Athlon XP, memory, bios problem.

Dr. Doughnut -- That solved my problem, and thanks very much.

You sound like the person who could answer this additional question: I
noticed when I booted up that my memory was reported as 266 (rather than
333, which it is). Do I understand correctly that in order to take advantage
of the 333 speed, I need an Athlon XP with 333 as well? Which I think is the
2500 or 2600 models and above?

Again, many thanks for your prompt help.


"Doughnut" <doughnut DeleteThis @doughnut.com> wrote in message
news:400a10fa$0$9390$cc9e4d1f@news-text.dial.pipex.com...
 > set the fsb to 133 , it seems to be set at 100.
 >
 >
 >
 >
 > "Bruce" <bhight DeleteThis @REMOVETHISaustin.rr.com> wrote in message
 > news:B4nOb.6470$Md6.3405@fe2.texas.rr.com...
  > > I have an Athlon XP 1700 CPU with 512 mb of PC 2700 memory that ran at
1.4
  > > mhz just fine until today. That was what was originally installed and
 > that's
  > > what all system settings in Windows XP showed.
  > >
  > > Then, I installed a Turtle Beach Santa Cruz sound card and drivers
 > today --
  > > and it works fine. I'm listening to music from it now. No apparent
 > problems.
  > > Also, the computer still starts up fine and all programs are accessible.
 > I'm
  > > able to send this e-mail.
  > >
  > > However, this evening, upon returning home, I was confronted with a blue
  > > screen of death that said there was a new CPU. There wasn't. And another
  > > message, upon rebooting, said a secondary IDE cable was not connected. I
  > > doublechecked all connections and they seemed to be okay; that message
 > went
  > > away. But that didn't solve the other problem with the CPU.
  > >
  > > Now, according to my bios (Award) as I boot up, I have a simple Athlon
CPU
  > > that operates at 1.1 mhz. And my memory is given simply as "200."
  > >
  > > I've gone into my bios, which is supposed to detect the proper CPU
  > > automatically but obviously is not. I don't see any settings there that
  > > would allow me to choose an Athlon XP.
  > >
  > > I have an IWill XP333-R Ver. 2.1 motherboard which has performed
 > flawlessly.
  > > I have not upgraded the bios; it is the original. This is a no-name
brand
  > > computer that is two years old and worked great until now.
  > >
  > > Any suggestions on what the problem could be and what I should do about
 > it?
  > > Although I'm able to use my computer, it's not quite as perky as it
should
  > > be.
  > >
  > > Thanks in advance.
  > >
  > >
  > >
  > >
 >
 ><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Athlon XP, memory, bios problem. 

Bruce wrote:
 > noticed when I booted up that my memory was reported as 266 (rather than
 > 333, which it is). Do I understand correctly that in order to take
 > advantage of the 333 speed, I need an Athlon XP with 333 as well? Which I
 > think is the 2500 or 2600 models and above?

Not necessarily, you can probably set your memory at a ratio to your FSB,
it's probably set to 100% or 1/1 at the moment, you'll need 166MHz/133MHz
which is 5/4 or 125%.

As to whether that will actually improve or decrease performance depends on
many factors. Often it's best to set the FSB and RAM to run at the same
speed (known as synchronous).

Probably best to benchmark your system. The difficulty is choosing a
benchmark representative of what you generally do. There will be a huge
variation as to which combination is better (1/1 or 5/4) depending on the
relative memory and I/O operations performed in your "typical" application.

For example, a pure memory bandwidth benchmark program will basically use
the CPU to do data transfers to RAM which will saturate the FSB at 133MHz,
yet your memory may be 166MHz. A multimedia benchmark will take into
account DMA transfers from your I/O devices such as sound card and hard
drive, which may be able to saturate the memory bus at 166MHz.

You need to have an understanding of the architecture to understand this,
basically the Northbridge is the centre of all data transfers, not the CPU.
(at least on an Athlon XP, the Athlon 64s have an onboard memory
controller, so the architecture is a bit different). This means that for
CPU to read memory, it must request the data which will be sent from the
memory to the memory controller on the Northbridge, via the memory bus.
From the Northbridge, it will be sent to the CPU vie the FSB. So both bus
speeds are relevant, and you clearly cannot go faster then the slowest bus.
Converting from 133MHz to 166MHz (or vice-versa) may introduce additional
latency which will slow down the transfers... this would not be so if both
buses were 133MHz.

Conversely, many PCI (AGP is basically PCI) devices can do DMA transfers (a
form of bus mastering) which means that they transfer the data from the
device to the Northbridge (via the PCI bus) and then to memory via the
memory bus, the CPU does not get involved in this process... this is where
having memory running faster than the FSB is useful. Whether it outweighs
the additional latency introduced in converting time-bases is obviously a
complicated interaction of many factors.

