Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
nayt wrote:
> Hi,
>
> Would it make a lot of difference if I upgrade my cpu from an AMD XP
> 2600+ to an AMD XP 3200+
>
> The 2600+ is overclocked slightly at 200 + 10.5 (2100 mhz).
It would not make a lot of difference, you would see a slight performance
increase.
--
Don Burnette
"When you decide something is impossible to do, try to stay out of the
way of the man that's doing it."
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
On Sun, 27 Feb 2005 23:11:08 -0600, Don Burnette
<d.burnette@clothes.comcast.net> wrote:
> nayt wrote:
>> Hi,
>>
>> Would it make a lot of difference if I upgrade my cpu from an AMD XP
>> 2600+ to an AMD XP 3200+
>>
>> The 2600+ is overclocked slightly at 200 + 10.5 (2100 mhz).
>
>
> It would not make a lot of difference, you would see a slight performance
> increase.
Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11 (2200
mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
and found the result better then a 3000+ but not quite that of a 3200+.
Guess I just have to wait and save some money untill I can afford a PCIE
motherboard and the AMD 64 proc.
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
"nayt" <nayt@dart.net.au> wrote in message newspsmv6ypcgx9krye@ragnarok...
> On Sun, 27 Feb 2005 23:11:08 -0600, Don Burnette
> <d.burnette@clothes.comcast.net> wrote:
>
>> nayt wrote:
>>> Hi,
>>>
>>> Would it make a lot of difference if I upgrade my cpu from an AMD XP
>>> 2600+ to an AMD XP 3200+
>>>
>>> The 2600+ is overclocked slightly at 200 + 10.5 (2100 mhz).
>>
>>
>> It would not make a lot of difference, you would see a slight performance
>> increase.
>
> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11 (2200
> mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
> and found the result better then a 3000+ but not quite that of a 3200+.
>
> Guess I just have to wait and save some money untill I can afford a PCIE
> motherboard and the AMD 64 proc.
Which XP2600 do you have? There were three. If you have a Barton 512K L2
chip then don't worry about Sandra's comparison with the 3200 if it seems a
bit slow. On the other hand, if your CPU is the 12.5x166 256L2 CPU then
ignore the following as, clock for clock, the CPUs with the lesser L2 bench
a bit slower in Sandra (at least from my tests).
Run Sandra a few times and the results will vary. Your benchmark in Sandra
can be slower if you preload a lot of stuff - you usually have a lot of
stuff in your systray when stuff is preloaded. Other than that, Sandra can
be somewhat of a random number generator when it comes to benchmarks. For
best results, run Sandra after a reboot as you will get more consistent
results. Run 5 or so tests in one module, note the scores and reboot and
run the next type of tests. Running Sandra after the computer has been
running for a while and there have been various programs open and closed
will get lower benchmarks.
Sandra's reference scores may be a bit higher than what you get when you
system is set to the same settings but Sandra is handy when you are trying
out various settings on your system as a comparison. Just run each test
about 5 times and average out the results, rebooting between different test
types.
Various background tasks can slow the system down and give poor results. I
had a dual boot system for a while and ran Win98SE and WinXP. Win98Se was
faster than WinXP in nearly all tests, using the same Sandra install. WinXP
runs a lot of services that most people don't need. If, during your tests,
you were running a firewall and virus scanner then you results may be down a
bit because of them as well.
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
On Mon, 28 Feb 2005 10:40:25 GMT, Dave <dave2972@hotmail.com> wrote:
>
> "nayt" <nayt@dart.net.au> wrote in message
> newspsmv6ypcgx9krye@ragnarok...
>> On Sun, 27 Feb 2005 23:11:08 -0600, Don Burnette
>> <d.burnette@clothes.comcast.net> wrote:
>>
>>> nayt wrote:
>>>> Hi,
>>>>
>>>> Would it make a lot of difference if I upgrade my cpu from an AMD XP
>>>> 2600+ to an AMD XP 3200+
>>>>
>>>> The 2600+ is overclocked slightly at 200 + 10.5 (2100 mhz).
>>>
>>>
>>> It would not make a lot of difference, you would see a slight
>>> performance
>>> increase.
>>
>> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11
>> (2200
>> mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
>> and found the result better then a 3000+ but not quite that of a 3200+.
>>
>> Guess I just have to wait and save some money untill I can afford a PCIE
>> motherboard and the AMD 64 proc.
