Anyway to only partly reduce the GPU load from SVP's GPU acceleration?

I already said that my TV already has great 720p --to-> 1080p upscaling, so spending GPU resources on any sort of upscaling would be quite wasteful...

Better, but still maxing out the GPU.

With this my SVP index is like 0.75x. For reference, when SVP is set to "CPU only", my SVP index is like 0.95x.

And to clarify, this is in fact being tested with a 1280x720 video.

One thing I'm concerned about is that doing this sort of thing will result in me having to reduce my SVP profile settings for 960x540 and lower resolution videos.  In particular, I can run pretty much run maxed out sane settings (read: not 8px or lower) for 960x540 videos (uniform + complicated + half pixel + 12 pixel...etc).

That's not upscaling, they're sharpening filters.

If you use MPC, and you can't get Svp to use it all, you should use MadVR anyway, even if you don't want use the super intensive scalers.

Turns out that switching my LAVfilters decoder from DXVA2 (copy-back) to Nvidia CUVID gives better performance...but that alone isn't quite enough.

Oh I'm an idiot; I know that x5 is quite a bit more GPU-intensive than x4, so I should have expected that x2.5 would be more demanding on the GPU than x2!  That makes 24 & 25fps content work great and I can crank up the settings to much higher levels (uniform + complicated + one pixel + 8px; half pixel is too CPU-intensive).

However, 720p 30fps AVC content seems to still be too demanding...but I've got something for that as well.

You see, to really determine if I was in fact GPU-limited, I had underclocked the GPU shaders and my SVP performance dropped accordingly...so I got thinking, since my temperatures are definitely cool enough (I've got a 120mm fan blowing directly onto the GPU and I even tested with furmark), what if I overclocked the GPU?

Sure enough, once I set the shader to around 1650MHz, 720p 30fps AVC "just worked" (sometimes as low as 1610MHz would work, but with my test video 1650MHz always worked).  Not only that, but it seems like the GPU is stable even up to above 1800MHz (maybe because it's a Quadro and therefore is binned higher?), so using something like 1700MHz should actually be quite within the stable range while still giving me a decent amount headroom.  Oh, and this is all without even altering the voltage (which cannot be changed anyway).

But there is still some bad news - 1080p 30fps downscaled to 720p is still too demanding.  For reference, using SVP to crop the video to 720p works perfectly fine, but downscaling does not.  Is there anyway to just do some light-and-fast CPU-based bilinear downscaling or something? (you know, similar to what MPC-HC's non-PS2.0 bilinear resizer)

This doesn't seem to make a difference when using x2, or if it does then it's very minor (both uniform and 1m lagged at 1600MHz GPU shader but worked at 1650MHz), and using 1m + x2.5 interpolation is too slow even with my GPU shader overclocked to 1800MHz.

Since my display can't handle 48Hz but can handle 50Hz (and 60Hz), I may just look into using Reclock or similar to speed up 24fps content to 25fps.

Lastly, may I ask for more details on just what those two override.js settings you provided are doing?  In particular, if I comment out "smooth.linear = false;" the performance is still fine with my shader at 1650MHz, but if I instead comment out "smooth.cubic = 0;" then performance isn't good unless I use a shader clock of like 1775MHz (same goes for if I comment out both settings).

OK, but what does it actually do?  Like, what visual difference is there if it's disabled vs enabled?

Also, maybe you missed it in that massive post, but is there anyway to set SVP's resizer to be a very light-and-fast CPU-based algorithm? (much like MPC-HC's non-PS2.0 "bilinear").  Alternatively, what are my choices for lighter-on-the-GPU scaling algorithms in SVP?

What about the "smooth.linear" setting?  What does it do?

Awww.   Thing is, using SVP for 1080p is completely out of the question for this PC as the performance is just nowhere even close to adequate.

I hate to triple post, but it's been nearly 5 days, my "new" HTPC is fully up and running, and there's been no response to my previous post... Maybe I'm simply asking the wrong question?

If that's the case, then let me ask this - is the massive visual difference in the interpolated frames for "cubic enabled" and "cubic disabled" the expected result?

In other words, is my so-called "issue" that I am running into actually not a problem at all and is in fact normal operation?

> is the massive visual difference in the interpolated frames for "cubic enabled" and "cubic disabled" the expected result?

not really

> What about the "smooth.linear" setting?  What does it do?

does frame composition in "linear light"

> is there any way I can add the smooth.cubic setting

> why variable framerate + non-cubic via GPU = seizures

seriously, I've no idea
the method of frame composition is not connected to vfr in any way

> I'm disappointed
I'll try harder to work better!

I haven't found a foolproof solution to the flickering issue, but I have figured out a lot about what it's deal is.

Simply put, at least on the NVS 3100m, the flicker goes hand-in-hand with SVP using more GPU utilization than it should, and unfortunately this seems to occur almost randomly (seemingly triggered by specific video files or something that don't really always have anything in common).

I had a situation where the flicker was occurring on a specific video, I only changed SVP to "repeat frame", relaunched the video, and the flicker was gone; GPU utilization was around 50%.  I changed SVP back to "blend adjacent frames", relaunched the video, and the flicker was back; GPU utilization was around 85%.  So I set SVP to "repeat frame" again, launched the very same video one more time, and the flicker was still there; GPU utilization was still around 85%.

Normally I'm able to do x3 interpolation on 20fps videos...though definitely not with the flicker/high-GPU-load issue.  Now what was interesting is that, when the flicker/high-GPU-load issue was occurring, I was able to greatly reduce the flickering by setting SVP to do only x2 interpolation which, as we all now know, reduces the GPU utilization; sure enough, the resulting GPU utilization was only around 80% (was ~95% with the flicker issue @ x3, no idea what GPU load at x3 without the flicker issue is).  To add weight to this theory, reducing my GPU's shader clock not only increased the GPU load resulting from SVP but also increased the amount of flickering.

This makes me wonder if people with powerful GPUs never notice this issue of higher-than-normal GPU utilization because on something like a Fury X it could simply be the difference between 10% and 12% GPU utilization; this could very well explain why I seem to be on my own here even though I've ran into this issue on two very different combinations of PC hardware...though now that I think of it, both PCs have one thing in common - relatively weak GPUs.

EDIT: Haha, seemingly random is right.  The above-attached "problematic" MP4 that was always triggering the flicker issue when I made my previous post is working perfectly fine.  However I've now run into another video that keeps having the issue, but I think I've found a repeatable method of making it go away for at least a single video-playback session:

hit "stop" in MPC-HC, close MPC-HC; re-open the video - assuming resume is enabled, it should resume playing right where it was stopped but without any flickering and with normal GPU utilization.

EDIT 2: I think I may have finally solved the flicker issue once and for all, but before I say what it is that I did, I would prefer to give it a few days worth of testing before we jump to any final conclusions.

Nevertheless, I would recommend to any SVP devs to not bother spending any time on this issue and/or with my bug reports until after I can confirm/deny whether or not I have in fact finally figured out what solves it - otherwise you devs may very well just be wasting your time chasing a dead end.

try this (32-bit plugin)