I wrote a simple 2D renderer in OpenGL that works pretty well. I can render about 16,000k sprites in 1 draw call if it uses the same texture. I wanted to refactor this code out and put it in its own module to make it easier to work with.

With this, I can only draw about 100 sprites before things get really slow:

mat4* v_mat, *p_mat, *mvp_mat;
GLint model_mat_loc;
GLint view_mat_loc;
GLint proj_mat_loc;
GLint mvp_matrix_loc;
GLint pos_loc;
GLint col_loc;
GLint tex_loc;

GLuint vao;
GLuint vert_buff;
GLuint ind_buff;

int sprite_count;
int idx;
GLuint texture_id;

float* v_buff;
short* i_buff;
size_t vbo_size_in_bytes;

tran3 rot, scal, trns, tmp;

cg_cam* camera;
Shader* shader;

//creating a spritebatch object
cg_spritebatch* out = calloc(1, sizeof(cg_spritebatch));
out->shader = calloc(1, sizeof(Shader));
out->shader = s;
out->sprite_count = 16000;

out->pos_loc = get_attrib_location(out->shader, "a_pos");
out->col_loc = get_attrib_location(out->shader, "a_col");
out->mvp_matrix_loc = get_uniform_location(out->shader, "u_mvp_mat");
out->model_mat_loc = get_uniform_location(out->shader, "u_mod_mat");
out->view_mat_loc = get_uniform_location(out->shader, "u_view_mat");
out->proj_mat_loc = get_uniform_location(out->shader, "u_proj_mat");
out->tex_loc = get_uniform_location(out->shader, "u_sprite_tex");

out->vbo_size_in_bytes =
    ((cg_sprite_get_sizeof_vert() * out->sprite_count) +
     (cg_sprite_get_sizeof_col() * out->sprite_count) +
     (cg_sprite_get_sizeof_tex_coord() * out->sprite_count));

out->v_buff = calloc(1, (out->vbo_size_in_bytes));
out->i_buff = calloc(1, (cg_sprite_get_sizeof_ind() * out->sprite_count));

glGenVertexArrays(1, &out->vao);

glGenBuffers(1, &out->vert_buff);
glBindBuffer(GL_ARRAY_BUFFER, out->vert_buff);
glBufferData(GL_ARRAY_BUFFER, (out->vbo_size_in_bytes), out->v_buff,

glVertexAttribPointer(0, 3, GL_FLOAT, GL_FALSE, 9 * sizeof(float),

glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, 9 * sizeof(float),
                      (GLvoid*)(3 * sizeof(float)));

glVertexAttribPointer(2, 2, GL_FLOAT, GL_FALSE, 9 * sizeof(float),
                      (GLvoid*)(7 * sizeof(float)));

int len = out->sprite_count * 6;
short j = 0;
for (int i = 0; i < len; i += 6, j += 4) {
    out->i_buff[i] = j;
    out->i_buff[i + 1] = (short)(j + 1);
    out->i_buff[i + 2] = (short)(j + 2);

    out->i_buff[i + 3] = (short)(j + 2);
    out->i_buff[i + 4] = (short)(j + 3);
    out->i_buff[i + 5] = j;
glGenBuffers(1, &out->ind_buff);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, out->ind_buff);
             out->sprite_count * cg_sprite_get_sizeof_ind(), out->i_buff,

glBindBuffer(GL_ARRAY_BUFFER, 0);

out->idx = 0;
out->texture_id = -1;

return out;

The main functions of this module:

void cg_spritebatch_begin(cg_spritebatch* b, cg_cam* cam) {
    b->camera = cam;
    b->idx = 0;

void flush(cg_spritebatch* b) {



    glUniform1i(b->tex_loc, 0);
    glBindTexture(GL_TEXTURE_2D, b->texture_id);

    cg_cam_get_matrices(&b->v_mat, &b->p_mat, &b->mvp_mat, b->camera);

    glUniformMatrix4fv(b->view_mat_loc, 1, GL_FALSE, vmathM4GetData(b->v_mat));
    glUniformMatrix4fv(b->proj_mat_loc, 1, GL_FALSE, vmathM4GetData(b->p_mat));
    glUniformMatrix4fv(b->mvp_matrix_loc, 1, GL_FALSE,

    glBindBuffer(GL_ARRAY_BUFFER, b->vert_buff);
    glBufferData(GL_ARRAY_BUFFER, (b->vbo_size_in_bytes), b->v_buff,