>
> Which XP2600 do you have? There were three. If you have a Barton 512K
> L2
> chip then don't worry about Sandra's comparison with the 3200 if it
> seems a
> bit slow. On the other hand, if your CPU is the 12.5x166 256L2 CPU then
> ignore the following as, clock for clock, the CPUs with the lesser L2
> bench
> a bit slower in Sandra (at least from my tests).
>
> Run Sandra a few times and the results will vary. Your benchmark in
> Sandra
> can be slower if you preload a lot of stuff - you usually have a lot of
> stuff in your systray when stuff is preloaded. Other than that, Sandra
> can
> be somewhat of a random number generator when it comes to benchmarks.
> For
> best results, run Sandra after a reboot as you will get more consistent
> results. Run 5 or so tests in one module, note the scores and reboot and
> run the next type of tests. Running Sandra after the computer has been
> running for a while and there have been various programs open and closed
> will get lower benchmarks.
>
> Sandra's reference scores may be a bit higher than what you get when you
> system is set to the same settings but Sandra is handy when you are
> trying
> out various settings on your system as a comparison. Just run each test
> about 5 times and average out the results, rebooting between different
> test
> types.
>
> Various background tasks can slow the system down and give poor
> results. I
> had a dual boot system for a while and ran Win98SE and WinXP. Win98Se
> was
> faster than WinXP in nearly all tests, using the same Sandra install.
> WinXP
> runs a lot of services that most people don't need. If, during your
> tests,
> you were running a firewall and virus scanner then you results may be
> down a
> bit because of them as well.
>
> Dave
>
>
Hi,
Thanks for the tip. I got the 12.5x166 256L2 CPU. As I mentioned in my
earlier post I've overclocked it to 11x200 (2200 mhz).
Its fine when playing doom3, UT2004, Empire earth, sims2, NWN and also
when I ran prime95 stress test for about 3 hours, however it crashes when
playing Half-life2. The whole computer does not lock up meaning
half-life2 quits after a while and I am back to the desktop.
At this point if I launch other games or launch hl2 again its fine but
everytime I play hl2 at this speed it crashes after a while. The time it
takes to crash is random too.
So I was just wondering if you have the same processor how far could you
push your 2600+.
For cooling I have a 3d gigabyte cooler with active silver paste. I
checked the temps when hl2 crashed and I am back to the desktop and its
about 51c. I have an epox nforce2 motherboard.
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
"nayt" <nayt@dart.net.au> wrote in message newspsmv6ypcgx9krye@ragnarok...
> On Sun, 27 Feb 2005 23:11:08 -0600, Don Burnette
> <d.burnette@clothes.comcast.net> wrote:
>
> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11 (2200
> mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
> and found the result better then a 3000+ but not quite that of a 3200+.
>
> Guess I just have to wait and save some money untill I can afford a PCIE
> motherboard and the AMD 64 proc.
Don't bother with PCIE unless you plan on getting a board with 2 slots and 2
vid cards. PCIE isn't any faster than AGP 8x, which itself is only a tiny
fraction faster than AGP 4x.
In theory, PCIE is faster, but in reality, it's not. There just isn't that
much interaction between the vid card and system memory to make the
differences noticeable, or worth the price difference. In theory, PCIE is 4
times faster than AGP 4x, but if it only comes into play 1% of the time, is
it really "faster"? No. Check out the review of PCIE at Tom's Hardware.
He does a nice comparison of PCIE and AGP. And yes, I was just as shocked
as you will be after reading it.
Bill G
AMD 2700+ (running at 11x220)
BFG 6800GT OC (OC'd to 420/1150)
1 GB Crucial Ballistix PC4000 (2-2-2-5 at DDR400)
22" Viewsonic pf225b
Klipsch 2.1
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
> Don't bother with PCIE unless you plan on getting a board with 2 slots and
> 2
> vid cards. PCIE isn't any faster than AGP 8x, which itself is only a tiny
> fraction faster than AGP 4x.
>
> In theory, PCIE is faster, but in reality, it's not. There just isn't
> that
> much interaction between the vid card and system memory to make the
> differences noticeable, or worth the price difference. In theory, PCIE is
> 4
> times faster than AGP 4x, but if it only comes into play 1% of the time,
> is
> it really "faster"? No. Check out the review of PCIE at Tom's Hardware.
