[Back to Index]

[00:02] --> kurtwr2 joined #scummvm.
[00:02] <-- D0SFreak left irc: Ping timeout: 246 seconds
[00:02] --> clone2728 joined #scummvm.
[00:04] --> Schnaks_ joined #scummvm.
[00:05] --> Gentle joined #scummvm.
[00:06] --> gus_ joined #scummvm.
[00:08] <-- clone2727 left irc: Disconnected by services
[00:08] Nick change: clone2728 -> clone2727
[00:08] #scummvm: mode change '+o clone2727' by ChanServ!ChanServ@services.
[00:09] sirlemonhead (bduncan22@95.44.164.183) left #scummvm.
[00:13] Schnaks (~Schnaks@p4FC4BA0D.dip0.t-ipconnect.de) got netsplit.
[00:13] kurtwr (~kurtwr@c-98-208-17-184.hsd1.ca.comcast.net) got netsplit.
[00:13] Guest66841 (~tompsson@h-238-224.a199.priv.bahnhof.se) got netsplit.
[00:13] dhewg (~dhewg@adiza.nexticom.net) got netsplit.
[00:14] Gentle` (~tier@quassel/contributors/gentle) got netsplit.
[00:14] gus (~quassel@2401:1801:7800:101:be76:4eff:fe18:1911) got netsplit.
[00:14] DrMcCoy (~drmccoy@scummvm/undead/drmccoy) got netsplit.
[00:14] _sev (~sev@scummvm/undead/sev) got netsplit.
[00:14] dhewg (~dhewg@adiza.nexticom.net) returned to #scummvm.
[00:14] #scummvm: mode change '+o dhewg' by ChanServ!ChanServ@services.
[00:24] Guest66841 (~tompsson@h-238-224.a199.priv.bahnhof.se) got lost in the net-split.
[00:24] kurtwr (~kurtwr@c-98-208-17-184.hsd1.ca.comcast.net) got lost in the net-split.
[00:24] Schnaks (~Schnaks@p4FC4BA0D.dip0.t-ipconnect.de) got lost in the net-split.
[00:25] gus (~quassel@2401:1801:7800:101:be76:4eff:fe18:1911) got lost in the net-split.
[00:25] Gentle` (~tier@quassel/contributors/gentle) got lost in the net-split.
[00:25] _sev (~sev@scummvm/undead/sev) got lost in the net-split.
[00:25] DrMcCoy (~drmccoy@scummvm/undead/drmccoy) got lost in the net-split.
[00:29] Nick change: gus_ -> gus
[00:36] <-- ST left irc: Disconnected by services
[00:36] --> ST1 joined #scummvm.
[00:36] Nick change: ST1 -> ST
[00:37] #scummvm: mode change '+o ST' by ChanServ!ChanServ@services.
[00:42] <-- Kirben left irc:
[01:00] --> Kirben joined #scummvm.
[01:00] #scummvm: mode change '+o Kirben' by ChanServ!ChanServ@services.
[01:05] --> DrMcCoy joined #scummvm.
[01:05] #scummvm: mode change '+o DrMcCoy' by ChanServ!ChanServ@services.
[01:07] --> ImperatorPrime joined #scummvm.
[01:13] <-- Dominus left irc: Ping timeout: 246 seconds
[01:13] --> Dominus joined #scummvm.
[01:28] Nick change: Javacat -> NotJavacat
[01:28] <-- NotJavacat left irc: Quit: Please, try the fish
[01:33] <-- clone2727 left irc: Quit: brb
[01:39] --> clone2727 joined #scummvm.
[01:39] #scummvm: mode change '+o clone2727' by ChanServ!ChanServ@services.
[01:42] --> Vampire0_ joined #scummvm.
[01:46] <-- Vampire0 left irc: Ping timeout: 272 seconds
[02:29] --> Jon_God joined #scummvm.
[02:57] <-- meunierd_ left irc: Quit: meunierd_
[03:17] --> RUBICN64 joined #scummvm.
[03:34] <-- clone2727 left irc: Quit: later

[03:56] <-- Ramal left irc: Ping timeout: 272 seconds
[03:57] --> Ramal joined #scummvm.
