23 Jul
2004
23 Jul
'04
12:23 a.m.
->Sounds like you have ver. 1.* of the A7N8X. I don't know much
about them. You might want to follow the guide on the RME website.<-
Actually, I have a DFI board.
M
23 Jul
23 Jul
3:23 a.m.
New subject: [linux-audio-user] CPU clock - beware
Matthew Barber wrote:
->Sounds like you have ver. 1.* of the A7N8X. I
don't know much
about them. You might want to follow the guide on the RME website.<-
Actually, I have a DFI board.
M
Sorry, I got people you confused with someone else.
Rick B
12:23 p.m.
New subject: [linux-audio-user] CPU clock - beware
Hmm, another very useful thread. It prompted me to have a look at my BIOS,
which I'm generally a bit nervous about looking at too closely.
I'm using an AOpen MX3W motherboard with a Celeron Coppermine 600MHz
processor. I wondered if the CPU speed was set rather low at 66.8 x 3.5, so I
upped the multiplier to 8 (max) to bring the result to 566, which I think
works with having an FSB of 100MHz. My system seems perfectly happy so far,
in fact I'd almost swear it was running cooler, but I've not checked. I don't
think the increase in speed of the delivery of graphics to the screen is my
imagination. Sorry, that's a bit of a fluffy explanation. If I'm obviously
heading for trouble here, I'd appreciate the warning.
In any case thanks for the discussion. From here, I think with a faster Hard
Drive and some more memory I'm going to feel like I've got a whole new
machine!
Thanks again.
tim hall
5:45 p.m.
New subject: [linux-audio-user] CPU clock - beware
tim hall wrote:
Hmm, another very useful thread. It prompted me to have
a look at my BIOS,
which I'm generally a bit nervous about looking at too closely.
I'm using an AOpen MX3W motherboard with a Celeron Coppermine 600MHz
processor. I wondered if the CPU speed was set rather low at 66.8 x 3.5, so I
upped the multiplier to 8 (max) to bring the result to 566, which I think
works with having an FSB of 100MHz. My system seems perfectly happy so far,
in fact I'd almost swear it was running cooler, but I've not checked. I don't
think the increase in speed of the delivery of graphics to the screen is my
imagination. Sorry, that's a bit of a fluffy explanation. If I'm obviously
heading for trouble here, I'd appreciate the warning.
In any case thanks for the discussion. From here, I think with a faster Hard
Drive and some more memory I'm going to feel like I've got a whole new
machine!
Thanks again.
tim hall
Yes, I would think that you would notice a big difference. If
before adjusting you had a 66.8mhz fsb x 3.5 multiplier = 233.8mhz and
after you have a 66.8mhz fsb x 8.0 = 566mhz. You might want to check and
see if you have a fan on your cpu (I have a 566mhz that doesn't), if not
and you do notice instability adding one would probably fix it. Also if
you dont have a fan now, adding one for $10 is a cheap way to overclock
it. I'm wondering though, where does the 100mhz fsb you mentioned come
in? You will either have a 66.8mhz fsb times a multiplier or a 100mhz
fsb times a multiplier
As far as memory goes, I think most people that read this list
would agree that you can never have to much. When you run out of
physical ram and have to swap to disk it slows the machine waaayy down
(and causes dropouts).
Rick B
8:57 p.m.
New subject: [linux-audio-user] CPU clock - beware
Last Friday 23 July 2004 22:32, Rick B was like:
Yes, I would think that you would notice a big
difference. If
before adjusting you had a 66.8mhz fsb x 3.5 multiplier = 233.8mhz and
after you have a 66.8mhz fsb x 8.0 = 566mhz.
Great, thanks for confirming that.
You might want to check and
see if you have a fan on your cpu (I have a 566mhz that doesn't), if not
and you do notice instability adding one would probably fix it. Also if
you dont have a fan now, adding one for $10 is a cheap way to overclock
it.
I do have a fan on it, although I could probably fit a bigger quieter one in.
I'm not interested in overclocking as such, just getting the straightforward
best performance.
I'm wondering though, where does the 100mhz fsb
you mentioned come
in? You will either have a 66.8mhz fsb times a multiplier or a 100mhz
fsb times a multiplier
These fluffy figures probably come from my misunderstanding of the manual.
CPU Freq: 600MHz
CPU Ratio: 6x (?)
CPU FSB: 100MHz
It also says: Warning: Intel 810/810DC-100 chipset supports a maximum of
100MHz CPU FSB blah ...
I think from the jumper settings that the display cache runs at the same
speed, there's nothing to suggest there's any value in pushing it.
So far I'm not seeing any instability, so I'll stick with this and see how it
goes.
As far as memory goes, I think most people
that read this list
would agree that you can never have to much. When you run out of
physical ram and have to swap to disk it slows the machine waaayy down
(and causes dropouts).
