Author |
Topic: LBB runs 7 times slower than LB on my GBRL-Sender (Read 1554 times) |
|
xtal
New Member
member is offline


Gender: 
Posts: 27
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #15 on: Dec 22nd, 2017, 5:49pm » |
|
I modified the test pgm, added back the print window, and I only have to click on the window for the program to speed up?
I was trying to verify the 10Ms , and LB setting Bucketdelay using run program on LB and LBB 10-15 I get 15 ms delay 16 I get 25 ms delay 17-? I get 31 ms
LBB 10-12 I get 15 ms delay [must click print window 2 speedup on LBB but not on LB]] 13 - ? time starts increasing
I didn't expect to see these #'s Could be Win10 Pro issue with unexpected delays?
Tried profiler , but I can't determine squat from it other than a big count at the wait .....
I think I lucked out getting My GCode sender working .
|
|
Logged
|
|
|
|
Richard Russell
Administrator
member is offline


Posts: 1348
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #16 on: Dec 22nd, 2017, 10:12pm » |
|
on Dec 22nd, 2017, 5:49pm, xtal wrote:| I only have to click on the window for the program to speed up? |
|
Clicking on a window activates it, and brings the process to the foreground. So this suggests that non-foreground processes are running much more slowly than they should. Have you checked with Task Manager or Resource Monitor to see what processes are using CPU time? Is it possible that you have some 'zombie' LB or LBB programs still running (invisibly), for example because of not including a trapclose?
Quote:| Tried profiler , but I can't determine squat from it other than a big count at the wait ..... |
|
If the profiler is not recording the delays, multimedia interrupts (timeSetEvent) are not running properly, and if multimedia interrupts - which have a high priority - are not running, something is seriously broken!
Richard.
|
|
|
|
Rod
Full Member
member is offline


Gender: 
Posts: 110
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #17 on: Dec 23rd, 2017, 10:22am » |
|
These delays are exactly what you should expect from the Windows timer. I always describe the timer as a clock face that only changes every 16ms or so.
Now this is a problem for the timer command because it has to look at the clock face and react when the desired time is shown. It seems we cannot see the time change faster than 16ms so it is pointless coding timer delays below this value.
Also because we start mid cycle it is variable, lengthening the delay does not help as you have seen because you tend to go up in 16ms increments and so jump from 16ms to 32ms.
So at the timings you are trying to get the timer is useless to you. Drop it and try SLEEP or simply loop.
That way you will lose your massive WAIT dely.
|
|
Logged
|
|
|
|
xtal
New Member
member is offline


Gender: 
Posts: 27
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #18 on: Dec 23rd, 2017, 1:20pm » |
|
If your thinking the 28 sec is a massive wait, it is not, it's just an indicator that the program is running without any unexpected delays, The bucketdelay wait is being used to process any serial data. Sometimes I have to stop the timer when going GCode processing / file reading etc, because invalidtag tag [ckbuf] is intermittantly displayed, then restart timer when leaving the code. Other than that the GCode pgm runs fine under LB.
|
|
Logged
|
|
|
|
Rod
Full Member
member is offline


Gender: 
Posts: 110
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #19 on: Dec 23rd, 2017, 2:00pm » |
|
No, I was thinking 16ms is a massive delay if you have a lot to do.
|
|
Logged
|
|
|
|
Richard Russell
Administrator
member is offline


Posts: 1348
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #20 on: Dec 23rd, 2017, 5:30pm » |
|
on Dec 23rd, 2017, 2:00pm, Rod wrote: No, I was thinking 16ms is a massive delay if you have a lot to do. |
|
As far as I can see the OP is using the timer appropriately. As I understand it he is using it to process incoming serial data, so the important thing is that the serial input buffer cannot overflow between one timer 'tick' and the next. 16ms is not a "massive" amount of time from this perspective: even at 115,200 baud the maximum amount of data that can be received in 16ms is fewer than 200 bytes, which is much smaller than the size of the input buffer.
Richard.
|
|
Logged
|
|
|
|
Richard Russell
Administrator
member is offline


