwell).

Polaris fixes these issues BTW smile.gif
Edited by Mahigan - 2/24/16 at 8:07pm
Kn0wledge
(20 items)
 
Pati3nce
(14 items)
 
Wisd0m
(10 items)
 
Reply
Kn0wledge
(20 items)
 
Pati3nce
(14 items)
 
Wisd0m
(10 items)
 
Reply
post #85 of 1175
Quote:
Originally Posted by BeerPowered View Post

Just like Mantle, it will most likley be short lived.

If your argument hangs on this supposition, it's pretty short-lived as well. Mantle was kind of a big deal for all the reasons we're talking about here.
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 #86 of 1175
Quote:
Originally Posted by BeerPowered View Post

I wouldn't put too much stock in Async Compute going very far. Just like Mantle, it will most likley be short lived. Plus Ashes is an AMD backed game which is a HUGE bias. Just like the Nvidia backed games have a huge bias over AMD.

AMD vs Nvidia is ALWAYS back and forth. Some games AMD has been better at, other games Nvidia will be better at. Its just how the cookie crumbles.
Async-compute is an integral part of DX12. That's like saying tesselation would be short lived.
post #87 of 1175
Quote:
Originally Posted by BeerPowered View Post

I wouldn't put too much stock in Async Compute going very far. Just like Mantle, it will most likley be short lived. Plus Ashes is an AMD backed game which is a HUGE bias. Just like the Nvidia backed games have a huge bias over AMD.


You've lost the plot, I fear.

Fact 1) Dx12 and Vulkan were both heavily informed by Mantle, thus Mantle's feature set is implemented into both APIs. This includes Async Compute.
Fact 2) Gamers have been clamoring for full core utilization for ages and both APIs give developers the ability to utilize that. Any dev that doesn't take advantage of that is going to be lambasted.
Fact 3) As soon as AMD swept the console board, people have been saying this is coming. nVidia can't ride the coat tails of GameWorks forever and if they try nobody is going to buy Pascal or following generations.
Fact 4) Even if nVidia wanted to do such a thing, Microsoft's recent habit of rapid deprecation of legacy software won't allow them. They're in Microsoft's sandbox and when MS decides it's no longer going to support Dx11 then nVidia will be forced to make the stark choice of stepping up or stepping out.

When you get right down to it, AMD played the long game and ended up flipping the script on everybody. Not only that, but I don't think MS is going to allow a 2 year phase in of Dx12. I figure a year on the outside.
post #88 of 1175
Quote:
Originally Posted by Fyrwulf View Post

You've lost the plot, I fear.

Fact 1) Dx12 and Vulkan were both heavily informed by Mantle, thus Mantle's feature set is implemented into both APIs. This includes Async Compute.
Fact 2) Gamers have been clamoring for full core utilization for ages and both APIs give developers the ability to utilize that. Any dev that doesn't take advantage of that is going to be lambasted.
Fact 3) As soon as AMD swept the console board, people have been saying this is coming. nVidia can't ride the coat tails of GameWorks forever and if they try nobody is going to buy Pascal or following generations.
Fact 4) Even if nVidia wanted to do such a thing, Microsoft's recent habit of rapid deprecation of legacy software won't allow them. They're in Microsoft's sandbox and when MS decides it's no longer going to support Dx11 then nVidia will be forced to make the stark choice of stepping up or stepping out.

When you get right down to it, AMD played the long game and ended up flipping the script on everybody. Not only that, but I don't think MS is going to allow a 2 year phase in of Dx12. I figure a year on the outside.

Its all about the money $$$$$. Not about whats best for gaming. Right now AMD is broke and indebted, they owe $2 Billion. Nvidia took most of the Market with their 80% slice of it. Nvidia has $$$$$.

You think Devs will do whats best for gaming? While catering to the smaller market? lachen.gif

Nvidia will keep pulling the same slick tricks they always have and give Devs $$$ reasons to keep them on top. AMD can buy a couple games but Nvidia can buy the whole table.
post #89 of 1175
The bench just shows how awful AMD DX11 overhead is.

