5
$\begingroup$

I have written a deferred renderer that can use either a fragment shader or a compute shader to execute the shading pass. Unfortunately, the compute shader implementation runs slower. I'm trying to understand why.

I believe I understand the proximate cause: memory locality when accessing textures. Somehow the fragment shader's accesses are significantly more coherent than the compute shader's.


To demonstrate that, I removed everything except the shadowmapping code and then changed that to sample randomly. Something like (GLSL pseudocode):

uniform sampler2D tex_shadowmap;

uniform float param;

#ifdef COMPUTE_SHADER
layout(local_size_x=8, local_size_y=4, local_size_z=1) in;
#endif

struct RNG { uint64_t state; uint64_t inc; } _rng;
void rand_seed(ivec2 coord) { /*seed `_rng` with hash of `coord`*/ }
float rand_float() { /*return random float in [0,1]*/ }

void main() {
    rand_seed(/*pixel coordinate*/);

    vec4 light_coord = /*vertex in scaled/biased light's NDC*/;
    vec3 shadowmap_test_pos = light_coord.xyz / light_coord.w;

    float rand_shadow = 0.0;
    for (int i=0;i<200;++i) {
        vec2 coord = fract(mix( shadowmap_test_pos.xy, vec2(rand_float(),rand_float()), param ));
        float tap = textureLod(tex_shadowmap,coord,0.0).r;
        rand_shadow += clamp(shadowmap_test_pos.z,0.0,1.0)<=tap+0.00001 ? 1.0 : 0.0;
    }
    vec4 color = vec4(vec3(rand_shadow)/200.0,1.0);

    /*[set `color` into output]*/
}

When param is set to 0, the shadowmap is sampled at shadowmap_test_pos, and we get correct hard shadows for the scene. In this case, the shadowmap texture lookup locations are somewhat correlated to the pixel coordinate, so we expect good performance. When param is set to 1, we get a completely random texture coordinate vec2(rand_float(),rand_float()), and so the texture lookups are not at all correlated to the pixel coordinate, and we expect bad performance.

Something very interesting happens when we try some more values for param and measure the latency of the shading pass with a timer query:

plot of param vs. latency

As one can see, when working with completely random coordinates (param=1, right side), the fragment shader and compute shader have the same performance. However, as the coordinates become less random, whatever the fragment shader is doing that makes it more coherent starts to come into play. When the coordinates are deterministic and correlated to screen position (param≈0, left side), the fragment shader wins by a factor of 2 (note: the param=0 case was omitted since the GLSL compiler optimizes out the loop).

What's especially weird is that the fragment shader being faster seems to depend on the texture sample coordinate being correlated to the pixel coordinate. For example, if instead of shadowmap_test_pos.xy I use vec2(0.5) as the deterministic coordinate, then the effect disappears, and the two shaders have the same performance for any param.

Both the sources and compiled codes of these shaders are essentially the same. Apart from some setup and writing the data out (which one expects to vary a bit), the shaders are identical. You can see a diff I made of the PTX disassemblies here. Most of the loop body is taken up with the inlined RNG, but the salient point is that it's the same loop.

Note: tested hardware was NVIDIA GTX 1080 with current (446.14) driver.


My question is basically: what can I do about this? I'm working in 8⨯4 tiles in the compute shader, but who knows what the fragment shader is doing. Yet, I would not expect whatever magical secret shading order the fragment shader does to be so much better that you would get a >2⨯ performance difference when you're running the same actual code. (FWIW I've tried different group sizes, to no real change in the above behavior.)

There are a few general discussions about how the different shaders work, but I haven't found anything that can explain this. And, while in the past driver issues have caused weird behavior, compute shaders have now been in core GL for almost 8 years, and using them for deferred shading is an obvious, arguably common, use-case that I would expect to work well.

What am I missing here?

$\endgroup$
  • 2
    $\begingroup$ "using them for deferred shading is an obvious, arguably common, use-case that I would expect to work well" Is it? I've never heard of it. There's little reason to use a CS for this when the CS and FS both have the same expressive power and the FS has access to blending operations, depth testing, and other useful functionality that can be used to improve overall rendering performance. Basically, the CS is never going to be faster than the FS, and the FS can do everything the CS can do, so why bother? $\endgroup$ – Nicol Bolas Jun 19 at 14:48
  • $\begingroup$ @NicolBolas As in the linked ref., the CS has way more expressive power than the FS, and since it bypasses the entire graphics pipeline above and the ROPs below, it can theoretically be faster. Indeed, in this very shader, the CS is ≈10% faster if I replace these texture accesses with an expensive BRDF that touches no textures. As to why you'd prefer the CS: if that isn't enough, it's also much cleaner to set up, and the CS's added flexibility allows you to subsequently extend the FS's functionality after you've replicated it. $\endgroup$ – imallett Jun 19 at 15:43
  • $\begingroup$ "since it bypasses the entire graphics pipeline above and the ROPs below, it can theoretically be faster." The cost of a graphics operation will be the cost of the slowest stage of that operation. Vertex processing will involve very few vertices, and ROP operations won't matter next to all of the FS processing. So the overall performance will be dominated by the slowest step: the fragment shader. Maybe if your FS is tiny you might get some performance benefit from a CS, but in general, no. Not for deferred rendering. $\endgroup$ – Nicol Bolas Jun 19 at 15:56
  • $\begingroup$ "the CS has way more expressive power than the FS" Expressive power that just doesn't matter for the specific case of a deferred renderer. Shared data storage and invocation intercommunication is irrelevant, as deferred lighting doesn't need intercommunication with other invocations (post-effects like SSAO, blur effects, and the like are not specific to deferred rendering and thus don't count here). And control of the work group size is likewise immaterial. So whatever expressive power exists is kind of irrelevant. $\endgroup$ – Nicol Bolas Jun 19 at 15:59
  • 4
    $\begingroup$ @NicolBolas You're shifting goalposts and being dismissive. And, while I actually do agree with many of the things you're saying, provided they were rephrased more carefully and as general heuristics, the question isn't about that. I don't care to get into it, and that's not what the comments section is for anyway . . . $\endgroup$ – imallett Jun 19 at 16:32
4
$\begingroup$

