TheBackShed.com

Forum Index : Microcontroller and PC projects : Armmite H7: V5.05.03, Arcs and curves

Author

Message

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 10572

Posted: 09:49am 10 Jan 2019

Copy link to clipboard

Print this post

Please find attached V5.05.03

2019-01-10_194559_Armmite1.3.zip

2019-01-10_194625_Armmite_H7_Manual.pdf

This includes arc and Bezier curve drawing commands as per my recently listed CFunctions. Command syntax in the manual. It also removes the restriction on arrays having dimensions > 32767 elements

CLS
bezier 10,10,106,80,212,80,309,10,rgb(yellow)
bezier 10,230,106,150,212,150,309,230,rgb(yellow)
bezier 10,10,80,80,80,160,10,230,rgb(green)
bezier 309,10,239,80,239,160,309,230,rgb(green)
text 85,110,"Cubic Bezier Curves"
for i=0 to 99 step 4
Arc 600,120,20+i,,325+2*i,5+5*i,rgb(magenta)
next i
arc mm.hres\2,350,100,110,315,45,rgb(red)
arc mm.hres\2,350,100,110,45,315,rgb(green)

Edited by matherp 2019-01-12

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 10572

Posted: 03:49pm 18 Jan 2019

Copy link to clipboard

Print this post

Bug fix - please install if you are using H7 otherwise you may see issues with programs passing arrays to subroutines

2019-01-19_014847_Armmite1.3.zip

matherp
Guru

Joined: 11/12/2012
Location: United Kingdom
Posts: 10572

Posted: 06:41pm 27 Jan 2019

Copy link to clipboard

Print this post

Big performance improvement for programs with strings or arrays and re-write of serial processing to fix bug and improve performance. It transpires STM32 chips hate non-aligned data (i.e. an integer should be on a 4-byte boundary). Performance on this benchmark is now much greater than 2x MMX and nearly 5x MM+ at 100MHz.

2019-01-28_043616_Armmite1.3.zip

Coming soon port for STM32F767ZI (since I had one and wanted to evaluate how portable STM32 code really is)

Edited by matherp 2019-01-29

lizby
Guru

Joined: 17/05/2016
Location: United States
Posts: 3470

Posted: 07:27pm 27 Jan 2019

Copy link to clipboard

Print this post

matherp said Big performance improvement ...

Yay!

matherp said Coming soon port for STM32F767ZI

Double Yay!!

I hope in time there are some suitable Nucleo-F series boards in the DIP style--32, 48, and 64 pins. They would be great micromite platforms.

PicoMite, Armmite F4, SensorKits, MMBasic Hardware, Games, etc. on fruitoftheshed

viscomjim
Guru

Joined: 08/01/2014
Location: United States
Posts: 925

Posted: 08:03pm 27 Jan 2019

Copy link to clipboard

Print this post

This is amazing! When do you sleep????

gadgetjack
Senior Member

Joined: 15/07/2016
Location: United States
Posts: 173

Posted: 08:08pm 27 Jan 2019

Copy link to clipboard

Print this post

Another big YAY from me! I miss ordered one of those when the H7 was launched and have had no use for it till I heard that! Waiting on pins and needles......
Thank you.

viscomjim
Guru

Joined: 08/01/2014
Location: United States
Posts: 925

Posted: 08:33pm 27 Jan 2019

Copy link to clipboard

Print this post

@matherp, is it possible to port to this STM32? They seem to be readily available and super cheap for some reason...

They seem to be called the "blue pill".

lizby
Guru

Joined: 17/05/2016
Location: United States
Posts: 3470

Posted: 08:49pm 27 Jan 2019

Copy link to clipboard

Print this post

>Stm32f103c8t6

Only 64k bytes flash. I assume at least 512K is needed for gui features.

PicoMite, Armmite F4, SensorKits, MMBasic Hardware, Games, etc. on fruitoftheshed

KeepIS

Guru

Joined: 13/10/2014
Location: Australia
Posts: 1945

Posted: 12:24am 02 Feb 2019

Copy link to clipboard

Print this post

matherp said Big performance improvement for programs with strings or arrays and re-write of serial processing to fix bug and improve performance. It transpires STM32 chips hate non-aligned data (i.e. an integer should be on a 4-byte boundary).

I forgot to report back on this change on the ARM-H7 - what surprised me was the speed improvement in some simple test code I was running in a do loop waiting for an ADC process to finish:

I was flashing a GUI led with in a crude way:

[code]
do
FlashCount = FlashCount + 1
if FlashCount = 4000 then CtrlVal(RedLED) = 1

if FlashCount >= 8000 then
CtrlVal(RedLED) = 0
FlashCount = 0
endif
loop while ADC_1 = 0

[/code]

Before the firmware change the FlashCounts were 300 and 600, I had to increase them to 4000 and 8000 with the latest firmware to get roughly the same flash rate. I assume the increase was due to underling MMBasic GUI code implementation speed increase?

AGAIN in case someone missed it: This was just a quick bit of CRUDE code to get a visual indication of the loop still running as I was developing it, in real world code the flash rate (if used) would be controlled from a timer flag, BUT then I wouldn't have noticed the execution speed change.

There is no other code, timers or interrupts running except for ADC acquisition complete interrupt.

I am yet to load the new firmware into my CNC type controller, it will be interesting to see what speed changes I see in that very big and complex program.

NANO Inverter: Full download - Only Hex Ver 8.1Ks

Print this page

To reply to this topic, you need to log in.