Skip to content
Snippets Groups Projects
Commit 31f4613c authored by Eric Kooistra's avatar Eric Kooistra
Browse files

Remove description of TAsk_1275 branch efforts, because it finished. Add how... 

Remove description of TAsk_1275 branch efforts, because it finished. Add how to do rebuild from SVN.
parent 5a2996d4
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
applications/apertif/doc/apertif_fpga_firmware_overview_erko.txt
+ 10
− 288
View file @ 31f4613c
......@@ -76,295 +76,17 @@ $RADIOHDL/applications/apertif/designs/
python ~/svnroot/UniBoard_FP7/RadioHDL/trunk/tools/oneclick/base/quartus_config.py -t unb1
run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank; run_qcomp unb1 apertif_unb1_fn_beamformer_trans --seed=1;run_rbf unb1 apertif_unb1_fn_beamformer_trans; run_qcomp unb1 apertif_unb1_correlator_full --seed=1;run_rbf unb1 apertif_unb1_correlator_full
ls
ll
7) Summary for clean build from SVN
7) SVN branching
mkdir branches
svn add branches/
commit -m "Start branching." branches/
svn info .
# The ^ is the root of the repository in SVN
svn ls ^/RadioHDL
# Define simple editor for SVN commit message editing
export SVN_EDITOR=vim
echo $SVN_EDITOR
# Create branch from revison 16579 of the trunk and call it Task_<Redmine issue number>
# Start the commit message with Task #<Redmine issue number>: to provide link for Redmine.
# Optinal use r<SVN revision number> in the commit message
svn cp ^/RadioHDL/trunk@16579 ^/RadioHDL/branches/Task_1275
cd branches
svn up Task_1275
rbf:17728 resynthesis of Hajee image 16579 on Task_1275 branch works OK.
cd ~/svnroot/UniBoard_FP7/RadioHDL/branches/Task_1275
diff -r . ../../trunk_hajee # shows diff inside files
diff -rq . ../../trunk_hajee # quiet, shows only diff files names
cntr-alt-F1
killall -9 gvimdiff
cntr-alt-F7
cd ../branches/Task_1275/
svn log -v -r 16588:HEAD ^/RadioHDL/trunk | more
svn merge -c 16592 ^/RadioHDL/trunk # merge changes
svn merge -r 16596:16658 ^/RadioHDL/trunk # merge a range of revisions
svn diff --diff-cmd diff # toont Merged /RadioHDL/trunk:r16580-16592
8) To do:
- ppsh.vhd decrease c_pipeline_output : NATURAL := 100;
9) Task_1275
9a) Possible suspects: WPFB, ADUH, node, PPSH, DP
r 17739 : Merged /RadioHDL/trunk:r16580-16658 and compaan stuff
r ..... : Merged technology, base, io (so including ppsh, aduh)
> cd technology
> svn merge ^/RadioHDL/trunk/libraries/technology
> cd base
> svn merge ^/RadioHDL/trunk/libraries/base
> cd io
> svn merge ^/RadioHDL/trunk/libraries/io
svn up -r # # update to -r
svn up # update to HEAD
svn status -q # status van working copy tov laatste svn up -r #
svn status -u -q # status van working copy tov HEAD
9b) Delete and recreate branch
Merge per function instead of in sequence, because not every revision yields a stable image. Start with only
merging the technology library.
svn rm ^/RadioHDL/branches/Task_1275 # remove branch directly in SVN, so no need to do svn commit afterwards
cd RadioHDL/branches
rm -rf Task_1275/ # remove the working copy with all its local dir and files (e.g. build)
svn cp ^/RadioHDL/trunk@16579 ^/RadioHDL/branches/Task_1275/ # recreate the branch
31-10-2017
# Merge technology/ from HEAD to Task_1275 working copy
cd Task_1275/
svn merge ^/RadioHDL/trunk/libraries/technology libraries/technology/
generate IP, run_sopc, run_qcomp --> rbf 17744 fails on HW, the mesh BSN monitors at the FN show timeout.
Can be an issue with the BSN source, because often one or more BN show an offset in BSN compared to the rest
or it can be an issue in tr_nonbonded, because at the FN the top 16 bits of the Rx data are often zero,
except for aaaaaaab and 3c3c3c3c, however these words also are sometimes corrupted, possibly this could be
due to a corrupt data path with missing sop? The PPSH seems ok, although the BN do not always all have the
same toggle level (which may be ok, because only the toggle edges are used and these are aligned).