Posts: 1348
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #21 on: Dec 23rd, 2017, 9:21pm » |
|
on Dec 23rd, 2017, 1:20pm, xtal wrote:| invalidtag tag [ckbuf] is intermittantly displayed |
|
LB 4 has a bug which means that you cannot (safely) use the timer with a SUB handler but only with a branch label. This means that, as you found, it is possible for a timer event to fire when the target label is out-of-scope (e.g. when your program's execution is inside a sub or function), causing an error to be reported.
You can work around this by ensuring that your program never executes a SCAN or WAIT statement when inside a function or sub, that way the timer handler will only be called when its target is in scope. But that's a serious limitation; for example it prevents you encapsulating a GUI window and all its control handlers inside a SUB, which is often desirable from a structured programming viewpoint.
Fortunately LBB does not have that bug. You can safely use a SUB handler for your timer events, and since SUBs are always in scope you can use SCAN or WAIT statements anywhere you like within your program without any risk of an error. Of course you will probably want to make all your other event handlers SUBs as well, but that's good practice anyway (and works in LB).
Richard.
|
|
|
|
xtal
New Member
member is offline


Gender: 
Posts: 27
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #22 on: Dec 24th, 2017, 3:18pm » |
|
modified my complete program Many buttons etc...
LB = RUN LBB = RUN Program ''display TIC value - stop timer - set timer [LoopExit] ' Display tic amount = 15 that I'm setting #main.GCwindow, "Set TIC=";bucketdelay timer 0 timer bucketdelay, [ckbuf] ' every 15 ms ckbuf wait ' Set cnt to 99 instead of 1775 if Qcnt >99 then '' COMPUTE TIC Time in millisec mstime=int((etime/100)*10000)/10 LB = Run Opened Dec 24, 2017 - 10:15:27 <10:15:28> <100> <1.659sec> <16.5msTIC> <10:15:30> <100> <1.732sec> <17.3msTIC> <10:15:32> <100> <1.697sec> <16.9msTIC> <10:15:34> <100> <1.627sec> <16.2msTIC> Closed Dec 24, 2017 - 10:15:34 Task mgr chrome 0.1-.3% lbpro 1-6% notepad 0.1% taskmgr .3-1.1%
LBB = Run Program Opened Dec 24, 2017 - 10:30:22 <10:34:52> <100> <314.259sec> <3142.5msTIC> <10:40:03> <100> <311.731sec> <3117.3msTIC> <10:45:16> <100> <312.729sec> <3127.2msTIC> <10:50:27> <100> <310.503sec> <3105msTIC> <10:55:37> <100> <310.218sec> <3102.1msTIC> <11:00:49> <100> <311.957sec> <3119.5msTIC> Closed Dec 24, 2017 - 11:05:06 --wow 3 second TIC time SB 15+ ms Task mgr chrome 0.1-.3% LB Booster .1-1.3% notepad 0.1% taskmgr .3-1.1%
Earlier I saw BBCwrun6 in task mgr apps @ 30% I have no Idea what that was , but now its gone ... Again this is W10 pro Tried compatility run whit same results
|
| « Last Edit: Dec 24th, 2017, 3:22pm by xtal » |
Logged
|
|
|
|
Richard Russell
Administrator
member is offline


Posts: 1348
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #23 on: Dec 24th, 2017, 4:10pm » |
|
Please show us the profiler report (you can edit out all the lines that show zero time, to reduce the size).
Richard.
|
|
Logged
|
|
|
|
xtal
New Member
member is offline


