squall-leonhart
 
 
Both there, and at the very start.
 
haven't played with ffxv all that much, i don't remember any hitching in the demo's menu but it does cycle a background song, doesn't seem to correlate with this looping at all does it?

There is a performance leak when you alt+tab out from FFXV to the desktop. Usually you have to turn full screen off from the game menu to be able to get access to the desktop, then (at least on my machine) when alt+tabbing back to game and re-enabling full screen mode the fps turns into a stutter fest, so bad I have to force close. This had occurred in the recent video I shared fighting that boss. The lag in the menu is also after I had alt+tabbed out from full screen to the desktop and back. There'd definitely an issue there... I can't recall the background music looping but it may have..
 
I think using Windows Game Bar for recording gameplay causes issues too. It seems to force the game out off full screen mode or something. My monitor has it's own fps counter and when I enable Game Bar recording the monitors fps counter is disengaged while MSI Afterburners OSD continues to read the correct fps.
 

Sajin
Heaven runs smooth for me through the entire thing except for in scene 19 when looking over the wall onto the village I get a little stutter. 

Well I ended up having that stall at the exact same spot 2-3 times now but on some runs it doesn't stutter there at all or throughout the whole benchmark :/ 
 
I don't know where to go with this (my card/system). I clean (DDU) installed the 411.63 driver this afternoon, then I loaded up my test spot in FFXV where there a few sets of monster battles before one final boss. Things seem a bit smoother, the fps still drops as low as 30, 40, 50 fps (somewhat randomly) during fights but they seem more dynamic, meaning they seem to recover from the dips quicker. I didn't have one stall, well not until I loaded up another save in a different area of the world. It stalled momentarily about 40 seconds after running around in the loaded up area :/

Sajin
If you can't reproduce this issue in every single game you own I would just chalk it up to bad game optimization. If you can reproduce it in every single game you own you most likely have a software/hardware issue. Reinstalling windows fresh should eliminate any software problems. After that you're stuck with replacing hardware until you find the issue.

The stalls can be reproduced over 2-3 different games so it's not a game unique issue. I most likely will look at an OS reinstall soon/next, not sure to wait until Microsoft October update.

ThrashZone
 
Did you use the king|pin fix yet ?
As stated there are about three place in heaven that will stall probably more than that but I'll show with screen shots exactly where in a little while I'm pushing time spy at the moment which is another benchmark you can also use it tests the gpu and cpu pretty well.
But might be nice to see what you're scoring in it.

Does the king|pin fix basically unlock the power target to what it should be on the slave Bios? Is that it all it fixes? I will test again before I apply it because I think I have seen my power go over 117%. Will do a Time Spy bench a little later also.
 

Graphite8five
So can something other than the GPU itself potentially cause the GPU usage and Power% to drop? Perhaps a Windows background process or some other software issue?
 

If the cpu isn't pushing the data to the gpu it could cause the gpu usage to drop out.

@ThrashZone My 1080 Ti hit 124% power in Time Spy, do I still need the king|pin mod/fix thingy?
 

Hi,
PCI-e link speed looks like it's hitting max 8GT/s like it's supposed to be hitting
I found if I use memory blck at 100.3 it drops to 5GT/s very weird yes.
 
You didn't post what gpu clocks you were using on that time spy score
Power and gpu activity drops I've found is usually the game/ benchmark bug like in heaven and FF XV just the benchmark if you've ever used it will show many instances od stalls horrible textures..... it's a basket case.
Adding just a little more gpu core voltage can help but often pushing the gpu memory speed can help more first to try.
Ultimately though a stall and those items dropping make sense because no activity no need for the voltage and activity to remain where it was.
 
Your gpu temp hit 68c on that screen shot fan speed never maxed out only about 80%
It's not terrible but these ftw3's like it cool and closer to 60c is ideal.
Did this one yesterday
 

Hi,
Post a screen shot like I did I edited my last post to show my clocks and stuff with a time spy score.
 

http://www.mediafire.com/..._Spy_29-09-18.png/file

ThrashZone

Graphite8five
http://www.mediafire.com/..._Spy_29-09-18.png/file

Hi,
Yeah
Drop you Voltage down to 30% not maxed out
Power limit drop down to 124%
Gpu core clock +65
Memory clock +666
See what happens.

Freeze/Crash, but.. I got stable with GPU + 60, Mem + 600, Voltage +50%.. Better score than yesterday too!
 
http://www.mediafire.com/..._Spy_30-09-18.png/file

keep in mind it all works in steps of 13 for both the clock and memory
 
My water cooled 1080ti FTW3 likes running with all sliders maxed (in AB) and +52 core and +650 on memory. I do not have FFXV so I cannot give you any input there. I do know that Steve at Gamers Nexus has said that in it's current state it is crap and not a viable bench mark for them, even with the new 20** series cards.
 
Just to note the above is for everyday usage and gaming. When running benchmarks with all the radiator fans on high I can get away with +65 core and as high as +843 on the memory.
 
Edited for typo on memory offset

From what Steve was saying they have it rendering stuff that is off screen when it doesn't need to be - a bit messy

Graphite8five

GGTV-Jon
From what Steve was saying they have it rendering stuff that is off screen when it doesn't need to be - a bit messy

Do you have a link? I'm interested to see this.

Here is the main one where they talk about the issues  - https://youtu.be/0eXbbh1f52I
 
In the past week or so in one of the 20** series cards benchmark videos he mentioned that FFXV was still off the table as a valid benchmark tool. Unfortunately I do not recall exactly which video it was as I had a few queued up on YouTube and listened to them while driving to and from work