SOFTWARE
Expert phone numbers
Threshold changes
Set 6 production triggers as of sunday:
mb,
zb,
bht2-mb @ 5GeV
bjp1-mb @ 7.8GeV
eht2-mb @ 3.5GeV,
ejp1-mb @ 8.0GeV
Tier1 file changes
Sampled Luminosity = [(# of mb evts)(25)(Run Control mb Prescale value)]/(25 x 10**6) (units of nb-1) An example calculation, for Run 7080002 is: - # mb evts = 2,074 - Run Control mb Prescale value = 4,064 Sampled luminosity = [(2074)(25)(40640]/(25 x 10**6) = 8.43 nb-1
mip_adc[i] = avgMip[etaBin]*avgSlope[etaBin]/slope[i] scale set to 56 GeV max gain[i] = 0.261*(1.+0.056*eta[i]*eta[i])/(sin(theta[i])*mip_adc[i]) so in that case your XXX[etaBin] is XXX[etaBin] = 0.261*(1.+0.056*eta2)/(sin(theta)*avgMip[etaBin]) I'll post the numbers in this format a little later, but you can also calculate it yourself using avgMip ADC numbers from http://www.star.bnl.gov/protected/spin/kocolosk/barrel_calibration/ 2006/2006_mip.txt
Also, http://www.star.bnl.gov/webdatanfs/pub/overall.html
is up-to-date as per indicating where FastOflfine is (and this for the past few hours). I will update as well http://www.star.bnl.gov/STAR/comp/prod/fast-offline.html
ppEmcCheck contained 4 triggers:
BHT1 & Minbias
BJP1 & Minbias
EHT1 & Minbias
EJP1 & Minbias
ppEmcBackgroundCheck contained 5 triggers:
BHT2 & BY sync (no minbias)
BJP2 & BY
EHT2 & BY
EJP2 & BY
minbias (prescaled @ ~10hz)
On Thu, 16 Mar 2006, chris perkins wrote: > > hi - i made a small change to trgStructures.h in CVS. I added the > > following defines: > > > > #define L2RESULTS_OFFSET_TRG 0 > > #define L2RESULTS_OFFSET_EMC_CHECK 1 > > #define L2RESULTS_OFFSET_JPSI 2 > > #define L2RESULTS_OFFSET_UPS 8 > > #define L2RESULTS_OFFSET_DIJET 14 > > #define L2RESULTS_OFFSET_EMC_PED 19 > > #define L2RESULTS_OFFSET_PIG 20 > > > > NO modifications were made to the structure itself so I left the version > > number at 0x22
Gerard Subsys DC-chan Cr-ID Device(for ref. only) ------ ------- ----- ------ BTOW 0 0x12 EC6W 1 0x11 EC4W 2 0x10 EC2W 3 0x1e EC30W 4 0x1d EC28W 5 0x1c EC26W 6 0x1b EC24W 7 0x1a EC22W 8 0x19 EC20W 9 0x18 EC18W 10 0x17 EC14W 11 0x16 EC13W 12 0x15 EC12W 13 0x14 EC10W 14 0x13 EC7W 15 0x01 EC2E 16 0x0f EC30E 17 0x0e EC28E 18 0x0d EC26E 19 0x0c EC24E 20 0x0b EC22E 21 0x0a EC20E 22 0x09 EC18E 23 0x08 EC14E 24 0x07 EC13E 25 0x06 EC12E 26 0x05 EC10E 27 0x04 EC7E 28 0x03 EC6E 29 0x02 EC4E ETOW 0 0x01 crate_1 1 0x02 crate_2 2 0x03 crate_3 3 0x04 crate_4 4 0x05 crate_5 5 0x06 crate_6 ESMD 0 0x40 12S1 1 0x41 12S2 2 0x42 12S3 3 0x43 12P1 4 0x44 1S1 5 0x45 1S2 6 0x46 1S3 7 0x47 1P1 8 0x48 2S1 9 0x49 2S2 10 0x4a 2S3 11 0x4b 2P1 12 0x4c 3S1 13 0x4d 3S2 14 0x4e 3S3 15 0x4f 3P1 16 0x50 4S1 17 0x51 4S2 18 0x52 4S3 19 0x53 4P1 20 0x54 5S1 21 0x55 5S2 22 0x56 5S3 23 0x57 5P1 24 0x58 6S1 25 0x59 6S2 26 0x5a 6S3 27 0x5b 6P1 28 0x5c 7S1 29 0x5d 7S2 30 0x5e 7S3 31 0x5f 7P1 32 0x60 8S1 33 0x61 8S2 34 0x62 8S3 35 0x63 8P1 36 0x64 9S1 37 0x65 9S2 38 0x66 9S3 39 0x67 9P1 40 0x68 10S1 41 0x69 10S2 42 0x6a 10S3 43 0x6b 10P1 44 0x6c 11S1 45 0x6d 11S2 46 0x6e 11S3 47 0x6f 11P1