And dropouts, right. This is now my biggest (but not insurmountable) problem,
which can obviously be fixed by throwing some cash at it when I get some.
That's a Win AFAIC :-)
Thanks for increasing my understanding of this subject.
tim hall
10:42 p.m.
New subject: [linux-audio-user] CPU clock - beware
Ah,
I just re-read your mail and then re-read my motherboard manual.
So I'm running with an effective fsb of 66.8MHz [penny]
That's what I get with the default settings of JP29 & JP23 on the motherboard
[duh] If I was to jumper them at, say 5-6 & 3-4 respectively, this would give
me host clock 3x (100 - 124MHz) [er] which I assume would then show up in the
BIOS? [oh] could I then set it to the specified 100MHz x 6 ?:-]
I assume that's what my system would prefer.
[ps] One day I might get the hang of this hardware stuff. [clunk]
I realise this is probably elementary stuff for most of you.
It's a flippin' revelation for me :-)
Little note for Steve Harris:
If I'm right, this means I've been successfully running JAMin on a severely
_underclocked_ Celeron 600, so stick that in your minimum system
requirements ;-]
thanks for your patience & assistance.
tim hall
Last Saturday 24 July 2004 01:42, I went:
Last Friday 23 July 2004 22:32, Rick B was like:
Yes, I would think that you would notice a
big difference. If
before adjusting you had a 66.8mhz fsb x 3.5 multiplier = 233.8mhz and
after you have a 66.8mhz fsb x 8.0 = 566mhz.
Great, thanks for confirming that.
You might want to check and
see if you have a fan on your cpu (I have a 566mhz that doesn't), if not
and you do notice instability adding one would probably fix it. Also if
you dont have a fan now, adding one for $10 is a cheap way to overclock
it.
I do have a fan on it, although I could probably fit a bigger quieter one
in. I'm not interested in overclocking as such, just getting the
straightforward best performance.
I'm wondering though, where does the 100mhz
fsb you mentioned come
in? You will either have a 66.8mhz fsb times a multiplier or a 100mhz
fsb times a multiplier
These fluffy figures probably come from my misunderstanding of the manual.
CPU Freq: 600MHz
CPU Ratio: 6x (?)
CPU FSB: 100MHz
It also says: Warning: Intel 810/810DC-100 chipset supports a maximum of
100MHz CPU FSB blah ...
I think from the jumper settings that the display cache runs at the same
speed, there's nothing to suggest there's any value in pushing it.
So far I'm not seeing any instability, so I'll stick with this and see how
it goes.
As far as memory goes, I think most
people that read this list
would agree that you can never have to much. When you run out of
physical ram and have to swap to disk it slows the machine waaayy down
(and causes dropouts).
And dropouts, right. This is now my biggest (but not insurmountable)
problem, which can obviously be fixed by throwing some cash at it when I
get some. That's a Win AFAIC :-)
Thanks for increasing my understanding of this subject.
tim hall 
24 Jul
24 Jul
12:32 a.m.
New subject: [linux-audio-user] CPU clock - beware
On Fri, 2004-07-23 at 19:27, tim hall wrote:
Ah,
I just re-read your mail and then re-read my motherboard manual.
So I'm running with an effective fsb of 66.8MHz [penny]
That's what I get with the default settings of JP29 & JP23 on the motherboard
[duh] If I was to jumper them at, say 5-6 & 3-4 respectively, this would give
me host clock 3x (100 - 124MHz) [er] which I assume would then show up in the
BIOS? [oh] could I then set it to the specified 100MHz x 6 ?:-]
I assume that's what my system would prefer.
[ps] One day I might get the hang of this hardware stuff. [clunk]
I realise this is probably elementary stuff for most of you.
It's a flippin' revelation for me :-)
Little note for Steve Harris:
If I'm right, this means I've been successfully running JAMin on a severely
_underclocked_ Celeron 600, so stick that in your minimum system
requirements ;-]
Wow! I luv JAMin. But it makes my Athalon XP 2500 with 512MB RAM break a
sweat when I run it with Ardour. :)
R~
thanks for your patience & assistance.
tim hall
Last Saturday 24 July 2004 01:42, I went:
Last Friday 23 July 2004 22:32, Rick B was like:
Yes, I would think that you would notice a
big difference. If
before adjusting you had a 66.8mhz fsb x 3.5 multiplier = 233.8mhz and
after you have a 66.8mhz fsb x 8.0 = 566mhz.
Great, thanks for confirming that.
You might want to check and
see if you have a fan on your cpu (I have a 566mhz that doesn't), if not
and you do notice instability adding one would probably fix it. Also if
you dont have a fan now, adding one for $10 is a cheap way to overclock
it.
I do have a fan on it, although I could probably fit a bigger quieter one
in. I'm not interested in overclocking as such, just getting the
straightforward best performance.