Sorry, that was a bit technical... I'm bored.

Ben
--
A7N8X FAQ: <a style='text-decoration: underline;' href="http://www.ben.pope.name/a7n8x_faq.html" target="_blank">www.ben.pope.name/a7n8x_faq.html</a>
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Athlon XP, memory, bios problem. 

The memory will show as 266mhz as the cpu is running at that, as Ben says
you can run it asynchronous to keep the cpu at the correct speed and run
your memory at spec ie 333mhz. If you just want your system to run at the
proper cpu speed then id leave all alone but if you do want the memory upto
spec then as we say adjust it in the bios.

doughnut




"Ben Pope" <spam.DeleteThis@hotmail.com> wrote in message
news:budsjs$g1ple$1@ID-191149.news.uni-berlin.de...
 > Bruce wrote:
  > > noticed when I booted up that my memory was reported as 266 (rather than
  > > 333, which it is). Do I understand correctly that in order to take
  > > advantage of the 333 speed, I need an Athlon XP with 333 as well? Which
I
  > > think is the 2500 or 2600 models and above?
 >
 > Not necessarily, you can probably set your memory at a ratio to your FSB,
 > it's probably set to 100% or 1/1 at the moment, you'll need 166MHz/133MHz
 > which is 5/4 or 125%.
 >
 > As to whether that will actually improve or decrease performance depends
on
 > many factors. Often it's best to set the FSB and RAM to run at the same
 > speed (known as synchronous).
 >
 > Probably best to benchmark your system. The difficulty is choosing a
 > benchmark representative of what you generally do. There will be a huge
 > variation as to which combination is better (1/1 or 5/4) depending on the
 > relative memory and I/O operations performed in your "typical"
application.
 >
 > For example, a pure memory bandwidth benchmark program will basically use
 > the CPU to do data transfers to RAM which will saturate the FSB at 133MHz,
 > yet your memory may be 166MHz. A multimedia benchmark will take into
 > account DMA transfers from your I/O devices such as sound card and hard
 > drive, which may be able to saturate the memory bus at 166MHz.
 >
 > You need to have an understanding of the architecture to understand this,
 > basically the Northbridge is the centre of all data transfers, not the
CPU.
 > (at least on an Athlon XP, the Athlon 64s have an onboard memory
 > controller, so the architecture is a bit different). This means that for
 > CPU to read memory, it must request the data which will be sent from the
 > memory to the memory controller on the Northbridge, via the memory bus.
 > From the Northbridge, it will be sent to the CPU vie the FSB. So both bus
 > speeds are relevant, and you clearly cannot go faster then the slowest
bus.
 > Converting from 133MHz to 166MHz (or vice-versa) may introduce additional
 > latency which will slow down the transfers... this would not be so if both
 > buses were 133MHz.
 >
 > Conversely, many PCI (AGP is basically PCI) devices can do DMA transfers
(a
 > form of bus mastering) which means that they transfer the data from the
 > device to the Northbridge (via the PCI bus) and then to memory via the
 > memory bus, the CPU does not get involved in this process... this is where
 > having memory running faster than the FSB is useful. Whether it outweighs
 > the additional latency introduced in converting time-bases is obviously a
 > complicated interaction of many factors.
 >
 > Sorry, that was a bit technical... I'm bored.
 >
 > Ben
 > --
<font color=purple> > A7N8X FAQ: <a style='text-decoration: underline;' href="http://www.ben.pope.name/a7n8x_faq.html</font" target="_blank">www.ben.pope.name/a7n8x_faq.html</font</a>>
 > Questions by email will likely be ignored, please use the newsgroups.
 > I'm not just a number. To many, I'm known as a String...
 >
 ><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Athlon XP, memory, bios problem. 

Thank you both, Dr. Doughnut and Dr. Pope. I think I understand all this
much better now, though you know how dangerous a little knowledge can be. It
sounds like there's really not much point upping the memory speed to
asynchronous mode. But it also sounds like that, if I ever upgrade my
processor, I should get the faster Athlon XP series at 333 mhz.

I'm tempted to upgrade the CPU, although I probably don't need to (how often
do upgrades REALLY involve a "need to" versus just want to do it because
it's fun and a chance to learn?) But that raises another question that maybe
you can help answer: Isn't it quite ridiculous that, with all these
incredibly complex, high-performance CPUs, an upgrader is reduced to
slopping goo on them to make sure they don't burn up? I know it's absolutely
necessary because of the heat, but all the instructions I've read in
magazines and books for installing a new processor talk about the need to
put that goop on there, slap on the fan/heat sink and ensure the whole
contraption works right before going any further. Shouldn't all that be done
at the factory?

Anyway, thanks again to you both for your help on my problem and giving me
explanations.


