Best .hack // G.U Settings?
#11
Back up, delete, and recreate your inis.

Some of the old outdated stuff being in the ini has caused effects like this. So try that and see. If it doesn't help, just restore your inis back that you backed up.

Likely not that but worth a shot.
[Image: XTe1j6J.png]
Gaming Rig: Intel i7 6700k @ 4.8Ghz | GTX 1070 TI | 32GB RAM | 960GB(480GB+480GB RAID0) SSD | 2x 1TB HDD
Reply

Sponsored links

#12
It's a fresh download really. Everytime I download a new PCSX2 folder, and just copy my saves over, so it's nothing like that.
|| Core i7 8700K @ 5.0GHZ || Gigabyte AUROS 1080 Ti 11GB @ 2050MHZ/12GHZ Memory || 32GB DDR4 @ 3000MHZ || 1TB Samsung Evo 850 SSD + 128GB 830 SSD || 4TB HDD + 1TB HDD || Windows 10 64-bit Professional || EVGA G3 850w PSU || MB: Z370 AORUS Gaming 5 ||
Reply
#13
Just tried the latest build, and sure enough the artifacts are there. This is a VU issue, using superVU solved the problem. Guess either ref or ssakash broke something with their latest VU commits.
Reply
#14
Then @K.F or @Mkilbride -

Please use the binary search method to determine which revision broke it so we can make a report.

Copied from a post by Bositman:


So with your awesome testing skills you've found a bug, yay! (not so hard now is it? Tongue)

But in order to fix it, it helps tremendously to find out which code commit caused it (not always possible, but we will assume so here).

To do this, we will use the binary search method, which guarantees us that we will find the code commit we are looking for in 5 tries maximum.

How does it work?

Let's say for example that we currently are at revision 500 of github (we will use build bot revisions since github hashes don't help much in this case) and the bug is there.

Our range is revision 1 to revision 500 and we know that rev 500 has the issue.

We start by testing the revision between our range, exactly in the middle, so revision 250.

If the issue occurs in rev 250, it means the bug is somewhere between 1 - 250, else it's between 250 - 499

a) 1- 250 we repeat the process so we test rev 125
b) 250 -500 we repeat the process so we test rev 375 ( 250+500 / 2 )

and so on and so forth until we end up with a revision that doesn't have the bug and the exact next revision has the bug.
The revision we are looking for is the last one that didn't have the bug so both of the above statements must be true.
a)Revision X doesn't have the bug
b)Revision X+1 (the exact next one) has it.


Here is a visual example of the algorithm, while searching for number 76.

[Image: 08fig02.jpg]

Our range is 2 to 94 and we start our search from the middle, 47.
Since 76 is larger, we then search from 51 to 94 and start from the middle, 77
Since 76 is smaller, we then search from 51 to 76 and start from the middle, 64.

Since 76 is larger, we then search from 76 to 76 ( Tongue ) Oh wait we found it! Wink
[Image: XTe1j6J.png]
Gaming Rig: Intel i7 6700k @ 4.8Ghz | GTX 1070 TI | 32GB RAM | 960GB(480GB+480GB RAID0) SSD | 2x 1TB HDD
Reply
#15
I managed to take a screenshot of it...seems someone else already confirmed it though.


[Image: OesZc2m.jpg]

So wait, yo uwant tus to search through a few hundred builds? (Mine is 1050, the newest one is 1250 or so...we're not sure which ones it worked on / at. Hmm.

It sounds like a lengthy process!

I can however, confirm SuperVU fixed it for me as well, as he said above. Still no shadows though! Oh well. I do have ghosting now though, I've tried with and without Half-Pixel offset, I wonder...

Tried:

Half-Pixel Off-set

Skipdraw 1, 2

Alpha Hack.