After more analysis, the TL;DR here is that, yes, the slowdown is due to memory locality, and yes the pixel order is to blame. More interestingly, by writing the shader differently, we can greatly surpass the fragment shader's performance—though we obviously shouldn't rely on being able to do that regularly.


First, to expand on the analysis: the best way to figure out what's going on in the GPU is to ask it. In this case, the relevant tool is NVIDIA NSight. After some fiddling, I got directly comparable results, which indicated that in both cases, memory is the bottleneck, and in the compute shader's case, it is worse.

Since the actual shader code is substantively identical at the assembly level (see above), and (slightly-better-than-)equal performance can be achieved by removing memory from the equation by altering the shading code, we can be confident that the pixel shading order is to blame.


Perhaps we can find a better shading order?

Spoiler alert: we can. After some experimentation, consider a new shader where there is a global queue of tiles, and each warp grabs a tile and shades the pixels within it in scanline order. This turns out to be 50% faster than the fragment shader!


Here's an animation from my HPG paper's presentation this week, which touched on this issue: enter image description here (It can be embiggened, which you may wish to do if you're having trouble reading the text.)

This summarizes the results of these experiments, along with the performance numbers for each and a visualization of what I surmise is going on behind the scenes (simplified: only one warp is shown, it is 8 wide, and latency-hiding is not visualized).

On the left, we have the fragment shader, labeled "Vendor Magic Goes Here". We don't know what the vendor is doing for their fragment shader pixel traversal order (though we could get hints e.g. by writing out atomic variables, etc.), but overall it works really well.

In the middle, we have the original compute shader I described (with param = 0), which divides the framebuffer into rectangular work groups. Notice that the work groups are probably executed in a mostly reasonable order exactly to mitigate these caching effects, but are not guaranteed to be in any order at all—and indeed will not be due to latency hiding: this explains why the groups are walking over the framebuffer in a mostly coherent fashion, but still skipping around a little bit. This is half the speed of the fragment shader, and I believe the possible skipping around is a reasonable starting guess for the additional memory incoherency revealed within the profile.

Finally, we have the tiles version. Because tiles are processed in a queue of tiles (defined by a global counter, visualized above the tile), the tiles and pixels are processed more in order (neglecting latency hiding and other thread groups). I believe this is a reasonable starting guess as to why this result turns out to be 50% faster than the fragment shader.

It is important to stress that, while these results are correct for this particular experiment, with these particular drivers, these results do not necessarily generalize. This is likely specific to this particular scene, view, and platform configuration, and this behavior may actually even be a bug. This is definitely interesting to play with, but don't go ripping apart your renderer (only) because of one datapoint from a narrowly-defined experiment.


Indeed, what kicked off this whole investigation was that the performance of a (more-complex) compute shader had declined in relative performance since it was last profiled in 2018, using the same code on the same hardware. The only difference was an updated driver.

The lesson is simple: pixel shading orders are hard, and as much as possible they are best left to the GPU vendor to determine. Compute shaders give us the option to do shading-like operations, but we should not expect to be able to reliably exceed the performance of fragment shaders (even if, occasionally, we spectacularly can) because our implementations are not based on insider knowledge of how to optimize for the particular GPU—even when there is a single particular GPU at all.

So, if you're thinking about shading orders, that's really something the GPU should be doing for you: take it up with the vendor. The main reason to use a compute shader would be if you want the convenience or flexibility. Of course, if you thoroughly profile and see a performance gain, and you have reason to expect that the GPU infrastructure you are building on top of will not shift beneath your feet (e.g. you are writing for a console), then maybe using a compute shader is the right choice.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Very nice experimant! how have you changed the CS to tiles/groups rendering? What exactly does that mean? is it only the group size? $\endgroup$ – Thomas Jul 17 at 12:32
  • $\begingroup$ @Thomas See the diagram on the right for a visualization; the strong red box represents individual thread groups. $\endgroup$ – imallett Jul 17 at 19:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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