Re: [dev] [st] Slow ST character typing on i686/pentium3 platforms]

From: Roger <>
Date: Sun, 8 Mar 2015 12:18:32 -0400

> On Tue, Mar 03, 2015 at 09:10:40PM -0800, Ariel A wrote:
>On March 3, 2015 5:21:04 PM PST, Roger <> wrote:
>>ST appears to compile and run fine on 64 bit and 32 bit platorms,
>>except on 32
>>bit (i686/pentium3), typing is extremely slow.
>Try adjusting xfps and actionfps. When I had this issue on a 32 bit machine, I seem to remember changing fps worked. I cannot confirm because I don't have the machine anymore.

I performed some differientials of the ST code along with runing the different
versions/snapshots available for download from the suckless website on my
Pentium3 (or i686) platform yesterday.

The following is my experience of the performance of ST under this bug on i686.

st-0.1* Slow character output to terminal, including key repeat.

st-0.2* Character output to terminal was excellent. No delays were seen and
key repeat worked as expected!

st-0.3 Very Slow character output to terminal.

st-0.4 Very Slow character output to terminal.

st-0.4.1 Extremely Slow character output to terminal!

Furthermore, adjusting the following line and the statically defined timeouts
within the code to static values appeared to have zero or no effect at all:
#define REDRAW_TIMEOUT (80*1000) /* 80 ms */

Now I did notice the drop from 80ms to 20ms delay within the st-0.2* versions,
but changing this variable definition from 80ms to 20ms within the other
versions had no effect. (Think I tested st-0.1*, st-0.3* and st-0.5*, changing
80ms to 20ms.)

So, it looks like I'm moving forward with checking to see if the delay
variables are actually using the expected delay using gdb (or printf), along
with further comparing the changes between st-0.1*, st-0.2* and st-0.3.

Whomever coded these early versions, might be able to very easily locate the
slowdown explicitly at this point.

Received on Sun Mar 08 2015 - 17:18:32 CET

This archive was generated by hypermail 2.3.0 : Sun Mar 08 2015 - 17:24:07 CET