Gender: 
Posts: 27
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #24 on: Dec 24th, 2017, 6:11pm » |
|
[copy] Start Copy Profiler report for C:\Users\msw10\AppData\Roaming\Liberty BASIC Pro v4.5.1\GRBL-NU5DggmZ10h-lbb.bas
Sun. 24 Dec 2017, 12:48:14
Figures in the first column indicate approximate time in milliseconds spent in each program line.
Figures in the second column indicate approximate percentage of the time spent in each program line.
Time spent profiling: 350.024 seconds.
82: 0.02 '** Liberty BASIC 4.51 pro - 3/14/2017 GCode Sender using GRBL 1.1F Program with simple Graphics 2: dim info$(BrowseLen,BrowseLen) ' folder and file 8: Menu #main, "DROreset", "Reset MPX", [ZX], "Reset MPY", [MZY], "Reset MPZ", [MZZ], "ResetALL", [MZR] 6: Menu #main, "HELP", "Q/A.txt", [HELPx],"LibertyBasic.chm",[HELPchm],"LB-Forum",[HELPforum] 8: Menu #main, "DEBUG","--NA1--",[XBUG1],"--NA2--",[XBUG2],"DLY15/35",[DLY32],"DLY25/45",[DLY48],"DLY55/75",[DLY64],"DLY70/90",[DLY80],"DLY85/105",[DLY96] 2: button #main.params,"$$ Stat", [PARAMS], UL, 5, 400, 45, 20 2: button #main.JZMa,"--",[JZMa],UL,125,549,12,13 ' 2: button #main.JZFg,"--",[JZFg],UL,151,624,14,13 ' 4: texteditor #main.WinStatus, 55,125, 420, 80 ' LBB adjust ' Top left Corner - over, down 2: texteditor #main.GCwindow, 55, 210, 420, 260 ' LBB adjust ' bottom right - width, down 4: bmpbutton #main.EMRRESET, BMPdir$;"estop6.bmp",[ESTOP],UL, 188,575 ' EM STOP / CLOSE 2: button #main.mac6, "Mac6", [mac6], UL, 687, 0, 40, 20 2: button #main.RunGcode, "Run GCode", [RunCode], UL, 940, 25, 62, 20 2: button #main.MZR, "ZR", [MZR], UL, 195, 0, 25, 100 ' 4: texteditor #main.PopUp, 500, 250, 475,350 ' 1124: 0.32 Open "Liberty Basic GRBL GUI" for Window as #main ' 4: #main.Status "setfocus" 'MUST setfocus to graphicbox ' 2: #main.PopUp, "!hide" ' 2: #main.selbox2, "hide" ' hide this box 2: #main.Filebox,"backcolor green" ' 4: #main.yellow,"!font roman 6 10" ' 4: #main.green,"!font roman 6 10" ' 2: #main.yellow,"Folder" ' 2: #main.green,"File" ' 6: #main.JX, str$(JXvar) : #main.JY, str$(JYvar) : #main.JZ, str$(JZvar) 6: #main.JXF, "F";str$(JXFvar) : #main.JYF,"F";str$(JYFvar) : #main.JZF,"F";str$(JZFvar) 2: magnify=1 : #main.Mag1,1 ' show magnifer value 2: plotxy=2 : #main.PLOTa,"TL" ' 2: #main.GCwindow, "TotalScreen Height = "; DisplayHeight ' 2: if len(P$)>0 then #main.Comx, P$ ' 2: if len(BaudRate$)>0 then #main.Baud, BaudRate$ ' 26: 0.01 #main.GCwindow, "Set TIC=";bucketdelay 32504: 9.29 wait ' 2: #main.GCwindow, Xtime$ ' 74: 0.02 #GRBL, "?"; '' Keep semicolen so ok-ok won't display ' approx 100 ms interval '' not sure how semicolen does this 8: #main.Dly1, int((AvgTime/25)+.5) ' 2: comlof=lof(#GRBL) ' lof >1 data is in buffer 12: if comlof>= 1 then GRBLin$ = GRBLin$ + input$(#GRBL,comlof) ' save comm data to GRCLin$ 4: timer 0 ' stop timer until line processed ***** 2: Ze=instr(WIN$,"|",2) : winswitch = 1 ' <Idle|MPos 2: Ze$=mid$(WIN$,2,Ze-2) ' isolate Status 2: call DROWCOUPDATE ' [WCO_Update][MPOS_Update] [WPOS_Update] 2: else ' " <~~~~~~~~~~~~~~~~~~~~~~~> " 62: 0.02 #main.WinStatus, "<";Qcnt;"> ";win$ ' print <~~~~~~~> msg in status window 4: #GRBL, "M5" ' ensure spindle off 34: 0.01 Q=DelayMS(500) ' wait 1/2 second 34: 0.01 Q=DelayMS(500) ' wait 1/2 second 2: close #GRBL ' comm port 36: 0.01 close #main ' main window 31032: 8.87 #main.DROWCx, "!font arial 14" ' 30944: 8.84 #main.DROWCy, "!font arial 14" ' 164: 0.05 #main.DROWCy, WCY ' update WCOy 172: 0.05 #main.DROWCz, WCZ ' update WCOz 31016: 8.86 #main.DROMPx, "!font arial 14" ' 166: 0.05 #main.DROMPx, MPX ' load MPOSx 31102: 8.89 #main.DROMPy, "!font arial 14" ' 166: 0.05 #main.DROMPy, MPY ' load MPOSy 156: 0.04 #main.DROMPz, MPZ ' load MPOSz 31018: 8.86 #main.DROWPx, "!font arial 14" ' 166: 0.05 #main.DROWPx, MPX - WCX ' load WPOSx 31072: 8.88 #main.DROWPy, "!font arial 14" ' 31014: 8.86 #main.DROWPz, "!font arial 14" ' 912: 0.26 while time$("milliseconds") < td ' 18: 0.01 wend ' 31834: 9.09 #main.Status, "font Courier_New ";fontsize$ ' 126: 0.04 #main.Status, "cls" ' clr graphics box 94: 0.03 #main.Status, "fill ";color$ ' fill yellow 16: #main.Status, "backcolor ";color$ ' text background 4: #main.Status, "place " l$;" " d$ ' from top left over-> own 346: 0.10 #main.Status, "\" ata$ ' print data 54: 0.02 #main.GCwindow, "CommPort <" ;"> checked at ";time$() ' 94: 0.03 open "Com";str$(p);":115200,n,8,1" for random as #com ' use this for LBB 4: close #com ' 2: [trap] ' 2: #main.Baudrate, "!" ' clr listbox 'BaudRate$ 8: open P$;":";BaudRate$;",n,8,1" for random as #GRBL ' LBB 2: #main.Baudrate, "!";BaudRate$ ' combo select box 2: #main.Comx, P$ ' testbox 2: #main.Baud, BaudRate$ ' test box 2: #main.GCwindow, P$;" WAS / IS OPEN " ' 4: #GRBL, CTLx$ ' 4: #GRBL, "$$" ' request params 276: 0.08 wait ' 2: #main.GCwindow, "GcodeDir$ = ";GcodeDir$ ' 4: #main.GCwindow, "BMPdir$ = ";BMPdir$ ' 2: #main.GCwindow, "EditPath$ = ";EditPath$ ' 2: #main.GCwindow, "GUI-1Line-[Mac";x;"] = "; Macro$(x) ' 8: #main.GCwindow, "<";x;"> ";MP1$(x) ' 2: #main.GCwindow, "<";x;"> ";MP2$(x) ' 4: #main.GCwindow, "<";x;"> ";MP3$(x) ' 2: #main.GCwindow, "<";x;"> ";MP4$(x) ' 2: StatusDly=55 : wait ' approx 64 392: 0.11 Calldll #user32, "ScreenToClient",_ 758: 0.22 (Unaccounted for) Finish Copy
[/code]
|
| « Last Edit: Dec 24th, 2017, 6:15pm by xtal » |
Logged
|
|
|
|
xtal
New Member
member is offline


