Re: [linrad] cpu load benchmarks

[w3sz <w3sz@xxxxxxxxxxx>]

Hi, Jeffrey,

Can you list the results that you got with your computer:

1. with

and

2. without

forcing VESA in the svgalib config file?

That is what we are wanting to see.

Thanks in advance,

Roger
W3SZ

On Fri, 18 Feb 2005 13:29:20 -0800 (PST), Jeffrey Pawlan  
<jpawlan@xxxxxxxxxx> wrote:

>
>
> On Thu, 17 Feb 2005 leif@xxxxxxxxxx wrote:
>> Avg      CPU
>> 100       12%
>>  10       33%
>>   3       87%
>>   2       96%
>>   1      not possible.
>
>
> I did this experiment using the small computer on my webpage. The  
> graphics card
> is a Sapphire Radeon ATI 7000 AGP 32MB.
>
> with 100 or more averages the cpu loading was around 12% as Leif found.
> But with averaging set to 1 (disabled) for both the spectrum and the  
> waterfall,
> and also the waterfall sized to 75% of the screen area, and the display
> resolution set to 1200x1024 I got a cpu loading of only 62%. That was  
> with
> 96000Hz sampling rate and one channel a/d selected. When I selected 4  
> channels
> the cpu usage went up to 72%.  This is still quite acceptable. My cpu  
> speed is
> 3.0GHz which is only slightly higher than Leif's 2.66GHz.
>
> Bug found:  If you select a screen resolution of 1200x1024 (mode 13) in  
> the 'S'
> menu BEFORE you select the 'U' menu then I got lockup after selecting  
> dsp00. If
> I selected 1024x768 (mode 12) then there was no problem setting up the  
> a/d and
> d/a dsp choices. Then once 'W' is hit to save the settings, you can go  
> back into
> 'S' and change the screen resolution to mode 13 and everything works!  
> Leif
> believes this may be an interrupt timing conflict problem in either OSS  
> or
> svgalib.
>
> 73,
>
> Jeffrey Pawlan, WA6KBL
>
>



-- 
Roger Rehr
W3SZ
http://www.qsl.net/w3sz

LINRADDARNIL