> He does a nice comparison of PCIE and AGP. And yes, I was just as shocked
> as you will be after reading it.
>
> Bill G
> AMD 2700+ (running at 11x220)
> BFG 6800GT OC (OC'd to 420/1150)
> 1 GB Crucial Ballistix PC4000 (2-2-2-5 at DDR400)
> 22" Viewsonic pf225b
> Klipsch 2.1
But with your nice new PCIe you also get nForce4 and AMD 64, sounds good to
me. PCIe cards also seem cheaper than the AGP cards.
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
"TonyC" <nospam@thanks.com> wrote in message
news:42250a2b$0$10943$cc9e4d1f@news-text.dial.pipex.com...
> > Don't bother with PCIE unless you plan on getting a board with 2 slots
and
> > 2
> > vid cards. PCIE isn't any faster than AGP 8x, which itself is only a
tiny
> > fraction faster than AGP 4x.
> >
> > In theory, PCIE is faster, but in reality, it's not. There just isn't
> > that
> > much interaction between the vid card and system memory to make the
> > differences noticeable, or worth the price difference. In theory, PCIE
is
> > 4
> > times faster than AGP 4x, but if it only comes into play 1% of the time,
> > is
> > it really "faster"? No. Check out the review of PCIE at Tom's
Hardware.
> > He does a nice comparison of PCIE and AGP. And yes, I was just as
shocked
> > as you will be after reading it.
> >
> > Bill G
> > AMD 2700+ (running at 11x220)
> > BFG 6800GT OC (OC'd to 420/1150)
> > 1 GB Crucial Ballistix PC4000 (2-2-2-5 at DDR400)
> > 22" Viewsonic pf225b
> > Klipsch 2.1
>
>
> But with your nice new PCIe you also get nForce4 and AMD 64, sounds good
to
> me. PCIe cards also seem cheaper than the AGP cards.
You can have AMD 64 with AGP. And while the cards may seem cheaper, they
actually are not. Check the prices. And the chipset isn't all that
important. An nForce3 will do just fine. Or maybe the new ATI chipset. I
haven't gone to the 64 yet (sometime this summer), but my board will have
AGP.
Spend your money however you want, but I'll stick with my AGP card until
someone proves to me that PCIE is faster and worth the money.
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
nayt wrote:
> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11 (2200
> mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
> and found the result better then a 3000+ but not quite that of a 3200+.
>
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
>>>>>
>>>>> Would it make a lot of difference if I upgrade my cpu from an AMD XP
>>>>> 2600+ to an AMD XP 3200+
>>>>>
>>>>> The 2600+ is overclocked slightly at 200 + 10.5 (2100 mhz).
>>>>
>>>>
>>>> It would not make a lot of difference, you would see a slight
>>>> performance
>>>> increase.
>>>
>>> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11
>>> (2200
>>> mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
>>> and found the result better then a 3000+ but not quite that of a 3200+.
>>>
>>> Guess I just have to wait and save some money untill I can afford a PCIE
>>> motherboard and the AMD 64 proc.
>>
>> Which XP2600 do you have? There were three. If you have a Barton 512K
>> L2
>> chip then don't worry about Sandra's comparison with the 3200 if it
>> seems a
>> bit slow. On the other hand, if your CPU is the 12.5x166 256L2 CPU then
>> ignore the following as, clock for clock, the CPUs with the lesser L2
>> bench
>> a bit slower in Sandra (at least from my tests).
>>
>> Run Sandra a few times and the results will vary. Your benchmark in
>> Sandra
>> can be slower if you preload a lot of stuff - you usually have a lot of
>> stuff in your systray when stuff is preloaded. Other than that, Sandra
>> can
>> be somewhat of a random number generator when it comes to benchmarks.
>> For
>> best results, run Sandra after a reboot as you will get more consistent
>> results. Run 5 or so tests in one module, note the scores and reboot and
>> run the next type of tests. Running Sandra after the computer has been
>> running for a while and there have been various programs open and closed
>> will get lower benchmarks.
>>
>> Sandra's reference scores may be a bit higher than what you get when you
>> system is set to the same settings but Sandra is handy when you are
>> trying
>> out various settings on your system as a comparison. Just run each test
>> about 5 times and average out the results, rebooting between different
>> test
>> types.
>>
>> Various background tasks can slow the system down and give poor results.
