TheBackShed.com

Forum Index : Microcontroller and PC projects : GFXterm AUTOSAVE speed....

Author

Message

Grogster

Admin Group

Joined: 31/12/2012
Location: New Zealand
Posts: 9753

Posted: 07:49am 04 Sep 2018

Copy link to clipboard

Print this post

This is interesting.

I now use Rob's excellent GFXterm for all my MM console talk, and up till now I have been using 2102 USB modules to transfer the program to the MM. I was using 230k4 baud to get the program into the MM quickly, as it is more then 1,500 lines of code.

Purely by chance while setting up a new E64 1D module to test, pasting the code from the clipboard to the 1455 USB interface, it transfers about twice as fast as the 2102 did at 230k4 baud - and this is all at the standard 38k4 baud!

Very interesting. Anyone got any ideas why that should be?
Not that I care, as I am delighted that the 1455 seems much faster.
But it is curious.

Smoke makes things work. When the smoke gets out, it stops!

robert.rozee
Guru

Joined: 31/12/2012
Location: New Zealand
Posts: 2464

Posted: 12:47pm 04 Sep 2018

Copy link to clipboard

Print this post

are you sure what you are seeing is not because you've switched to the latest release of GFXterm?

the latest release of GFXterm (26-aug-2018) implements a much faster paste method - in fact it is too fast. whereas earlier versions had fixed time delays (35 to 50mS) at the end of each line, the latest version has no such delays. instead, it looks at the data echoed back from the micromite.

each <CR> sent out to the micromite increments a counter, while each <CR> received back decrements the counter (the counter is never allowed to drop below zero). when the counter reaches 2, GFXterm idles until the counter drops below 2. a 150mS timeout also resets the counter to zero.

this creates a 'sliding window' effect, that keeps things ticking along at a fair speed. i just did a few tests with an MX170/1455@38k4 and a 42k/636 line program i use for testing upload speeds, using different versions of GFXterm:

13-june-2018 ... 51 seconds ... (35-50mS line delay)
26-aug-2018 .... 19 seconds ... (sliding <CR> window, size=2)
4-sept-2018 .... 37 seconds ... (sliding <LF> window, size=1)
4-sept-2018 .... 32 seconds ... (sliding <LF> window, size=2)

(note that micromites may pause when sending <CR><LF> back, the likely cause of the difference between 19 and 32 seconds. i'm picking this will be far less pronounced with the MX470 version of mmbasic)

i guess now i'll have to make the sliding window size user adjustable!

cheers,
rob :-)Edited by robert.rozee 2018-09-06

Grogster

Admin Group

Joined: 31/12/2012
Location: New Zealand
Posts: 9753

Posted: 12:44am 05 Sep 2018

Copy link to clipboard

Print this post

Ahhhh, yes, that could be it.

It's great anyway.

Smoke makes things work. When the smoke gets out, it stops!

Print this page

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