[03:58] --> Cheeseness joined #scummvm.
[04:04] <-- Poly-C left irc: Ping timeout: 272 seconds
[04:31] <-- RUBICN64 left irc: Quit: -a- Android IRC 2.1.15
[04:33] <-- MrSiggler left irc: Quit: Textual IRC Client: www.textualapp.com
[04:43] <-- LordHoto left irc: Quit: leaving
[05:02] --> bgKa joined #scummvm.
[05:02] #scummvm: mode change '+o bgKa' by ChanServ!ChanServ@services.
[06:01] --> waltervn joined #scummvm.
[06:01] <waltervn> morning
[06:09] <-- bgKa left irc: Ping timeout: 255 seconds
[06:36] --> ajax16384 joined #scummvm.
[06:36] #scummvm: mode change '+o ajax16384' by ChanServ!ChanServ@services.
[06:45] <-- Lightkey left irc: Ping timeout: 260 seconds
[06:47] --> h00ligan joined #scummvm.
[06:47] #scummvm: mode change '+o h00ligan' by ChanServ!ChanServ@services.
[06:50] <-- ajax16384 left irc: Ping timeout: 258 seconds
[06:58] --> Lightkey joined #scummvm.
[07:37] --> edheldil joined #scummvm.
[07:43] --> _sev|work joined #scummvm.
[07:43] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[07:46] --> _sev|work_ joined #scummvm.
[07:46] <-- _sev|work_ left irc: Changing host
[07:46] --> _sev|work_ joined #scummvm.
[07:46] #scummvm: mode change '+o _sev|work_' by ChanServ!ChanServ@services.
[07:47] <-- _sev|work left irc: Ping timeout: 258 seconds
[07:51] --> _marc` joined #scummvm.
[07:59] --> Poly-C joined #scummvm.
[07:59] --> UukGoblin joined #scummvm.
[07:59] <-- UukGoblin left irc: Max SendQ exceeded
[07:59] --> UukGoblin joined #scummvm.
[08:00] Nick change: Poly-C -> Guest42996
[08:01] <-- Schnaks_ left irc: Ping timeout: 272 seconds
[08:05] --> Schnaks joined #scummvm.
[08:12] --> Strangerke|work joined #scummvm.
[08:12] #scummvm: mode change '+o Strangerke|work' by ChanServ!ChanServ@services.
[08:12] <Strangerke|work> hi guys
[08:20] --> TMM joined #scummvm.
[08:25] <-- Jon_God left irc: Remote host closed the connection
[08:42] --> Kirben_ joined #scummvm.
[08:42] #scummvm: mode change '+o Kirben_' by ChanServ!ChanServ@services.
[08:46] <-- Kirben left irc: Ping timeout: 272 seconds
[08:51] <-- somaen left irc: Ping timeout: 245 seconds
[08:53] <waltervn> hey Strangerke|work
[08:53] <fuzzie> shh, it's still early.
[09:01] <waltervn> right, sorry fuzzie
[09:02] --> t0by joined #scummvm.
[09:02] #scummvm: mode change '+v t0by' by ChanServ!ChanServ@services.
[09:03] <t0by> Hi darlings
[09:09] <Strangerke|work> Hi evil one
[09:27] --> somaen joined #scummvm.
[09:57] <t0by> Hi somaen
[09:58] <t0by> I think I have something you'd like to see, here.
[10:01] <wjp> do tell :-)
[10:02] <t0by> Short story short, I'm trying an anternative fallback approach for multirect
[10:03] <t0by> which consists in, well, the plain old single giga rect approach, but after splitting the viewport videowall style
[10:03] <t0by> think think of partitionaning the viewport in 4x4 smaller screens
[10:04] <t0by> and applying the gigarect approach there.
[10:04] <t0by> So far so good
[10:04] <t0by> yesterday I tried to hack it in just to see if it was feasible
[10:04] <t0by> http://tinypic.com/r/vhc2zb/8 <--- and well, a lot of stuff which should /not/ be dirty is dirty (=pinky).
[10:05] <t0by> Take for example the first rect of the second row.
