Discussion in 'MSI AfterBurner Application Development Forum' started by Unwinder, Feb 20, 2017.
Is that the g-sync demo? If you have g-sync, you don't need scanline sync
I just found the gsync demo good for quick test stuff, I would like to see scanline sync working with DSR.
Guys, please give new RTSS beta a try. I think I fixed the problem with display mode switching in Assasin's Creed: Odyssey and possibly other Ubisoft games (it was in fact resource leak in Uplay overlay, which incompletely "killed" itself after seeing unknown RTSS overlay installed) via altering hook trampoline in RTSS hook engine. As a pleasant side effect, 64-bit Uplay and Steam overlays should be more compatible now and have less chances to kill themselves when seeing RTSS injected prior to them. Due to this change they should work now even with delayed injection disabled (I still left it set to 15 seconds by default, but you may experiment with reducing or completely disabling it). The rest changes include the following:
- Hook engine is now using alternate double jump x64 hook trampoline to improve compatibility with third party 64-bit On-Screen Display applications
- Added compatibility profile switch for hooking IDXGISwapChain::ResizeBuffers via VTable instead of hotpatching
- Added hook epilogs support to 32-bit VTable hook handler
- Added exclusion profile for Forza Horizon 4. Please take a note that On-Screen Display is currently not supported in this game due to its protective system limitations
- Forcible graphics and compute queues synchronization is now disabled by default for Vulkan applications presenting frames from compute queue (AMD Vulkan rendering codepaths in DOOM and Wolfenstein II : The New Colossus). Due to this change, On-Screen Display will be invisible in those games on such platforms by default. Experienced users, understanding and accepting that On-Screen Display rendering will cause performance penalty, may reenable it with PresentFromCompute profile switch
- Improved implementation of On-Screen Display rendering from separate OpenGL context (profile compatibility switch used in certain OpenGL applications, e.g. Pyre) on AMD graphics cards
- Improved SDK:
o Improved RTSSFrametimePipeSample sample. Now the sample demonstrates frametime pipe connection for applications running with both full administrative and limited user rights
o Improved NVENC plugin. Added NVIDIA 416.xx drivers family support. The plugin was recompiled with newer NVENC encoder API headers, because NVIDIA stopped supporting legacy v4 NVENC API in release 416 and newer series drivers. Due to this change NVENC plugin no longer supports pre-release 358 NVIDIA drivers
Yaaay! I was probably one of the first to notice this, so thanks a lot. I can confirm that both uplay and steam overlays (at the same time) work with this RTSS beta in ACO now.
Resolution is changing without issues in both GR Wildlands and AC Odyssey with the new beta.
Excellent, thanks for testing!
Could you release a full zipped file with Afterburner + new RTSS beta?
You can just install the new RTSS, there's no need to reinstall Afterburner.
today i install the last update for win 10 insider pro build 18262
after i install the last rtss beta 6 + msi ab again with clean install
but ........ continues my system do not have OSD in DX12 mode (bench and games) DX 11 is ok !!
No, it is development beta for testers and it is not intended to be available for eveyone on the main Guru3D downloads page.
No need to continue it in this thread.
That doesnt happen anymore as far as I tested it in Assassins Creed Odyssee. Was always a problem to be honest in AC or Ghost Recon as far back as I remember. When you change ingame resolution, the scaling was totally off and you would only see a part of the screen. As I played yesterday that didnt happen anymore. FYI Well done!
Thanks very much! I will use the current version in the downloads section
Guys, sadly I must return to the same boring topic again. One more youtube channel joined the hype train and published investigation named "Does Riva Tuner Statistics Server Overlay Impact AMD Radeon Performance?" with fundamentally wrong conclusion. Which means that instead of developing something useful or instead of simply spending weekend with my family I need to deal with the next wave of AMD fanboys, who watch this video and start painting results in wrong colors, claiming that it is GPP again, declaring RTSS as a slow bloatware etc etc etc. So I'm forced to leave my official comments here.
Differences in .1% low are used as evidence of RTSS impacting the performance and conclusion is that due to that fact RTSS is not trustworthy to be used. This is fundamentally wrong methodology and wrong conclusion. 1% low/.1% low metrics became rather popular in different tech places during the last year, but they need explicit understanding of their nature to interpret them properly by reviewer. And that review is a good live example of interpreting such metrics improperly.
