 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : ArmmiteH7 V5.07.00b6 - Major performance hike
Page 1 of 3   |
|||||
| Author | Message | ||||
| matherp Guru  Joined: 11/12/2012 Location: United KingdomPosts: 10315 |
ArmmiteH7V5.07.00b6.zip This has the variable hash table look-up from the CMM2. This should give 30-40% improvement in performance e.g Solar eclipse calculation V5.05.11 23.73 seconds V5.07.00b5 16.91 seconds V5.07.00b6 10.86 seconds |
||||
| gadgetjack Senior Member  Joined: 15/07/2016 Location: United StatesPosts: 169 |
This version fixed the keyboard error and runs faster!!! Damn , you are good Peter. Only error I have found is if you type MEMORY , you get a hard interrupt failure. First I have ever seen that. Not a deal breaker , rest seems ok. Thank you for all the effort you are putting in this. Printing a new case for this as we speak. |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 10315 |
Thanks for the report - please try this ArmmiteH7V5.07.00b7.zip |
||||
CaptainBoing Guru Joined: 07/09/2016 Location: United KingdomPosts: 2170 |
In the morning, I will dig out my H7 and blow this new f/w into it - use it as my go-to for a while to exercise things a bit... |
||||
TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6283 |
Beta 7 is looking good. Thanks Peter. Jim VK7JH MMedit |
||||
KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
Just loaded Beta 7. Keyboard now works. I loaded my test program from the SD card, loads ok, type RUN and the program will not run and the H7 locks, no screen output, no CRL-C break, no Terminal coms. Tried both the V1 and V2 of the H7, both lock solid and need a hard reset or power cycle. All options list correctly. All MM info status queries return correctly. GUI calibrate is fine. GUI test is fine. Here are the memory and options display from the Z1 Test unit, notice this program is not small, it ran perfectly with the first 3 test versions of the updated firmware, Beta 7 broke it, I will do more testing on smaller programs when I get a chance. Program appears to List correctly. MEMORY Flash: 218K (56%) Program (8819 lines) 166K (44%) Free RAM: 0K ( 0%) 0 Variables 0K ( 0%) General 555K (100%) Free > Option List OPTION COLOURCODE ON OPTION DISPLAY 36,100 OPTION LCDPANEL SSD1963_8_16, LANDSCAPE OPTION TOUCH 56, 123 OPTION FLASHPAGES 2 OPTION SDCARD 87 OPTION LCDPANEL CONSOLE OPTION CONTROLS 300 OPTION USBKEYBOARD US, 91 > Mike. . Edited 2021-04-26 09:03 by KeepIS NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
Loaded a smaller 43K program for a graphical Alarm clock. Program ran and graphics all work, seconds hand is ballistic so likely changes to the code need to be made to accommodate some upgrades / changes in MM.basic. Fixed the problem by adding OPTION MILLISECONDS OFF. This program now runs correctly  Tried to EDIT the file: Editor works ok and I can traverse the file. Commented out a line and pressed F2 Display said saving 3 bytes. ?????? List returns nothing but a cursor. The program no longer exists in memory  Mike. Edited 2021-04-26 11:57 by KeepIS NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
I thought I had another problem but it looks like a memory issue with the size of the loaded bitmap, thought I'd post it anyway. The Alarm clock is currently running, I stop the program using CTRL-C. Type Files and: files Error : Not enough memory > CPU RESTART > files A:/ <DIR> AUTOMATE <DIR> CLOCK <DIR> DROSetup <DIR> JOBS 252661 AGV3.BAS 4 directories, 1 file > RUN !!!!! I Stopped running program again via CTRL-C > files Error : Not enough memory > MEMORY Flash: 36K ( 9%) Program (1428 lines) 348K (91%) Free RAM: 21K ( 3%) 130 Variables 390K (70%) General 144K (27%) Free > Obviously CPU-RESTART releases all RAM used by the program and so the files cmd works again. Mike. . Edited 2021-04-26 12:09 by KeepIS NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6283 |
I noticed that milliseconds was on also. I put it down to a leftover from a program I had just run but it looks like it now defaults to MILLSECONDS ON. Jim VK7JH MMedit |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
After loading the main large test program: > MEMORY Flash: 218K (56%) Program (8819 lines) 166K (44%) Free RAM: 1266532K (1485%) 1012 Variables 0K ( 0%) General 4293701319K (-1385%) Free After restart: > MEMORY Flash: 218K (56%) Program (8819 lines) 166K (44%) Free RAM: 0K ( 0%) 0 Variables 0K ( 0%) General 555K (100%) Free > Mike. . Edited 2021-04-26 12:15 by KeepIS NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
Yes certainly seems like that, knew it would be something simple and the first thing I thought of was the returned time string was now including MS and remembered the Option for that. Whish my main program was as easy to fix, really hard when the inbuilt editor keeps deleting the loaded image. I get the feeling there is something happening in the token "speed change" area of beta 7? stopping the program? Mike. . Edited 2021-04-26 13:21 by KeepIS NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
I loaded the main test file, compared the listing to the disk image and they were identical with the exception of a few leading case changes for if, then, else, etc. I placed a few print statement in the file: 1: The beginning of the variable declarations. 2: The end of the variable decelerations. 3: The end of the GUI definitions. The only print statement executed was 1, it never got past variable decelerations. Looks like some kind of memory or tokenised corruption issue. I can't do more that this. I can confirm the loss of the file from memory when in the editor if any change is made to the file and it's saved or run from the editor. Mike. . NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6283 |