[10:05] <wjp> sounds interesting; will look later. (Have to get off train in a minute)
[10:05] <t0by> I'm trying to see if it's my mistake
[10:05] <t0by> or the game is just dirtying a whole lot of stuff for no particular reason
[10:05] <t0by> Heh, no hurry.
[10:05] <t0by> It's probably my mistake somewhere.
[10:05] <t0by> Or probably not.
[10:06] <t0by> *maybe not
[10:06] <t0by> (BTW that's my 2825320e14e6fca0082bd749a306e013ebeb89c4 commit)
[10:06] --> D0SFreak joined #scummvm.
[10:10] <t0by> No, it's fine.
[10:10] <t0by> I am probably screwing up somewhere.
[10:24] <t0by> Oh, right.
[10:30] <-- Kirben_ left irc:
[10:32] <somaen> There _might_ be fully transparent particles there
[10:32] <t0by> nope, my bad
[10:32] <somaen> What I'd LOVE to see, is this solution actually not getting a full grid over the enter manual override box
[10:32] <somaen> But that'd require opaque filtering.
[10:33] <t0by> that i plan to try as well
[10:33] <somaen> Which, actually, might be a rather decent way to get some speed
[10:33] <t0by> yup, it's on my hit list
[10:33] <somaen> I had a vague idea yesterday about not storing rects as rects, but putting start/stop points in a bunch of "Line"-objects
[10:33] <t0by> yesterday night I felt more like trying out this one, too tired to think about the opaque thing
[10:34] <somaen> And then parsing those lines to create rects afterwards.
[10:34] <somaen> Might be super silly though.
[10:34] <t0by> uhm
[10:34] <t0by> wouldn't that equal to a gazillion rects with height 1?
[10:34] <somaen> Yes
[10:35] <somaen> BUT
[10:35] <t0by> but with overlap already cured.
[10:35] <somaen> Yes, that.
[10:35] <somaen> And, to reduce the load, you could keep track of "lines that are equal to this line" as long as they continue to be equal.
[10:36] <t0by> I think it's quite interesting.
[10:36] <t0by> Or at least quite interesting j.u.l.i.a.-wise
[10:36] <somaen> And, of course drop any rects added between a start and stop point
[10:36] <t0by> I don't know if it would amount to shooting ourselves in the foot with 8-rect games
[10:36] <somaen> The 8 rect games fall into low res territory usually though
[10:37] <somaen> for 800x600, you'd have a max of 600 such objects.
[10:37] <t0by> say 300 if we use height 2?
[10:37] <t0by> i don't we'll have a whole lot of 1px high prtes
[10:37] <t0by> *sprites
[10:37] <somaen> And well, you'd cut the N^2
[10:37] <t0by> even the stars in julia are 4px tall or something
[10:37] <t0by> interesting indeed.
[10:38] <somaen> I have a hunch that it'd be possible to create rects linearly from that structure
[10:38] <t0by> i have been thinking of ways of circumventing the problem as well
[10:38] <somaen> i.e. "Start from the top, create a rect as long as the width is the same"
[10:38] <t0by> but this didn't occur to me.
[10:39] <somaen> Worst-case memory usage should be around Xres*Yres/2
[10:39] <somaen> which would be something like a checkerboard update.
[10:40] <somaen> And well, if THAT happens, you can always just go F*!" it, let's draw the entire screen, it'll all go through the cache anyhow.
[10:41] <t0by> checkerboard is not working particularly well anyway
[10:41] <t0by> even with a 12x12 grid it's still lighting up like a christmas tree.
[10:42] <t0by> uh, er
[10:42] <somaen> The coarser your grid is, the more of it will be enabled.
[10:42] <t0by> of course
[10:42] <t0by> but 12x12 is a lot
[10:43] <somaen> Besides, given that we update row by row, it might be worth considering coarser X values than Y values
[10:43] <somaen> (since we'll be dragging stuff in the ROW into cache anyhow, but not the next row)
[11:02] <-- ImperatorPrime left irc: Quit: Konversation terminated!
[11:08] --> DOSFreak joined #scummvm.
