<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"></head><body ><div>Hello, thank you for your feedback :)</div>I just sent a patch that fixes the NaN issue in all places it can happen (at least in the GLSL backend).<div>Well, stupid drivers... It seems to work really well with Mesa, but I just saw some 1 fps benchmarks, so I guess it's a *nope*.</div><div>I'll do a test to make you happy, and try to fix the other things. :)</div><br><br><br><br><br><br><div>-------- Message d'origine --------</div><div>De : Józef Kucia <joseph.kucia@gmail.com> </div><div>Date :13/06/2016 20:32 (GMT+01:00) </div><div>À : wine-devel <wine-devel@winehq.org> </div><div>Cc : Guillaume Charifi <guillaume.charifi@sfr.fr> </div><div>Objet : Re: [PATCH] wined3d: Implement icb dcl support in GLSL. </div><div><br></div>This implementation has some issues, e.g. a generated GLSL shader will<br>produce a compilation error when the immediate constant buffer<br>contains an unsigned integer which represents NaN. Morever, from<br>testing a similar implementation, it seems that some drivers don't<br>handle shaders with huge arrays very well and such shaders perform<br>poorly. My current impression is that we should implement immediate<br>constant buffers using a uniform variable.<br><br>Also, a test would be nice.<br><br>On Mon, Jun 13, 2016 at 6:41 PM, Guillaume Charifi<br><guillaume.charifi@sfr.fr> wrote:<br>> Signed-off-by: Guillaume Charifi <guillaume.charifi@sfr.fr><br>> ---<br>> dlls/wined3d/glsl_shader.c | 15 ++++++++++++++-<br>> 1 file changed, 14 insertions(+), 1 deletion(-)<br>><br>> diff --git a/dlls/wined3d/glsl_shader.c b/dlls/wined3d/glsl_shader.c<br>> index 55affc1..def224c 100644<br>> --- a/dlls/wined3d/glsl_shader.c<br>> +++ b/dlls/wined3d/glsl_shader.c<br>> @@ -1886,6 +1886,19 @@ static void shader_generate_glsl_declarations(const struct wined3d_context *cont<br>> if (shader->limits->constant_bool > 0 && reg_maps->boolean_constants)<br>> shader_addline(buffer, "uniform bool %s_b[%u];\n", prefix, shader->limits->constant_bool);<br>><br>> + if (reg_maps->icb)<br>> + {<br>> + shader_addline(buffer, "const vec4 %s_icb[%u] = vec4[%u](\n",<br>> + prefix, reg_maps->icb->element_count / 4, reg_maps->icb->element_count / 4);<br>> + for (i = 0; i < reg_maps->icb->element_count / 4; i++) {<br>> + shader_glsl_append_imm_vec4(buffer, (float *)®_maps->icb->data[4 * i]);<br>> + if (i != reg_maps->icb->element_count / 4 - 1)<br>> + shader_addline(buffer, ",");<br>> + shader_addline(buffer, "\n");<br>> + }<br>> + shader_addline(buffer, ");\n");<br>> + }<br><br>I think it would be better to store the number of vectors instead of<br>the element count in "reg_maps->icb". Not that it matters much, but<br>this block of code would certainly look nicer.<br><br>> +<br>> for (i = 0; i < WINED3D_MAX_CBS; ++i)<br>> {<br>> if (reg_maps->cb_sizes[i])<br>> @@ -8561,7 +8574,7 @@ static const SHADER_HANDLER shader_glsl_instruction_handler_table[WINED3DSIH_TAB<br>> /* WINED3DSIH_DCL_GLOBAL_FLAGS */ shader_glsl_nop,<br>> /* WINED3DSIH_DCL_HS_FORK_PHASE_INSTANCE_COUNT */ NULL,<br>> /* WINED3DSIH_DCL_HS_MAX_TESSFACTOR */ NULL,<br>> - /* WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER */ NULL,<br>> + /* WINED3DSIH_DCL_IMMEDIATE_CONSTANT_BUFFER */ shader_glsl_nop,<br>> /* WINED3DSIH_DCL_INPUT */ shader_glsl_nop,<br>> /* WINED3DSIH_DCL_INPUT_CONTROL_POINT_COUNT */ NULL,<br>> /* WINED3DSIH_DCL_INPUT_PRIMITIVE */ shader_glsl_nop,<br>> --<br>> 2.7.4<br>><br>><br>><br></body>