Archived from groups: alt.comp.hardware.homebuilt (
More info?)
"kony" <spam@spam.com> wrote in message
news:77b1t0tpmlqdb2n90o0lpnt4p6itug6u5n@4ax.com...
> On Mon, 27 Dec 2004 18:42:35 -0500, "Dave C."
> <mdupre@sff.net> wrote:
>
>>
>>"Ruel Smith" <NoWay@NoWhere.com> wrote in message
>>news:Ff0Ad.13367$IZ2.2587@fe37.usenetserver.com...
>>> Robert wrote:
>>>
>>>> currently have a 2100xp looking to put in a 3200*400barton would i see
>>>> any
>>>> performance gains or would i be better putting the money towards a
>>>> 64bit
>>>> upgrade later on.
>>>
>>> Does your motherboard even support the 3200+? Can it do the 400 MHz FSB?
>>> It
>>> might by overclocking, but can it handle the multiplier too?
>>>
>>> If the answers are all yes, then consider this: The 3200+ costs $200,
>>> roughly, online. Using the Drystone benchmark for Sandra, the 2100+ gets
>>> a
>>> score of about 7063. The 3200+, a score of 9250. Now, using math, the
>>> Barton core processor is about 30% faster (9250/7063).
>>>
>>>
http://www.tomshardware.com/cpu/20041221/cpu_charts-22.html
>>>
>>> Now, assuming that most other benchmarks are in the ballpark, is it
>>> worth
>>> $200 to you to get a 30% speed increase?
>>>
>>
>>Your argument is great, but it fails when you consider that the 30% speed
>>increase is for the CPU only. That does NOT equate to a 30% speed
>>increase.
>>At best, you are looking at maybe 10%. And that is AT BEST. The CPU is
>>an
>>important part of a computer system, but speeding up the CPU 30% does not
>>speed up your system 30%. -Dave
>>
>
>
> Both of your aruguments aren't quite right.
>
> Judging CPU performance by a synthetic benchmark like
> Sandra's CPU test, which is so bad that it will even allow a
> Celeron 1.0 GHz to appear faster than a P3 933, won't be
> telling of much.
>
> On the other hand, presuming that it was a 30% difference
> but that it was only CPU speed, isn't right either, as the
> FSB and memory bus would be higher too.
>
> It could easily exceed 25% performance gain on any if not
> all applications that are CPU performance-bound. On some
> tasks like SETI, it should be faster than 30% due to the
> larger L2 cache. However paying $200 for an end-of-life
> platform upgrade when it wouldn't cost much if any more to
> buy a new motherboard and Sempron 3100, is poor bang for
> buck. The best buy for a Barton is a XP2500 (or XP2400
> Mobile) then overclocking the motherboard past 200MHz FSB.
Thanks all will be going down the xp2500m route if i can find out if the
ga-7nnxp will be able to adjust enough to take advantage of it