Issue when load the first map

Hello every one.
I got the blackmessa, and when i load the game it start fine, with the main screen scenario and the menu
but when i do start a new game i get this error message

failed to lock vertex buffer in cmeshdx8 lockvertexbuffer

I do own counter-strike source and i made an autoexec.cfg for best gfx quality and the game run fine. I put the autoexec on blackmessa autoexec too, here are the specs of the pc, the command line i use and the autoexec for blackmesa.

Pentium D 235 @3,3Ghz
4GB DDR2
ATI Radeon HD4350 1GB edition
Latest drivers
Windows 7 x64 SP1 Ultimate

Command line trough steam:
-novid -heapsize 1572864 -refresh 75 +mat_hdr_level 2 +mat_hdr_enabled 1 +mathdrenabled 1

Autoexec:

exec config.cfg
//exec threading.cfg
mat_dxlevel 98
mat_queue_mode 2
mat_forceaniso 16
mat_trilinear 1
mat_vsync 1
mat_colorcorrection 1
mat_hdr_level 2
mat_bloomscale 1.35 //1.25 //1.85
mat_picmip -10
mat_clipz 1
mat_bumpmap 1
mat_specular 1
mat_fastnobump 0
mat_fastspecular 1
mat_reducefillrate 0
mat_mipmaptextures 1
mat_compressedtextures 0
mat_motion_blur_enabled 1
mat_antialias 8
mat_aaquality 2
r_occlusion 1
r_renderoverlayfragment 1
r_waterforceexpensive 1
r_waterdrawreflection 1
r_Waterdrawrefraction 1
r_waterforcereflectentities 1
r_forcewaterleaf 0
r_cheapwaterstart 1
r_cheapwaterend 1
r_eyes 1
r_teeth 1
r_shadows 1
r_shadowrendertotexture 1
r_shadowmaxrendered 32
r_lod 0
r_rootlod -10
r_3dsky 1
r_propsmaxdist 2048
r_worldlights 1
r_dynamic 1
host_thread_mode 1
rate 35000
r_fastzreject -1
mp_usehwmmodels 1
mp_usehwmvcds 1
cl_new_impact_effects 1
r_radiosity 4
mat_parallaxmap 1
r_threaded_particles 1
r_threaded_client_shadow_manager 1
cl_threaded_bone_setup 1
cl_threaded_client_leaf_system 1
mp_usehwmvcds 1
cl_c4dynamiclight 1
mat_filtertextures 1
mat_filterlightmaps 0
muzzleflash_light 1
r_decals 2048
cl_ragdoll_collide 1
r_avglight 3
cl_detaildist 99999
cl_detailfade 99999
cl_phys_props_max 1024
mat_postprocess_x 8
mat_postprocess_y 8
mat_software_aa_quality 2
mat_software_aa_strengt 2
mat_software_aa_edge_threshold 0.8
mat_software_aa_strength_vgui 2
mat_use_compressed_hdr_textures 0
cl_showsplashes 1
mat_bloom_scalefactor_scalar 5.75 //2.65 //2.90
violence_agibs 1
violence_hgibs 1
violence_hblood 1
violence_ablood 1
cl_show_bloodspray 1
mat_hdr_enabled 1
mat_hdr_level 2
mat_hdr_manual_tonemap_rate 0.35 //0.45 //0.35
mathdrenabled 1
mat_software_aa_strength -1
mat_software_aa_blur_one_pixel_lines 0.5
mat_software_aa_edge_threshold “0.8”
mat_software_aa_quality 2
mat_software_aa_strength 2
mat_software_aa_strength_vgui 2
mat_managedtextures 1
mat_queue_priority 2

bind p bloom1
alias bloom1 “mat_bloomscale 0.35; bind p bloom2”
alias bloom2 “mat_bloomscale 1.35; bind p bloom1”

Please help :slight_smile:

Remove mat_compressedtextures 0 and it should work. I want to get that working but haven’t found a way, I only got it to work on some maps.

+mat_hdr_enabled 1 +mathdrenabled 1 are bogus and should be removed. It’s not the problem cuase though. mat_hdr_enabled is a read-only cvar flag to be used internally by the engine to inform game scripts and what else that it is using FPxx buffers for rendering. mathdrenabled simply do not exist at all - thus it is invalid trying to set it.

Do not ever have this one on a constant basis in your config file. dxlevel to use is expected to be changed by using one-time use command switch “-dxlevel XX” in source games. Read more in Question 6 of FAQ or head on directly to the Valve developers wiki if you want gory details.

Miltithreaded materials handling is broken in Source Base SDK 2007 and would cause troubles and crashes in BM. Set it to 0, don’t touch in future unless requested to do so by BM devs.

Possible trouble cause. Normally should work but in 2007 version of the engine correct values for this one are in -1…4 range. Anything else might cause problems. Safe default is to use 0 here.

