Re: xPC taget CPU - Athlon or Intel
- From: "Adam Hersbach" <adam.hersbach@xxxxxxxxx>
- Date: Mon, 12 Sep 2005 04:29:37 -0400
Keith,
I would love the answers to exactly the same questions you are
asking, but so far, I haven't been able to find out these answers,
and have resorted to doing the tests myself.
I have ordered an Athlon 64 FX processor and will post the bench test
results once I have the chance to test it. Check back in about 2
weeks.
Adam
Keith Williams wrote:
>
>
> Hello,
>
> With regards to this issue of AMD vs. Intel and XPC Target, I have
> two questions:
>
> First, can I use an Athlon 64-equipped machine as my target? I
> would
> like to build a new target box to outperform my old AMD T-Bird
> machine, but I didn't want to buy a 64-bit cpu (say socket 939) and
> then find out that I couldn't use it except as a desktop.
>
> Second, assuming that I can use the 64-bit cpu, does the 64-bit
> behavior handicap the system in any way, as compared to if I was
> using an Athlon XP? My purchase choices will be between an Athlon
> XP
> 3200+ (333 MHz FSB) and an Athlon 64 3000+ ("scalable?" FSB).
>
> I recognize that you may not be able to provide buying advice; I'm
> just looking to see if I first, be able to use an Athlon 64 and
> second, if I will be hurting myself by going that route vs. an
> Athlon
> XP.
>
> Any thoughts would be appreciated.
>
> Regards,
>
> Keith Williams
>
> Gordon Weast wrote:
>>
>>
>> Adam,
>>
>> Unfortunately, we haven't been able to reproduce the results
that
>> we found for that one 3 Ghz P4 machine. We borrowed it from
our
> IT
>> department, ran the benchmark and then returned it. Now it's
in
>> use
>> by someone as their desktop machine. We're sure we got those
>> results,
>> but it seems like we have to send it to the journal of
>> irreproducible
>> results because we can't find another machine that performs as
>> well.
>>
>> It is likely that a machine that performs well for high end
> gaming
>> might also perform well with xPC Target in realtime use.
> Graphics
>> performance is irrelevant to xPC Target but CPU/Cache/memory
>> performance is important to both.
>>
>> I suppose if we got enough results that included CPU speed, FSB
>> speed
>> and chipset type, we might see some trends. At the present
time,
>> we
>> don't have that information. Do any current machines still use
>> that
>> configuration where they let the graphics board use part of
main
>> memory? I would expect that to kill performance by adding wait
>> states
>> when the CPU needs to fill a cache line.
>>
>> Gordon Weast
>> xPC Target Development
>> The Mathworks
>>
>> Adam Hersbach wrote:
>>
>>> Gordon,
>>>
>>> I have recently tested two Intel Pentium4 3.0 GHz CPU's
from
>>> different manufacturers (using .NET03 compiler). The
> benchmark
>> tests
>>> show a 20% difference in performance, and the best machine
>> performs
>>> at only 75% of what the Mathworks reports in the xpcbench
> test
>> for a
>>> P4 3.0 GHz machine.
>>>
>>> Through our experience with xPC target and recent support
> from
>> the
>>> Mathworks, the difference between the two machines that I
> have
>> tested
>>> is due to the chipset/motherboard used by the different
>>> manufacturers, not the CPU itself, as you have commented.
>>>
>>> The performance varies so markedly, that we cannot run our
>>> simulations on these machines. We need something that
> performs
> at
>>> least slightly better than the reported performance of the
> AMD
>> 2800+
>>> .NET03 (and if possible as good as the reported performance
> of
>> the P4
>>> 3.0 GHz .NET03).
>>>
>>> Can you suggest anything?
>>>
>>> Adam
>>>
>>> Gordon Weast wrote:
>>>
>>>>
>>>>If anyone does have some benchmark results for
processors,
> please
>>>>include the front side bus speed. That will make a
> difference.
>>>>I would expect the chipset that a given motherboard uses
> to
> make
>>>>some difference as well.
>>>>
>>>>There are just too many combinations for us to test all
of
> them
>>>>so input from users is important to us as well.
>>>>
>>>>Gordon Weast
>>>>xPC Target Development
>>>>The Mathworks
>>>>
>>>>Adam Hersbach wrote:
>>>>
>>>>
>>>>>Hi all,
>>>>>
>>>>>Does anyone have xpcbench data for any of the
> following
> CPU's:
>>>>>
>>>>>Athlon XP 3000+ (2.167 GHz)
>>>>>Athlon XP 3200+ (2.200 GHz)
>>>>>Athlon 64 4000+ (2.4 GHz)
>>>>>Athlon 64 FX-55 (2.6 GHz)
>>>>>Athlon 64 FX-57 (2.8 GHz)
>>>>>
>>>>>Intel Pentium4 > 3.0 GHz
>>>>>
>>>>>Would greatly appreciate this data as we are trying
to
> find a
>>>
>>> CPU
>>>
>>>>>which is as fast as possible.
>>>>>
>>>>>Cheers,
>>>>>Adam
>>>>
>>
.
- References:
- Re: xPC taget CPU - Athlon or Intel
- From: Keith Williams
- Re: xPC taget CPU - Athlon or Intel
- Prev by Date: Computable general equilibrium model
- Next by Date: Re: subtracting column vector from every column in matrix ...
- Previous by thread: Re: xPC taget CPU - Athlon or Intel
- Next by thread: c-gateways, fortran-sources, VC 6 - how to?
- Index(es):
Relevant Pages
|
