SHORT Main memory bandwidth

EVENTSET
FIXC0   INSTR_RETIRED_ANY
FIXC1   CPU_CLK_UNHALTED_CORE
FIXC2   CPU_CLK_UNHALTED_REF
WBOXFIX UNCORE_CLOCK
MBOX0C0 FVC_EV0_BBOX_CMDS_READS
MBOX0C1 DRAM_CMD_CAS_WR_OPN
MBOX1C0 FVC_EV0_BBOX_CMDS_READS
MBOX1C1 DRAM_CMD_CAS_WR_OPN


METRICS
Runtime (RDTSC) [s] time
Runtime unhalted [s] FIXC1*inverseClock
Clock [MHz]  1.E-06*(FIXC1/FIXC2)/inverseClock
Uncore Clock [MHz]  1.E-06*(WBOXFIX)/time
CPI  FIXC1/FIXC0
Memory read bandwidth [MBytes/s] 1.0E-06*(MBOX0C0+MBOX1C0)*64.0/time
Memory read data volume [GBytes] 1.0E-09*(MBOX0C0+MBOX1C0)*64.0
Memory write bandwidth [MBytes/s] 1.0E-06*(MBOX0C1+MBOX1C1)*64.0/time
Memory write data volume [GBytes] 1.0E-09*(MBOX0C1+MBOX1C1)*64.0
Memory bandwidth [MBytes/s] 1.0E-06*(MBOX0C0+MBOX1C0+MBOX0C1+MBOX1C1)*64/time
Memory data volume [GBytes] 1.0E-09*(MBOX0C0+MBOX1C0+MBOX0C1+MBOX1C1)*64

LONG
Formulas:
Uncore Clock [MHz] = 1.E-06*(UNCORE_CLOCK)/time
Memory read bandwidth [MBytes/s] = 1.0E-06*(SUM(FVC_EV0_BBOX_CMDS_READS))*64.0/time
Memory read data volume [GBytes] = 1.0E-09*(SUM(FVC_EV0_BBOX_CMDS_READS))*64.0
Memory write bandwidth [MBytes/s] = 1.0E-06*(SUM(MDRAM_CMD_CAS_WR_OPN))*64.0/time
Memory write data volume [GBytes] = 1.0E-09*(SUM(DRAM_CMD_CAS_WR_OPN))*64.0
Memory bandwidth [MBytes/s] = 1.0E-06*(SUM(FVC_EV0_BBOX_CMDS_READS)+SUM(DRAM_CMD_CAS_WR_OPN))*64.0/time
Memory data volume [GBytes] = 1.0E-09*(SUM(FVC_EV0_BBOX_CMDS_READS)+SUM(DRAM_CMD_CAS_WR_OPN))*64.0
-
Profiling group to measure memory bandwidth drawn by all cores of a socket.
On Nehalem EX it is not possible to measure the write operations with the
FVC_EV0_BBOX_CMDS_WRITES event at the same time as the FVC_EV0_BBOX_CMDS_READS
because they set contrary bits. The DRAM_CMD_CAS_WR_OPN is an alternative but
it only measures write operations to open pages, hence writes to closed pages
are not included here.