Re: [Pan-users] Re: Pan v0.14.2.91 crashing X hard (black screen)/messing up NVIDIA card badly...

[Csv4Me2]

On Sunday 27 August 2006 03:30, Duncan wrote:
> Csv4Me2 <address@hidden> posted
> address@hidden, excerpted below, on  Sun, 27 Aug 2006
>
> 02:10:53 +0200:
> > As long as the current versions don't load the startrek group succesfully
> > i won't change versions. All new versions fail that simple test. The max
> > size of  a succesfull group download seem to depend solely on the
> > available amount of system RAM. As soon as pan (0.1xx) causes the system
> > to swap it slows down horrendously and never (?) finishes. This is IMHO
> > not a bug but a design/implementation limitation.
> > Adding to, removing from and sorting databases that are way larger than
> > the amount of core memory is a challenge! But it can be done and has been
> > done. See the sort program. Perhaps a v1.x issue ?
>
> I wonder if you are running into the 0.109 bug I just ran into (see the
> tasklist optimization thread for the gory details).  It's not a memory
> issue (I've 8 gig memory here, and seldom fill it all up with cache let
> alone start flushing to swap) but a CPU bottleneck. Darren says it has been
> bugged and patched for 0.110.  I hope so.  I'm about ready to go to sleep
> now, but if 0.110 isn't out tomorrow, I'll likely go try to find the bug
> and apply the patch and rebuild 0.109, as pan's behavior trying to queue
> tens of thousands of small posts at a time with 0.109 is beyond
> ridiculous, at least here.

No, I'm not.
When loading the complete startrek group pan claims about 1.8 G of memory 
(oscillating a bit and eventually claiming the full 2G of swap space) on my 
1.5 G RAM box. The swap drive is exerting itself nicely averaging at 10 
MB/sec or so. Definitely a memory starvation issue AND a working set problem. 
The last CVS version of pan loads this group just fine, takes ages though 
because old pan uses a single connection for the header download.

C