.1% low for typical 30 seconds benchmarking session is covering just 30 milliseconds range of the most slowest frames during whole test run. For such testing time .1% low is normally representing just a single slowed frame, and for overlay usage scenario it is always just the first frame when overlay is getting loaded and initialized. So it is neither measuring nor reflecting the performance hit at all. Quite opposite, it is showing display driver related differences in load times (when typical Direct3D resources are created, textures are uploaded, shaders are compiled etc) for AMD and NVIDIA platforms. And that's exact reason why ingame benchmarks never take load times into account.
Stay tuned and be smart!
You should not even embed the video and send them actual traffic.
Simply ignore those people. Sadly a bad rumor will keep going on, even if it was proven 1000 times wrong. (there are a lot of other examples)
You do a great work, no, THE BEST WORK, for GPU OC and OSD tools. There is nothing on the market, not even if you want to pay, that is so reliable as AF+RTSS.
By the way: Family is more important, than Haters!!! So THANKS for you work and an have some nice days with you family.
Do let me know in the comments DOWN BELOW, and don't forget to like and subscribe. All three of you.
Did I mention the comment section is DOWN BELOW? The comments are DOWN BELOW. They are DOWN, and BELOW. In the downwards direction. Up is towards the sky, down is the opposite of that. So the comment section is DOWN, the opposite of UP. It's DOWN BELOW.
If you have any problems locating the comment section that is DOWN BELOW, let me know in the comments DOWN BELOW.
(This guy pretty much is the personification of everything I hate about most tech youtubers. First they talk stuff they've no clue about, and then describe to me the relative geometrical location of the comment section, as if I'm retarded. Which it seems I have to be in order to be able to watch any of this drivel.)
Comments and likes do push the algorithm for recommended and front-page videos does it not? Thus ad revenue and videos being 10 minute 1 second long at a minimum ha ha.
Would be nice if it could be ignored but I don't think Unwinder can do that when the discussion spreads all over the net and RTSS now has a perceived anti-AMD agenda or optimization issues which is in the drivers but it's turned into being a problem with Rivatuner and this misconception is all over the place now even if Unwinder is trying to clear it up.
And that comes with the usual stuff about bias and being the author of the software and ignore and unwillingness and yada yada.
Can't have been fun, especially having it start when as I recall you were on vacation and having to go online and try and refute some of these statements but it just keeps popping up. Which is ridiculous considering you've already answered in full detail what and where the problem is but that's ignored entirely and you get these click bait articles and headline videos and then it's back in force yet again.
Would be nice if AMD fixed the problem properly next time they overhaul the Vulkan driver but yeah I don't expect that to happen for a while but that's probably what it's going to take to stop this. (Well, at least reduce it hopefully but even that might be optimistic at this point.)
EDIT: Well not like I understand the software and tech and the driver in full or anything but Unwinder's earlier explanation is right here in this topic and that has a explanation for the entire thing but it's unfortunately being ignored or disregarded or whatever. (And these articles and videos will probably pop up until the core issue is fixed from AMD's side.)
EDIT: That or it's just how these engines handle AMD's Vulkan driver, I have more reading to do myself it would seem. DOOM and Wolfenstein New Colossus are both running on Vulkan 1.0 after all with maybe DOOM Eternal being the first idTech6 (7?) release on the newer 1.1 version and whatever SDK it'll end up using.
Lots of things changed or became standard instead of going with some AMD GCN stuff and early async compute support.
(Not my area at all, low-level API's and putting more in the hands of the developers.)
Bit of both perhaps, AMD's noted performance deficiency already covered in a earlier release note for RTSS and then these games and what they are doing. Eh whatever the case the results are this constant back and forth about how RTSS is harming AMD's performance when it's not the software's fault but it keeps coming back up every few weeks.
EDIT: Though the new beta stopping monitoring unless the user overrides the config might help put a stop to it and if it doesn't well maybe MSI themselves can help out if some of these accusations keep mounting and it starts affecting MSI Afterburner itself and their brand. Although it'd be nice if it calmed down instead of escalated to that point.
(Though I suppose Rivatuner isn't their software and it's more of a affiliation with Afterburner too. Well hopefully these articles can come to a stop sooner rather than later but I wouldn't be surprised to see this new beta release being spun into some additional articles or videos too unfortunately, easy clicks and ad revenue and all that.)
@Unwinder No offense, but why you keep finding issue with those low subscriber content creators, but you never pinpoint a fella like JayzTwoCents.
"Don't do the third party like GURU3D or anything like that"