Gender: 
Posts: 27
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #25 on: Dec 24th, 2017, 6:17pm » |
|
Ran LBB on older laptop with WinXP and it also has the big delays
|
|
Logged
|
|
|
|
tsh73
Full Member
member is offline


Gender: 
Posts: 210
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #26 on: Dec 24th, 2017, 8:19pm » |
|
Hello xtal Looking to profiler report - long time is spent changing fonts. Are you by chance changing font in a loop? Is there any reason to change font at all - not to set it right after opening window, once and for all?
|
|
Logged
|
|
|
|
Richard Russell
Administrator
member is offline


Posts: 1348
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #27 on: Dec 24th, 2017, 9:43pm » |
|
on Dec 24th, 2017, 8:19pm, tsh73 wrote:| long time is spent changing fonts |
|
Exactly. As I suggested from the start, the profiler makes the cause of the slowness obvious.
I expect LB 4 checks to see whether the selected font happens to be the same as what was previously selected, and ignores the command if so. LBB doesn't do that, because I had no reason to suspect that anybody would keep selecting the same font repetitively.
All in all, a bit of a wild (Christmas) goose chase!
Richard.
|
|
|
|
xtal
New Member
member is offline


Gender: 
Posts: 27
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #28 on: Dec 24th, 2017, 10:10pm » |
|
I noticed I was setting the same font on the DRO, when the profiler was running, I just set the font every time I entered a procedure.
I will remove some of the redundant font settings, and see what happens, but keep in mind that the shortened program was having issues, and it had very little font changes... 
thx all
|
|
Logged
|
|
|
|
Richard Russell
Administrator
member is offline


Posts: 1348
|
 |
Re: LBB runs 7 times slower than LB on my GBRL-Sen
« Reply #29 on: Dec 24th, 2017, 11:12pm » |
|
on Dec 24th, 2017, 10:10pm, xtal wrote:| keep in mind that the shortened program was having issues |
|
Sort out your coding flaws first: eliminate unnecessary repetitions, set the font just once when the window is created, don't update controls faster than somebody can read the changes etc.
Then, and only then, check whether there are any remaining issues. Until that point I'm not "keeping anything in mind" because as far as I'm concerned there is nothing wrong with LBB.
Richard.
|
|
Logged
|
|
|
|
|