tested both versions on the same file 1.5GB
2.38
average download speed 1.5MB/s
average CPU load 95%

2.34
average download speed 3.3MB/s
average CPU load 35%

both versions are portable and run from the same folder (different exe filenames)

P.S also noticed that incomplete naming works only on newly added torrents

For me it's the same with CPU utilization.  
I have an old intel Dual Core as my Torrent box and version Tixati 2.38 uses around 70-80% of one core while 2.34 uses 5-10% of one core.

I haven't checked if there's a difference in bandwidth utilization since I've already gone back to 2.34.

results of proc monitor for both versions downloading the same torrent

2.38
http://i.piccy.info/i9/5d7c175d0fea5414bae8408e0e1bbfb9/1462377157/29125/1030941/Clipboard_2.png


2.34
http://i.piccy.info/i9/4b90cfef16a2df07921c77a2286f7d80/1462377207/24323/1030941/Clipboard_4.png

tested 2.39 Alpha-1
issue is not resolved

v 2.41
extremely high CPU usage
starting from 2.38 tixati turned unusable

What does the Proc Monitor : Xfers Detail graph show?

I tried to recreate this myself and my CPU is actually pretty low these days, even when I'm downloading a lot.

On my Proc Monitor: Normal view, my Xfers line rides not much above my Network line, much like your 2.34 pic.  But this is what mine looks like for 2.41 too.

Is there any unusual settings you are using?  Categories?  Tree-mode layout?  Are you downloading to a flash-drive?

It might help to email the exact torrent you made the test with to the dev team (support@tixati.com) with a link to this forum post, so they can test the exact same situation.  Or will this happen with every torrent you have tried?

santos, i use Tixati as portable, i save to HDD not flash-drive, i haven't changed any settings while upgrading to 2.38 and i have problems with all torrents i have
my PC is quite old, only one core

What does the Help > Diagnostics > Proc Monitor (Transfers Detail) graph show?

2.34
(mod note: Link removed and sent to devs)

2.42
(mod note: Link removed and sent to devs)

That's not valid as a test at all.

You can't just run a few transfers at any given moment, note down the CPU/memory/whatever usage and then compare, like that.

Torrenting is dynamic. What you're running at any given time - ammount of transfers, ammount of peers, ammount of bandwidth used, ammount of hard drive stress caused by file writing, other programs using your computer resources, etc - will determine how it uses up your system resources.

You could as well run that "test" two times with the same exact bittorrent client version (Tixati or any other) and get completely different results.

To achieve any semblance of validity in that kind of test, you would need to run for days or weeks in a row, noting down averages and equating the whole to come to anything that could be called a statistic.

you think it's possible to "run tests for days or weeks in a row" with constant 100% CPU usage?
my PC will burn up in a couple of hours :)