[Tfug] tfug Digest, Vol 69, Issue 6

[Paul Steinbach]

tfug-request at tfug.org wrote:
> Send tfug mailing list submissions to
> 	tfug at tfug.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> 	http://www.tfug.org/mailman/listinfo/tfug_tfug.org
> or, via email, send a message with subject or body 'help' to
> 	tfug-request at tfug.org
>
> You can reach the person managing the list at
> 	tfug-owner at tfug.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of tfug digest..."
>
>
> Today's Topics:
>
>    1. Re: Looking for a cable (Bexley Hall)
>    2. Guys, what's a good way to test video performance? (Jim March)
>    3. Re: Guys, what's a good way to test video performance?
>       (John Gruenenfelder)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Sat, 4 Apr 2009 17:41:08 -0700 (PDT)
> From: Bexley Hall <bexley401 at yahoo.com>
> Subject: Re: [Tfug] Looking for a cable
> To: Tucson Free Unix Group <tfug at tfug.org>
> Message-ID: <482547.25515.qm at web32907.mail.mud.yahoo.com>
> Content-Type: text/plain; charset=iso-8859-1
>
>
>   
>> Interesting.? Just replaced a couple of 3-port Linksys
>> PS units at a client - 
>> the in-use unit was bad and the spare they bought off eBay
>> wouldn't work.? 
>> Should have thought to try WC.
>>     
>
> Perhaps a dozen of the HP variety (single port like the 300EX)
> probably went in the "recycle" pile today as there are typically
> more donated than there is demand.  <frown>
>
> I think there may be a metric butt-load of single port
> *terminal* servers there, as well.  Dunno.  I didn't
> have time to research them (Joe was more interested in
> salvaging the plastic bags that they were packed in than
> the devices themselves  :-/  odd priorities)
>
>
> Noticed one single port HP EX3? print server in a box out in the yard sale out at WorldCare Sat.      
>
>
>
> ------------------------------
>
> Message: 2
> Date: Sun, 5 Apr 2009 02:40:00 -0700
> From: Jim March <1.jim.march at gmail.com>
> Subject: [Tfug] Guys, what s a good way to test video performance?
> To: Tucson Free Unix Group <tfug at tfug.org>
> Message-ID:
> 	<eee7e8570904050240n1aadf0efrc043b7ce30697f8 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> My lappy has an Intel 965 (aka "x3100") video chipset, I'm running
> Ubuntu Jaunty and I'm trying to tune video performance using the new
> "UXA" acceleration method.  Apparently it's stable for some people,
> not others.  Heh.
>
> Well I've run into something funky.  I've switched, the system seems
> OK, it "feels" fast as hell in Compiz rotations and the like.  But
> according to GLXGears, it's pulling about 450ish versus about 680
> under EXA.
>
> At the same time, GLXGears itself is giving me a weird message I've
> never seen before in the terminal window:
>
> ===
> jim at thecritter:~$ glxgears
> Running synchronized to the vertical refresh.  The framerate
> should be approximately 1/32 the monitor refresh rate.
> ===
>
> That seems to suggest GLXgears is giving an even more inaccurate view
> of video performance than usual...like it may be total BS.
>
> And I've seen numerous reference to GLXGears not being a valid test anyways.
>
> So what IS a good test suite to compare UXA and EXA settings in xorg.conf?
>
> Thanks!
>
> Jim
>
> PS: as I was typing this video froze completely so...yeah, I could
> still use a reliable video performance tester, BUT uxa ain't ready for
> prime time yet.
>
>
>
> ------------------------------
>
> Message: 3
> Date: Sun, 5 Apr 2009 06:09:48 -0700
> From: John Gruenenfelder <johng at as.arizona.edu>
> Subject: Re: [Tfug] Guys, what s a good way to test video performance?
> To: Tucson Free Unix Group <tfug at tfug.org>
> Message-ID: <20090405130948.GA27714 at as.arizona.edu>
> Content-Type: text/plain; charset=us-ascii
>
> On Sun, Apr 05, 2009 at 02:40:00AM -0700, Jim March wrote:
>   
>> My lappy has an Intel 965 (aka "x3100") video chipset, I'm running
>> Ubuntu Jaunty and I'm trying to tune video performance using the new
>> "UXA" acceleration method.  Apparently it's stable for some people,
>> not others.  Heh.
>>
>> Well I've run into something funky.  I've switched, the system seems
>> OK, it "feels" fast as hell in Compiz rotations and the like.  But
>> according to GLXGears, it's pulling about 450ish versus about 680
>> under EXA.
>>
>> ===
>>
>> That seems to suggest GLXgears is giving an even more inaccurate view
>> of video performance than usual...like it may be total BS.
>>
>> And I've seen numerous reference to GLXGears not being a valid test anyways.
>>
>> So what IS a good test suite to compare UXA and EXA settings in xorg.conf?
>>     
>
> Yeah, I don't think I would trust glxgears.  It's an extremely simple GL test
> and most any modern configuration will be able to run it very fast.  Also,
> being so simple, it doesn't even test a tenth of the capabilities of your GPU.
> glxgears is about as useful for benchmarking as the speed tests hdparm can do.
> bonnie++ is a much better drive/filesystem benchmark and there are far better
> tools for GPU testing.
>
> Some things to try:
>
> * A quick one: if you're running Compiz or some other compositing window
> manager, it's already pushing your GPU much harder than glxgears.  See if you
> can make it display its FPS value.  Not really a valid/repeatable benchmarking
> tool, but it can give you a quick answer about how the system "feels".
>
> * Install one of the available FOSS FPS games.  Most of these should have a
> demo mode than can be accessed via the game's console.  It's an automated and
> repeatable test and should test a lot of what OpenGL and your video driver can
> do.
>
>
>   


-- 
Paul Steinbach
MIS Manager
Sam Levitz Furniture

E-mail: MIS at samlevitz.com
Phone:  520.624.7443 X2571
Cell:   520.247.5730