Hi again, Carl, Just to pick up on your postscript, regarding how we get from where we're running now to desired jet trigger rates: 1) We're currently running bemc-jp0-etot-mb-L2 and eemc-jp0-etot-mb-L2 with prescale factors of 10.0. Combined, they give 15 Hz of L0 rate at 220 kHz BBC AND rate. Currently, only half the endcap is giving jet triggers with the tier1 file in use. So allowing for complete endcap and 325 kHz goal BBC AND rate, this gives anticipated L0 rate of dijet input triggers = 25 Hz, vs. desired 10 Hz. I think thresholds (jp0 at 4 GeV, Etot at 14 GeV) are already as high as we want to run them, so I suggest accomplishing the desired L0 rate reduction by a prescale by factor of 2. That is, for longitudinal running, prescale these triggers by 20, for transverse running prescale by 2. 2) For both barrel and endcap triggers, L2 is currently passing 25% of L0 input events. 10% are accepted at random. As Carl suggests, we should reduce this to 1% accepted at random. Another 3% are accepted as monojets (threshold at 8.0 GeV, equivalent to JP1 L0 thresholds), but ~half of these overlap with JP1 events. So reducing random rates and taking into account the overlap with JP1, L2 accept rate will be reduced to 15%. To get to desired 10%, we can raise current L2 thresholds from 3.1 GeV jet 1 and 3.0 GeV jet 2 to 3.6 GeV jet 1 and 3.5 GeV jet 2. With the above changes, I think we'll be very close to desired rates for dijet triggers. Steve
int l2_alg_emul_l0(TrgDataType *evt, int i0, int i1, int i2, int i3, int i4)
{
static unsigned short onbits;
static unsigned short offbits;
static unsigned short lastDSM;
onbits = i0;
offbits = i1;
// Does this need to be swapped?
lastDSM = evt->EvtDesc.DSMInput;
if(lastDSM & onbits != onbits) return 0;
if(~lastDSM & offbits != offbits) return 0;
if(i4 <= 1) return 1;
onbits = i2;
offbits = i3;
if(lastDSM & onbits != onbits) return 0;
if(~lastDSM & offbits != offbits) return 0;
return 1;
}
>> 2. Do I have to swap the value in : evt->EvtDesc.DSMInput, before
>> using it? I know it comes from the big-endian L1 cpu, but I don't know
>> if you swap the event descriptor data as a service for the L2 algorithms.
yes, the event descriptor (and other relevant pieces) are swapped for the
L2 algorithms. your code looks fine the way it is. i'll add it to L2
tonight and be around after the 10am meeting tuesday.
chris
right - boards 4,6,10,11,12 (sequence number, S, in jeff's previous email) are read out periodically throughout a run. boards 3,5,7,8,9 integrate over the entire run and are only written at the end of the run (these were the problematic ones). so you would need to look for scaler files from these boards to know that the scalers were written properly. i agree, tonko's log message is the simplest indication that all scaler files have been written chris On Thu, 6 Apr 2006, Jeff Landgraf wrote: >> Be careful about this conclusion. The filename convention for the >> scaler files is: >> >> runNNNNNNN_S_TTTTTTTT.sca >> >> where NNNNNN is run number, S a sequence number and TTTTTT a time.