Could somebady explain me why there is high difference in total speed if my gcode looks like

a)

G1 X1
G1 X1
G1 X1
...
G1 X1

100 times.
or (b) just:

G1 X100

Less gcode is faster? Why is the firmware not optimizing that?

@dc42 on SD-Card/RPI (Duet 3)

I found my error, and I think its Bug from the firmware. But, the firmware experts can probably tell me if this is intended....

If i call

; first step
M42 P4 S0.0
G4 P1
M42 P4 S1.0
G4 P1
; second step
M42 P4 S0.0
G4 P1
M42 P4 S1.0
G4 P1
....

everything is fine and required only a few miliseconds. But if call macros with exactly the same content (eg. M42 P4 S1.0), the programm needs several 100ms to perform a step.

M98 P"/macros/LOW.g"
G4 P1
M98 P"/macros/HIGH.g"
G4 P1
M98 P"/macros/LOW.g"
G4 P1
M98 P"/macros/HIGH.g"
G4 P1

@OwenD I think in RRF_3 the frequency is specified in M950 by Q and not anymore in M42, but I will try anway

To switch one of my H-bridge, I use two digital output

; definition
M950 P4 C"^io4.out" Q200                        
M950 P6 C" ^io6.out" Q200

Then if switch between two settings (1,2) with the "highest" possible frequency

; setting 1
M42 P6 S0.0
M42 P4 S1.0

;setting 2
M42 P6 S1.0
M42 P4 S0.0

; again setting 1
M42 P6 S0.0
M42 P4 S1.0

; again setting 2
M42 P6 S1.0
M42 P4 S0.0

..... repeat

The problem is that i only get a frequency about 3Hz between the changes from setting 1 to setting 2. I would expect a much higher one... Also a small break with G4 P1 does not work

I am running out of Ideas.... Some Hints for me?

[CORRECTION:] I was not using exactly this code, the code was refactored in macros and that was also the problem

@wilriker Ok I tried. I took the latest commit on the dev branch (Hash: 7b7c9291aaa1e77bd570c2503f981a73410a0f32)

I changed:

cmd_conn.connect('/var/run/dsf/dcs.sock')

I ran the example file with python 3.7 and the RPI and sudo rights (dcs.sock requires it)

and I got the following error

File "/home/shares/pi/HandwritingRoboter/src/DSF-APIs/pydsfapi/pydsfapi.py", line 234, in connect
  server_init_message.EXPECTED_SERVER_VERSION, server_init_message.version))
AttributeError: 'ServerInitMessage' object has no attribute 'EXPECTED_SERVER_VERSION'

Probably you have an Idea what that could be?

I am searching for the socket file "duet.sock".
Is it not under "/var/run/"? Can't find it there or anywhere on my raspberry.

@arhi I share arhi's opinion. That could the job of the firmware to avoid these unnessesary decelerations.

I just wanted to check, if it would be possible to replace G02 and G03 command by interpolating and making small G01 steps. But this is now significant slower and therefor a big disadvantages even if I am able to perfectly approximate the arc with my "lines".

Is there a way to suppress deceleration? Lets assume I want to move straight with G1 and want to keep the current speed without decelerating and you want to change in a (tangent) arc. How would this be possible?

@wilriker Thanks. I will try it directly tomorrow and give feedback to you. Just for others who are searching: these changes are still on the dev Branch.

Could somebady explain me why there is high difference in total speed if my gcode looks like

a)

G1 X1
G1 X1
G1 X1
...
G1 X1

100 times.
or (b) just:

G1 X100

Less gcode is faster? Why is the firmware not optimizing that?