If u serious about gaming now AMD is not a recommended buy until DirectX12 game gain some market share.

IMO, I dont think I will even care for AMD DirectX12 advantage until DirectX12 game market share hit 50%.
Gaming Rig
(10 items)
 
Work/Web Rig
(11 items)
 
Web Rig
(8 items)
 
CPUMotherboardGraphicsRAM
i5-2500K @4GHz stock volt OC Asus P8Z77-V LK Palit GTX 750Ti StormX Dual Corsair Vengence 1600MHz_CL8 4GBx4 
Hard DriveHard DriveCoolingOS
Seagate 80GB 7200rpm Maxtor 250GB 7200rpm Cooler Master Hyper 212+ Windows 10 Pro 
MonitorCase
BenQ XL2720z @ 1920x1080 144Hz Silverstone Ps06 
CPUMotherboardGraphicsRAM
Core 2 Quad Q9650 Gigabyte EP41-UD3L Asus 1GB Radeon 7790 DirectCU OC Corsair Gaming Ram 2GBx2 DDR2-800 
Hard DriveHard DriveCoolingOS
Samsung 80GB 7200rpm Western Digital 200GB 7200rpm Cooler Master Hyper TX3 Windows 7 pro 
MonitorPowerCase
Samsung 226bw 1680x1050 Acbel iPower 510w (450w PSU) LianLi PS05 
CPUMotherboardGraphicsRAM
AMD E350 MSI E350IA-E45 AMD Radeon 6310 4GB Kingston DDR3 1333 
Hard DriveOSMonitorPower
Western Digital 80GB Linux Slackware 14.1 Samsung 22" 1680x1050 LCD Acbel 300w PSU 
  hide details  
Reply
Gaming Rig
(10 items)
 
Work/Web Rig
(11 items)
 
Web Rig
(8 items)
 
CPUMotherboardGraphicsRAM
i5-2500K @4GHz stock volt OC Asus P8Z77-V LK Palit GTX 750Ti StormX Dual Corsair Vengence 1600MHz_CL8 4GBx4 
Hard DriveHard DriveCoolingOS
Seagate 80GB 7200rpm Maxtor 250GB 7200rpm Cooler Master Hyper 212+ Windows 10 Pro 
MonitorCase
BenQ XL2720z @ 1920x1080 144Hz Silverstone Ps06 
CPUMotherboardGraphicsRAM
Core 2 Quad Q9650 Gigabyte EP41-UD3L Asus 1GB Radeon 7790 DirectCU OC Corsair Gaming Ram 2GBx2 DDR2-800 
Hard DriveHard DriveCoolingOS
Samsung 80GB 7200rpm Western Digital 200GB 7200rpm Cooler Master Hyper TX3 Windows 7 pro 
MonitorPowerCase
Samsung 226bw 1680x1050 Acbel iPower 510w (450w PSU) LianLi PS05 
CPUMotherboardGraphicsRAM
AMD E350 MSI E350IA-E45 AMD Radeon 6310 4GB Kingston DDR3 1333 
Hard DriveOSMonitorPower
Western Digital 80GB Linux Slackware 14.1 Samsung 22" 1680x1050 LCD Acbel 300w PSU 
  hide details  
Reply
post #90 of 1175
Quote:
Originally Posted by BeerPowered View Post

You think Devs will do whats best for gaming? While catering to the smaller market? lachen.gif

Devs aren't going to have a choice. Their choice is Dx12 or Vulkan. It doesn't matter what nVidia wants, because they missed the vote.
Quote:
Nvidia will keep pulling the same slick tricks they always have and give Devs $$$ reasons to keep them on top. AMD can buy a couple games but Nvidia can buy the whole table.

Yep, right up until the FTC crushes them. American companies must follow American laws and nVidia doing that is a massive violation of anti-trust law.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Software News
Overclock.net › Forums › Industry News › Software News › [Anand] Ashes of the Singularity Revisited: A Beta Look at DirectX 12 & Asynchronous Shading