My FTP server program runs well. The only changes were to a few pin numbers and com port choice. There is definitely a problem with the MEMORY command. After NEW and CPU RESTART, the memory looks correct. After loading the FTP server program but before running, the memory RAM results are way out. After running the program and exiting cleanly, the memory command is correct again. > new > memory Flash: 0K ( 0%) Program (0 lines) 256K (100%) Free RAM: 0K ( 0%) 0 Variables 0K ( 0%) General 562K (100%) Free > cpu restart > memory Flash: 0K ( 0%) Program (0 lines) 256K (100%) Free RAM: 0K ( 0%) 0 Variables 0K ( 0%) General 562K (100%) Free ' load a program Saved 22526 bytes > memory Flash: 22K ( 8%) Program (1049 lines) 234K (92%) Free RAM: 1460177K (6072%) 403 Variables 0K ( 0%) General 4293507681K (-5972%) Free > ' run the program Shutting down... > memory Flash: 22K ( 8%) Program (1049 lines) 234K (92%) Free RAM: 16K ( 2%) 55 Variables 0K ( 0%) General 546K (98%) Free > Jim VK7JH MMedit |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 10315 |
ArmmiteH7V5.07.00b8.zip Fixes the memory command - this was just a silly artifact and didn't represent an underlying problem Defaults milliseconds correctly Fixes ERASE and CLEAR commands Fixes EDIT Mike: I can't do anything about your issue if it still exists in this version unless you let me have the code to work with. The biggest program I have is much smaller than yours. Send me a PM if you don't want to post online. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
Thanks for fixing the Editor, I've been using it to narrow down the problem. The problem is the number of variables. If I comment out about 8 to 10 lines of DIM statements, the GUI loads and the program runs until a missing variable is encountered. I cut out just the variable declarations and placed them in a file with a Print "END" statement followed by END. The same error is encountered, comment out some variables and when enough have been commented out the simple Print "END" statement is reached and the header .BAS file drops to the cursor without locking up the H7. I can send you this header if you like, but it is simply locking after a certain number of variables. There are around 1300 variables, the header is 15K in size. Mike. NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 10315 |
Mike This is a major issue where you may need to stick with the earlier firmware or do some major rework. The hashing mechanism allows for up to 512 global variables and 512 local variables. On the CMM2 no-one has come up against this limit. This allocates 64Kb of memory for the variables (memory for arrays is separate) Note also that with more than about 256 variables performance will start to drop because of hash key conflicts Your code with so many variables is also very inefficient with the original firmware as variable location is a sequential search. Using array elements would probably be quicker Edited 2021-04-26 20:01 by matherp |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
Dam, I'm about 74 variables over, doubt I can drop that many, I assume if not using many local variables then global variables can use remaining memory allocation or am I barking up the wrong tree? Mike. NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 10315 |
No, the limit is fixed in each category. Maximum 512 global variables and up to 512 local variables active at any one time in any nested structure of calls. |
||||
| KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1882 |
Ok understand, I'm including all constants in that count and it's strange that I'm only 70 old over 512, I'll have to go back and count them manually. BTW, I understand the slowness of the number of variables, but in this case the variables used in the high speed part of the code are around 60 and are placed and named with a common unique starting character at the start of the file, the rest are used in setup screens, so speed is of no consequence in this part of the code. The application (machine automation part) runs very very fast in the previous version. Of course with the way the CMM2 works this may well be very inefficient as you point out. Thanks for the information. Mike. Edited 2021-04-26 20:16 by KeepIS NANO Inverter: Full download - Only Hex Ver 8.1Ks |
||||
thwill Guru Joined: 16/09/2019 Location: United KingdomPosts: 4311 |
Hi Peter, So far I've come nowhere near the CMM2 limits and 512 globals sounds "excessive" considering the CMM2's target use. However I assume that global CONSTs count against that limit and I can imagine use cases for lots of those (to emulate enumerations for example). Use of those can I think be mitigated by using #define (is that present on the Armmite ?) but we only have 64 of those, perhaps you might increase that limit - just a thought, like I say I'm nowhere near that limit and will probably have wandered off into Forth and excessive CSUB usage before getting there. Best wishes, Tom MMBasic for Linux, Game*Mite, CMM2 Welcome Tape, Creaky old text adventures |
||||
| Page 1 of 3 |
|||||
 Print this page |
 |
The Back Shed's forum code is written, and hosted, in Australia. | © JAQ Software 2025 |
