Doom 3 engine release and game code

[qbism]

Looks like it works well. But at the moment I can't speak about performance, my laptop is stuck in "power save" mode.

[revelator]

ok then ill upload the changes.

Stuck in powersave hmm could try downloading advanced system care and let it run a repair, usually it does a quite ok job fixing those things free version does not autorepair so you will have to do it manually but it has a ton of tools for fixing weird stuff like that.

Performance wise it should be the same if not a little faster (better optimization in some parts of the code).

[revelator]

https://github.com/revelator/Revelator-Doom3

Comitted all changes.

Maybe better to just download the repo as the changes make the patch pretty freaking huge :S

[revelator]

Forgot to mention i made wrapper executables for doom3 doom3 resurrection of evil including sikkmod and doom3 in hell.
Sikkmod's game code is also included in the neo\sikkmod folder. To use it just replace the files in neo with the ones in the sikkmod folder.
Remember to backup the old files so you can go back.

Im filtering out portalsky for ssao and it fixes someof the bugs the ssao mod has but i probably need to include heathaze and a few other things to make it look just right.

[revelator]

Having a deep hard look at xreals render code and its kinda close to the doom3 renderer but its fully glsl still have to port it to C++ but i bet if someone got that far
it would make for a nice upgrade. Only problem is that it would be 100% incompatible with Doom3's arb shaders ouch :S.
Best bet is to keep the old backend for compatibility atleast untill theres a full set of shaders for the xreal render backend.

[motorsep]

I wonder if mh would like to add D3D rendering path into id Tech 4 ?

[revelator]

Cant say but maybe

Btw i tried the small codepice for vertex buffering function here.

Code: Select all

vertCache_t   *idVertexCache::AllocFrameTemp( void *data, int size )
{
    vertCache_t   *block;

    if ( size <= 0 )
    {
        common->Error( "idVertexCache::AllocFrameTemp: size = %i\n", size );
    }

    if ( dynamicAllocThisFrame + size > frameBytes )
    {
        // if we don't have enough room in the temp block, allocate a static block,
        // but immediately free it so it will get freed at the next frame
        tempOverflow = true;
        Alloc( data, size, &block );
        Free( block);
        return block;
    }

    // this data is just going on the shared dynamic list

    // if we don't have any remaining unused headers, allocate some more
    if ( freeDynamicHeaders.next == &freeDynamicHeaders )
    {

        for ( int i = 0; i < EXPAND_HEADERS; i++ )
        {
            block = headerAllocator.Alloc();
            block->next = freeDynamicHeaders.next;
            block->prev = &freeDynamicHeaders;
            block->next->prev = block;
            block->prev->next = block;
        }
    }

    // move it from the freeDynamicHeaders list to the dynamicHeaders list
    block = freeDynamicHeaders.next;
    block->next->prev = block->prev;
    block->prev->next = block->next;
    block->next = dynamicHeaders.next;
    block->prev = &dynamicHeaders;
    block->next->prev = block;
    block->prev->next = block;

    block->size = size;
    block->tag = TAG_TEMP;
    block->indexBuffer = false;
    block->offset = dynamicAllocThisFrame;
    dynamicAllocThisFrame += block->size;
    dynamicCountThisFrame++;
    block->user = NULL;
    block->frameUsed = 0;

    // copy the data
    block->virtMem = tempBuffers[listNum]->virtMem;
    block->vbo = tempBuffers[listNum]->vbo;
   
   // mh code start
   if ( block->vbo )
   {
      qglBindBufferARB (GL_ARRAY_BUFFER, block->vbo);

      // use GL_ARB_map_buffer_range if availiable
      if (glConfig.ARBMapBufferRangeAvailable)
      {
         void *dst = NULL;

         if ((dst = qglMapBufferRange (GL_ARRAY_BUFFER, block->offset, (GLsizeiptr) size, BUFFER_MAP_BITS)) != NULL)
         {
            SIMDProcessor->Memcpy ((byte *) dst, data, size);

            qglUnmapBufferARB (GL_ARRAY_BUFFER);

            return block;
         }
      }
      qglBufferSubDataARB (GL_ARRAY_BUFFER, block->offset, (GLsizeiptr) size, data);

      return block;
   }

   // fall through to the temp buffer ?
   SIMDProcessor->Memcpy ((byte *) block->virtMem + block->offset, data, size);

   return block;
   // mh code end
}


Seems to work fine but not really noticing any speedup :/ though i guess thats why he mentioned that it needed a total rewrite.

Edit: heres the missing #define

// let's not have code stretching 4 miles across the screen
#define BUFFER_MAP_BITS (GL_MAP_WRITE_BIT | GL_MAP_UNSYNCHRONIZED_BIT | GL_MAP_INVALIDATE_BUFFER_BIT)

just yank it somewhere at the top of VertexCache.cpp

[motorsep]

What's with doom3world.org ? I can't get to it :/

[revelator]

Aye same here :/ seems the site is down for maintainance i guess.

[revelator]

Hmm still down :S wonder whats happened

[revelator]

Ok is back up

And some updates to my code.

Removed deprecated functions light and specular caches (engine now requires vertex or fragment programs).
Upped to OpenGL3 requirement.
removed old slow virtualmem stuff from renderer everything is now handled with VBO's.
Fixed a number of functions which had some or the other defect according to pvs-studio.
Added mh's glMapBufferRange optimization (might need some more work in other places but atleast it does not break anything).
Upped a number of settings in the autodetect routine like LOD bias -3 for ultra (makes a huge difference in detail).
Fixed some NULL on int mistakes (int is 0 not NULL).
Added a number of missing detructors and fixed a few to delete[] on arrays.
Replaced a number of strlen(somestring) == 0 to somestring[0] == 0 for speed.

Ill upload changes later

[andrewj]

Replaced a number of strlen(somestring) == 0 to somestring[0] == 0 for speed.

Did you really get any speed improvement from that?

Just admit you did it for programming aesthetics

[taniwha]

If it's in a loop, then yes, it will make a difference. Why test 2+ chars (with loop and maybe function call overhead) when one is enough?

[revelator]

Noticed a slight speedup but nothing earthshattering biggest hurdle atm seems to be the vbo code (needs some optimizing badly).
Btw anyone noticed the r_vertexbuffermegs cvar ? i had kindof a chuckle from reading that setting it higher would speedup the game
the sad thing is that the block that should handle the memory amount set with this is disabled in code and it was unfinished so it newer worked lol.
I i tried removing it completely and no change at all xD Doom3 works just as before.

btw its somestring[0] == '\0' forgot the '\'

[revelator]

other stuff pvs studio says might speed it up are using & on arrays unfortunatly that does not work atleast not with msvc idvec3 &somearray will not compute.