>> I
>> had a dual boot system for a while and ran Win98SE and WinXP. Win98Se
>> was
>> faster than WinXP in nearly all tests, using the same Sandra install.
>> WinXP
>> runs a lot of services that most people don't need. If, during your
>> tests,
>> you were running a firewall and virus scanner then you results may be
>> down a
>> bit because of them as well.
>>
>> Dave
>>
>>
>
>
> Hi,
>
> Thanks for the tip. I got the 12.5x166 256L2 CPU. As I mentioned in my
> earlier post I've overclocked it to 11x200 (2200 mhz).
From what I have read here and other places, your overclock is quite good
for a t'bred CPU. Some will go to 2300MHZ or a bit higher with a lot of
extra voltage but whether that will damage the CPU only time will tell. I
read on the AMD site about 1.5 years ago that the maximum recommended
voltage for T'breds was 1.85V so I don't go much above that for any longer
than a few weeks as it summer here at the moment.
> Its fine when playing doom3, UT2004, Empire earth, sims2, NWN and also
> when I ran prime95 stress test for about 3 hours, however it crashes when
> playing Half-life2. The whole computer does not lock up meaning
> half-life2 quits after a while and I am back to the desktop.
>
> At this point if I launch other games or launch hl2 again its fine but
> everytime I play hl2 at this speed it crashes after a while. The time it
> takes to crash is random too.
>
> So I was just wondering if you have the same processor how far could you
> push your 2600+.
>
> For cooling I have a 3d gigabyte cooler with active silver paste. I
> checked the temps when hl2 crashed and I am back to the desktop and its
> about 51c. I have an epox nforce2 motherboard.
Does HL2 force you to reboot after a crash? If so, go into BIOS and check
the temps there. Sometimes the temps in BIOS can a bit higher immediately
after the crash. If the CPU temp is under the mid to low 60s C then that
should be OK. On my aBit board the CPU temp can drop by over 8C between
checking in BIOS and once the desktop is loaded, depending on how long it
sat in BIOS for.
If the temp is OK then it could be that your CPU does not have enough
voltage or you are running the RAM too fast.
Your CPU is rated at 1.65V standard but it could require around 1.75V or
more to run at 2200MHz. Run Prime95 torture test 1 (maximum FPU stress).
If it runs for 12 hours or longer then the CPU is OK for games. If it halts
with an error within 20 mins then you could need to add upto an extra 0.1V,
experiment. Aim to get past 12 hours on the Prime95 test.
If you are running PC2700 RAM overclocked to PC3200 speeds then the crashes
could be caused by that. Some PC2700 modules have SPD chips that only go to
166FSB and have no settings for 200FSB so if you increase the FSB from 166
to 200 they get flaky as they use the 166 settings. Generally, DDR modules
are made to run on 2.5V so increasing the voltage to 2.6V of 2.7V may make
them stable, newer low latency PC3200+ modules may need 2.6-2.7V before
overclocking.
Download CPU-Z, the "Memory" tab will tell you what your current memory
settings are and the "SPD" tab will tell you what is in the SPD chips for
various FSBs. Run Memtest86+ to check for RAM errors. Some Value RAM
PC2700 modules are actually PC3200 modules that would not run at 2.5V.
Using 2.6V or 2.7V may let them work OK at PC3200 speeds. If you can't
increase the voltage to the RAM then try slower timings, eg if rated at
2.5,3,3,7 at 166 FSB then try 3.0,3,3,8 or even 3.0,4,4,9, etc.
Last but not least, are you overclocking your video card? If you get
artefacts on screen then the overclock is to high. Artefacts range from
random bright pixels on dark screens (you don't tend to notice them on
brighter screens), light/white polygons where they should not be, to various
rainbow and moire effects, etc.
My CPU is an XP2000 B, rated at 12.5x133 @ 1.6V for 1666MHz. On my current
board I am running it at 10.5x200 @ 1.85V for 2100MHz. I have also run it
at 11x200 @ 1.93V for 2200MHz, 10.5x207 @ 1.9V for 2174MHz using PC3200 RAM.
On a previous board I could run it at 13.5x154 @ 1.75V for 2079MHz using
PC2100 RAM. That PC2100 RAM was also used in my current board and with
voltage increase and slower timings I got it to 185FSB reliably. It helped
that they were down-rated PC2700 modules. The speeds quoted in this
paragraph were Prime95 and Memtest86+ stable.
