|
|
#1
|
|||
|
|||
![]()
I re-wrote a lot of the internal handling of compiled code a few builds ago. This is one of the few games that came off worse for wear for it. I was debugging in to it today to see if I could understand what was going on with the game and why it behaved worse.
I have worked out why the game is worse off. Where in 1.6 I could use protected memory and the game worked great. I currently have to turn on other self modification methods that slow the game down significantly. What the game is doing is it has 3 buffers with code in memory and one virtual address pointing to these code blocks. It constantly is changing the buffer that it points to, so turning on self mod methods detects this change and constantly recompiles these buffers. What broke it in latter beta builds is that I found the virtual address and said this code already existed and replaced it with the old one. So it had both virtual locations pointing to the same physical location. I had built the original solution on how games like golden eye and conkers bad fur day used the TLB. I am going to re-write some of the function lookup to make use of Physical address as well as virtual address. I hope I should be able to fix this problem in the next beta build. Read here to get access to the beta development of pj64 |
#2
|
|||
|
|||
![]()
Thats a pretty nice fix
![]() Any luck btw with the bug I described a few months ago (with the "weird" error, and what other people have encountered with the emulator core)? |
#3
|
|||
|
|||
![]()
Which bug in particular, maybe bump it up the beta forum so I remember.
|
#4
|
|||
|
|||
![]()
great, i'm getting news from the blog now!
![]() that makes perfect sense, and why the game needs TLB unmapping? AFAIK only the two Castlevanias affected. (i think this should be in Developer Diaries btw ) |
#5
|
|||
|
|||
![]()
have you tried the other Castlevania ?
um, i hope this won't be another control - or a hack? ![]() edit: the caching sometimes makes an old version of your post show after editing. spooky ![]() |
#6
|
|||
|
|||
![]() Quote:
|
#7
|
|||
|
|||
![]()
woah. leading on from this zilmar seems to have just a made a performance breakthrough - which imho we needed - core should be 50% faster (!) for the next build (due sometime next week?), applies to many of the difficult rare games.. maybe he'll post about this too (it lead on from this issue).
|
#8
|
|||
|
|||
![]() Quote:
|
#9
|
|||
|
|||
![]()
probaly about 3 games only :P
and still need to confirm it on other peoples PC's |
#10
|
|||
|
|||
![]()
right but those games are AA titles (Goldeneye, Conker, PD? etc..) and currently the games most in need of a speedup.. then that's pretty bloody good.
|