[11:11] <-- D0SFreak left irc: Ping timeout: 258 seconds
[11:23] <t0by> I'm sorry, can anybody refresh my memory? What do you pass to configure to enable gprof profiling?
[11:23] <t0by> --enable-profiling
[11:23] <t0by> nevermind.
[11:24] --> Vanfanel joined #scummvm.
[11:25] --> droid2727 joined #scummvm.
[11:25] #scummvm: mode change '+o droid2727' by ChanServ!ChanServ@services.
[11:30] <-- SylvainTV left irc: Ping timeout: 246 seconds
[11:34] --> SylvainTV joined #scummvm.
[11:34] #scummvm: mode change '+o SylvainTV' by ChanServ!ChanServ@services.
[11:37] --> uruk-hai joined #scummvm.
[11:37] #scummvm: mode change '+o uruk-hai' by ChanServ!ChanServ@services.
[11:48] <-- ^Styr_ left irc: Ping timeout: 245 seconds
[11:48] <t0by> Uhm.
[11:50] <t0by> somaen, even if patchwork is almost-linear we still manage to waste a truckload of time in addDirtyRect. The good news is that it it seemingly is almost as fast as the original - ~36 FPS and ~25-20 with the wakeup video.
[11:51] <t0by> Only problem is, "almost as fast as the original" is a rather sorry excuse for an "optimization"
[11:51] <t0by> \Lunch now
[11:51] <t0by> see you later
[11:51] <-- t0by left irc: Quit: Leaving
[12:01] --> Styrbjorn joined #scummvm.
[12:04] <-- DOSFreak left irc: Remote host closed the connection
[12:19] <-- TAS_2012v left irc: Read error: Connection reset by peer
[12:21] --> TAS_2012v joined #scummvm.
[12:30] --> D0SFreak joined #scummvm.
[12:41] --> t0by joined #scummvm.
[12:41] #scummvm: mode change '+v t0by' by ChanServ!ChanServ@services.
[12:42] <-- Styrbjorn left irc: Ping timeout: 258 seconds
[12:42] <-- Tomaz^W2 left irc:
[12:46] <-- t0by left irc: Ping timeout: 246 seconds
[12:59] --> Tomaz^W joined #scummvm.
[13:01] --> Styrbjorn joined #scummvm.
[13:02] --> t0by joined #scummvm.
[13:02] #scummvm: mode change '+v t0by' by ChanServ!ChanServ@services.
[13:02] --> GitHub50 joined #scummvm.
[13:02] <GitHub50> [scummvm] urukgit pushed 1 new commit to master: http://git.io/PAd9Dw

[13:02] GitHub50 (GitHub50@192.30.252.36) left #scummvm.
[13:06] <-- droid2727 left irc: Ping timeout: 260 seconds
[13:08] --> GitHub121 joined #scummvm.
[13:08] <GitHub121> [scummvm] urukgit pushed 1 new commit to master: http://git.io/Milkiw

[13:08] GitHub121 (GitHub121@192.30.252.44) left #scummvm.
[13:28] <Strangerke|work> Hum, removing unused variables in WIP engines may be a bad idea
[13:28] --> droid2727 joined #scummvm.
[13:28] #scummvm: mode change '+o droid2727' by ChanServ!ChanServ@services.
[13:28] <Strangerke|work> in this case, it shouldn't be a problem though
[13:41] <uruk-hai> duh, sorry :D
[13:41] <uruk-hai> i was overdiligent
[13:41] <Strangerke|work> As I mentioned, it's not a problem in that case
[13:42] <Strangerke|work> but be cautious with WIP engines, you never know if those useless variables aren't something that still has to be partially implemented
[13:42] <uruk-hai> kk, thank you :)
[14:04] <-- Strangerke|work left irc: Quit: Bbl
[14:06] <t0by> somaen, nope. even with the patchwork approach full screen blitting is faster.
[14:06] <t0by> marginally, but faster.
[14:06] <t0by> I put them both in a cpu capped vm
[14:07] <t0by> and I get something like ~6 +- 2 fps vs ~8 +- 2 fps
[14:07] <t0by> let's see if the opaque thing helps
[14:12] <t0by> this is depressing.
