It looks like exercise gpgpu-3 requires a total of three buffers (one front and two back for the two previous states), but currently only two buffers are being created. I think the original author intended to use the numBuffers variable, which is set to 3, but accidently hardcoded 2 in the createStateBuffers call.
numBuffers was also declared but unused in the heat equation example, so I also updated that for consistency.
It looks like exercise gpgpu-3 requires a total of three buffers (one front and two back for the two previous states), but currently only two buffers are being created. I think the original author intended to use the
numBuffersvariable, which is set to 3, but accidently hardcoded 2 in thecreateStateBufferscall.numBufferswas also declared but unused in the heat equation example, so I also updated that for consistency.