While looking at motherboards on the web I ran across these Dual
Socket ones. I looked on wikipedia and it said a computer with two
CPUs. That's sort of what I expected, but what's the purpose of two?
Do they multiply the power? For example, if I installed two 1000mhz
CPUs would I now have 2000mhz, or would it just double the power or
the speed, or what? Are both CPUs working at the same time, or do
they take turns? I guess I dont really understand the purpose of two.
Also, does this mean that twice the power supply current and twice the
memory are needed too?
scotttews@noemail.com wrote:
> While looking at motherboards on the web I ran across these Dual
> Socket ones. I looked on wikipedia and it said a computer with two
> CPUs. That's sort of what I expected, but what's the purpose of two?
> Do they multiply the power? For example, if I installed two 1000mhz
> CPUs would I now have 2000mhz, or would it just double the power or
> the speed, or what? Are both CPUs working at the same time, or do
> they take turns? I guess I dont really understand the purpose of two.
>
> Also, does this mean that twice the power supply current and twice the
> memory are needed too?
>
> Scott
Computers with two sockets can multiply performance in two ways.
1) Start a program on one socket. Start a program on the other
socket. Both programs are running at the same time. You
get twice as much work done, in a given period of time.
You could shrink a DVD and use Winrar at the same time,
and have them finish in their normal time, even though
you are doing two things at once. If you had a single socket,
they would fight for a share of the time on the single
CPU.
2) Use a program like Photoshop. Work on an image. Press the
filter button. When you do, half the image is computed on
one socket and half the image is computed on the other
socket. Your filter operation happens in half the time.
These are ideal situations. There will be lots of situations
where the second socket isn't helping you at all. For example,
most of the software I own, is type (1). Hardly any of my software
is like Photoshop, the type (2) above. Much recent multimedia
software has been improved, to work like (2).
Using two sockets, uses twice the power of one socket (if the
processors are both being kept busy at the same time). If
one socket is idle, and the other one busy, the power might
be 30% more than with just one socket. An idle CPU still
draws power.
You use the normal amount of memory. That part doesn't
change.
Two sockets is faster. Under worst case conditions, it is
only 1x as fast. Under best case conditions, it is 2x as fast.
So it is not a "2x winner" all the time. Only some of the
time.
On 2007-05-18, Paul <nospam@needed.com> wrote:
>
> Two sockets is faster. Under worst case conditions, it is
> only 1x as fast. Under best case conditions, it is 2x as fast.
> So it is not a "2x winner" all the time. Only some of the
> time.
While your explanation of multiprocessing was generally very good
I would take issue with this last paragraph. There's always an
overhead when MP is involved. In the extreme case this can mean
a performance reduction over a single processor configuration
(unusual but it does happen). At the other end of the scale, you'll
never get twice the performance. With a good configuration and
appropriate software an 80-90% increase is a possibility but you'll
never double it.
Of course, the processor(s) is/are only one part of the overall
system in any case, and for many (most?) desktop users the processor
is idling the vast majority of the time waiting for you to do
something or waiting on the disk/network/whatever. MP in such
circumstances is of questionable value.
Andrew Smallshaw wrote:
> On 2007-05-18, Paul <nospam@needed.com> wrote:
>> Two sockets is faster. Under worst case conditions, it is
>> only 1x as fast. Under best case conditions, it is 2x as fast.
>> So it is not a "2x winner" all the time. Only some of the
>> time.
>
> While your explanation of multiprocessing was generally very good
> I would take issue with this last paragraph. There's always an
> overhead when MP is involved. In the extreme case this can mean
> a performance reduction over a single processor configuration
> (unusual but it does happen). At the other end of the scale, you'll
> never get twice the performance. With a good configuration and
> appropriate software an 80-90% increase is a possibility but you'll
> never double it.
>
> Of course, the processor(s) is/are only one part of the overall
> system in any case, and for many (most?) desktop users the processor
> is idling the vast majority of the time waiting for you to do
> something or waiting on the disk/network/whatever. MP in such
> circumstances is of questionable value.
>
It was an attempt at a broad brush answer. There
is always room for a little more detail :-)
On Fri, 18 May 2007 01:17:34 -0400, Paul <nospam@needed.com> wrote:
>scotttews@noemail.com wrote:
>> While looking at motherboards on the web I ran across these Dual
>> Socket ones. I looked on wikipedia and it said a computer with two
>> CPUs. That's sort of what I expected, but what's the purpose of two?