I'm wondering though, where does the 100mhz
fsb you mentioned come
in? You will either have a 66.8mhz fsb times a multiplier or a 100mhz
fsb times a multiplier
These fluffy figures probably come from my misunderstanding of the manual.
CPU Freq: 600MHz
CPU Ratio: 6x (?)
CPU FSB: 100MHz
It also says: Warning: Intel 810/810DC-100 chipset supports a maximum of
100MHz CPU FSB blah ...
I think from the jumper settings that the display cache runs at the same
speed, there's nothing to suggest there's any value in pushing it.
So far I'm not seeing any instability, so I'll stick with this and see how
it goes.
As far as memory goes, I think most
people that read this list
would agree that you can never have to much. When you run out of
physical ram and have to swap to disk it slows the machine waaayy down
(and causes dropouts).
And dropouts, right. This is now my biggest (but not insurmountable)
problem, which can obviously be fixed by throwing some cash at it when I
get some. That's a Win AFAIC :-)
Thanks for increasing my understanding of this subject.
tim hall
8:41 a.m.
New subject: [linux-audio-user] CPU clock - beware
Last Saturday 24 July 2004 05:17, Russell Hanaghan was like:
Wow! I luv JAMin. But it makes my Athalon XP 2500 with
512MB RAM break a
sweat when I run it with Ardour. :)
Ah yes, I should point out that's xrun free when I run it on its own (+JACK)
to process external sources (in 192MB RAM). So far I've only managed to use
it in conjunction with ReZound, glitch free. Even ecasound balked at it
previously. I doubt very much that I could run it with Ardour. Perhaps I'll
have more luck now.
If not, then I really will have to get some more memory ;-)
cheers
tim hall
6:41 a.m.
New subject: [linux-audio-user] CPU clock - beware
tim hall schrieb:
So I'm running with an effective fsb of 66.8MHz
[penny]
That's what I get with the default settings of JP29 & JP23 on the motherboard
[duh] If I was to jumper them at, say 5-6 & 3-4 respectively, this would give
me host clock 3x (100 - 124MHz) [er] which I assume would then show up in the
BIOS? [oh] could I then set it to the specified 100MHz x 6 ?:-]
uhm ... be careful processors have a specific fsb that they are
designed for. For early Celerons like yours that's 66 mhz.
And you could run into problems with your RAM if it only supports 66mhz.
I looked through your manual and I think you can't really go any higher
without doing anything that could be considered overclocking *shrug*.
You could for example up your fsb to 75 mhz but your memory and
everything else may not like it. Don't blame me if you blow up your
computer. :D
I've found an interesting table that says that they don't even recommend
running your cpu on that board. :) (probably just means that you won't
get full speed out of it)
http://english.aopen.com.tw/tech/report/cpuref/skt370.htm
hth
8:36 a.m.
New subject: [linux-audio-user] CPU clock - beware
Last Saturday 24 July 2004 11:27, Stefan Scheffler was like:
uhm ... be careful processors have a specific fsb
that they are
designed for. For early Celerons like yours that's 66 mhz.
And you could run into problems with your RAM if it only supports 66mhz.
Thanks for the warning and the link.
I'm as sure as can be that the processor is supported at 100 FSB - i.e. this
is stated quite clearly in the hardcopy manual. I'm that nervous about
fiddling with hardware that I wouldn't bother with a move like this unless
there were strong indications that this is _the_ right setting.
I looked through your manual and I think you can't
really go any higher
without doing anything that could be considered overclocking *shrug*.
You could for example up your fsb to 75 mhz but your memory and
everything else may not like it. Don't blame me if you blow up your
computer. :D
Well, I'm running at 100MHz x 6 right now.
I take full responsibility for my actions.
Are there any signs I should watch out for?
I've found an interesting table that says that
they don't even recommend
running your cpu on that board. :) (probably just means that you won't
get full speed out of it)
I think it's probably more a question of me explaining badly. I think I've
just set it to the maximum I can get with this mobo & processor. The one
weirdness is that the processor is now being detected as 900MHz, which is not
true. Perhaps the 1800 bogomips reading is, I don't know.
Well, I shall now put it through its paces.
Thanks for the perspective, it's all good learning :-)
tim hall
8:50 a.m.
New subject: [linux-audio-user] CPU clock - beware
On Sat, 2004-07-24 at 13:21, tim hall wrote:
Well, I'm running at 100MHz x 6 right now.
I take full responsibility for my actions.
Are there any signs I should watch out for?
if at all possible, you should run memtest86 (or similar) for a few
hours to ensure the memory performance hasn't suffered. i keep a
memtest86 on my grub boot menu at all times.
the poor man's test would be to stress the hell outta your box with lots
of heavy lifting - perhaps recompiling the kernel over, and over, and
over again to see if strange gcc errors occur. such strange errors are
usually an indication of an unhappy cpu or memory.
you'd probably want to shell script the repetition and go out for a few
hours:)
good luck
R
9:41 p.m.
