Hi Guys,

In the process of adding Vector support to b0, I've done up the attached cheat sheet for SSE2 instructions.

While I undestand how SSE works, I've done very little with it in real life, so would appreciate if someone could look at the xls spreadsheet and offer any feedback. eg any missing instructions, or additional mappings, etc.

I've delibrately left out MMX, as all the operations should only operate on either mem128 or xmm registers. (no GPRs, or MMX regs). (b0 will only offer a subset of SSE, not the full range). Additionally I only want to support packed data types as well, since b0 offers full x87 support (but no MMX support).

Are there any operations that I've missed, or is it rather complete? (The shuffle commands are either 1 position left or 1 position right. I will add in the additional shuffle offsets in the actual implementation). For operations that I've missed, how would they best fit into the attached spreadsheet.

PS. It really seems that the only decent support in SSE2 is for 4x packed integer dwords, 4x single precision values or 2x packed double precision values. Without looking at SSE3/SSE4 I would expect those missing instructions to handle the other data types to have been filled in.

Very cool Chewy509, thanks for sharing!