From 86d37e06f961c55d4ad02d5aac8788cd15e5190b Mon Sep 17 00:00:00 2001 From: Christoph Oelckers Date: Tue, 19 Aug 2014 15:56:33 +0200 Subject: [PATCH] - lowered requirements of GL 2.x to OpenGL 3.3. There was one issue preventing the previous 2.0 betas from running under GL 3.x: The lack of persistently mapped buffers. For the dynamic light buffer today's changes take care of that problem. For the vertex buffer there is no good workaround but we can use immediate mode render calls instead which have been reinstated. To handle the current setup, the engine first tries to get a core profile context and checks for presence of GL 4.4 or the GL_ARB_buffer_storage extension. If this fails the context is deleted again and a compatibility context retrieved which is then used for 'old style' rendering which does work on older GL versions. This new version does not support GL 3.2 or lower, meaning that Intel GMA 3000 or lower is not supported. The reason for this is that the engine uses a few GL 3.3 features which are not present in the latest Intel driver. In general the Intel GMA 3000 is far too weak, though, to run the demanding shader of GZDoom 2.x, so this is no real loss. Performance would be far from satisfying. A command line option '-gl3' exists to force the fallback render path. On my Geforce 550Ti there's approx. 10% performance loss on this path. --- src/gl/data/gl_vertexbuffer.cpp | 99 +++++++++++++++++++++++++-------- src/gl/data/gl_vertexbuffer.h | 9 ++- src/gl/system/gl_interface.cpp | 11 +++- src/gl/system/gl_interface.h | 3 +- src/win32/win32gliface.cpp | 90 +++++++++++++++++++++--------- src/win32/win32gliface.h | 1 + 6 files changed, 159 insertions(+), 54 deletions(-) diff --git a/src/gl/data/gl_vertexbuffer.cpp b/src/gl/data/gl_vertexbuffer.cpp index 854750ecb..0e9065557 100644 --- a/src/gl/data/gl_vertexbuffer.cpp +++ b/src/gl/data/gl_vertexbuffer.cpp @@ -90,26 +90,60 @@ void FVertexBuffer::BindVBO() FFlatVertexBuffer::FFlatVertexBuffer() : FVertexBuffer() { - unsigned int bytesize = BUFFER_SIZE * sizeof(FFlatVertex); - glBindBuffer(GL_ARRAY_BUFFER, vbo_id); - glBufferStorage(GL_ARRAY_BUFFER, bytesize, NULL, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_COHERENT_BIT); - map = (FFlatVertex*)glMapBufferRange(GL_ARRAY_BUFFER, 0, bytesize, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_COHERENT_BIT); - mNumReserved = mIndex = mCurIndex = 0; + if (gl.flags & RFL_BUFFER_STORAGE) + { + unsigned int bytesize = BUFFER_SIZE * sizeof(FFlatVertex); + glBindBuffer(GL_ARRAY_BUFFER, vbo_id); + glBufferStorage(GL_ARRAY_BUFFER, bytesize, NULL, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_COHERENT_BIT); + map = (FFlatVertex*)glMapBufferRange(GL_ARRAY_BUFFER, 0, bytesize, GL_MAP_WRITE_BIT | GL_MAP_PERSISTENT_BIT | GL_MAP_COHERENT_BIT); - glBindVertexArray(vao_id); - glBindBuffer(GL_ARRAY_BUFFER, vbo_id); - glVertexAttribPointer(VATTR_VERTEX, 3,GL_FLOAT, false, sizeof(FFlatVertex), &VTO->x); - glVertexAttribPointer(VATTR_TEXCOORD, 2,GL_FLOAT, false, sizeof(FFlatVertex), &VTO->u); - glEnableVertexAttribArray(VATTR_VERTEX); - glEnableVertexAttribArray(VATTR_TEXCOORD); - glBindVertexArray(0); + glBindVertexArray(vao_id); + glBindBuffer(GL_ARRAY_BUFFER, vbo_id); + glVertexAttribPointer(VATTR_VERTEX, 