- Marquer comme nouveau
- Marquer
- S'abonner
- Sourdine
- S'abonner au fil RSS
- Surligner
- Imprimer
- Signaler un contenu inapproprié
Today I use iaca(a tool to analysis code), I find that the instruction "molvpd" and "movupd"
has difference numbers of uops .
the following is the instrucion I used in my assembly (AT&T style)
movlpd %mm1,(%rdx,%rcx,8)
movupd %mm1,(%rdx,%rcx,8)
movupd has 1 uops, while movlpd has 2 .
movlpd accupied port2(load port) and port 5(SIMD ALU).
I wonder why it occupied port 5?
I thinks the two instruction are almost the same .
but the result is they are not the same.
Why this could happend?
has difference numbers of uops .
the following is the instrucion I used in my assembly (AT&T style)
movlpd %mm1,(%rdx,%rcx,8)
movupd %mm1,(%rdx,%rcx,8)
movupd has 1 uops, while movlpd has 2 .
movlpd accupied port2(load port) and port 5(SIMD ALU).
I wonder why it occupied port 5?
I thinks the two instruction are almost the same .
but the result is they are not the same.
Why this could happend?
Lien copié
1 Répondre
- Marquer comme nouveau
- Marquer
- S'abonner
- Sourdine
- S'abonner au fil RSS
- Surligner
- Imprimer
- Signaler un contenu inapproprié
One difference is that for movlpd, "the high quadword of the xmm register remains unchanged", i.e. partof the old register value is still needed. How this is implemented depends on the micro-architecture.

Répondre
Options du sujet
- S'abonner au fil RSS
- Marquer le sujet comme nouveau
- Marquer le sujet comme lu
- Placer ce Sujet en tête de liste pour l'utilisateur actuel
- Marquer
- S'abonner
- Page imprimable