SHORT Overview of arithmetic and main memory performance

EVENTSET
FIXC0 INSTR_RETIRED_ANY
FIXC1 CPU_CLK_UNHALTED_CORE
FIXC2 CPU_CLK_UNHALTED_REF
TMP0  TEMP_CORE
PWR0  PWR_PKG_ENERGY
PWR3  PWR_DRAM_ENERGY
PMC0  FP_COMP_OPS_EXE_SSE_FP_PACKED_DOUBLE
PMC1  FP_COMP_OPS_EXE_SSE_FP_SCALAR_DOUBLE
PMC2  SIMD_FP_256_PACKED_DOUBLE
MBOX0C0 CAS_COUNT_RD
MBOX0C1 CAS_COUNT_WR
MBOX1C0 CAS_COUNT_RD
MBOX1C1 CAS_COUNT_WR
MBOX2C0 CAS_COUNT_RD
MBOX2C1 CAS_COUNT_WR
MBOX3C0 CAS_COUNT_RD
MBOX3C1 CAS_COUNT_WR

METRICS
Runtime (RDTSC) [s] time
Runtime unhalted [s] FIXC1*inverseClock
Clock [MHz]  1.E-06*(FIXC1/FIXC2)/inverseClock
CPI  FIXC1/FIXC0
Temperature TMP0
Energy [J]  PWR0
Power [W] PWR0/time
Energy DRAM [J]  PWR3
Power DRAM [W] PWR3/time
MFlops/s  1.0E-06*(PMC0*2.0+PMC1)/time
32b AVX MFlops/s  1.0E-06*(PMC2*4.0)/time
Packed MUOPS/s   1.0E-06*(PMC0+PMC2)/time
Scalar MUOPS/s 1.0E-06*PMC1/time
Memory Read BW [MBytes/s] 1.0E-06*(MBOX0C0+MBOX1C0+MBOX2C0+MBOX3C0)*64.0/time
Memory Write BW [MBytes/s] 1.0E-06*(MBOX0C1+MBOX1C1+MBOX2C1+MBOX3C1)*64.0/time
Memory BW [MBytes/s] 1.0E-06*(MBOX0C0+MBOX0C1+MBOX1C0+MBOX1C1+MBOX2C0+MBOX2C1+MBOX3C0+MBOX3C1)*64.0/time
Memory data volume [GBytes] 1.0E-09*(MBOX0C0+MBOX0C1+MBOX1C0+MBOX1C1+MBOX2C0+MBOX2C1+MBOX3C0+MBOX3C1)*64.0

LONG
Formula:
Power =  PWR_PKG_ENERGY / runtime
Power DRAM = PWR_DRAM_ENERGY / runtime
MFlops/s =  (FP_COMP_OPS_EXE_SSE_FP_PACKED*2 +  FP_COMP_OPS_EXE_SSE_FP_SCALAR)/ runtime
AVX MFlops/s =  (SIMD_FP_256_PACKED_DOUBLE*4)/ runtime
Memory Read BW [MBytes/s] 1.0E-06*(CAS_COUNT_RD+CAS_COUNT_RD+CAS_COUNT_RD+CAS_COUNT_RD)*64.0/time
Memory Write BW [MBytes/s] 1.0E-06*(CAS_COUNT_WR+CAS_COUNT_WR+CAS_COUNT_WR+CAS_COUNT_WR)*64.0/time
--
Profiling group to measure memory bandwidth drawn by all cores of a socket.
Since this group is based on uncore events it is only possible to measure on
a per socket base. Also outputs total data volume transfered from main memory,
SSE scalar and packed double precision flop rates as well as consumed energy and 
temperature. Also reports on packed AVX 32b instructions.  Please note that the 
current flop measurements on SandyBridge are potentially wrong. So you cannot trust these counters at the moment!