"Doughnut" <doughnut RemoveThis @doughnut.com> wrote in message
news:400a9cbe$0$9385$cc9e4d1f@news-text.dial.pipex.com...
 > The memory will show as 266mhz as the cpu is running at that, as Ben says
 > you can run it asynchronous to keep the cpu at the correct speed and run
 > your memory at spec ie 333mhz. If you just want your system to run at the
 > proper cpu speed then id leave all alone but if you do want the memory
upto
 > spec then as we say adjust it in the bios.
 >
 > doughnut
 >
 >
 >
 >
 > "Ben Pope" <spam RemoveThis @hotmail.com> wrote in message
 > news:budsjs$g1ple$1@ID-191149.news.uni-berlin.de...
  > > Bruce wrote:
   > > > noticed when I booted up that my memory was reported as 266 (rather
than
   > > > 333, which it is). Do I understand correctly that in order to take
   > > > advantage of the 333 speed, I need an Athlon XP with 333 as well?
Which
 > I
   > > > think is the 2500 or 2600 models and above?
  > >
  > > Not necessarily, you can probably set your memory at a ratio to your
FSB,
  > > it's probably set to 100% or 1/1 at the moment, you'll need
166MHz/133MHz
  > > which is 5/4 or 125%.
  > >
  > > As to whether that will actually improve or decrease performance depends
 > on
  > > many factors. Often it's best to set the FSB and RAM to run at the same
  > > speed (known as synchronous).
  > >
  > > Probably best to benchmark your system. The difficulty is choosing a
  > > benchmark representative of what you generally do. There will be a huge
  > > variation as to which combination is better (1/1 or 5/4) depending on
the
  > > relative memory and I/O operations performed in your "typical"
 > application.
  > >
  > > For example, a pure memory bandwidth benchmark program will basically
use
  > > the CPU to do data transfers to RAM which will saturate the FSB at
133MHz,
  > > yet your memory may be 166MHz. A multimedia benchmark will take into
  > > account DMA transfers from your I/O devices such as sound card and hard
  > > drive, which may be able to saturate the memory bus at 166MHz.
  > >
  > > You need to have an understanding of the architecture to understand
this,
  > > basically the Northbridge is the centre of all data transfers, not the
 > CPU.
  > > (at least on an Athlon XP, the Athlon 64s have an onboard memory
  > > controller, so the architecture is a bit different). This means that
for
  > > CPU to read memory, it must request the data which will be sent from the
  > > memory to the memory controller on the Northbridge, via the memory bus.
  > > From the Northbridge, it will be sent to the CPU vie the FSB. So both
bus
  > > speeds are relevant, and you clearly cannot go faster then the slowest
 > bus.
  > > Converting from 133MHz to 166MHz (or vice-versa) may introduce
additional
  > > latency which will slow down the transfers... this would not be so if
both
  > > buses were 133MHz.
  > >
  > > Conversely, many PCI (AGP is basically PCI) devices can do DMA transfers
 > (a
  > > form of bus mastering) which means that they transfer the data from the
  > > device to the Northbridge (via the PCI bus) and then to memory via the
  > > memory bus, the CPU does not get involved in this process... this is
where
  > > having memory running faster than the FSB is useful. Whether it
outweighs
  > > the additional latency introduced in converting time-bases is obviously
a
  > > complicated interaction of many factors.
  > >
  > > Sorry, that was a bit technical... I'm bored.
  > >
  > > Ben
  > > --
<font color=green>  > > A7N8X FAQ: <a style='text-decoration: underline;' href="http://www.ben.pope.name/a7n8x_faq.html</font" target="_blank">www.ben.pope.name/a7n8x_faq.html</font</a>>
  > > Questions by email will likely be ignored, please use the newsgroups.
  > > I'm not just a number. To many, I'm known as a String...
  > >
  > >
 >
 ><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Athlon XP, memory, bios problem. 

Bruce wrote:
 > Isn't it quite ridiculous that, with all
 > these incredibly complex, high-performance CPUs, an upgrader is reduced to
 > slopping goo on them to make sure they don't burn up? I know it's
 > absolutely necessary because of the heat, but all the instructions I've
 > read in magazines and books for installing a new processor talk about the
 > need to put that goop on there, slap on the fan/heat sink and ensure the
 > whole contraption works right before going any further. Shouldn't all
 > that be done at the factory?

Wel, no. If they put the goo on in the factory it'd just fill with crap
before you put the heatsink on (which would defeat the point). If you buy a
retail cpu it's possible they could assemble the whole lot, but then, they
use thermal pads rather than goop (thermal paste) anyway, whioch is much
easier. The thermal paste is messy, but good if done correctly.

Ben
--
A7N8X FAQ: <a style='text-decoration: underline;' href="http://www.ben.pope.name/a7n8x_faq.html" target="_blank">www.ben.pope.name/a7n8x_faq.html</a>
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Athlon XP, memory, bios problem.