mirror of
https://github.com/ClusterCockpit/cc-metric-collector.git
synced 2024-11-10 12:37:25 +01:00
53 lines
2.0 KiB
Plaintext
53 lines
2.0 KiB
Plaintext
|
SHORT QPI Link Layer data
|
||
|
|
||
|
EVENTSET
|
||
|
FIXC0 INSTR_RETIRED_ANY
|
||
|
FIXC1 CPU_CLK_UNHALTED_CORE
|
||
|
FIXC2 CPU_CLK_UNHALTED_REF
|
||
|
SBOX0C0 DIRECT2CORE_SUCCESS_RBT_HIT
|
||
|
SBOX1C0 DIRECT2CORE_SUCCESS_RBT_HIT
|
||
|
SBOX2C0 DIRECT2CORE_SUCCESS_RBT_HIT
|
||
|
SBOX0C1 TXL_FLITS_G0_DATA
|
||
|
SBOX1C1 TXL_FLITS_G0_DATA
|
||
|
SBOX2C1 TXL_FLITS_G0_DATA
|
||
|
SBOX0C2 TXL_FLITS_G0_NON_DATA
|
||
|
SBOX1C2 TXL_FLITS_G0_NON_DATA
|
||
|
SBOX2C2 TXL_FLITS_G0_NON_DATA
|
||
|
SBOX0C3 SBOX_CLOCKTICKS
|
||
|
SBOX1C3 SBOX_CLOCKTICKS
|
||
|
SBOX2C3 SBOX_CLOCKTICKS
|
||
|
SBOX0FIX QPI_RATE
|
||
|
SBOX1FIX QPI_RATE
|
||
|
SBOX2FIX QPI_RATE
|
||
|
|
||
|
METRICS
|
||
|
Runtime (RDTSC) [s] time
|
||
|
Runtime unhalted [s] FIXC1*inverseClock
|
||
|
Clock [MHz] 1.E-06*(FIXC1/FIXC2)/inverseClock
|
||
|
CPI FIXC1/FIXC0
|
||
|
QPI Speed Link 0 [GT/s] ((SBOX0C3)/time)*inverseClock*(8/1000)
|
||
|
QPI Speed Link 1 [GT/s] ((SBOX1C3)/time)*inverseClock*(8/1000)
|
||
|
QPI Speed Link 2 [GT/s] ((SBOX2C3)/time)*inverseClock*(8/1000)
|
||
|
QPI Rate Link 0 [GT/s] 1.E-09*SBOX0FIX
|
||
|
QPI Rate Link 1 [GT/s] 1.E-09*SBOX1FIX
|
||
|
QPI Rate Link 2 [GT/s] 1.E-09*SBOX2FIX
|
||
|
data from QPI to LLC [MByte] 1.E-06*(SBOX0C0+SBOX1C0+SBOX2C0)*8
|
||
|
QPI data volume [MByte] 1.E-06*(SBOX0C1+SBOX1C1+SBOX2C1)*8
|
||
|
QPI data bandwidth [MByte/s] 1.E-06*(SBOX0C1+SBOX1C1+SBOX2C1)*8/time
|
||
|
QPI link volume [MByte] 1.E-06*(SBOX0C1+SBOX1C1+SBOX2C1+SBOX0C2+SBOX1C2+SBOX2C2)*8
|
||
|
QPI link bandwidth [MByte/s] 1.E-06*(SBOX0C1+SBOX1C1+SBOX2C1+SBOX0C2+SBOX1C2+SBOX2C2)*8/time
|
||
|
|
||
|
LONG
|
||
|
Formulas:
|
||
|
QPI Speed Link 0/1/2 [GT/s] = ((SBOX_CLOCKTICKS)/time)*clock*(8/1000)
|
||
|
QPI Rate Link 0/1/2 [GT/s] = 1.E-09*(QPI_RATE)
|
||
|
data from QPI to LLC [MByte] = 1.E-06*(sum(DIRECT2CORE_SUCCESS_RBT_HIT)*64)
|
||
|
QPI data volume [MByte] = 1.E-06*(sum(TXL_FLITS_G0_DATA)*8)
|
||
|
QPI data bandwidth [MByte/s] = 1.E-06*(sum(TXL_FLITS_G0_DATA)*8)/runtime
|
||
|
QPI link volume [MByte] = 1.E-06*((sum(TXL_FLITS_G0_DATA)+sum(TXL_FLITS_G0_NON_DATA))*8)
|
||
|
QPI link bandwidth [MByte/s] = 1.E-06*((sum(TXL_FLITS_G0_DATA)+sum(TXL_FLITS_G0_NON_DATA))*8)/runtime
|
||
|
--
|
||
|
The Intel QPI Link Layer is responsible for packetizing requests from the caching agent (CBOXes)
|
||
|
on the way out to the system interface.
|
||
|
|