On 12/27/2017 07:31 AM, Kirill Batuzov wrote:
> I think something is wrong with instruction encodings here. Looks like
> tcg_out_mov(_ctx, TCG_TYPE_I64, TCG_REG_EBP, TCG_REG_XMM0)
> produces
> vmovq %xmm5, %rax
> instead.
Bah. The operands are swapped -- ebp == 5 and eax == 0.
r~
On Mon, 18 Dec 2017, Richard Henderson wrote:
> The x86 vector instruction set is extremely irregular. With newer
> editions, Intel has filled in some of the blanks. However, we don't
> get many 64-bit operations until SSE4.2, introduced in 2009.
>
> The subsequent edition was for AVX1,
The x86 vector instruction set is extremely irregular. With newer
editions, Intel has filled in some of the blanks. However, we don't
get many 64-bit operations until SSE4.2, introduced in 2009.
The subsequent edition was for AVX1, introduced in 2011, which added
three-operand addressing, and