22
Table Improvements & Near Term Plan Ted & Kirsten July 08, 2005 TDWG Introduction Table performance due to improved L2 JETs Initial Trigger Table Cleanup: possible items Near Term Plan Happy Birthday !!!

Table Improvements & Near Term Plan Ted & Kirsten July 08, 2005 TDWG Introduction Table performance due to improved L2 JETs Initial Trigger Table Cleanup:

  • View
    217

  • Download
    2

Embed Size (px)

Citation preview

Table Improvements & Near Term Plan

Ted & Kirsten July 08, 2005 TDWG

Introduction Table performance due to improved L2 JETs Initial Trigger Table Cleanup: possible items Near Term Plan

Happy Birthday !!! Veronica & Kirsten (July 4th and 11th)

TDWG Organization

TDWG Heads: Kirsten and TedTrigger Table Creation and Testing Coordinator: Veronica Sorin / Zeynep Gunay

Technical Support: Trigger SPLs Vadim Rusu/Roberto Carso/Veronique Boisvert Trigger Software Coordinator: Daniel Whiteson Trigger Database: Tom Wright Trigger Representatives: Exotics Group: Vadim Rusu and Oscar Gonzalez Lopez QCD Group: Mary Convery, Gene Flanagan and Mario Campanelli Top Group: Veronique Boisvert and Ambra Gresele EWK Group: Jiyeon Han B Physics: Konstantin Anikeev and Simone Donati

Now we have the FULL GROUP, thanks to all physics conveners! will have the 1st Trigger Rep meeting soon …

Reminder: Plan for Summer

Phase I: Get system in stable condition (with improvements)

(June - early July) we are now at step II

Phase II: Trigger Table Cleanup and Optimization

(the rest of July)

Phase III: Integrate new triggers

(August - Sept)

Phase I: Get system in stable condition

--> see Veronica’s talk for details

Step I: Improving L2 JETs. DONE L2 JETs: improved the “Junk Enhanced Trigger” ROF shoulder threshold at full scale Opened up single track trigger at luminosity below 40E30 (JET_CALIB) Take data for a few stores with no other change, L2 JETs cross section improved dramatically As a result: triggers with DPS at L2 released much earlier

Step II: Switch to new XTRP firmware (early July) On-going, EOS done last night! Check track trigger rate changes due to adjacent linker veto Mt cut trigger as test trigger with prescale, map out performace CLC multiplicity trigger as test trigger with prescale, map out performance Added new Z to BBbar and MET + b-jet triggers Take data for a few stores with no other change, data for offline analysis

Will see how much improvements the new track firmware will bring (next week)…

L2 JETs improvements

JET40:before JET40: Now

JET90: NowJET90:before

Overall L2A rate is lower at beginning of run

before Now

As a result, all L2 DPS released earlier!

Zbb: beforeZbb:Now

Fully released

met+b-jet: before

met+b-jet: Now

a glitch

One comment

These are good examples to show when one trigger with high growth term is improved, many other triggers will benefit right away …

This seems obvious but it may not to everyone

Other people’s triggers affect your triggers!

Keep an eye on other people’s triggers with fast

growth term, not just yours! Helping them to improve

will help yours.

Phase II: Table cleaning and optimizing

July to August: Removing useless or obsolete triggers Add a few mature fully tested triggers Open up more triggers earlier at lower luminosity Increase 20KHz threshold up to 25KHz To pave the road for CLCM trigger and Mt triggers and integrate other test triggers

This is the first step towards table unification Some details in this talk next….

Table cleaning

Our table is 5 years old, designed > 7 years ago.Has been used to serve food for > 600 physicists (lots of themare poor, hungry and hardworking young postdoc and students).Over years, it has accumulated dust, food stain (over “food

fights?”)… it has been burned by high luminosity beam quite many times and broken into two pieces…

“High Lumi Table” == “Panic Table” It is time to do a clean up, so the young people can eat healthy/clean food…. “clean it up, use strong detergent!” --- Dan Amidei Will start with water cleaning first. some examples next…

Table cleaning Example 1: L1_TEN_TRK2 drives L2_PS0_L1_TEN_TRK2

This is in both high and low lumi tables, doing nothing exceptusing extra L1 bandwidth …

Should be at least with PS250: L1_TEN_TRK2_PS250

Running at 6KHzdoing nothing, while overall L1A is 14KHz

Table cleaning Example 2: L1_TWO_TRK10_DPHI20_DPS (CHAMP trigger)

This trigger is in both high and low lumi tables, very high L1 rate at high luminosity when DPS is released.

would like to fully open this trigger with lumi enable at around ~60E30, without DPS.

running at 4KHz,mostly fake junk!wasting L1 bandwidth

Table cleaning Example 2 (cont): L1_TWO_TRK10_DPHI20_DPS (CHAMP trigger)

At low luminosity, it behaves ok at L2. Much cleaner/pure evts Remove the DPS at L1! fully open it below ~ 60E30 Will get more useful data, without DPS Everyone wins here

Running at~30Hz at L2,mostly rejected at L3! Wasting L2bandwidth…

Table cleaning Example 3 (already done): L1_TRK10_PS250 for JET_CALIB

This is in both high and low lumi tables, very low rate though.

Fully enable (no PS) this trigger at luminosity below 40E30, to get more data for JET_CALIB. (done already). Thanks Ken who worked closely with us.

Table cleaning Example 4: L1_MB_CLC_PS10K for MINIBIAS_1HZ

This trigger is rate limited at L2: L2_RL1.3Hz No point to run L1 at higher rate Will reduce the rate at L1 by a factor of ~10

Table cleaning Example 5: L1_TOF_14_MIP_&_CLC_PS10K for MINIBIAS_1HZ

This trigger is currently rate limited at L2: L2_RL0.5Hz Will increase the rate at L2 for Niccolo Moggi’s multiplicity trigger soon

--> he told me about this trigger three years ago when I first met him! Still, no need to run L1 at much higher rate than L2 Could reduce the rate at L1 by a factor of ~10

Table cleaning Example 6: L1_TWO_TRK2_PS200

This backup trigger is only in low lumi table It drives L2_PS1K_L1_TWO_TRK2 and L2_TWO_TRK2_D100_PS20 Ideally, would like to replace L1_TWO_TRK2_PS200 with PS4K But only four L1 bits can be prescaled by more than 255 May change the PS to 250 instead

Table cleaning Example 7: “unused?” datasets

Anyone else use these datasets? PLUG_DIJET

JET_20_CENTRAL

All DIJET_MASS (TWO_JET_xx_Mxxx)

QCD group is no longer using them (recent email from Mary):Hi Ted,

This is to confirm that the QCD group is no longer using the PLUG_DIJET, JET_20_CENTRAL, TWO_JET_xx_Mxxx (dijet mass) triggers and they can be removed from the physics table.

Regards

Mary

Table cleaning Example 8: save some L1 trigger bits: 5 possible

candidates

L1_EM8_&_MET15 replaced by L1_EM8 no need to use another L1 bit since L1_EM8 is not prescaled L2 requirement on MET15

L1_EM8_&_CMU1.5_PT1.5 replaced by L1_EM8 ditto Pulsar CPU is much faster than Alpha, no performance penalty

L1_EM8_&_CMX1.5_PT2 replaced by L1_EM8 ditto

L1_JET10_&_SUMET90 by L1_JET10 ditto

J/Psi --> ee running off by existing 4 GeV electron + track at L1 free up one L1 bit and two DIRAC bits keep L2/L3 the same

L1_EM8_&_MET15 replaced by L1_EM8

What’s going to happen in the coming two weeks

New trigger tables running with new track firmware (middle next week) Then we will implement next table with cleanup and with

add backup trigger for MET + b-jet: L1_MET15_JET5_TWO_TRK2 remove the old Z to BBbar trigger (once the new trigger data is ok) add a backup trigger for L1_MET15 itself anything else which is FULLY ready to go in

By next meeting, the new cleaned up table should be running as default table.Will then raise the L1A 20KHz to 25 KHz

That is the time we will push with FULL SPEED to integrate all long due test triggers (some of them have been waiting for one year! This is very sad!)

Meanwhile, Kirsten will push hard on getting L3 code ready for all tau triggersas well as multiplicity trigger (all LONG due), so that we will be able to integratethem as soon as we finish the table cleaning.

Phase III: Integrate new triggers

August to Sept:

Finish integrating all mature fully tested triggers If you have triggers which are fully ready, they will be integrated! If you have triggers need more tests, let’s test them soon All new/test trigger requests should be done by Trigger Reps.from now on Trigger Reps: please finalize your list/plan at next two meetings Will have trigger rep meeting starting next week (Wed 2-3pm?) The meeting will be dedicated for integrating new triggers

Expand L2 bits from 128 to 192 (“Non-beam test successful” two days ago) Daniel will give a status report soon Take full advantage of new EVB upgrade, Pulsar Phase II L2 system and SVT upgrade (partially installed by then) We should be able to handle 150E30 luminosity with all the improvements The table should be stable by then for this luminosity range

The rest of talks Ming: Study on L1_MET trigger without ROF

Ming, like others from exotic/QCD, are our heroes like others, he only has time to work on trigger issues on weekend!

Simone/Konstantin: B Physics trigger plan Two best trigger reps from B group we can think of!

Markus: EVB upgrade status report Markus, just like all other people working on hardware improvements,

(esp. Silicon SPLs!), they’re heroes working on hardware upgrade, who fight the war against deadtime at the front lines

Charles: Everything you want to know about XMON he is the one who makes sure we don’t drive blind !