New subject: [linux-audio-user] CPU clock - beware
Last Saturday 24 July 2004 13:40, robin fell was like:
On Sat, 2004-07-24 at 13:21, tim hall wrote:
Well, memtest86 sounds easy enough.
Thanks.
tim hall
Well, I'm running at 100MHz x 6 right now.
I take full responsibility for my actions.
Are there any signs I should watch out for?
if at all possible, you should run memtest86 (or similar) for a few
hours to ensure the memory performance hasn't suffered. i keep a
memtest86 on my grub boot menu at all times.
the poor man's test would be to stress the hell outta your box with lots
of heavy lifting - perhaps recompiling the kernel over, and over, and
over again to see if strange gcc errors occur. such strange errors are
usually an indication of an unhappy cpu or memory. 
9:52 p.m.
New subject: [linux-audio-user] CPU clock - beware
tim hall wrote:
Last Saturday 24 July 2004 13:40, robin fell was
like:
cpuburn was also mentioned and it's easy too (and generally more
intensive than compiling a kernel).
cheers,
chris
On Sat, 2004-07-24 at 13:21, tim hall wrote:
Well, memtest86 sounds easy enough. Well, I'm running at 100MHz x 6 right now.
I take full responsibility for my actions.
Are there any signs I should watch out for?
if at all possible, you should run memtest86 (or similar) for a few
hours to ensure the memory performance hasn't suffered. i keep a
memtest86 on my grub boot menu at all times.
the poor man's test would be to stress the hell outta your box with lots
of heavy lifting - perhaps recompiling the kernel over, and over, and
over again to see if strange gcc errors occur. such strange errors are
usually an indication of an unhappy cpu or memory.
10:26 a.m.
New subject: [linux-audio-user] CPU clock - beware
tim hall schrieb:
I'm as sure as can be that the processor is supported at 100 FSB - i.e. this
is stated quite clearly in the hardcopy manual. I'm that nervous about
fiddling with hardware that I wouldn't bother with a move like this unless
there were strong indications that this is _the_ right setting.
hmm ... are you sure about the type of processor? AFAIK only Celerons
+700 mhz support a 100mz fsb.
Intel has a utility to check cpu information bus speed and so on ..
sadly it's windows/dos only:
http://downloadfinder.intel.com/scripts-df/Product_Filter.asp?ProductID=441
Well, I'm running at 100MHz x 6 right now.
I take full responsibility for my actions.
Are there any signs I should watch out for?
freezes, crashes(segfaults usually), screen artifacts, corruption all
that interesting stuff. :)
another neat utility for stress testing is cpuburn you can get it here:
http://www.ibiblio.org/pub/Linux/system/hardware/
or as a simple benchmark glxgears in software mode.
I think it's probably more a question of me explaining badly. I think I've
just set it to the maximum I can get with this mobo & processor. The one
weirdness is that the processor is now being detected as 900MHz, which is not
true. Perhaps the 1800 bogomips reading is, I don't know.
now that is uh ... interesting ... so within two days you like almost
quadrupled your processor speed .. congratulations :D
could you post the output of "cat /proc/cpuinfo"? I'd love to see that. :)
just FYI:
"model name :" should tell you the name and aometimes the design
speed of the cpu
"cpu MHz" the real current speed
"bogomips" issome weird timing value I have no idea what it means exactly
10:28 p.m.
New subject: [linux-audio-user] CPU clock - beware
Last Saturday 24 July 2004 15:12, Stefan Scheffler was like:
Yeah, basically I suspect I've been driving with the handbrake on :-]
hmm ... are you sure about the type of processor?
AFAIK only Celerons
+700 mhz support a 100mz fsb.
I'm just reading the motherboard manual.
Intel has a utility to check cpu information bus speed
and so on ..
sadly it's windows/dos only:
http://downloadfinder.intel.com/scripts-df/Product_Filter.asp?ProductID=441
Won't be much use to me then ;-)
Thanks for the other suggestions tho'
I think
it's probably more a question of me explaining badly. I think I've
just set it to the maximum I can get with this mobo & processor. The one
weirdness is that the processor is now being detected as 900MHz, which is
not true. Perhaps the 1800 bogomips reading is, I don't know.
now that is uh ... interesting ... so within two days you like almost
quadrupled your processor speed .. congratulations :D could you post the output of "cat
/proc/cpuinfo"? I'd love to see that. :)
sure:
~$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 8
model name : Celeron (Coppermine)
stepping : 6
cpu MHz : 896.977
cache size : 128 KB
fdiv_bug : no
hlt_bug : yes
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
cmov pat pse36 mmx fxsr sse
bogomips : 1789.13
just FYI:
"model name :" should tell you the name and aometimes the design
speed of the cpu
"cpu MHz" the real current speed
"bogomips" issome weird timing value I have no idea what it means exactly
It's a semi-arbitrary benchmarking value, usually twice the processor speed.
AFAIU.
Last Saturday 24 July 2004 16:33, Matthew Barber was like:
Be careful doing something like that, especially with
old boards. I
think setting the fsb clock to "no man's land" will set the pci clock
(and AGP if you have AGP on your board) to something unusable, unless
BIOS locks the PCI/AGP clock to a certain range of values. PCI
generally wants to run at about 33Mhz (unless you have a very new board
with PCI-X or some such), and AGP at 66Mhz, and these values will
generally be a fraction of the fsb. So if your fsb is 66Mhz, PCI will
be 1/2FSB. If it's 100Mhz, PCI will be 1/3. Setting it to 75Mhz may
cause it to still be in the 66Mhz realm as far as the division is
concerned, and set PCI to around 38Mhz, which may cause a lot of
problems. I know some BIOS will take care of this by locking AGP and
PCI to a certain value, but I wouldn't count on it with an older
board/bios.
This is what I guessed, I'm using the values specified in the manual for a
Coppermine FC-PGA 600, so:
at 3x clock ratio I get:
CPU (I assume this is FSB) 100
PCI 33
AGP 66
The display cache runs at 100MHz too, a 1:1 ratio seems logical.
One of the reasons I only have 192MB of RAM is because I have already been
fussy enough only to use memory that is supposed to run at 100MHz too.
No-man's land FSBs are only relevant if overclocking, which is not the object
of the exercise for me. I'm trying to optimise.
I don't think I'm doing _too_ badly.
Last Saturday 24 July 2004 17:49, Ryan Underwood was like:
Where? Is there an online copy of this manual?
I found one here:
http://www.motherboards.org/mobot/manuals/All/0/50/
http://english.aopen.com.tw/products/mb/MX3WPro-V.htm
I think I might read it now. I'm a bit confused as to why my system thinks
it's running a 896MHz CPU, But so far there's no overt signs that it isn't
happy, so I'll prod it a bit and see ;-)
cheers
tim hall
11:06 p.m.
New subject: [linux-audio-user] CPU clock - beware
tim hall wrote:
Last Saturday 24 July 2004 15:12, Stefan Scheffler was
like:
Yeah, basically I suspect I've been driving with the handbrake on :-]
Yeah that's close enough to 900mhz for me. Some cpu's are noted as being
good overclockers, but, hell, that's a free 50% overclock! The new sweet
spot for OCer's is the mobile athlon 2500/2600 (~$90US), out of the box
they are ~1800 to 1900mhz and are easily hitting 2500mhz on air cooling
with just a little extra voltage. The Pentium 2.4b's were easily hitting
3Ghz.
Rick B
hmm ... are you sure about the type of processor?
AFAIK only Celerons
+700 mhz support a 100mz fsb.
I'm just reading the motherboard manual.
Intel has a utility to check cpu information bus
speed and so on ..
sadly it's windows/dos only:
http://downloadfinder.intel.com/scripts-df/Product_Filter.asp?ProductID=441
Won't be much use to me then ;-)
Thanks for the other suggestions tho'
I think
it's probably more a question of me explaining badly. I think I've
just set it to the maximum I can get with this mobo & processor. The one
weirdness is that the processor is now being detected as 900MHz, which is
not true. Perhaps the 1800 bogomips reading is, I don't know.
now that is uh ... interesting ... so within two days you like almost
quadrupled your processor speed .. congratulations :D
could you post the output of "cat
/proc/cpuinfo"? I'd love to see that. :)
sure:
~$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 8
model name : Celeron (Coppermine)
stepping : 6
cpu MHz : 896.977
cache size : 128 KB
fdiv_bug : no
hlt_bug : yes
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
cmov pat pse36 mmx fxsr sse
bogomips : 1789.13
just FYI:
"model name :" should tell you the name and aometimes the design
speed of the cpu
"cpu MHz" the real current speed
"bogomips" issome weird timing value I have no idea what it means exactly
It's a semi-arbitrary benchmarking value, usually twice the processor speed.
AFAIU.
Last Saturday 24 July 2004 16:33, Matthew Barber was like:
Be careful doing something like that, especially
with old boards. I
think setting the fsb clock to "no man's land" will set the pci clock
(and AGP if you have AGP on your board) to something unusable, unless
BIOS locks the PCI/AGP clock to a certain range of values. PCI
generally wants to run at about 33Mhz (unless you have a very new board
with PCI-X or some such), and AGP at 66Mhz, and these values will
generally be a fraction of the fsb. So if your fsb is 66Mhz, PCI will
be 1/2FSB. If it's 100Mhz, PCI will be 1/3. Setting it to 75Mhz may
cause it to still be in the 66Mhz realm as far as the division is
concerned, and set PCI to around 38Mhz, which may cause a lot of
problems. I know some BIOS will take care of this by locking AGP and
PCI to a certain value, but I wouldn't count on it with an older
board/bios.