3,GL_FLOAT, false, sizeof(FFlatVertex), &VTO->x); + glVertexAttribPointer(VATTR_TEXCOORD, 2,GL_FLOAT, false, sizeof(FFlatVertex), &VTO->u); + glEnableVertexAttribArray(VATTR_VERTEX); + glEnableVertexAttribArray(VATTR_TEXCOORD); + glBindVertexArray(0); + } + else + { + vbo_shadowdata.Reserve(BUFFER_SIZE); + map = &vbo_shadowdata[0]; + } + mNumReserved = mIndex = mCurIndex = 0; } FFlatVertexBuffer::~FFlatVertexBuffer() { - glBindBuffer(GL_ARRAY_BUFFER, vbo_id); - glUnmapBuffer(GL_ARRAY_BUFFER); - glBindBuffer(GL_ARRAY_BUFFER, 0); + if (gl.flags & RFL_BUFFER_STORAGE) + { + glBindBuffer(GL_ARRAY_BUFFER, vbo_id); + glUnmapBuffer(GL_ARRAY_BUFFER); + glBindBuffer(GL_ARRAY_BUFFER, 0); + } +} + +//========================================================================== +// +// immediate mode fallback for drivers without GL_ARB_buffer_storage +// +// No single core method is performant enough to handle this adequately +// so we have to resort to immediate mode instead... +// +//========================================================================== + +void FFlatVertexBuffer::ImmRenderBuffer(unsigned int primtype, unsigned int offset, unsigned int count) +{ + // this will only get called if we can't acquire a persistently mapped buffer. +#ifndef CORE_PROFILE + glBegin(primtype); + for (unsigned int i = 0; i < count; i++) + { + glVertexAttrib2fv(VATTR_TEXCOORD, &map[offset + i].u); + glVertexAttrib3fv(VATTR_VERTEX, &map[offset + i].x); + } + glEnd(); +#endif } //========================================================================== @@ -289,10 +323,24 @@ void FFlatVertexBuffer::UpdatePlaneVertices(sector_t *sec, int plane) void FFlatVertexBuffer::CreateVBO() { - vbo_shadowdata.Resize(mNumReserved); - CreateFlatVBO(); - mCurIndex = mIndex = vbo_shadowdata.Size(); - memcpy(map, &vbo_shadowdata[0], vbo_shadowdata.Size() * sizeof(FFlatVertex)); + if (gl.flags & RFL_BUFFER_STORAGE) + { + vbo_shadowdata.Resize(mNumReserved); + CreateFlatVBO(); + mCurIndex = mIndex = vbo_shadowdata.Size(); + memcpy(map, &vbo_shadowdata[0], vbo_shadowdata.Size() * sizeof(FFlatVertex)); + } + else if (sectors) + { + // set all VBO info to invalid values so that we can save some checks in the rendering code + for(int i=0;iGetHeightSec(); - if (hs != NULL) CheckPlanes(hs); - for(unsigned i = 0; i < sector->e->XFloor.ffloors.Size(); i++) - CheckPlanes(sector->e->XFloor.ffloors[i]->model); + if (gl.flags & RFL_BUFFER_STORAGE) + { + CheckPlanes(sector); + sector_t *hs = sector->GetHeightSec(); + if (hs != NULL) CheckPlanes(hs); + for (unsigned i = 0; i < sector->e->XFloor.ffloors.Size(); i++) + CheckPlanes(sector->e->XFloor.ffloors[i]->model); + } } \ No newline at end of file diff --git a/src/gl/data/gl_vertexbuffer.h b/src/gl/data/gl_vertexbuffer.h index de61fb7cb..b6d61f2da 100644 --- a/src/gl/data/gl_vertexbuffer.h +++ b/src/gl/data/gl_vertexbuffer.h @@ -83,7 +83,14 @@ public: void RenderArray(unsigned int primtype, unsigned int offset, unsigned int count) { drawcalls.Clock(); - glDrawArrays(primtype, offset, count); + if (gl.flags & RFL_BUFFER_STORAGE) + { + glDrawArrays(primtype, offset, count); + } + else + { + ImmRenderBuffer(primtype, offset, count); + } drawcalls.Unclock(); } diff --git a/src/gl/system/gl_interface.cpp b/src/gl/system/gl_interface.cpp index 4da1d9a42..7a745110b 100644 --- a/src/gl/system/gl_interface.cpp +++ b/src/gl/system/gl_interface.cpp @@ -117,9 +117,9 @@ void gl_LoadExtensions() else Printf("Emulating OpenGL v %s\n", version); // Don't even start if it's lower than 3.0 - if (strcmp(version, "3.3") < 0 || !CheckExtension("GL_ARB_buffer_storage")) + if (strcmp(version, "3.3") < 0) { - I_FatalError("Unsupported OpenGL version.\nAt least OpenGL 3.3 and the »GL_ARB_buffer_storage« extension is required to run " GAMENAME ".\n"); + I_FatalError("Unsupported OpenGL version.\nAt least OpenGL 3.3 is required to run " GAMENAME ".\n"); } // add 0.01 to account for roundoff errors making the number a tad smaller than the actual version @@ -130,7 +130,12 @@ void gl_LoadExtensions() if (CheckExtension("GL_ARB_texture_compression")) gl.flags|=RFL_TEXTURE_COMPRESSION; if (CheckExtension("GL_EXT_texture_compression_s3tc")) gl.flags|=RFL_TEXTURE_COMPRESSION_S3TC; - if (CheckExtension("GL_ARB_shader_storage_buffer_object")) gl.flags |= RFL_SHADER_STORAGE_BUFFER; + if (!Args->CheckParm("-gl3")) + { + // don't use GL 4.x features when running in GL 3 emulation mode. + if (CheckExtension("GL_ARB_shader_storage_buffer_object")) gl.flags |= RFL_SHADER_STORAGE_BUFFER; + if (CheckExtension("GL_ARB_buffer_storage")) gl.flags |= RFL_BUFFER_STORAGE; + } glGetIntegerv(GL_MAX_TEXTURE_SIZE,&gl.max_texturesize); glPixelStorei(GL_UNPACK_ALIGNMENT, 1); diff --git a/src/gl/system/gl_interface.h b/src/gl/system/gl_interface.h index 5d59e5560..af4c4f7c2 100644 --- a/src/gl/system/gl_interface.h +++ b/src/gl/system/gl_interface.h @@ -9,7 +9,8 @@ enum RenderFlags RFL_TEXTURE_COMPRESSION=1, RFL_TEXTURE_COMPRESSION_S3TC=2, - RFL_SHADER_STORAGE_BUFFER = 4, // to be used later for a parameter buffer + RFL_SHADER_STORAGE_BUFFER = 4, + RFL_BUFFER_STORAGE = 8 }; enum TexMode diff --git a/src/win32/win32gliface.cpp b/src/win32/win32gliface.cpp index ab6d15254..a9911460f 100644 --- a/src/win32/win32gliface.cpp +++ b/src/win32/win32gliface.cpp @@ -726,6 +726,32 @@ bool Win32GLVideo::SetupPixelFormat(int multisample) // //========================================================================== +bool Win32GLVideo::checkCoreUsability() +{ + // if we explicitly want to disable 4.x features this must fail. + if (Args->CheckParm("-gl3")) return false; + + // GL 4.4 implies GL_ARB_buffer_storage + if (strcmp((char*)glGetString(GL_VERSION), "4.4") >= 0) return true; + + // at this point GLEW has not been initialized so we have to retrieve glGetStringi ourselves. + PFNGLGETSTRINGIPROC myglGetStringi = (PFNGLGETSTRINGIPROC)wglGetProcAddress("glGetStringi"); + if (!myglGetStringi) return false; // this should not happen. + + const char *extension; + + int max = 0; + glGetIntegerv(GL_NUM_EXTENSIONS, &max); + + // step through all reported extensions and see if we got what we need... + for (int i = 0; i < max; i++) + { + extension = (const char*)myglGetStringi(GL_EXTENSIONS, i); + if (!strcmp(extension, "GL_ARB_buffer_storage")) return true; + } + return false; +} + bool Win32GLVideo::InitHardware (HWND Window, int multisample) { m_Window=Window; @@ -737,36 +763,50 @@ bool Win32GLVideo::InitHardware (HWND Window, int multisample) return false; } - m_hRC = NULL; - if (myWglCreateContextAttribsARB != NULL) + for (int prof = WGL_CONTEXT_CORE_PROFILE_BIT_ARB; prof <= WGL_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB; prof++) { - // let's try to get the best version possible. Some drivers only give us the version we request - // which breaks all version checks for feature support. The highest used features we use are from version 4.4, and 3.0 is a requirement. - static int versions[] = { 44, 43, 42, 41, 40, 33, -1 }; - - for (int i = 0; versions[i] > 0; i++) + m_hRC = NULL; + if (myWglCreateContextAttribsARB != NULL) { - int ctxAttribs[] = { - WGL_CONTEXT_MAJOR_VERSION_ARB, versions[i] / 10, - WGL_CONTEXT_MINOR_VERSION_ARB, versions[i] % 10, - WGL_CONTEXT_FLAGS_ARB, gl_debug ? WGL_CONTEXT_DEBUG_BIT_ARB : 0, - WGL_CONTEXT_PROFILE_MASK_ARB, WGL_CONTEXT_CORE_PROFILE_BIT_ARB, - 0 - }; + // let's try to get the best version possible. Some drivers only give us the version we request + // which breaks all version checks for feature support. The highest used features we use are from version 4.4, and 3.3 is a requirement. + static int versions[] = { 45, 44, 43, 42, 41, 40, 33, -1 }; - m_hRC = myWglCreateContextAttribsARB(m_hDC, 0, ctxAttribs); - if (m_hRC != NULL) break; + for (int i = 0; versions[i] > 0; i++) + { + int ctxAttribs[] = { + WGL_CONTEXT_MAJOR_VERSION_ARB, versions[i] / 10, + WGL_CONTEXT_MINOR_VERSION_ARB, versions[i] % 10, + WGL_CONTEXT_FLAGS_ARB, gl_debug ? WGL_CONTEXT_DEBUG_BIT_ARB : 0, + WGL_CONTEXT_PROFILE_MASK_ARB, prof, + 0 + }; + + m_hRC = myWglCreateContextAttribsARB(m_hDC, 0, ctxAttribs); + if (m_hRC != NULL) break; + } + } + + if (m_hRC == NULL && prof == WGL_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB) + { + Printf("R_OPENGL: Couldn't create render context. Reverting to software mode...\n"); + return false; + } + + wglMakeCurrent(m_hDC, m_hRC); + + // we can only use core profile contexts if GL_ARB_buffer_storage is supported or GL version is >= 4.4 + if (prof == WGL_CONTEXT_CORE_PROFILE_BIT_ARB && !checkCoreUsability()) + { + wglMakeCurrent(0, 0); + wglDeleteContext(m_hRC); + } + else + { + return true; } } - - if (m_hRC == NULL) - { - Printf ("R_OPENGL: Couldn't create render context. Reverting to software mode...\n"); - return false; - } - - wglMakeCurrent(m_hDC, m_hRC); - return true; + return false; } //========================================================================== diff --git a/src/win32/win32gliface.h b/src/win32/win32gliface.h index f61c31857..f425757f1 100644 --- a/src/win32/win32gliface.h +++ b/src/win32/win32gliface.h @@ -87,6 +87,7 @@ protected: void MakeModesList(); void AddMode(int x, int y, int bits, int baseHeight, int refreshHz); void FreeModes(); + bool checkCoreUsability(); public: int GetTrueHeight() { return m_trueHeight; }