They might, but relying on the CPUID is like playing whack a mole. You are
never up to date, and never finished. I know because we have some similar
code in our product, and every week or so, one of our customers complains
to tech support that their device is reporting incorrectly. Then the
developers have to go fix it (and we have more developers around than BOINC
does).
jm7
Raistmer
<raist...@mail.ru
> To
Sent by: "Lynn W. Taylor" <l...@buscom.net>,
<boinc_dev-bounce "BOINC dev"
s...@ssl.berkeley.ed <boinc_dev@ssl.berkeley.edu>
u> cc
Subject
09/30/2009 05:25 Re: [boinc_dev] [boinc_alpha] Card
AM Gflops in BOINC 6.10
I simple don't know if they have some analog of CPUID? From your answer I
infer - they have not, right ?
----- Original Message -----
From: "Lynn W. Taylor" <l...@buscom.net>
To: "Raistmer" <raist...@mail.ru>; "BOINC dev" <boinc_dev@ssl.berkeley.edu>
Sent: Wednesday, September 30, 2009 11:03 AM
Subject: Re: [boinc_dev] [boinc_alpha] Card Gflops in BOINC 6.10
> If we assume that all processors have an x86 heritage, sure.
>
> What about the PowerPC processor, or Sun SPARC, or Motorola 68000 family
> parts, or VAX machines?
>
> They aren't mainstream, but they're still around.
>
> Raistmer wrote:
>>> Just having a rough idea of system performance solves a lot of
problems.
>>>
>> BTW, some estimation of computer performance can be aquired from CPUID
>> instruction.
>> CPU type + freq would give not bad estimation of FLOPS it can do. At
>> least if we talk about error of few times of value itself accepted...
>>
>> _______________________________________________
>> boinc_dev mailing list
>> boinc_dev@ssl.berkeley.edu
>> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
>> To unsubscribe, visit the above URL and
>> (near bottom of page) enter your email address.
>>
>
_______________________________________________
boinc_dev mailing list
boinc_dev@ssl.berkeley.edu
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
boinc_dev@ssl.berkeley.edu
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
