Re: Keeping up to date with -current with a cross-compiler

[David Riley <fraveydank%gmail.com@localhost>]

On Dec 13, 2011, at 12:14 AM, Michael wrote:

> Hello,
> 
>> I've noticed that the 9500 is pretty slow on modern NetBSD, especially 
>> compared to the speed I remember from the 1.6 days.  I don't remember if 
>> that's nostalgia covering my memory or not, but disk I/O seems particularly 
>> slow (I'm using MESH).  I'd run a profiling build, but that seems to have 
>> trouble with build.sh (and the patch I saw online a year or so ago to add 
>> that to build.sh never seems to have made it into the mainline).
> 
> Are you by any chance using the same old kernel config from 1.6? Mesh had a 
> bug that prevented synchronous transfers which I fixed somewhere around 2.0, 
> older configs will disable sync negotiation - if your config is old the 
> disable sync bits probably never got removed.

I installed fresh from 5.0.2 a while ago.  So in theory, it shouldn't be that.  
I'll take a look at some point.

> Then, old SCSI disks are rather slow compared to anything halfway modern - 
> nothing prevents you from putting a cheap SATA card and a modern disk into 
> the 9500, I have this in my G4:

Well, yeah.  I've just been sitting on it because I'd like an OF-bootable one 
because I'm a stupid perfectionist. :-)  Thus the previous question about the 
Sonnet TSATA.  I've been holding off because it's $60, which is slightly 
outside of my "spend on a whim" budget.

> I used to build releases on a dual 2GHz G5 running 10.5 - it barely managed 
> to beat a dual 500MHz G4 running NetBSD thanks to OSX's slow-as-hell fork() ( 
> and probably NetBSD being much faster creating and deleting lots and lots of 
> small files )

Curious.  One of these days I need to get the power supply in my MDD G4 fixed 
so I can set it up as a proper server.  Seems all those machines developed a 
curious power supply failure right around the same time... there are a bunch of 
forum postings about it clustered around just a few months.


- Dave