lubuntu-desktop team mailing list archive

[∅ <maps.backward@xxxxxxxxx>]

On 11/07/2011 11:04 PM, amjjawad HOOHAA wrote:
> I was wondering if someone has noticed that LXTask actually shows 
> +100MB usage more than System Monitor and Conky so what is going on?
> I have System Monitor installed on Lubutnu 11.04 (was testing 
> something) and Conky as well. While each of System Monitor (GNOME APP) 
> and Conky are showing the same memory usage, LXTask add *100MB* to that.
> Example: System Monitor + Conky shows 287MB while Firefox Loaded with 
> two tabs. On the other hand, LXTask shows 387MB.

The immediate thing that comes to mind is I wonder if you're looking at 
the same thing. It's highly likely, but it's good to double check. 
Assumedly you're looking at either the virtual memory size or the 
resident memory size, which should be respectively ${top_mem mem_vsize 
n} and ${top_mem mem_res n} for conky (where n is the appropriate 
number); Virtual Memory and Resident Memory for gnome-system-monitor; 
VM-Size and RSS for lxtask; and VIRT and RES for top (just for grins), I 
come up with:

conky: resident 25.1MiB, virtual 176MiB
gnome-system-monitor: resident 292.2 MiB, virtual 874.8 MiB
lxtask: resident 292.2 MB, virtual 874.8 MB
top: resident 292m, virtual 874m

I took a screenshot of them all to reduce error.

Anyways, I guess to answer your question, it doesn't seem to be my 
experience that lxtask reports anything different from what other tools 
report for memory usage.

It seems like the only one that's funky is conky. I'm guessing that I'm 
wrong about how to set it up, but from the searching and documentation 
I've found it seems to be right. Sorry, I don't use conky normally.

wxl/walter

P.S. firefox with 11 tabs if it matters :D

P.P.S. When doing this benchmarking, we do need to let both programs run 
for a while. As I've said before, I feel firefox can be a bit memory 
leaky. I've also found Chromium gobbles up CPU on occassion (at least on 
the VM install I have at work). This is all sort of anecdotal, but it's 
worth testing. Maybe gauge CPU/RAM and then an hour later gauge again.