>> Do they multiply the power? For example, if I installed two 1000mhz
>> CPUs would I now have 2000mhz, or would it just double the power or
>> the speed, or what? Are both CPUs working at the same time, or do
>> they take turns? I guess I dont really understand the purpose of two.
>>
>> Also, does this mean that twice the power supply current and twice the
>> memory are needed too?
>>
>> Scott
>
>Computers with two sockets can multiply performance in two ways.
>
>1) Start a program on one socket. Start a program on the other
> socket. Both programs are running at the same time. You
> get twice as much work done, in a given period of time.
> You could shrink a DVD and use Winrar at the same time,
> and have them finish in their normal time, even though
> you are doing two things at once. If you had a single socket,
> they would fight for a share of the time on the single
> CPU.
>
>2) Use a program like Photoshop. Work on an image. Press the
> filter button. When you do, half the image is computed on
> one socket and half the image is computed on the other
> socket. Your filter operation happens in half the time.
>
>These are ideal situations. There will be lots of situations
>where the second socket isn't helping you at all. For example,
>most of the software I own, is type (1). Hardly any of my software
>is like Photoshop, the type (2) above. Much recent multimedia
>software has been improved, to work like (2).
>
>Using two sockets, uses twice the power of one socket (if the
>processors are both being kept busy at the same time). If
>one socket is idle, and the other one busy, the power might
>be 30% more than with just one socket. An idle CPU still
>draws power.
>
>You use the normal amount of memory. That part doesn't
>change.
>
>Two sockets is faster. Under worst case conditions, it is
>only 1x as fast. Under best case conditions, it is 2x as fast.
>So it is not a "2x winner" all the time. Only some of the
>time.
>
> Paul
Thanks for your detailed help.
Much Appreciated !!!!
In message <slrnf4rkd8.5bj.andrews@sdf.lonestar.org> Andrew Smallshaw
<andrews@sdf.lonestar.org> wrote:
>Of course, the processor(s) is/are only one part of the overall
>system in any case, and for many (most?) desktop users the processor
>is idling the vast majority of the time waiting for you to do
>something or waiting on the disk/network/whatever. MP in such
>circumstances is of questionable value.
While your explanation of multiprocessing was generally very good
I would take issue with this last paragraph.
Even though my CPUs/cores sit idle most of the time, when I do have one
of my CPUs doing something, it tends to be something both cores can do
at once. Perhaps that's just me?
--
If quitters never win, and winners never quit,
what fool came up with, "Quit while you're ahead"?
On 2007-05-19, DevilsPGD <spam_narf_spam@crazyhat.net> wrote:
> In message <slrnf4rkd8.5bj.andrews@sdf.lonestar.org> Andrew Smallshaw
><andrews@sdf.lonestar.org> wrote:
>
>>Of course, the processor(s) is/are only one part of the overall
>>system in any case, and for many (most?) desktop users the processor
>>is idling the vast majority of the time waiting for you to do
>>something or waiting on the disk/network/whatever. MP in such
>>circumstances is of questionable value.
>
> Even though my CPUs/cores sit idle most of the time, when I do have one
> of my CPUs doing something, it tends to be something both cores can do
> at once. Perhaps that's just me?
No, it's not just you. It's just that many users have no need for
even a dual-core system since a lot of software isn't gered up to
deal with them yet. As dual-core systems become common this will
probably change. Don't forget that many users do little more than
run Office and a web browser all day.
Speaking personally, I rarely hit 100% CPU usage. Even a large
compilation usually only hits ~70% CPU utilisation. The rest of
the time it's disk bound. If I played games all day the situation
might be different, but I'm not a gamer. I consider myself a power
user so how many people out there must there be that don't even
use their resources as much as me? I wasn't saying that there's
no use for MP, jsut that many users will make limited use of it.
It's a question of balancing how much use will be made of a second
CPU vs its additional cost.
In short, it comes down to the standard question when specifying
a new computer. What are you using it for?