As a rough rule, clock for clock, my CPU needs a higher voltage with a
higher FSB. EG, if the CPU runs at about 2100MHz then it needs more voltage
as the FSB is increased from 133 to 166 to 200 while the multiplier is
reduced to keep the CPU at that speed.
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
On Wed, 02 Mar 2005 11:27:08 -0600, Charles <charles97@dunjas.com> wrote:
> nayt wrote:
>
>> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11
>> (2200
>> mhz) which I believe is the same speed as the 3200+. Ran sandra sisoft
>> and found the result better then a 3000+ but not quite that of a 3200+.
>>
>
> What is "sandra sisoft" ???
> Tnx,
> Charles
>
>
Archived from groups: alt.comp.hardware.overclocking.amd (More info?)
On Wed, 02 Mar 2005 23:56:41 GMT, Dave <dave2972@hotmail.com> wrote:
>
>>>>>>
>>>>>> Would it make a lot of difference if I upgrade my cpu from an AMD XP
>>>>>> 2600+ to an AMD XP 3200+
>>>>>>
>>>>>> The 2600+ is overclocked slightly at 200 + 10.5 (2100 mhz).
>>>>>
>>>>>
>>>>> It would not make a lot of difference, you would see a slight
>>>>> performance
>>>>> increase.
>>>>
>>>> Thanks for the tip. I've actually overclocked the 2600+ at 200 x 11
>>>> (2200
>>>> mhz) which I believe is the same speed as the 3200+. Ran sandra
>>>> sisoft
>>>> and found the result better then a 3000+ but not quite that of a
>>>> 3200+.
>>>>
>>>> Guess I just have to wait and save some money untill I can afford a
>>>> PCIE
>>>> motherboard and the AMD 64 proc.
>>>
>>> Which XP2600 do you have? There were three. If you have a Barton 512K
>>> L2
>>> chip then don't worry about Sandra's comparison with the 3200 if it
>>> seems a
>>> bit slow. On the other hand, if your CPU is the 12.5x166 256L2 CPU
>>> then
>>> ignore the following as, clock for clock, the CPUs with the lesser L2
>>> bench
>>> a bit slower in Sandra (at least from my tests).
>>>
>>> Run Sandra a few times and the results will vary. Your benchmark in
>>> Sandra
>>> can be slower if you preload a lot of stuff - you usually have a lot
>>> of
>>> stuff in your systray when stuff is preloaded. Other than that, Sandra
>>> can
>>> be somewhat of a random number generator when it comes to benchmarks.
>>> For
>>> best results, run Sandra after a reboot as you will get more consistent
>>> results. Run 5 or so tests in one module, note the scores and reboot
>>> and
>>> run the next type of tests. Running Sandra after the computer has been
>>> running for a while and there have been various programs open and
>>> closed
>>> will get lower benchmarks.
>>>
>>> Sandra's reference scores may be a bit higher than what you get when
>>> you
>>> system is set to the same settings but Sandra is handy when you are
>>> trying
>>> out various settings on your system as a comparison. Just run each
>>> test
>>> about 5 times and average out the results, rebooting between different
>>> test
>>> types.
>>>
>>> Various background tasks can slow the system down and give poor
>>> results.
>>> I
>>> had a dual boot system for a while and ran Win98SE and WinXP. Win98Se
>>> was
>>> faster than WinXP in nearly all tests, using the same Sandra install.
>>> WinXP
>>> runs a lot of services that most people don't need. If, during your
>>> tests,
>>> you were running a firewall and virus scanner then you results may be
>>> down a
>>> bit because of them as well.
>>>
>>> Dave
>>>
>>>
>>
>>
>> Hi,
>>
>> Thanks for the tip. I got the 12.5x166 256L2 CPU. As I mentioned in my
>> earlier post I've overclocked it to 11x200 (2200 mhz).
>
> From what I have read here and other places, your overclock is quite good
> for a t'bred CPU. Some will go to 2300MHZ or a bit higher with a lot of
> extra voltage but whether that will damage the CPU only time will tell.
> I
> read on the AMD site about 1.5 years ago that the maximum recommended
> voltage for T'breds was 1.85V so I don't go much above that for any
> longer
> than a few weeks as it summer here at the moment.
