Agner`s CPU blog

Software optimization resources | E-mail subscription to this blog | www.agner.org

List Messageboards

Test results for AMD Bulldozer processor

Author: Agner

Date: 2012-03-21 01:59

Massimo wrote:

BD has a VERY slow REP MOVS

I think Sandy Bridge has a special implementation of REP MOVS, moving a whole cache line at a time under certain conditions. Many function libraries implement memcpy as a loop of aligned xmm moves, which is efficient on all processors.

You can still get a throughput of 4 instructions per clock on Bulldozer on a single thread if you mix integer and floating point instructions, og you mix different type of vector instructions.

Reply To This Message

Previous Message

Test results for AMD Bulldozer processor new - Agner - 2012-03-02

Test results for AMD Bulldozer processor new - Massimo - 2012-03-13

Test results for AMD Bulldozer processor new - Agner - 2012-03-14

Test results for AMD Bulldozer processor new - Alex - 2012-03-14

Test results for AMD Bulldozer processor new - fellix - 2012-03-15

Test results for AMD Bulldozer processor new - Agner - 2012-03-16

Test results for AMD Bulldozer processor new - Massimo - 2012-03-16

Test results for AMD Bulldozer processor new - Agner - 2012-03-17

Test results for AMD Bulldozer processor new - avk - 2012-03-17

Test results for AMD Bulldozer processor new - Massimo - 2012-03-17

Test results for AMD Bulldozer processor new - Agner - 2012-03-17

Test results for AMD Bulldozer processor new - Massimo - 2012-03-20

Test results for AMD Bulldozer processor - Agner - 2012-03-21

Cache WT performance of the AMD Bulldozer CPU new - GordonBGood - 2012-06-05

Test results for AMD Bulldozer processor new - zan - 2012-04-03

Multithreads load-store throughput for bulldozer new - A-11 - 2014-06-27

Multithreads load-store throughput for bulldozer new - Bigos - 2014-06-28

Multithreads load-store throughput for bulldozer new - A-11 - 2014-07-04

Store forwarding stalls of piledriver new - A-11 - 2014-09-07

List Messageboards