This is what I guessed, I'm using the values specified in the manual for a
Coppermine FC-PGA 600, so:
at 3x clock ratio I get:
CPU (I assume this is FSB) 100
PCI 33
AGP 66
The display cache runs at 100MHz too, a 1:1 ratio seems logical.
One of the reasons I only have 192MB of RAM is because I have already been
fussy enough only to use memory that is supposed to run at 100MHz too.
No-man's land FSBs are only relevant if overclocking, which is not the object
of the exercise for me. I'm trying to optimise.
I don't think I'm doing _too_ badly.
Last Saturday 24 July 2004 17:49, Ryan Underwood was like:
Where? Is there an online copy of this manual?
I found one here:
http://www.motherboards.org/mobot/manuals/All/0/50/
http://english.aopen.com.tw/products/mb/MX3WPro-V.htm
I think I might read it now. I'm a bit confused as to why my system thinks
it's running a 896MHz CPU, But so far there's no overt signs that it isn't
happy, so I'll prod it a bit and see ;-)
cheers
tim hall
25 Jul
25 Jul
8:49 a.m.
New subject: [linux-audio-user] CPU clock - Coppermining
Last Sunday 25 July 2004 03:53, Rick B was like:
>I think I might read it now. I'm a bit
confused as to why my system thinks
>it's running a 896MHz CPU, But so far there's no overt signs that it
isn't
>happy, so I'll prod it a bit and see ;-)
Yeah that's close enough to 900mhz for me. Some
cpu's are noted as being
good overclockers, but, hell, that's a free 50% overclock! The new sweet
spot for OCer's is the mobile athlon 2500/2600 (~$90US), out of the box
they are ~1800 to 1900mhz and are easily hitting 2500mhz on air cooling
with just a little extra voltage. The Pentium 2.4b's were easily hitting
3Ghz.
From what I've just read it seems the Celeron
Coppermines may not be rated
performance-wise, but they overclock to 900MHz like it
was a perfectly
natural thing to do. If my clock ratio is fixed to 9x, this would explain.
It's running a little warmer than at 66.8MHz x 8 = 566, (? clock ratio) but
nothing you could cook on. I've installed cpuburn to my HD, really I want to
run it from a floppy, like memtest86. Just in case ;-)
AFAICT, so long as I keep it cool enough, it could well be stable at this
setting. TBH I'd say it's actually performing like a well tuned 600. Which is
what I want. We will see :-] I think it helps that I'm using harmonic ratios
based on the recommended defaults.
How hot is too hot? I've got my warning level set to 50deg in the BIOS and I
still don't think I'm running under 40. It's hard to tell without rebooting,
unfortunately although I have both the hardware and software for i2c & hwmon
they're not talking at present. Again, it's something I never really bothered
with before. Most of the heat is coming off the power supply.
What started out as an idle comment has turned into a bit of a journey of
discovery! Thanks for all the comments and suggestions.
cheers
tim hall
7:22 p.m.
New subject: [linux-audio-user] CPU clock - Coppermining
tim hall wrote:
Last Sunday 25 July 2004 03:53, Rick B was like:
50 degrees C is not very hot, from my experience. Especially if you did
not turn up the voltage on your CPU. Most processors have thermal
cuttoffs that kick in around 80C. Under high load (~95% CPU utilization)
my Athlon 2500 that is OCed will run at about 50C and that is with the
default voltage, and that is about the norm from what I have read.
Without a load it runs ~40C. Raising your CPU voltage is what will get
you into trouble, as part of the extra power is turned into heat. That's
the reason some of the hardcore overclockers will use water and peltier
methods to keep their CPU's cool. Although the water and peltier combo
results in very low CPU temps (~around the freezing point *sometimes*)
it results in condensation problems.
Rick B
>I think I might read it now. I'm a bit
confused as to why my system thinks
>it's running a 896MHz CPU, But so far there's no overt signs that it
isn't
>happy, so I'll prod it a bit and see ;-)
>
>
Yeah that's close enough to 900mhz for me.
Some cpu's are noted as being
good overclockers, but, hell, that's a free 50% overclock! The new sweet
spot for OCer's is the mobile athlon 2500/2600 (~$90US), out of the box
they are ~1800 to 1900mhz and are easily hitting 2500mhz on air cooling
with just a little extra voltage. The Pentium 2.4b's were easily hitting
3Ghz.
From what I've just read it seems the Celeron
Coppermines may not be rated
performance-wise, but they overclock to 900MHz like it
was a perfectly
natural thing to do. If my clock ratio is fixed to 9x, this would explain.