>
>> Its fine when playing doom3, UT2004, Empire earth, sims2, NWN and also
>> when I ran prime95 stress test for about 3 hours, however it crashes
>> when
>> playing Half-life2. The whole computer does not lock up meaning
>> half-life2 quits after a while and I am back to the desktop.
>>
>> At this point if I launch other games or launch hl2 again its fine but
>> everytime I play hl2 at this speed it crashes after a while. The time
>> it
>> takes to crash is random too.
>>
>> So I was just wondering if you have the same processor how far could you
>> push your 2600+.
>>
>> For cooling I have a 3d gigabyte cooler with active silver paste. I
>> checked the temps when hl2 crashed and I am back to the desktop and its
>> about 51c. I have an epox nforce2 motherboard.
>
> Does HL2 force you to reboot after a crash? If so, go into BIOS and
> check
> the temps there. Sometimes the temps in BIOS can a bit higher
> immediately
> after the crash. If the CPU temp is under the mid to low 60s C then that
> should be OK. On my aBit board the CPU temp can drop by over 8C between
> checking in BIOS and once the desktop is loaded, depending on how long it
> sat in BIOS for.
>
> If the temp is OK then it could be that your CPU does not have enough
> voltage or you are running the RAM too fast.
>
> Your CPU is rated at 1.65V standard but it could require around 1.75V or
> more to run at 2200MHz. Run Prime95 torture test 1 (maximum FPU stress).
> If it runs for 12 hours or longer then the CPU is OK for games. If it
> halts
> with an error within 20 mins then you could need to add upto an extra
> 0.1V,
> experiment. Aim to get past 12 hours on the Prime95 test.
>
> If you are running PC2700 RAM overclocked to PC3200 speeds then the
> crashes
> could be caused by that. Some PC2700 modules have SPD chips that only
> go to
> 166FSB and have no settings for 200FSB so if you increase the FSB from
> 166
> to 200 they get flaky as they use the 166 settings. Generally, DDR
> modules
> are made to run on 2.5V so increasing the voltage to 2.6V of 2.7V may
> make
> them stable, newer low latency PC3200+ modules may need 2.6-2.7V before
> overclocking.
>
> Download CPU-Z, the "Memory" tab will tell you what your current memory
> settings are and the "SPD" tab will tell you what is in the SPD chips for
> various FSBs. Run Memtest86+ to check for RAM errors. Some Value RAM
> PC2700 modules are actually PC3200 modules that would not run at 2.5V.
> Using 2.6V or 2.7V may let them work OK at PC3200 speeds. If you can't
> increase the voltage to the RAM then try slower timings, eg if rated at
> 2.5,3,3,7 at 166 FSB then try 3.0,3,3,8 or even 3.0,4,4,9, etc.
>
> Last but not least, are you overclocking your video card? If you get
> artefacts on screen then the overclock is to high. Artefacts range from
> random bright pixels on dark screens (you don't tend to notice them on
> brighter screens), light/white polygons where they should not be, to
> various
> rainbow and moire effects, etc.
>
> My CPU is an XP2000 B, rated at 12.5x133 @ 1.6V for 1666MHz. On my
> current
> board I am running it at 10.5x200 @ 1.85V for 2100MHz. I have also run
> it
> at 11x200 @ 1.93V for 2200MHz, 10.5x207 @ 1.9V for 2174MHz using PC3200
> RAM.
> On a previous board I could run it at 13.5x154 @ 1.75V for 2079MHz using
> PC2100 RAM. That PC2100 RAM was also used in my current board and with
> voltage increase and slower timings I got it to 185FSB reliably. It
> helped
> that they were down-rated PC2700 modules. The speeds quoted in this
> paragraph were Prime95 and Memtest86+ stable.
>
> As a rough rule, clock for clock, my CPU needs a higher voltage with a
> higher FSB. EG, if the CPU runs at about 2100MHz then it needs more
> voltage
> as the FSB is increased from 133 to 166 to 200 while the multiplier is
> reduced to keep the CPU at that speed.
>
>
Hi I increased the voltage to 1.68 and I played HL2 for about 4 hours
without any problems. So it looks like the increasing the voltage did the
trick.
Thanks for the tip about how high it could go but I don't think I'll bump
it up that high.
I am using 2 x Geil 512 pc 3200 ram. Thanks for the tip. Really
appreciate it. Now running happily at 3200+ speed, HL2 and all.
psmv6ypcgx9krye@ragnarok...