On Sat, 19 May 2007 14:42:44 +0000 (UTC), Andrew Smallshaw
<andrews@sdf.lonestar.org> wrote:
>On 2007-05-19, DevilsPGD <spam_narf_spam@crazyhat.net> wrote:
>> In message <slrnf4rkd8.5bj.andrews@sdf.lonestar.org> Andrew Smallshaw
>><andrews@sdf.lonestar.org> wrote:
>>
>>>Of course, the processor(s) is/are only one part of the overall
>>>system in any case, and for many (most?) desktop users the processor
>>>is idling the vast majority of the time waiting for you to do
>>>something or waiting on the disk/network/whatever. MP in such
>>>circumstances is of questionable value.
>>
>> Even though my CPUs/cores sit idle most of the time, when I do have one
>> of my CPUs doing something, it tends to be something both cores can do
>> at once. Perhaps that's just me?
>
>No, it's not just you. It's just that many users have no need for
>even a dual-core system since a lot of software isn't gered up to
>deal with them yet. As dual-core systems become common this will
>probably change. Don't forget that many users do little more than
>run Office and a web browser all day.
>
>Speaking personally, I rarely hit 100% CPU usage. Even a large
>compilation usually only hits ~70% CPU utilisation. The rest of
>the time it's disk bound. If I played games all day the situation
>might be different, but I'm not a gamer. I consider myself a power
>user so how many people out there must there be that don't even
>use their resources as much as me? I wasn't saying that there's
>no use for MP, jsut that many users will make limited use of it.
>It's a question of balancing how much use will be made of a second
>CPU vs its additional cost.
>
>In short, it comes down to the standard question when specifying
>a new computer. What are you using it for?
I have an Althon 2500+ and I will be upgrading to Dual Core soon.
At present I run MS Excel with speadsheets around 28mb in size and
they run the CPU @100% for best part of 2 mins to open and close.
While doing that I have 50 web pages open in Firefox, Email Progarm
(not Outlook) & Agent Newsreader.
I am expecting that a dual core will show considerable improvement,
particularly while opening and closing Excel worksheets.
On Sun, 20 May 2007 12:58:26 +1000, DJT
<dtope@hotmail.com.au> wrote:
>I have an Althon 2500+ and I will be upgrading to Dual Core soon.
> At present I run MS Excel with speadsheets around 28mb in size and
>they run the CPU @100% for best part of 2 mins to open and close.
>While doing that I have 50 web pages open in Firefox, Email Progarm
>(not Outlook) & Agent Newsreader.
>
>I am expecting that a dual core will show considerable improvement,
>particularly while opening and closing Excel worksheets.
I'm suspecting you don't have a CPU bottleneck causing your
result, perhaps insufficient memory?
Try rebooting the system then using it typically, and after
running such a job and closing it when you see these long
lags, you should open Task Manager and compare the "Commit
Charge", "Peak" value to the "Physical Memory", "Total"
value. The Peak should be at least a couple hundred MB
below the Total for good performance (even more than a
couple hundred really, considering the level of
multitasking).
On Sat, 19 May 2007 23:56:55 -0400, kony <spam@spam.com> wrote:
>On Sun, 20 May 2007 12:58:26 +1000, DJT
><dtope@hotmail.com.au> wrote:
>
>
>>I have an Althon 2500+ and I will be upgrading to Dual Core soon.
>> At present I run MS Excel with speadsheets around 28mb in size and
>>they run the CPU @100% for best part of 2 mins to open and close.
>>While doing that I have 50 web pages open in Firefox, Email Progarm
>>(not Outlook) & Agent Newsreader.
>>
>>I am expecting that a dual core will show considerable improvement,
>>particularly while opening and closing Excel worksheets.
>
>I'm suspecting you don't have a CPU bottleneck causing your
>result, perhaps insufficient memory?
>
>Try rebooting the system then using it typically, and after
>running such a job and closing it when you see these long
>lags, you should open Task Manager and compare the "Commit
>Charge", "Peak" value to the "Physical Memory", "Total"
>value. The Peak should be at least a couple hundred MB
>below the Total for good performance (even more than a
>couple hundred really, considering the level of
>multitasking).
Thanks for the comments.
I have 1gig Memory installed
Commit Charge Peak is 960644K
Physical Memory 1048048k
This less than 200mb free. Do you think I lack memory even with 1 gig
installed
What's the purpose of a Dual Socket Mobo? 
Linear Mode