It's running a little warmer than at 66.8MHz x 8 = 566, (? clock ratio) but
nothing you could cook on. I've installed cpuburn to my HD, really I want to
run it from a floppy, like memtest86. Just in case ;-)
AFAICT, so long as I keep it cool enough, it could well be stable at this
setting. TBH I'd say it's actually performing like a well tuned 600. Which is
what I want. We will see :-] I think it helps that I'm using harmonic ratios
based on the recommended defaults.
How hot is too hot? I've got my warning level set to 50deg in the BIOS and I
still don't think I'm running under 40. It's hard to tell without rebooting,
unfortunately although I have both the hardware and software for i2c & hwmon
they're not talking at present. Again, it's something I never really bothered
with before. Most of the heat is coming off the power supply.
What started out as an idle comment has turned into a bit of a journey of
discovery! Thanks for all the comments and suggestions.
cheers
tim hall
26 Jul
26 Jul
11:38 a.m.
New subject: [linux-audio-user] CPU clock - Coppermining
Last Monday 26 July 2004 00:08, Rick B was like:
50 degrees C is not very hot, from my experience.
Especially if you did
not turn up the voltage on your CPU. Most processors have thermal
cuttoffs that kick in around 80C. Under high load (~95% CPU utilization)
my Athlon 2500 that is OCed will run at about 50C and that is with the
default voltage, and that is about the norm from what I have read.
Without a load it runs ~40C. Raising your CPU voltage is what will get
you into trouble, as part of the extra power is turned into heat.
Last Monday 26 July 2004 03:44, Ryan Underwood was like:
ust for another data point, I have a 533MHz Coppermine
Celeron which
has been running at 800Mhz (100MHz * 8) for over 3 years now. Never
given me a lick of trouble at that speed. I did bump up Vcc by
1/10 of a volt to ward off any potential marginal behavior. Perhaps you
could try the same. I would not recommend more than a .2V increment,
because it starts to be a rather large percentage increase.
OK, the voltage is automatically set by my motherboard, so I don't think I
could touch it even if I wanted to. From what I've read so far, it's possible
that these processors will run stably at this speed, it's just not a
guarantee, so they've been rated very conservatively. I'll keep watching for
signs to the contrary.
cheers
tim hall
25 Jul
25 Jul
10:59 p.m.
New subject: [linux-audio-user] CPU clock - Coppermining
On Sun, Jul 25, 2004 at 01:34:02PM +0100, tim hall wrote:
AFAICT, so long as I keep it cool enough, it could well be stable at this
setting.
Just for another data point, I have a 533MHz Coppermine Celeron which
has been running at 800Mhz (100MHz * 8) for over 3 years now. Never
given me a lick of trouble at that speed. I did bump up Vcc by
1/10 of a volt to ward off any potential marginal behavior. Perhaps you
could try the same. I would not recommend more than a .2V increment,
because it starts to be a rather large percentage increase.
--
Ryan Underwood, <nemesis(a)icequake.net>
11:05 p.m.
New subject: [linux-audio-user] CPU clock - beware
On Sun, Jul 25, 2004 at 03:13:37AM +0100, tim hall wrote:
Depends on the chip. It is used to calculate the amount of time spent
in a particular kernel delay loop. Some chips run this code faster or
slower than others. It's not a benchmark as much as a calibration
value.
sure:
~$ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 8
model name : Celeron (Coppermine)
stepping : 6
cpu MHz : 896.977
cache size : 128 KB
fdiv_bug : no
hlt_bug : yes
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
cmov pat pse36 mmx fxsr sse
bogomips : 1789.13
Yeah, definitely overclocked. There are no Celerons with 128KB cache
that run with a 100Mhz bus, AFAIK.
"cpu
MHz" the real current speed
"bogomips" issome weird timing value I have no idea what it means exactly
It's a semi-arbitrary benchmarking value, usually twice the processor speed.
AFAIU. This is what I guessed, I'm using the values
specified in the manual for a
Coppermine FC-PGA 600, so:
Oops. They could be referring to a PIII Coppermine, you know...
at 3x clock ratio I get:
CPU (I assume this is FSB) 100
PCI 33
AGP 66
The display cache runs at 100MHz too, a 1:1 ratio seems logical.
1:1 CPU<->AGP ratio? That's not going to work well at 100MHZ FSB.
I think I might read it now. I'm a bit confused as
to why my system thinks
it's running a 896MHz CPU, But so far there's no overt signs that it isn't
happy, so I'll prod it a bit and see ;-)
897MHz is just the FSB * 9. It doesn't run exactly at 100MHz because it
would be impossible to do so while maintaining synchronicity with a
33MHz PCI bus. 33MHz is actually 33.something, so you end up with
99.something when the FSB is running at 3x the PCI frequency.
