-*- ftp-quik -*-

Top-level index file with pointers to internal notes and other relevant information
related to GHC GUM/SMP/Eden development.

Latest stable version (used for IFL13 paper results):
/home/hwloidl/www/hackspace/ghc-6.12-eden-gumsmp-MSA-IFL13


Jost's latest Eden modules etc
/u1/others/berthold/eden7/nbody-iterationcode


HaWos Notes:
/u1/staff/hwloidl/GPH/GUM6-TestPrograms/parfib

HaWos repo:
/u1/staff/hwloidl/INPLACE/ghc-6.12-eden-gum-msa62-fresh

HaWos aliases for PVM configuration:
/u1/staff/hwloidl/bin/beotools/beo-utils.sh

Clone of the HEAD of the Eden repo
/u1/staff/hwloidl/INPLACE/ghc-eden-HEAD


Jost's clone of the Eden repo:
/u1/others/berthold/eden7/ghc-eden-HEADMR


Malaks NOTES and test programs:
/home/msa62/GPH/GumSmp-TestPrograms/NOTES

Malak's basic GUMSMP repo:
/home/msa62/GumSmpWV/ghc-6.12-eden-gum

Malak's repo with own extensions:
/home/msa62/GumSmpImp/ghc-6.12-eden-gum

Malak's repo with own extensions (LATEST VERSION)
/home/msa62/INPLACE/GumSmpMSA4/ghc-6.12-eden-gum


Malak's ToDo list
/home/msa62/GPH/GumSmp-TestPrograms/TODO

Malak's Current issues list (post IFL13)
  /scpc:hwloidl@lxhwloidl.macs.hw.ac.uk:/home/msa62/GPH/GumSmp-TestPrograms:

Measurements for the IFL'13 paper:
/home/msa62/GPH/GumSmp-TestPrograms/parfib/GumSmp_pre_results/EXP1
/home/msa62/GPH/GumSmp-TestPrograms/parfib/GumSmp_pre_results/EXP2


Evgenij's NOTES, ISSUES, and TODO:
/u1/pg/eb96/public_html/hackspace/NOTES
/u1/pg/eb96/public_html/hackspace/TODO
/u1/pg/eb96/public_html/hackspace/ISSUES
 current issues/next steps:
 - P2: spark colocation (update TSO with ancestry info) -> see. ISSUES => branch: colocation
   - measure performance on up to 256 cores (bwlf) and perhaps a setup of 4x64 core servers
   - TODO: use more benchmark applications: use parEnc with coins, worpitzky, minimax, maze, soda9 etc.
   - TODO: use different matching functions (shortest match, limit number of compared symbols; limit number of considered sparks etc)
 - P3: using thread and spark pool size to detect generators and temporarily switch to work pushing  
 ? MONITORING+TUNING MEMO
 v P2: introduce sparkInfo_t
 v GUM vs GUMSMP vs GHCSMP baseline comparison 
 v APPLICATION CALSSIFICATION
 v MERGE to get the baseline GUM for app. classification -> ATPS15
 - #sent fetches > #received resumes (sometimes; for qp<n> with n > 2)
(v) double free corrution during shutdown/cleanup (oftern; for qp< n>2 >)
 (- #sent acks > #received schedules  (on PE1 for qp< n>2 >): solved, some acks are sent in respons to resume messages with GAs)

Evgenij's repos (currently inplace builds):
/u1/pg/eb96/SRCS/ghc-6.12-eden-gumsmp-NEWEST/(GUMSMP with per-TSO ParTicky profiling and branches for fishing history and spark colocation)
/u1/pg/eb96/SRCS/ghc-6.12-eden-gum-msa62/    (cloned from /u1/staff/hwloidl/INPLACE/ghc-6.12-eden-gum-msa62-fresh)
/u1/pg/eb96/SRCS/GUM-6.12-032013-WV/         (same clone, using /u1/staff/hwloidl/BUILDS/pvm3/x86_64-unknown-linux)

PAR_TICKY for SMP
/home/eb96/SRCS/ghc-6.12-eden-gumsmp-NEWEST/

Latest commits :
 commit TODO    (ParTicky port to GHCSMP and GUMSMP finalised and tested)
 commit 465d5da (ParTicky for GUMSMP and GHCSMP (not much tested yet))
Last commit before the pull:
 commit 2248868247b65f3b04460131856a9c0da9cea79c
(commit 65aa368c1d6c6926faa24340d19573c8e58b5544)

git
Jost's Eden clone with separated serialisation
/u1/others/berthold/eden7/BLD-HEADclone/rts
../../ghc-eden-HEADclone/


Example of compiling:
/u1/staff/hwloidl/INPLACE/ghc-6.12-eden-gum-msa62/inplace/bin/ghc-stage1 -fforce-recomp -fglasgow-exts -cpp    -parpvm -debug   -osuf="pp_debug_o" -hisuf="pp_debug_hi"  -i/u1/staff/hwloidl/GPH/par-modules/0 -i/u1/staff/hwloidl/GPH/par-modules/1 --make  -o dnc_pp_debug dnc.hs
./dnc_pp_debug 33 11 +RTS -qPg -qp3 -qD3 -DS    -G1 -T1 -qG1 -qT0 -qf500 -qqf5 -qF1 -qQ1024 -K64M -qW9  -sstderr