It is developer-only cvar that bypasses correct rendering of the specular highlights. It do not increase performance and should not be used normally. Intended use (just as for mat_fastnobump) is to check how would look specular highlight with vertex-based Gouraud lighting instead of pixel-shaders based Phong or Blinn-Phong approach. Set this cvar to 0.

Could be a problem cause. Try changing to 1 and check if it helps.

mat_aaqulity is known not to work correctly with 2007 version of the Source engine. If you really want to enable 32xCSAA (which is 8xMSAA + 24 extra coverage samples) - set in-game AA to 2x MSAA and use “Enchance application AA setting” in the GPU driver control panel.

No point in using these as you won’t ever get “cheap” water with your config at all - you force water rendering to always use costly renderpath.

You’d better of leaving this at their defaults. Changing rootlod from default 0 to significant negative numbers is known to cause crashes in geometry processing pipeline of the Source engine sometimes. It is pretty safe to have r_lod set to -1 though if you want models to became less detailed at a larger distance.

Could cause problems with BM, messes up maximum draw distance from the expected default.

Could cause crashes and/or jutters due to the same reasons mat_queue_mode 2 could cause it. It is not warranted to cause problems (while mat_queue_mode is) so you could try playing with setting it either to 0 or to 2 and check what works best for you.

It is safe and good to set it to 1 with modern GPUs. But it is also safe to have it set to -1 and let the engine to automatically enable it if the GPU supports early depth test.

Known to work but are one of the candidates to set to 0 in case of troubles.

Do not exist in BM -> invalid, should be removed.

Gives no performance benefits but worse quality, set to 1.

IIRC it is a boolean flag, having anything other than 0 here produces the same effect.

Changing these from defaults (dist -> 1200, fade -> 400, phys_props_max -> 300) is directly asking for troubles like “texture ringing” glitches on the distant objects due to detail textures being visible but not affected with aniso filtering. Also, detailfade sets the size of the zone where detail textures fade from being fully visible up to totally invisible. In your case you set them to start fading out right from the player’s position and up to the 99999 distance. Not the best way to conserve performance forcing GPU to alpha-blend detail-textures almost for every surface visible on the screen (assuming that every surface has detail texture applied to it).

a) software aa - which is a post-processing shader Source engine use on x360 to perform AA (MSAA is not viable there due to performance constraints) is known to be buggy and de-facto not working properly on PC. Expect troubles enabling it.
b) You have a typo there - strengt should be strength. Thus you do not have it enabled. Good for you.
c) There are only two levels of quality recognized: 0 (5-tap filter) and 1 (9-tap filter). Setting quality to 2 is bogus.

Could be trouble cause. And is redundant with your config because you already disable compressed textures usage whatsoever.

Bogus. You can’t set read-only internal engine flag.

Nonexistent cvar. Bogus.

Copy-pasting things from other configs without looking/thinking? These are duplicates, all the above comments about them apply here.

Nonexistent in Source Base SDK 2007. Bogus.

I’d suggest you not to try to fix the thing that isn’t broken.
If should have worked without major problems with the default config in place but you had messed the things up by mindlessly copy-pasting things without really trying to figure out what they do.

TL;DR: Revert config to default and don’t mess with the things you don’t know internals of.

Hello, thanks for the reply.
I wasnt tryng to fix it or nothing, dont get me wrong, i just thot that the source engine offer more quality then we can see in the video settings. :slight_smile:

I did took away almost all of the commands and here was what i left (before check the awnser)

mat_forceaniso 16
mat_trilinear 1
mat_vsync 1
mat_colorcorrection 1
mat_bloomscale 0.35 //1.25 //1.85
mat_picmip -10
mat_antialias 8
mat_aaquality 2
r_shadows 1
r_shadowrendertotexture 1
r_shadowmaxrendered 32

And its working :slight_smile:
tnx for the reply.

Glad to read you have it working now. And yes, it is possible to get slightly better render quality in the source engine by tweaking cvars but it is an advanced task so to do it properly you have to at least have a general knowledge about the way nowdays 3d game engines work and how does the rendered picture gets constructed. Having programming experience with OpenGL 1.x, 2.x and 3.x or with D3D 8.0/8.1, 9.0 and 10.0 APIs could also be of a great help allowing you to understand what does cvars like mat_clipz or r_fastzreject really do. Now, back to the setting you have now in your config. You could safely remove the following (cause they are expected to be controled by the entries in the game video settings dialogs): mat_forceaniso, mat_trilinear, mat_vsync, mat_picmip, mat_antialias, mat_aaquality, r_shadowrendertotexture.

On the other hand it is safe to left them in your config but don’t be surprised later on when the game would behave like it don’t save changes made to some things like shadow or texture details level when you change then using in-game video setting menu.

Founded in 2004, Leakfree.org became one of the first online communities dedicated to Valve’s Source engine development. It is more famously known for the formation of Black Mesa: Source under the 'Leakfree Modification Team' handle in September 2004.