Re: Reverse engineering of Intel branch predictors

On Mon, 11 Nov 2024 9:03:36 +0000, Thomas Koenig wrote:

MitchAlsup1 <mitchalsup@aol.com> schrieb:
>

For example: The classical way to do dense switches is a LD of the
target
address and a jump to the target. This requires verifying the address of
the target. Whereas if you predict as JTT does, you verify by matching
the index number (which is known earlier and since the table is
read-only
you don't need to verify the target address.
>
So, it is not that you don't predict, it is that the data used to
verify the prediction is more precise and available earlier.

>
Hmm... just wondering.
>
How does this handle
>
  switch(a)
    {
       case 1:
       case 3:
       case 4:
       case 7:
         /* Do something */
       break;
       case 2:
       case 5:
       case 11:
         /* Do something else */
       break;
       default:
         /* Something else entirely  */
    }
>
Would the values 1,3,4 and 7 be considered different for
prediciton purposes?

Probably.
You might take the target IPs and put them in an indirect prediction
target buffer, so that by the time one has (a) one has the first inst-
runctions at the target label. Then, verification is by index
comparison\
not by IP-address comparison.