[14:13] <wjp> did you say addDirtyRect itself is taking up much time?
[14:14] <t0by> as a rule no
[14:15] <t0by> with the patchwork approach that's where the job is done so yeah.
[14:17] <t0by> that said i'm a bit scared that the opaque thing won't help much either
[14:17] <t0by> the patchwork thing had ~10x10 rects total
[14:17] <t0by> ~n=100
[14:17] <t0by> and iterating over them in a linear fashion was a enough to have a performance drop
[14:18] <wjp> patchwork is what exactly?
[14:18] <t0by> videowall.
[14:18] <t0by> the approach for handling fallback I was talking about earlier.
[14:18] <wjp> a 10x10 dirty bitmap?
[14:18] <t0by> yes
[14:18] <t0by> tried various values for (n,m)
[14:19] <wjp> and that makes addDirtyRect noticable? Surprising
[14:19] <wjp> what do you do in there?
[14:21] <t0by> basically, /exactly/ the current single-rect thing (from master) on each of the 100 (or whatever) videowall displays.
[14:21] <t0by> https://github.com/tobiatesan/scummvm/blob/a8713e2720c26a9a5f50f26326900ac7cc3c364d/engines/wintermute/base/gfx/osystem/dirty_rect_container.cpp#L116
[14:21] <t0by> thing is, this approach is quite obviously opacity-agnostic
[14:22] <t0by> so I'm not sure the usual quadratic algorithm with an added priority queue for opaque rects can be any faster
[14:22] <wjp> why loop over everything? You know exactly which ones are touched from the min/max coords of the new rect
[14:22] <wjp> and _rectArray.size() is probably still expensive?
[14:23] <t0by> and _rectArray.size() is probably still expensive? ====> doesbn't that basically return the value of an attribute from Common::Array?
[14:23] <t0by> isn't that basically getSize()?
[14:23] <t0by> <wjp> why loop over everything? You know exactly which ones are touched from the min/max coords of the new rect ====> not following you, sorry
[14:24] <t0by> oh, i get it.
[14:24] <t0by> interesting.
[14:24] <t0by> very interesting.
[14:24] <t0by> didn't occur to me.
[14:24] <t0by> let me try.
[14:25] <t0by> this could actually work
[14:26] <wjp> oh, rectArray is a Common::Array? That doesn't sound like the right structure for this
[14:26] <t0by> Yeah, I don't ever do random access
[14:26] <t0by> a list is probably better.
[14:26] <t0by> *a dll
[14:26] <wjp> or don't store them at all in this case?
[14:27] <wjp> but maybe that's not so important
[14:27] <t0by> wjp, for the "pure" bitmap thing i don't need them, right
[14:27] <t0by> as a fallback for the traditional, i probably do
[14:27] <t0by> but you know what.
[14:28] <t0by> let me try it bitmap only.
[14:28] <wjp> but note if you go to List, size() will become expensive
[14:28] <wjp> (and now the push_backs are likely expensive)
[14:28] <t0by> doesn't Common::List cache size somehow?
[14:29] <t0by> if not, any reason why I souldn't extend it/wrap it?
[14:31] <wjp> just keep track of size separately yourself
[14:31] <wjp> (i.e., rectcount)
[14:33] <t0by> well, extending common::list would avoid me having to do what you say on a big chunk of code at a small price; what about I try that first, and if there is an actual improvement I'll add a rectcount?
[14:34] <wjp> sure
[14:34] <wjp> up to you
[14:35] <t0by> Common::List is a dll, right?
[14:36] <wjp> yes
[14:36] --> Deledrius_ joined #scummvm.
[14:38] <-- Deledrius_ left irc: Client Quit
[14:40] <-- Deledrius left irc: Ping timeout: 250 seconds
[14:40] <t0by> thanks wjp
[14:41] <t0by> i have a bunch of interesting ideas to try out
[14:41] <t0by> wouldn't be surprised if they helped a lot
[14:50] --> meunierd_ joined #scummvm.
[14:59] --> RUBICN64 joined #scummvm.
