[jadie kell]

Thanks for the input, I'll go try out the latest version of the plugin!

[Jordyn Youngman]

Just to confirm, this fixed the Helgen issues for me.



Considering my Core i7 930 @ 2.8 GHz is the Bottleneck when it comes to Skyrim and the mobile processors are much less powerful than the desktop versions, yes, it'd definately be worth trying the patch.

Ah, interesting-- I'm on a 950 with a 560TI card, the rest ist he same. So it'll probably help me considerably, too.

[WYatt REed]

Using Silverglade's second version patch, Helgen works as intended.

[Beast Attire]

@Silverglade

Arisu popped in to the PC subforum thread and this is what he said about your hotfix:



Maybe it will help you?

I implemented the fix (changed setnz to setz) as Arisu described and it appears to be working at Helgen's still.

Here is the file. It is version 1 with just the setnz to setz fix for Patches.replace.test_bit_30

I used 4shared to host it as someone suggested.

http://www.4shared.com/zip/y9_xxsOw/TESVAL_v1_with_fix.html?

Please let us know how if you find any bugs.

[Rude_Bitch_420]

Heh, was going to stop by to report the same thing. The test is reversed, that should fix the problem.

[{Richies Mommy}]

So.... wich version should we use ? Silverglade fixed version or GreatWalrus one? They are the same thing?


BTW I'm using silverglade 2nd fixed ver. and the game runs just fine

[Rob Davidson]

Yeah, kind of confused. Silverglade has 2 different ones, and now a GreatWalrus one.

[CArla HOlbert]

Well mine implements the fix as suggested by Arisu and Silverglade's uses a workaround. Both may work, but mine needs testing on the bugs we initially discovered (I tested Helgen's and it works).

Here is the quote above from Arisu:

Hey guys,

as noted yesterday, I don't have much free time during Christmas. However, I can confirm that the original code for Patches.replace.test_bit_30 was flawed.

The "setnz" should've been "setz". The fixed version posted above [Silverglade's 2nd version] also solves this problem by simply inverting the wrong value (the neg transforms 0 -> 0 or 1 -> -1, the add transforms from 0 -> 1 or -1 -> 0).

So mine just changed "setnz" to "setz" as he described.

[Alkira rose Nankivell]

I was just about to report about the invisible forge issue and saw the update, works great, thanks

[Laura Shipley]

Okay, I have uploaded another version that tries to add back the functionality of the original code. I've added the code back into the zip so people can take a look at it. It would be nice if someone like ianpatt, Alexander, Timeslip, etc. would look over the code change and make sure I did it correctly. It works for me; everyone can try it and let me know if they have any problems.

http://www.megaupload.com/?d=9VPUID4C

Version from Silverglade works in Helgen but he has stuff code there . Use the version from GreatWalrus , he did the fix correctly - in fact test_bit_30 must be named as bit_30_not_set , as ianpatt said - func result is reversed

[Fam Mughal]

Thanks for the clarification

I put GreatWalrus's into my public Dropbox so people don't need to use megaupload/4shared: http://dl.dropbox.com/u/1016838/TESVAL%20v1%20with%20fix.zip

[Peetay]

In order to show everyone how minimal the differences are here are the three versions of the significant code segment.
Original by Arisu:

; v1.3.10.0: $0051E340@@: Patches.replace.test_bit_30:   test     dword [ecx+152], $20000000   setnz    al   retn.size = $ - @b

2nd fix by Silverglade:

; v1.3.10.0: $0051E340@@: Patches.replace.test_bit_30:   test     dword [ecx+152], $20000000   setnz    al   movzx    eax, al   neg      eax   sbb      eax, eax   add      eax, 1   retn.size = $ - @b

And finally GreatWalrus' fix as suggested by Arisu (notice the setz):

; v1.3.10.0: $0051E340@@: Patches.replace.test_bit_30:   test     dword [ecx+152], $20000000   setz    al   retn.size = $ - @b

You won't get any of the two fixes at Nexus at the moment, but here in this thread.

[Darren Chandler]

Yes, that is all there is, Telegraf. I won't bother renaming the function as I don't want to confuse people with even more files :X

[Hairul Hafis]

I don't really care which one you publish; they effectively do the same thing. The setz version will save a few clock cycles. But as was noted earlier, the name should probably be changed to test_bit_30_not_set.

[Anthony Diaz]

Ok, then renaming the function could be done when this gets a dedicated page at skse.silverlock.org. I wonder if setz fixes all broken scripts now.
Anyway since other bottlenecks have been found already, this is going to be an interesting project. Thanks to you all guys, especially Arisu!

[Philip Lyon]

Thank you very much for your work earlier, Silverglade. You got it working for all of us while we waited to hear from Arisu.

[Anthony Diaz]

Just chucking in the general "works for me, too". (Using GreatWalrus' fix)

Got a Phenom II BE Clocked at 3,4 GHz and had about 35% increase (from 21 to 29 FPS in Whiterun looking down onto the town) - took me by surprise so to speak - installed the SKSE/plugins folder wrong the first time (I put it into the same directory as the skse_launcher instead of the data folder - duh! )

Thanks a lot, I'm also looking forward what (and if) further optimizations will be made.

[Melis Hristina]

Gonna join in on the kudos
I've just installed the latest nVidia beta driver (290.53) which for my increased outdoor fps by at least 20 but it still didn't feel right and I kinda thought that my Phenom X4 975 might be the bottleneck so that tweak/fix finally got my rig to run Skyrim on 1920, decent AF + AA and Ultra presets. You guys are my heroes

[Sabrina Steige]

for those that is not having any increase in their performance,check again if you guys are using any ENB patch (any version) that has .dll and enbpatch.ini in the patch..if you are,try to remove the enbpatch files and the start the game again with skse_loader.exe

it works for me after i delete the files...

NOTE:it seems enb's overide the TESV accel code even when the enb was there before you put the skse and the plugin...hope it works for you guys..

[Nicole Elocin]

So are we ready to upload a version on to Nexus? It sounds like both Silver's fix and Walrus' fixes take different approaches. Would it be better to combine them or would that break everything?

[Emmi Coolahan]

Go with GreatWalrus' version. It's a proper fix instead of a workaround fix.

[renee Duhamel]

Someone should post up a version with the test_bit_30_not_set change... I am clueless how to compile the dll using the src files.

[Ann Church]

So are we ready to upload a version on to Nexus? It sounds like both Silver's fix and Walrus' fixes take different approaches. Would it be better to combine them or would that break everything?

According to the responses so far, it seems to work fine, so I'd say yes. The fix by GreatWalrus is actually how it was intended by me; it removes the "n" from setnz instead of working around it by a value transformation of the wrong value, which war the earlier approach to fix it.

[u gone see]

I don't really care which one you publish; they effectively do the same thing. The setz version will save a few clock cycles. But as was noted earlier, the name should probably be changed to test_bit_30_not_set.

What exactly is the deal with switching to "test_bit_30_not_set"? Is it just a proper naming convention thing, or does it actually impact function?

[kelly thomson]

I gained 20 fps in cities after installing this! It's like magic! =D

Though after reading through this thread i understand that the mod is causing some kind of problem? What exactly?