--
Ryan Underwood, <nemesis(a)icequake.net>
26 Jul
26 Jul
11:30 a.m.
New subject: [linux-audio-user] CPU clock - Coppermining
Last Monday 26 July 2004 03:50, Ryan Underwood was like:
It's what the manual recommends and it seems happy enough so far.
Some of the screensavers are making me a bit giddy though ;-)
The maximum temperature I've achieved is 45 degC running xscreensaver for
several hours with the case on, which won't even keep my coffee warm.
I'm running it with the lid off just in case anyway, until I get round to
putting a fan on the ?north bridge? as my housemate suggests - I assume
that's the bit that gets vaguely warm. Memtest86 is perfectly happy too. I'm
not sure when I'll get time to properly stress test it, so I might just put
the jumpers back the way they were for now, fiddly inaccessible things though
they are in this tiny micro ATX case.
Yup, I figured.
Thanks again for your suggestions.
tim hall
Yeah, definitely overclocked. There are no Celerons
with 128KB cache
that run with a 100Mhz bus, AFAIK.
> This is what I guessed, I'm using the values
specified in the manual for
> a Coppermine FC-PGA 600, so:
Oops. They could be referring to a PIII Coppermine,
you know...
Ah.
It's listed under the heading of Intel Pentium II - Is this different from
what I have? I have to admit to have been very trusting of the set-up,
naively as it had been set up by the local computer shop. Perhaps they did
know what they were doing after all and decided to leave it set to defaults
and give it away rather than bothering trying to resell it :-] I'm confused.
at 3x clock
ratio I get:
CPU (I assume this is FSB) 100
PCI 33
AGP 66
The display cache runs at 100MHz too, a 1:1 ratio seems logical.
1:1 CPU<->AGP ratio? That's not going to work well at 100MHZ FSB. I think I
might read it now. I'm a bit confused as to why my system
thinks it's running a 896MHz CPU, But so far there's no overt signs that
it isn't happy, so I'll prod it a bit and see ;-)
897MHz is just the FSB * 9. It doesn't run exactly at 100MHz because it
would be impossible to do so while maintaining synchronicity with a
33MHz PCI bus. 33MHz is actually 33.something, so you end up with
99.something when the FSB is running at 3x the PCI frequency.
6:15 p.m.
New subject: [linux-audio-user] CPU clock - Coppermining
On Mon, Jul 26, 2004 at 04:15:04PM +0100, tim hall wrote:
Well, it's a PII (P6) architecture, but that's about the only similarity
to your chip. The package is different, the core voltage is different,
your chip has SSE instructions, less cache but running at full core
clockspeed, etc.
Some video cards tolerate it, some don't. You probably got lucky...
--
Ryan Underwood, <nemesis(a)icequake.net>
Oops. They could be referring to a PIII
Coppermine, you know...
Ah.
It's listed under the heading of Intel Pentium II - Is this different from
what I have? I have to admit to have been very trusting of the
set-up,
naively as it had been set up by the local computer shop. Perhaps they did
know what they were doing after all and decided to leave it set to defaults
and give it away rather than bothering trying to resell it :-] I'm confused.
I worked at a place where they sold overclocked machines regularly.
They saw nothing wrong with it as long as the machine seemed stable and
they didn't falsely advertise the speed of the CPU. I disagreed with
this approach, but I have to admit I saw fewer problems than I would
have thought otherwise.
1:1
CPU<->AGP ratio? That's not going to work well at 100MHZ FSB.
It's what the manual recommends and it seems happy enough so far.
24 Jul
24 Jul
1:04 p.m.
New subject: [linux-audio-user] CPU clock - beware
On Sat, Jul 24, 2004 at 01:21:01PM +0100, tim hall wrote:
I'm as sure as can be that the processor is supported at 100 FSB - i.e. this
is stated quite clearly in the hardcopy manual.
Where? Is there an online copy of this manual?
Well, I'm running at 100MHz x 6 right now.
But a 600MHz Celeron has a 9x multiplier, and it's locked....
I think it's probably more a question of me
explaining badly. I think I've
just set it to the maximum I can get with this mobo & processor. The one
weirdness is that the processor is now being detected as 900MHz, which is not
true.
I think the processor is being run at 900MHZ because you have set it to
a 100MHz bus. With the locked 9x multiplier, that gives you 900Mhz. If
you don't have proper cooling or your chip isn't up to such
overclocking, that could cause some strange behavior.
Can you paste in your /proc/cpuinfo here?
--
Ryan Underwood, <nemesis(a)icequake.net>
7531
days inactive
7534
days old
linux-audio-user@lists.linuxaudio.org
23 comments
8 participants
participants (8)
-
Chris Pickett -
Matthew Barber -
Rick B -
robin fell -
Russell Hanaghan -
Ryan Underwood -
Stefan Scheffler -
tim hall