8 Bit textures on / off

OpenGL / High Blending mode / DX11. No fix. Hmm...
|| Core i7 8700K @ 5.0GHZ || Gigabyte AUROS 1080 Ti 11GB @ 2050MHZ/12GHZ Memory || 32GB DDR4 @ 3000MHZ || 1TB Samsung Evo 850 SSD + 128GB 830 SSD || 4TB HDD + 1TB HDD || Windows 10 64-bit Professional || EVGA G3 850w PSU || MB: Z370 AORUS Gaming 5 ||
Reply
#16
Nah I wouldn't ask you to search that many. As the text says - the binary method would enable you to do it in a lot less time(though IDK about max 5 as Bosit said, unless I'm doing it wrong)

Example. Current revision is 1358. Let's pretend the bug is introduced in 996(but we don't know that yet). We know that the bug DOES exist in 1358. Let's assume we are testing them ALL, so all the way back to 0.

We first try the midpoint, which is 679. It doesn't have the bug. Our new range is 679-1358. Always go to the midpoint and round down on halves.
We try 1018. It has the bug. Our range is 679-1018
We try 848. No bug. (848+1018)/2 is 933
We try 933. No bug. (933+1018)/2 is 975.5
We try 975. No bug. (975+1018)/2 is 996.5
We try 996. Bug! (975+996)/2 is 985.5
We try 985. No bug. Midpoint is 990
We try 990. No bug. Midpoint is 993
We try 993. No bug. Midpoint is 994
We try 994. No bug.
We try 995 as it's all that's left. No bug.

So the revision it was introduced in was 996. We searched 1358 builds in 10 tries.

It's a fair bit of work yeah but with a good save you should be able to handle it in 30 minutes or so. Also you don't have to start from 0 if you know a build that DOESN'T have the bug, you can start with it as the zero point, 1358 as the top, and make a midpoint.
[Image: XTe1j6J.png]
Gaming Rig: Intel i7 6700k @ 4.8Ghz | GTX 1070 TI | 32GB RAM | 960GB(480GB+480GB RAID0) SSD | 2x 1TB HDD
Reply
#17
Or just find the commits that deals with VUs and test them and the the ones before them, there aren't many of these lately.

The issue is not present at 956, but appears at 963, there are no builds in the buildbot between these. So I guess the problem is from commit 957, since it is the only one between these that deals with the vu.

Quote:v1.3.1-957-g9c7d417 refractionpcsx2 mVU: Tweak the flag hack again so it plays nice with Bloody Roar (and any other which started experiencing issues). - Now makes sure the previous 4 calculations are taken.
Reply
#18
^ Nice.

I will file a report on github.
[Image: XTe1j6J.png]
Gaming Rig: Intel i7 6700k @ 4.8Ghz | GTX 1070 TI | 32GB RAM | 960GB(480GB+480GB RAID0) SSD | 2x 1TB HDD
Reply
#19
Awesome. This is all awesome, finding a bug and hopefully getting it fixed.

Curious if there is any way to get shadows / remove the ghosting in it? Or does the VU fix it? because I didn't notice it before switching to SuperVU, the ghosting that is. Shadows have always been gone, except in SW mode. I mean I can deal without them, but it'd be cool to have em'.

Hmm, nope, ghosting is there in MicroVU as well. Curious. Skidraw does not help, SKipdraw2 in fact breaks the game hard.
|| Core i7 8700K @ 5.0GHZ || Gigabyte AUROS 1080 Ti 11GB @ 2050MHZ/12GHZ Memory || 32GB DDR4 @ 3000MHZ || 1TB Samsung Evo 850 SSD + 128GB 830 SSD || 4TB HDD + 1TB HDD || Windows 10 64-bit Professional || EVGA G3 850w PSU || MB: Z370 AORUS Gaming 5 ||
Reply
#20
Done: https://github.com/PCSX2/pcsx2/issues/939

Thanks for finding it.
[Image: XTe1j6J.png]
Gaming Rig: Intel i7 6700k @ 4.8Ghz | GTX 1070 TI | 32GB RAM | 960GB(480GB+480GB RAID0) SSD | 2x 1TB HDD
Reply




Users browsing this thread: 1 Guest(s)