Dolphin, the GameCube and Wii emulator - Forums

Full Version: ZTP speed hack?
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3
Thank you guys for looking into this...and apologize for the disaster which became the "Dolphin 4.0 vs 3.0 thread.

After doing a thorough re-check, I have confirmed that the latest build of Dolphin in fact plays flawlessly on all games (Even Online if the correct files are extracted from the Wii OS).

My only issue at this point is Twilight Princess as it slows down horrendously compared to the previous builds. If the hack is gone, I highly suggest no one actually play in the huge areas, as i've confirmed it to run my CPU up to a dangerous 88 degrees. My Rig is below:

Windows 7 64 bit OS
Geforce GTX 570
Sandy Bridge Core I7 2700k (@ 4.5 ghz)
8 Gigs of Ram
Cool Black Full ATX Tower case
NVIDIA STICKERS

Thanks,
The hack has been removed
(05-19-2014, 02:08 AM)ExtremeDude2 Wrote: [ -> ]The hack has been removed
Um, I'm pretty sure everyone knows that

(05-18-2014, 03:54 PM)MaJoR Wrote: [ -> ]Actually the ZtP hack was removed because not only was it a hack, but mainly because it was a hack that finally got in the way of something the developers wanted to implement. Hence why it was removed then.
I beg to differ, this code works the same as when it's off since it's removal.
Code:
    if (g_ActiveConfig.bZTPSpeedHack)
    {
        if (!mapTexFound)
        {
            if (bp.address != BPMEM_TEV_COLOR_ENV && bp.address != BPMEM_TEV_ALPHA_ENV)
            {
                numWrites = 0;
            }
            else if (++numWrites >= 100)    // seem that if 100 consecutive BP writes are called to either of these addresses in ZTP,
            {                                // then it is safe to assume the map texture address is currently loaded into the BP memory
                mapTexAddress = bpmem.tex[0].texImage3[0].hex << 5;
                mapTexFound = true;
                WARN_LOG(VIDEO, "\nZTP map texture found at address %08x\n", mapTexAddress);
            }
            FlushPipeline();
        }
        else if ((bpmem.tex[0].texImage3[0].hex << 5) != mapTexAddress ||
            bpmem.tevorders[0].getEnable(0) == 0 ||
            bp.address == BPMEM_TREF)
        {
            FlushPipeline();
        }
    }  // END ZTP SPEEDUP HACK
    else
    {
        if (((s32*)&bpmem)[bp.address] == bp.newvalue)
        {
            if (!(bp.address == BPMEM_TRIGGER_EFB_COPY
                || bp.address == BPMEM_CLEARBBOX1
                || bp.address == BPMEM_CLEARBBOX2
                || bp.address == BPMEM_SETDRAWDONE
                || bp.address == BPMEM_PE_TOKEN_ID
                || bp.address == BPMEM_PE_TOKEN_INT_ID
                || bp.address == BPMEM_LOADTLUT0
                || bp.address == BPMEM_LOADTLUT1
                || bp.address == BPMEM_TEXINVALIDATE
                || bp.address == BPMEM_PRELOAD_MODE
                || bp.address == BPMEM_CLEAR_PIXEL_PERF))
            {
                return;
            }
        }

        FlushPipeline();
    }
It got in the way of a potential change to the BPStructs code; I do not know if that was merged or not.
purpasmart96: I know that the ztp speedhack could still be applied, but iirc the speedup won't be there as common. It will still speed up a bit, but iirc by far not as much as on 4.0. So did you compare the performance of current development builds + this speedhack vs 4.0 ?

btw: the second commit breaking the design of this speedhack is 6b0183952564a11cd4eee42715e61ea2962caff3 .
(05-19-2014, 07:21 AM)degasus Wrote: [ -> ]purpasmart96: I know that the ztp speedhack could still be applied, but iirc the speedup won't be there as common. It will still speed up a bit, but iirc by far not as much as on 4.0. So did you compare the performance of current development builds + this speedhack vs 4.0 ?

btw: the second commit breaking the design of this speedhack is 6b0183952564a11cd4eee42715e61ea2962caff3 .

I haven't ben able to build the dev build plus ztp hack applied, it doesn't want to compile, although it has nothing to do with the ztp hack, git is somehow not being recognized and SCMRevGen project doesn't want to compile.
I get this in MVS 2013. which causes 3 more errors since this one failed.
Error 101 error MSB3073: The command "cscript /nologo /E:JScript "make_scmrev.h.js"
:VCEnd" exited with code 1. C:\Program Files (x86)\MSBuild\Microsoft.Cpp\v4.0\V120\Microsoft.CppCommon.targets 122 5 SCMRevGen

here's the source with it added, it would help me out a lot if someone could build this for me
https://anonfiles.com/file/ed561d65b27ce...fc6f2a098e
Guys! I managed to compile it, with no problems, although the speed hack doesn't' work in d3d, it only works under opengl now, hyrule field was 10-12 fps now it is 20-29 fps.
I'll upload the build as soon as possible.
here is the link to the build
https://anonfiles.com/file/e925017b6c52f...6858f8c100
also bumping this
wait a sec I cant delete my own posts? Now I've quad posted...
You cant extract these files. You get:
! C:\file path removed by me\1400547657633.zip: Unknown method in x64\Dolphin.exe
(05-20-2014, 11:35 PM)haddockd Wrote: [ -> ]You cant extract these files. You get:
! C:\file path removed by me\1400547657633.zip: Unknown method in x64\Dolphin.exe

Are you using winrar or winzip? because that might be your problem. It extracts perfectly for me.
EDIT: I can confirm that using the windows explorer to extract the file gives the error
Use 7zip, it's faster and compresses better.
Pages: 1 2 3