Discussion in 'Videocards - AMD Radeon Drivers Section' started by MerolaC, Sep 12, 2019.
Shader cache should help with hitching not cause it.
Mmm, this makes sense since frame presentation is handled directly by direct3d in DirectX9..
So for anyone having freesync issues you can still use this fix(becouse AMD still didnt) Its working on Polaris too.
Make sure its set to 1.
So is this driver any good for my poor little notebook/laptop. 3200U AMD Ryzen 3 Mobile Processor with Radeon Vega Graphics
Any update on Drivers for the beta launch of Modern Warfare which starts today?
I get a warning about driver (still on 19.5.2) at startup.
Clicking OK opens the page https://support.activision.com/articles/en_US/FAQ/Updating-Video-Card-Drivers-on-a-PC
Most likely, updating to 19.9.2 should solve the problem, but still I hope for 19.9.3 with optimizations for the MW.
So far 19.9.2 is working in MW. Hope for optimized version soon though.
MW is not either popular anymore nor heavy, to seek special attention by driver resources. If team is to push out a driver it will be from today until Monday, as the ending of third week of September. (19.9.3) We'll see.
and... MPC-HC + lav filters flickering issues which were fixed in 19.9.1 are back again -.-
maybe @nevcairiel can tell them how to fix it XD
The Vulkan loader or well the runtime software for it was updated again.
(More of a thing for Linux or efforts like Steam Proton and using this https://github.com/doitsujin/dxvk/releases but still noteworthy.)
EDIT: 19.9.2 and I think it was 19.8.2 or possibly 19.9.1 updated to the included VLK files from 220.127.116.11 to 18.104.22.168 too but these can be installed separately and it is backwards compatible.
Seems odd that they won't support RIS on Vulkan on Win 7 or the usual on Win 1607.
Up to now many people see Navi as a failure due to aggressive underclock. It is so sad ATI with so many years of experience just to justify horrendous beta consumption degrades performance.
They couldn't even surpass Nvidia favored games like assassin's creed series. Still on those games old 980 gtx wins all rx 580, 480 and the equally old r9 390X. Does amd team knows how to do driver coding? I can't stand high school coders.
Something is changing around in the driver though but that's not something I would be able to say much about what's going on in the actual code here.
The older GPU models use this from the INF:
Navi10 uses this:
Which changes the driver .dll like so:
atiumd64.dll -> amdxn64.dll
atiumdag.dll -> amdxn32.dll
(AMD DirectX Next perhaps or Navi or could mean anything really but that's changed.)
Might explain some of the issues if they're working on a newer code branch here though that's far above my knowledge level for what's going on in here, could be for leveraging RDNA instead of the previous GCN or it could be the start of a more extensive revamp and code overhaul but that's going to take a while if it's anything like that and that seems a bit early to stick in a shipped release driver and not some internal beta though who knows how long this has been in development really.
For Assassin's Creed specifically well now we're back to DirectX 11 and D3D11.1 and driver command lists and other features though I am curious if there might be changes to AMD's support and performance here later on as more titles make use of additional threading and oh let's see driver command list deferred render contexts and multi threaded rendering and well other complex things I only have a very basic grasp about.
Navi's keeping up well with NVIDIA though from benchmarks used via Assassin's Creed Origins and Odyssey but it's a bit of a uphill struggle until Anvil Next gets support for DirectX 12 or there's further work on the D3D11 driver code but that is a massive undertaking and could also risk regressions and other breaking changes if not done carefully and it would also take a lot of time, NVIDIA spent around a year retooling their driver code and that's a larger team size and one that had more resources to invest all of which likely factors in for what gets weighted and prioritized for the complexity involved in the development and maintaining of a GPU display driver.
Would like to see changes happen of course but not expecting it to be something that can just be done without diverting a lot of time and resources to this task which from the earlier interview of what happened to the OpenGL driver code around 2004 and then over a year or two it's not without it's problems and risks. (DOOM 3 was indeed faster at first and everything else was kinda broken and then that got fixed and now DOOM 3 was not much faster than before as a very short summary.)
EDIT: Also and this isn't just something that happens immediately either but a change like that and diverging code branches could also be a cut off point for earlier GPU model support although it's not like AMD could just drop the 7000 series up until Vega entirely and just like that or anything particularly for Polaris which is occupying the low cost bracket as Navi's still a bit costly and availability isn't quite there either yet.
Could still support both of these for several more years but something I'm wondering how that will continue and Navi is also just the start and then we'll see what happens first with Navi10 and then beyond that.
^^ IMhO it's for sure -> New Code branch for DX applications.
Congratulation for another gloriously FAILed driver - see attachment!
This driver never idles (as usual) and keeps temps/fan up.
Your problem is that you barely understand what you are looking at. You complain about small clock spikes while actual power draw is tiny.
My RX-580 ate 3 times as much at idle due to 240Hz screen.
I likely understand much more than you keen picking a fight?
This is what the bus interface should display in idle (x16 1.1)
any later driver than 19.4.3 shows me x16 3.0 permanently never seeing 1.1 again
and btw I got a single 60HZ display here hence nothing fancy!
In that picture it's showing a maximum power draw of 13.6 watts which is virtually nothing. The card is only 2 degrees celsius higher than the idle CPU. I don't see anything out of the ordinary in that. If you have hardware acceleration on in your browser that would account for that wattage.
I don't use hardware acceleration on firefox and let the cpu do that work. I have all the desktop graphics features like transparent windows turned off also, just use font smoothing. My card is between 3-9 watts on the desktop.