[15:02] <-- _marc` left irc: Ping timeout: 255 seconds
[15:05] --> Javacat joined #scummvm.
[15:10] --> NotJavacat joined #scummvm.
[15:11] <-- Javacat left irc: Ping timeout: 272 seconds
[15:11] Nick change: NotJavacat -> Javacat
[15:14] <t0by> Oh, well
[15:15] <t0by> i just finished replacing all arrays with lists and now it compiles.
[15:15] <t0by> now for some debugging.
[15:15] <t0by> aka: in which screaming and cursing in italian happens.
[15:17] <uruk-hai> :D
[15:17] <uruk-hai> i do the same in hungarian, don't worry
[15:21] <t0by> actually
[15:21] <t0by> it works fine.
[15:21] <t0by> it's just as dog slow as the array version.
[15:21] <t0by> teach me to curse in hungarian, for my mother tongue just doesn't cut it.
[15:29] <-- Guest42996 left irc: Changing host
[15:29] --> Guest42996 joined #scummvm.
[15:29] <-- meunierd_ left irc: Quit: meunierd_
[15:29] Nick change: Guest42996 -> Polynomial-C
[15:33] <-- RUBICN64 left irc: Quit: AndroidIrc Disconnecting
[15:34] --> Littleboy joined #scummvm.
[15:34] #scummvm: mode change '+o Littleboy' by ChanServ!ChanServ@services.
[15:37] <t0by> i hate everybody
[15:38] <fuzzie> hi.
[15:38] Nick change: t0by -> conspicuously-ev
[15:38] Nick change: conspicuously-ev -> very-evil-t0by
[15:39] <fuzzie> the first rule of linked lists is that arrays are always better
[15:39] <very-evil-t0by> honestly i don't think the data structure at hand matters much given what the profiler shows
[15:39] <fuzzie> due to inconvenient real-world co siderations
[15:40] <fuzzie> sure :-)
[15:40] <very-evil-t0by> it' a quadratic algorithm with a ton of input
[15:40] <very-evil-t0by> saving something on push_back isn't saving me here.
[15:41] <fuzzie> mmhm
[15:41] <very-evil-t0by> (BTW I vaguaely remember push_back in vector having O(1) amortized cost, I suppose that wouldn't be unheard of for Common::Array)
[15:41] <very-evil-t0by> *vageuly
[15:41] <very-evil-t0by> darn
[15:41] <very-evil-t0by> *vaguely
[15:42] <fuzzie> irrelevant anyway
[15:50] Nick change: very-evil-t0by -> t0by
[15:52] <uruk-hai> t0by, i don't think this channel is the appropriate place to swear in hungarian :D
[15:53] --> Deledrius joined #scummvm.
[15:53] <-- D0SFreak left irc: Ping timeout: 246 seconds
[15:54] --> ny00123 joined #scummvm.
[16:16] --> _marc` joined #scummvm.
[16:19] <-- _sev|work_ left irc: Quit: This computer has gone to sleep
[16:27] --> bgKa joined #scummvm.
[16:27] #scummvm: mode change '+o bgKa' by ChanServ!ChanServ@services.
[16:45] --> D0SFreak joined #scummvm.
[16:48] <-- TMM left irc: Quit: Ex-Chat
[17:00] --> WooShell joined #scummvm.
[17:03] <WooShell> meow =^.^=
[17:20] --> _sev joined #scummvm.
[17:20] #scummvm: mode change '+o _sev' by ChanServ!ChanServ@services.
[17:30] --> _sev|work joined #scummvm.
[17:30] #scummvm: mode change '+o _sev|work' by ChanServ!ChanServ@services.
[17:34] <-- t0by left irc: Ping timeout: 250 seconds
[17:53] --> HypnZ joined #scummvm.
[18:08] --> t0by joined #scummvm.
[18:08] #scummvm: mode change '+v t0by' by ChanServ!ChanServ@services.
[18:32] --> surix joined #scummvm.
[18:42] <surix> im trying to write my own c decompiler, are there any resources or articles someone would suggest?
[18:42] <surix> specifically a strategy to solve branching
[18:47] <wjp> t0by: this bitmap approach isn't quadratic
[18:47] <t0by> nope, it's not
[18:47] <t0by> who said otherwise?