(ssh lcu-rtc -X "python $UPE/peripherals/util_tr_nonbonded.py --unb 4:7 --fn 0:3 -n 6 -s MESH")
I think the BSN offset is the cause, but why does it start about 274363 different than on the other BN?
Offset 302114 also occurs and then about the same on two BN. The offset can also occur on three BN on
different Uniboards and the offset BSN are still nearly similar. The offset can appear as nof_sop in BSN monitor.
1-11-2017
# Revert Task_1275 and remove any dependencies on trunk:
- temporarily rename trunk into trunky
- corrected user_components.ipx to use only code from $RADIOHDL
- corrected applications/aartfaac/libraries/rsp_terminal/hdllib.cfg to use $RADIOHDL instead of $SVN/RadioHDL/trunk
(minor issue that is already fixed on trunk)
generate IP, run_sopc, run_qcomp --> rbf 17746 works on HW (cmd 72 passed)
# Retry merge technology/ from HEAD to Task_1275 working copy
- svn status -q
- svn revert -R .
- cd ..
- rm -rf Task_1275/
- svn up Task_1275
- cd Task_1275/
- svn merge ^/RadioHDL/trunk/libraries/technology libraries/technology/
- svn merge ^/RadioHDL/trunk/applications/aartfaac/libraries/rsp_terminal/hdllib.cfg applications/aartfaac/libraries/rsp_terminal/hdllib.cfg
- svn diff applications/aartfaac/libraries/rsp_terminal/hdllib.cfg
> cd $RADIOHDL/libraries/technology/ip_stratixiv
> ./generate-all-ip.sh
> python $RADIOHDL/tools/oneclick/base/modelsim_config.py -t unb1
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17747 works on HW (cmd 72 OKE)
--> Verify with Arts-0 using dish 7 (FB 17747.rbf, BF 17128.rbf) and Arts-SC1 16748.rbf OKE
- svn commit .
common_pkg r16850 | schuur | 2017-02-27 13:26:10 +0100 (Mon, 27 Feb 2017) --> ceil_log2(0) := 0
6-11-2017
# Merge aduh/, common/ from HEAD to Task_1275 working copy
- svn merge ^/RadioHDL/trunk/libraries/base/common libraries/base/common
- svn merge ^/RadioHDL/trunk/libraries/io/aduh libraries/io/aduh
> python $RADIOHDL/tools/oneclick/base/modelsim_config.py -t unb1
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> python $RADIOHDL/tools/oneclick/base/modelsim_regression_test_vhdl.py -t unb1 -v 5 -l aduh -r # OKE
--> strangely when trunk is there then bsn source start is sometimes offset for some BN therefore used mv trunk/build trunk/buildy as trial
--> rbf 17747_buildy detects pulsar
- svn commit .
8-11-2017
# Merge base/, io/ from HEAD to Task_1275 working copy, comment wpfb_unit_dev and tb in hdllib.cfg
--> rbf 17767
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
! ssh lcu-rt7 -X "python $UPE/peripherals/util_system_info.py --unb 0:7 --fn 0:3 --bn 0:3 -n 2" # read system info
> ssh lcu-rtc -X "python $UPE/peripherals/util_bsn_monitor.py --unb 0:7 --fn 0:3 -n 1 -r 0" # verify FN bsn align
! ssh lcu-rt7 -X "python $UPE/peripherals/util_bsn_source.py --unb 0:7 --bn 0:3 -n 7" # read current BSN @ sync, all equal
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_bsn_scheduler.py --unb 0:7 --bn 0:3 --cmd 0" # read current BSN immediate, last - first ~< 300 (1.5 us)
! ssh lcu-rt7 -X "python $UPE/peripherals/util_aduh_quad.py --unb 0:7 --bn 0:3 --sp 0:3 -n 0" # verify ADUH lock
> ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:7 --bn 0:3 --sp 0:3 --cmd 40 -v 5" # read ADUH mean sum: ~< +-2
> ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:7 --bn 0:3 --sp 0:3 --cmd 41 -v 5" # read ADUH power: sum effective WG ampl ~15
> ssh lcu-rt7 -X "python $UPE/peripherals/util_aduh_monitor.py --unb 3,7 --bn 2 --sp 0 -n 3 -v 5" # read ADUH buffer X,Y pol SP 56
> ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 3,7 --bn 2 --sp 0 --cmd 43 -v 5" # plot ADUH buffer X,Y pol samples for SP 56
> ssh lcu-rt7 -X "python $UPE/peripherals/util_ppsh.py --unb 0:7 --bn 0:3 -n 8" # verify PPS stable
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:7 --bn 0:3 --sp 0:3 --cmd 31 -v 4" # verify F_sub FIR coefficients