    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, b->ind_buff);
                 b->sprite_count * cg_sprite_get_sizeof_ind(), b->i_buff,

    glDrawElements(GL_TRIANGLES, b->sprite_count * cg_sprite_get_vert_count(),
                   GL_UNSIGNED_SHORT, 0);


    b->idx = 0;

static void update_texture(cg_spritebatch* b, GLuint t) {
    b->texture_id = t;

void cg_spritebatch_draw(cg_spritebatch* b, cg_sprite* sp) {
    int idx = b->idx;
    int i = idx;
    idx = 0;

    if (sp->texture_id != b->texture_id) {
        update_texture(b, sp->texture_id);
    } else if (b->idx == (36 * b->sprite_count)) {


    vmathT3MakeScale(&b->scal, &sp->scale);
    vmathT3MakeRotationZYX(&b->rot, &sp->angl);
    vmathT3MakeTranslation(&b->trns, &sp->pos);
    vmathT3Mul(&b->tmp, &b->trns, &b->scal);
    vmathT3Mul(&b->tmp, &b->tmp, &b->rot);

    vmathM4MakeFromT3(&sp->m_mat, &b->tmp);

    cg_quad_getquadverts(&sp->iv0, &sp->iv1, &sp->iv2, &sp->iv3, sp->quad);
    vmathM4MulV4(&sp->ov0, &sp->m_mat, &sp->iv0);
    vmathM4MulV4(&sp->ov1, &sp->m_mat, &sp->iv1);
    vmathM4MulV4(&sp->ov2, &sp->m_mat, &sp->iv2);
    vmathM4MulV4(&sp->ov3, &sp->m_mat, &sp->iv3);
    /* vmathV4Prints(&sp->ov0, "v0"); */

    // v0
    b->v_buff[idx++] = sp->ov0.x;
    b->v_buff[idx++] = sp->ov0.y;
    b->v_buff[idx++] = sp->ov0.z;

    b->v_buff[idx++] = sp->quad->colors[0];
    b->v_buff[idx++] = sp->quad->colors[1];
    b->v_buff[idx++] = sp->quad->colors[2];
    b->v_buff[idx++] = sp->quad->colors[3];

    b->v_buff[idx++] = sp->quad->tex_coords[0];
    b->v_buff[idx++] = sp->quad->tex_coords[1];

    // v1
    b->v_buff[idx++] = sp->ov1.x;
    b->v_buff[idx++] = sp->ov1.y;
    b->v_buff[idx++] = sp->ov1.z;

    b->v_buff[idx++] = sp->quad->colors[4];
    b->v_buff[idx++] = sp->quad->colors[5];
    b->v_buff[idx++] = sp->quad->colors[6];
    b->v_buff[idx++] = sp->quad->colors[7];

    b->v_buff[idx++] = sp->quad->tex_coords[2];
    b->v_buff[idx++] = sp->quad->tex_coords[3];

    // v2
    b->v_buff[idx++] = sp->ov2.x;
    b->v_buff[idx++] = sp->ov2.y;
    b->v_buff[idx++] = sp->ov2.z;

    b->v_buff[idx++] = sp->quad->colors[8];
    b->v_buff[idx++] = sp->quad->colors[9];
    b->v_buff[idx++] = sp->quad->colors[10];
    b->v_buff[idx++] = sp->quad->colors[11];

    b->v_buff[idx++] = sp->quad->tex_coords[4];
    b->v_buff[idx++] = sp->quad->tex_coords[5];

    // v3
    b->v_buff[idx++] = sp->ov3.x;
    b->v_buff[idx++] = sp->ov3.y;
    b->v_buff[idx++] = sp->ov3.z;

    b->v_buff[idx++] = sp->quad->colors[12];
    b->v_buff[idx++] = sp->quad->colors[13];
    b->v_buff[idx++] = sp->quad->colors[14];
    b->v_buff[idx++] = sp->quad->colors[15];

    b->v_buff[idx++] = sp->quad->tex_coords[6];
    b->v_buff[idx++] = sp->quad->tex_coords[7];



void cg_spritebatch_end(cg_spritebatch* b) {  flush(b); }

I use these functions like this:

cg_spritebatch_begin(sb, ce_get_default_camera());
for (int i = 0; i < sc; i++) {
    cg_spritebatch_draw(sb, sprites[i]);

With this code, I can draw about 100 sprites at most before things get really terrible. With the same code outside of those loops, I can draw 16,000 sprites with no problem.

For example, here is the same code outside of the spritebatch setup that works really well:

edit fixed the unrolled loop for drawing 16,000 sprites in one draw call.

void variable_render(double alpha) {
    glClearColor(0.2f, 0.3f, 0.3f, 1.0f);

    cg_sprite_back_2_front_tex_id(sprites, sc);

    cg_spritebatch_begin(sb, ce_get_default_camera());
    for (int i = 0; i < sc; i++) {
        cg_spritebatch_draw(sb, sprites[i]);

    idx = 0;
    for (int i = 0; i < sc; i++) {
        //--------------- start update vertex data ---------------------

        sp = sprites[i];
        /* printf("%d %d\n",sp->texture_id, sp->z_index); */

        vmathT3MakeScale(&scal, &sp->scale);
        vmathT3MakeRotationZYX(&rot, &sp->angl);
        vmathT3MakeTranslation(&trns, &sp->pos);
        vmathT3Mul(&tmp, &trns, &scal);  // scale then trnslate
        vmathT3Mul(&tmp, &tmp, &rot);    // scale then translate then rotate

        vmathM4MakeFromT3(&sprites[i]->m_mat, &tmp);

        cg_quad_getquadverts(&sp->iv0, &sp->iv1, &sp->iv2, &sp->iv3, sp->quad);
        vmathM4MulV4(&sp->ov0, &sp->m_mat, &sp->iv0);
        vmathM4MulV4(&sp->ov1, &sp->m_mat, &sp->iv1);
        vmathM4MulV4(&sp->ov2, &sp->m_mat, &sp->iv2);
        vmathM4MulV4(&sp->ov3, &sp->m_mat, &sp->iv3);
        /* --------------- finish update vertex data --------------------- */

        /* --------------- start packing data into buffers----------------- */
        /* sp = sprites[i]; */
        // v0
        v_buff[idx++] = sp->ov0.x;
        v_buff[idx++] = sp->ov0.y;
        v_buff[idx++] = sp->ov0.z;

        v_buff[idx++] = sp->quad->colors[0];
        v_buff[idx++] = sp->quad->colors[1];
        v_buff[idx++] = sp->quad->colors[2];
        v_buff[idx++] = sp->quad->colors[3];

        v_buff[idx++] = sp->quad->tex_coords[0];
        v_buff[idx++] = sp->quad->tex_coords[1];

        // v1
        v_buff[idx++] = sp->ov1.x;
        v_buff[idx++] = sp->ov1.y;
        v_buff[idx++] = sp->ov1.z;

        v_buff[idx++] = sp->quad->colors[4];
        v_buff[idx++] = sp->quad->colors[5];
        v_buff[idx++] = sp->quad->colors[6];
        v_buff[idx++] = sp->quad->colors[7];

        v_buff[idx++] = sp->quad->tex_coords[2];
        v_buff[idx++] = sp->quad->tex_coords[3];

        // v2
        v_buff[idx++] = sp->ov2.x;
        v_buff[idx++] = sp->ov2.y;
        v_buff[idx++] = sp->ov2.z;

        v_buff[idx++] = sp->quad->colors[8];
        v_buff[idx++] = sp->quad->colors[9];
        v_buff[idx++] = sp->quad->colors[10];
        v_buff[idx++] = sp->quad->colors[11];

        v_buff[idx++] = sp->quad->tex_coords[4];
        v_buff[idx++] = sp->quad->tex_coords[5];

        // v3
        v_buff[idx++] = sp->ov3.x;
        v_buff[idx++] = sp->ov3.y;
        v_buff[idx++] = sp->ov3.z;

        v_buff[idx++] = sp->quad->colors[12];
        v_buff[idx++] = sp->quad->colors[13];
        v_buff[idx++] = sp->quad->colors[14];
        v_buff[idx++] = sp->quad->colors[15];

        v_buff[idx++] = sp->quad->tex_coords[6];
        v_buff[idx++] = sp->quad->tex_coords[7];
        /* printf("my idx:%d\n",idx*sc); */

    //--------------- finish packing data into buffers --------------------




    glUniform1i(tex_loc, 0);
    glBindTexture(GL_TEXTURE_2D, sp->texture_id);

    cg_cam_get_matrices(&v_mat, &p_mat, &mvp_mat, ce_get_default_camera());

    // projection * view * model * vertex_pos;
    glUniformMatrix4fv(model_mat_loc, 1, GL_FALSE, vmathM4GetData(&sp->m_mat));
    glUniformMatrix4fv(view_mat_loc, 1, GL_FALSE, vmathM4GetData(v_mat));

    glUniformMatrix4fv(proj_mat_loc, 1, GL_FALSE, vmathM4GetData(p_mat));

    glUniformMatrix4fv(mvp_matrix_loc, 1, GL_FALSE, vmathM4GetData(mvp_mat));

    glBindBuffer(GL_ARRAY_BUFFER, vert_buff);
    glBufferData(GL_ARRAY_BUFFER, (vbo_size_in_bytes), v_buff, GL_STREAM_DRAW);

    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, ind_buff);
    glBufferData(GL_ELEMENT_ARRAY_BUFFER, sc * cg_sprite_get_sizeof_ind(),
                 i_buff, GL_STREAM_DRAW);

    glDrawElements(GL_TRIANGLES, sc * cg_sprite_get_vert_count(),
                   GL_UNSIGNED_SHORT, 0);

    debug_opengl("render loop");

ok the above code on my machine can render 16000 sprites in 1 draw call but of course they all need to use the same texture.

I have 4 total textures which should be able to draw the 16000 sprites with their own unique texture instead of being forced to use only one like I am doing right now.

I also added two images enter image description here enter image description here

  • \$\begingroup\$ Your code for "outside the spritebatch setup" doesn't make sense because you are doing the "flush" part on every sprite. I think you are ending up with only one sprite in your vertex buffer instead of 16000. \$\endgroup\$
    – JS1
    Commented Jan 19, 2016 at 4:38
  • \$\begingroup\$ I just edited the question and fixed those issues. This is the reason why I would like to abstract this large chunk of code away somewhere that I won't be messing with it too often. \$\endgroup\$ Commented Jan 19, 2016 at 10:31

1 Answer 1



At the end of cg_spritebatch_draw(), this line:


should be:

b->idx = idx;

Otherwise you will clobber most of your vertex data on each new sprite. Also, this will cause your program to flush at 1/36 the rate that it should be flushing (maybe the cause of your performance problem).

Flush confusion

I'm not sure how to compare your batched version with your "unrolled" version because it looks like your unrolled version flushes the vertex buffer after each sprite. I'm not even sure that is what you intended to do because lines like this:

    glDrawElements(GL_TRIANGLES, sc * cg_sprite_get_vert_count(),
                   GL_UNSIGNED_SHORT, 0);

make it seem like you want to draw all of the sprites (sc) at once instead of just one. Perhaps you could clarify what your code is attempting to do in each of the two versions.

  • \$\begingroup\$ ***I just edited the above question fixing the errors and adding some images. *** In the unrolled version I am attempting to draw all the sprites in 1 draw call. I was doing that at one point but in an attempt to refactor seems I changed the code to single draw calls. The reason for this is that each sprite uses 1 or 4 textures and without drawing 1 at a time i'd only be able to draw everything with 1 texture so the other 3 textures weren't being used. \$\endgroup\$ Commented Jan 19, 2016 at 10:30
  • \$\begingroup\$ @user1610950 I can see you fixed the unrolled version. Did you try changing the code to b->idx = idx in your batched version? \$\endgroup\$
    – JS1
    Commented Jan 19, 2016 at 18:55
  • \$\begingroup\$ I made those changes as well but I am only drawing 1 sprite. \$\endgroup\$ Commented Jan 19, 2016 at 20:26
  • \$\begingroup\$ @user1610950 I thought you set out->sprite_count = 16000;? Doesn't that mean you are drawing 16000 sprites? Are you running code that you didn't show? \$\endgroup\$
    – JS1
    Commented Jan 19, 2016 at 22:44
  • \$\begingroup\$ no that's all the code, that I am running. I feel like i've coded myself into a corner with bad design decisions. \$\endgroup\$ Commented Jan 19, 2016 at 23:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.