[18:48] <t0by> I was referring to the plain old thing with lists tossed in in lieu of arrays
[18:48] <t0by> I'm trying to do something with you bitmap approach right now
[18:49] <t0by> Which kind of requires a rewrite, though, since things are much easier if you have the "patch" size fixed instead of the number of patches
[18:49] <t0by> (that way you either need the cliprect to be an exact multiple, or do some ugly calculations)
[18:55] --> criezy joined #scummvm.
[18:55] #scummvm: mode change '+o criezy' by ChanServ!ChanServ@services.
[18:58] --> MrSiggler joined #scummvm.
[19:04] <-- Akz- left irc: Ping timeout: 250 seconds
[19:10] --> salty-horse joined #scummvm.
[19:10] #scummvm: mode change '+o salty-horse' by ChanServ!ChanServ@services.
[19:27] <-- D0SFreak left irc: Ping timeout: 272 seconds
[19:37] --> meunierd_ joined #scummvm.
[19:45] <-- meunierd_ left irc: Ping timeout: 272 seconds
[19:49] <-- h00ligan left irc: Read error: Connection reset by peer
[19:51] --> HypnZ_ joined #scummvm.
[19:51] --> RUBICN64 joined #scummvm.
[19:52] <-- HypnZ left irc: Ping timeout: 258 seconds
[19:56] --> ST1 joined #scummvm.
[19:56] <-- ST left irc: Disconnected by services
[20:09] <-- TAS_2012v left irc: Read error: Connection reset by peer
[20:10] --> TAS_2012v joined #scummvm.
[20:18] --> D0SFreak joined #scummvm.
[20:26] --> meunierd_ joined #scummvm.
[20:26] <-- bgKa left irc: Ping timeout: 272 seconds
[20:29] <t0by> nope
[20:30] <t0by> apparently not even the wjp-style patchwork is working.
[20:30] <t0by> you can find it on my multirect_2014_patchwork branch
[20:30] <t0by> there's some kind of computation error that makes it so that some rects are not properly cleaned up
[20:31] <t0by> but I'm not even bothering to fix that
[20:31] <t0by> I get ~60% of the fullscreen framerate
[20:31] <t0by> At this point I really think that multirect is a nice thing in 8 rect games, but it won't do anything for J.u.l.i.a. and the like.
[20:33] <t0by> "It's a nice thing in 8 rect games" as in "it is irrelevant in comparison to blitting and can really help squeezing some more FPS, so you can play Rosemary on your dad's 486 with that"
[20:35] <t0by> I'm shelving this for the time being.
[20:36] <t0by> unless somebody has some kind of idea.
[20:36] <t0by> Of course, it's still a nice thing to play 8 rect games on awfully slow machines.
[20:37] <t0by> I think I'll have a look to that audio desyncing issue that really gets on my nerves.
[20:38] --> sirlemonhead joined #scummvm.
[20:41] <Lightkey> apropos: J.U.L.I.A.: Among the Stars has been released

[20:42] <t0by> which, to make things more interesting, uses a custom fork of WME
[20:51] <madmoose> surix: What do you mean by 'solve branching'?
[20:55] --> Akz- joined #scummvm.
[20:56] <surix> turning jumps into if then while for switch
[20:59] <madmoose> surix: Find Christina Cifuentes' thesis
[20:59] <surix> thanks, found it
[21:01] <-- blitter_ left irc: Ping timeout: 250 seconds
[21:04] <madmoose> surix: If you have anything to show, please do. I'd be interested in seeing it :)
[21:05] <madmoose> Haven't had time to work on mine for a long time.
[21:15] <-- uruk-hai left irc: Ping timeout: 272 seconds
[21:16] <somaen> t0by: Why is it slower than master though?
[21:16] <surix> im still designing mine, what i have written so far just splits it into continuous blocks and puts comments after some lines of what the c code should be, mostly it just detects datastructures specific to the project im working on, and inserts function names that ive identified so far
[21:17] <somaen> And well, the opaque thing would mean a backwards pass every time an opaque draw is issued, that's true, but it's not likely to be close to a fully quadratic algorithm, as ONLY the opaque calls trigger it.
[21:17] <somaen> Anyhow, the desynch thing is the reason why I haven't announced Dirty Split officially yet
[21:18] <somaen> That and the super slow scroll in the white chamber are currently the two things that's holding back the most games
[21:18] <-- somaen left irc: Changing host
[21:18] --> somaen joined #scummvm.
[21:18] #scummvm: mode change '+o somaen' by ChanServ!ChanServ@services.
[21:21] <-- waltervn left irc: Quit: Textual IRC Client: www.textualapp.com
[21:26] <t0by> super slow scroll.
[21:26] <t0by> that's new
[21:26] <t0by> to me, I mean.
[21:27] <-- salty-horse left irc: Quit: Leaving
[21:28] <t0by> <somaen> And well, the opaque thing would mean a backwards pass every time an opaque draw is issued, that's true, but it's not likely to be close to a fully quadratic algorithm, as ONLY the opaque calls trigger it. ----> But to do thisin the first place... you need to /do rects/. Or did I misunderstand your idea?
[21:29] <t0by> somaen, are you doing something for desync atm? I'd like to spend the last couple free weeks I have having a look at it, especially if this would mean getting DS supported.
[21:29] <t0by> (i love that one)
[21:36] <t0by> nite everybody
[21:36] <-- t0by left irc: Quit: Leaving
[21:43] Nick change: ST1 -> ST
[21:43] #scummvm: mode change '+o ST' by ChanServ!ChanServ@services.
[21:46] <-- WooShell left irc: Quit: Walking upside down in the sky, between the satellites passing by. Gliding along the black rainbow, I fly away with my shadow. Scratching the moon like a DJ, the night follows its odyssey.
[21:53] <-- HypnZ_ left irc:
[22:05] --> blitter joined #scummvm.
[22:05] <-- Schnaks left irc:
[22:13] <-- Javacat left irc: Ping timeout: 260 seconds
[22:14] --> Jon_God joined #scummvm.
[22:14] --> Javacat joined #scummvm.
[22:17] <-- Javacat left irc: Client Quit
[22:19] <-- RUBICN64 left irc: Quit: AndroidIrc Disconnecting
[22:22] <-- droid2727 left irc: Ping timeout: 272 seconds
[22:29] --> Javacat joined #scummvm.
[22:30] <-- _sev left irc: Quit: Leaving
[22:30] --> droid2727 joined #scummvm.
[22:30] #scummvm: mode change '+o droid2727' by ChanServ!ChanServ@services.
[22:40] --> _sev joined #scummvm.
[22:40] <-- _sev left irc: Changing host
[22:40] --> _sev joined #scummvm.
[22:40] #scummvm: mode change '+o _sev' by ChanServ!ChanServ@services.
[23:26] --> Kirben joined #scummvm.
[23:26] #scummvm: mode change '+o Kirben' by ChanServ!ChanServ@services.
[23:30] sirlemonhead (bduncan22@95.44.164.183) left #scummvm.
[23:35] <-- meunierd_ left irc: Ping timeout: 272 seconds
[23:50] --> broosky joined #scummvm.
[23:51] <-- borosky left irc: Ping timeout: 240 seconds
[23:52] _marc` (~marc@p20030046081F9CFCF01C9922453B63D3.dip0.t-ipconnect.de) got netsplit.
[23:52] Littleboy (~littleboy@c-174-62-174-218.hsd1.ma.comcast.net) got netsplit.
[23:52] edheldil (~edheldil@2001:1488:fffe:1:505b:b2fd:6b64:2453) got netsplit.
[23:52] Vampire0_ (~Vampire@jEdit/Vampire) got netsplit.
[23:53] _marc` (~marc@p20030046081F9CFCF01C9922453B63D3.dip0.t-ipconnect.de) returned to #scummvm.
[23:53] --> Vampire0_ joined #scummvm.
[23:54] --> edheldil joined #scummvm.
[23:54] Nick change: Vampire0_ -> Guest72212
[00:00] --- Tue Sep 16 2014