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:7 --bn 0:3 --cmd 5 -v 4" # verify R_sub
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:7 --fn 0:3 --cmd 53 -v 4 --scheme arts_sc1" # verify R_beam
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:7 --bn 0:3 --sp 0:3 --cmd 10 -v 5" # plot SST
> ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:3 --bn 0:3 --sp 0:3 --cmd 11 -v 5" # log SST
> ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:3 --bn 0:3 --sp 0:3 --cmd 12 --pol 0 --globalsp 0:63 -v 5" # log SST maximum and effictive WG ampl
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 0:3 --bn 0:3 --fn 0:3 --sp 0:3 --pol 0 --globalsp 0:39 --weight 32767,0 --cmd 72 -v 3" # verify X-pol BST
! ssh lcu-rt7 -X "python $UPE/peripherals/pi_apertif_system.py --unb 4:7 --bn 0:3 --fn 0:3 --sp 0:3 --pol 1 --globalsp 0:39 --weight 32767,0 --cmd 72 -v 3" # verify Y-pol BST
--> all oke
--> SC1 Pulsar detected
9-11-2017
# Merge tools/, boards/, applications/apertif from HEAD to Task_1275 working copy
- svn merge ^/RadioHDL/trunk/tools tools
- svn merge ^/RadioHDL/trunk/boards boards
- svn merge ^/RadioHDL/trunk/applications/apertif applications/apertif
- svn revert applications/apertif/designs/apertif_unb1_bn_filterbank/src/vhdl/node_apertif_unb1_bn_filterbank.vhd
- temporarily define function func_wpfb_set_nof_block_per_sync() in tb_apertif_unb1_bn_filterbank_stimuli_wg.vhd
> python $RADIOHDL/tools/oneclick/base/modelsim_config.py -t unb1
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17772
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
--> SC1 Pulsar detected
10-11-2017
- On HEAD
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17777
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
# On branch merge fft/ from HEAD to Task_1275 working copy
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17772_fft
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
--> SC1 Pulsar detected
13-11-2017
- svn merge ^/RadioHDL/trunk/libraries/dsp/filter libraries/dsp/filter/
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17781
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15 --opt noflash
--> First time one BN had wrong BSN (delta 233681) therefore restarted main.py an then all BSN were equal
--> SC1 Pulsar detected
14-11-2017
- svn merge ^/RadioHDL/trunk/libraries/dsp/wpfb libraries/dsp/wpfb
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17782
--> First time UNB-4, BN-2: read_current_bsn = 1180213750618702 --> diff 304772
--> SC1 Pulsar detected
15-11-2017
- svn merge ^/RadioHDL/trunk/applications applications
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17783
>>> pol 0 : 1180266288357487 - 1180266288652369 = -294882
>>> pol 1 : 1180266288357405 - 1180266288652388 = -294983
285351
266642
257682
Try power cycle dish 7 uniboards, does not help
279232
275006
Try main.py pol 0 # BSN oke
Try main.py pol 1 # BSN oke
Try main.py pol 0,1 # now BSN oke, so continue
--> SC1 Pulsar detected
- rm -rf build/unb1/quartus
- use wpbf_unit_dev in node_apertif_unb1_bn_filterbank.bak_17783_dev, but with g_big_endian_wb_in=false
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17783_dev
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
--> SC1 Pulsar not detected!
16-11-2017
- cp applications/apertif/designs/apertif_unb1_bn_filterbank/src/vhdl/node_apertif_unb1_bn_filterbank.vhd applications/apertif/designs/apertif_unb1_bn_filterbank/src/vhdl/node_apertif_unb1_bn_filterbank.bak_17783_dev
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
-rbf 17783_astrunk
--> SC1 Pulsar not detected! (as expected because branch is now same as trunk)
- rm -rf build/unb1/quartus
- use wpbf_unit_dev with old blockgen and node_apertif_unb1_bn_filterbank.bak_17783_dev (so with g_big_endian_wb_in=false)
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17783_blkgen
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
--> SC1 Pulsar detected!
17-11-2017
- rm -rf build/unb1/quartus
- use wpbf_unit_dev with new blockgen and node_apertif_unb1_bn_filterbank.bak_17783_dev (so with g_big_endian_wb_in=false)
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
--> rbf 17783_newblkgen
> python main.py --app arts_sc1 --tel 7 --pol 0,1 --unb 8:15
--> SC1 Pulsar not detected --> so it must be due to the local BSN that is not recognized by SC1 pipeline
20-11-2017
- Created dp_bsn_restore_global.vhd and use it in dp_block_gen_valid_arr
- rm -rf build/unb1
> python $RADIOHDL/tools/oneclick/base/modelsim_config.py -t unb1
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_sopc unb1 apertif_unb1_fn_beamformer_trans
> tb_apertif_unb1_bn_filterbank_stimuli_wg --> ok
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank; run_qcomp unb1 apertif_unb1_fn_beamformer_trans --seed=1;run_rbf unb1 apertif_unb1_fn_beamformer_trans
--> fb rbf 17790
--> bf rbf 17790
--> SC1 Pulsar fb 17790 + bf 17128 --> Pulsar detected!
--> SC1 Pulsar fb 17790 + bf 17790 --> Pulsar detected!
Conclusion: #1275 is now fixed
# Optionally recreate the IP
> cd $RADIOHDL/libraries/technology/ip_stratixiv
> ./generate-all-ip.sh
#1350 BSN start bug
- sometimes BSN start not the same on all BN, one or a few BN then have an offset BN
- for FB 17352 it occurs about once per 500 BN, for FB 17783 it occurs about once per 10 BN, so it is FPGA
syhthesis dependend
- ssh lcu-rtc -X "python $UPE/peripherals/util_ppsh.py --unb 0:7 --bn 0:3 -n 8" PPS stable is ok for all BN,
so it does not seem to be an PPS input capture issue
- ssh lcu-rtc -X "python $UPE/peripherals/pi_bsn_scheduler.py --unb 0:7 --bn 0:3 --cmd 0" shows the offset.
The offset can be e.g. (note 800000 = 1.024 s)
285351
266642
257682
279232
275006
300636
- the offset is nearly the same for all BN, even if they are on different subracks, so there is some central
relation, because how could all those offst BN have (nearly) the same offset
- power cycle dish uniboards, does not help
- ssh lcu-rtc -X "python $UPE/peripherals/util_bsn_source.py --unb 0:7 --bn 0:3 -n 7" reads current BSN @ sync
and these are all equal.
- ssh lcu-rtc -X "python $UPE/peripherals/util_bsn_source.py --unb 0:7 --bn 0:3 -n 2" stop BSN source
ssh lcu-rtc -X "python $UPE/peripherals/util_bsn_source.py --unb 0:7 --bn 0:3 -n 11 -r 0" init BSN
ssh lcu-rtc -X "python $UPE/peripherals/util_bsn_source.py --unb 0:7 --bn 0:3 -n 4" restart BSN source on PPS also
keeps on failing to restart all BN at PPS with same BSN
- is the bsn_init value written correctly? this cannot be readback.
- check synthesis report on 'latch', 'sensitivity'
# Optionally recreate the build directory and a rbf
> rm -rf build/unb1/quartus
> python $RADIOHDL/tools/oneclick/base/modelsim_config.py -t unb1
> python $RADIOHDL/tools/oneclick/base/quartus_config.py -t unb1
> run_sopc unb1 apertif_unb1_bn_filterbank
> run_qcomp unb1 apertif_unb1_bn_filterbank --seed=1;run_rbf unb1 apertif_unb1_bn_filterbank
ssh lcu-rtc -X "python $UPE/peripherals/util_ppsh.py --unb 0 --bn 0 -n 4 -v 0;\
python $UPE/peripherals/pi_bsn_scheduler.py --unb 0 --bn 0 --cmd 0;\
python $UPE/peripherals/util_ppsh.py --unb 0 --bn 0 -n 4 -v 0;\
python $UPE/peripherals/pi_bsn_scheduler.py --unb 0 --bn 0 --cmd 0"
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment