XP2500@1870Mhz or XP2100@2158Mhz? Which is better?

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

I have to give one cpu away tomorrow afternoon and i'm trying to
figure out which one i want to keep for myself. Which am i better off
with on my Gigabyte 7VAX mobo. It does not have a 1/6 divider for the
pci/agp bus so i can't go to 200Mhz FSB. The main purpose is to play
games.

The XP2500 i have is brand new and seems to be one of the hard locked
ones. Nothing, not even cutting bridges or pinmods, can unlock the
multiplier. And overclocking the bus only goes up to about 170.

The XP2100 is completely unlocked. I haven't tried overclocking any
more than 13x166=2158Mhz.

XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
or
XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
clock higher.

I used a pinmod to make my mobo use the 166Mhz bus for the XP2100
which is really a 133Mhz bus cpu.

Anyone know which is the better one for me to keep? I know the
answer should be "try out your games and see" but i don't have any
installed right now. I just reinstalled XP and i'm wondering if
there's any real reasons why one might be better than the other.

Any thoughts appreciated.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sat, 15 May 2004 06:08:24 +0000, kevins_news2 wrote:

> XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
> or
> XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
> clock higher.
>
The 2100+ will give you the better performance. It's running as a 2700+
while the 2500+ is not even up to 2600+ speed.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sat, 15 May 2004 06:55:49 GMT, Wes Newell
<w.newell@TAKEOUTverizon.net> wrote:

>On Sat, 15 May 2004 06:08:24 +0000, kevins_news2 wrote:
>
>> XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>> or
>> XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>> clock higher.
>>
>The 2100+ will give you the better performance. It's running as a 2700+
>while the 2500+ is not even up to 2600+ speed.

Strange my 3dmark 2003 scores with each are nearly identicle? I would
have expected the CPU score for a 2158Mhz cpu to be better than that
for a 1835Mhz (11x166 when i tested on the XP2500).

Damn. Now i need to do more testing.

 

[Ed]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sat, 15 May 2004 15:39:45 GMT, kevins_news2
<kevins_news2@rogers.n0spam.com> wrote:

>On Sat, 15 May 2004 06:55:49 GMT, Wes Newell
><w.newell@TAKEOUTverizon.net> wrote:
>
>>On Sat, 15 May 2004 06:08:24 +0000, kevins_news2 wrote:
>>
>>> XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>>> or
>>> XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>>> clock higher.
>>>
>>The 2100+ will give you the better performance. It's running as a 2700+
>>while the 2500+ is not even up to 2600+ speed.
>
>Strange my 3dmark 2003 scores with each are nearly identicle? I would
>have expected the CPU score for a 2158Mhz cpu to be better than that
>for a 1835Mhz (11x166 when i tested on the XP2500).
>
>Damn. Now i need to do more testing.

The CPU isn't the bottle neck , that's why the scores are similar.

 

[augustus]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

> XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
> or
> XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
> clock higher.

You must have the only Barton 2500 in the entire world that can only be
overclocked to a maximum of 4Mhz above stock 166Mhz. I've never heard of
this.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sat, 15 May 2004 16:29:29 GMT, "Augustus" <tiberius@weeik.com>
wrote:

>> XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>> or
>> XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>> clock higher.
>
>You must have the only Barton 2500 in the entire world that can only be
>overclocked to a maximum of 4Mhz above stock 166Mhz. I've never heard of
>this.
>

Could be my memory. It was cheap stuff when i got it a year and a
half ago.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sat, 15 May 2004 11:09:13 -0500, Ed <nobox@hotmail.com> wrote:

>On Sat, 15 May 2004 15:39:45 GMT, kevins_news2
><kevins_news2@rogers.n0spam.com> wrote:
>
>>On Sat, 15 May 2004 06:55:49 GMT, Wes Newell
>><w.newell@TAKEOUTverizon.net> wrote:
>>
>>>On Sat, 15 May 2004 06:08:24 +0000, kevins_news2 wrote:
>>>
>>>> XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>>>> or
>>>> XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>>>> clock higher.
>>>>
>>>The 2100+ will give you the better performance. It's running as a 2700+
>>>while the 2500+ is not even up to 2600+ speed.
>>
>>Strange my 3dmark 2003 scores with each are nearly identicle? I would
>>have expected the CPU score for a 2158Mhz cpu to be better than that
>>for a 1835Mhz (11x166 when i tested on the XP2500).
>>
>>Damn. Now i need to do more testing.
>
>The CPU isn't the bottle neck , that's why the scores are similar.

No, i mean just the CPU score. I'm getting a cpu score of 299 with
the XP2500 and 303 with the overclocked XP2100. All the video scores
are equal (since i'm using the same vid card). In fact both those
scores seem low for this range of processor. Now i'm even more
confused. I didn't think much could screw up your CPU score.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

kevins_news2 wrote:
[...]
> The XP2500 i have is brand new and seems to be one of the hard locked
> ones. Nothing, not even cutting bridges or pinmods, can unlock the
> multiplier.

If you're into cutting and joining bridges, you can mobile the 2500 to be
able to change the multiplier in your OS of choice. In which case it would
almost certainly beat the XP2100, once you got it up to speed.

[...]

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sat, 15 May 2004 19:35:36 +1200, "Michael Brown"
<see@signature.below> wrote:

>kevins_news2 wrote:
>[...]
>> The XP2500 i have is brand new and seems to be one of the hard locked
>> ones. Nothing, not even cutting bridges or pinmods, can unlock the
>> multiplier.
>
>If you're into cutting and joining bridges, you can mobile the 2500 to be
>able to change the multiplier in your OS of choice. In which case it would
>almost certainly beat the XP2100, once you got it up to speed.
>
>[...]

So i'm reading
http://groups.google.co.nz/groups?selm=g_Emc.1412%24XI4.66669%40news.xtra.co
..nz
in the hopes of mobilizing my cpu.

Do you also have a link to a really good explanation of the best
current way join and how to cut the bridges?

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

kevins_news2 wrote:
[...]
> Do you also have a link to a really good explanation of the best
> current way join and how to cut the bridges?

I haven't yet found any site that gives an easy way to do the mods, mainly
because I think the "easiest" way varies depending on the equipment
available and the patience of the modder. Personally, I do bridge cuts under
a 100x microscope with a 100W lamp about 20cm from the chip, using a sharp
needle to scrape across the bridges. For filling bridges, I tend to do a
similar thing but move the lamp back a lot so that it doesn't dry the paint
too much and the microscope at lower magnification. I use a similar needle,
and squeeze a bit of the paint (I'm using a pen) onto a piece of plastic,
then "scoop up" the paint with the tip of the pin. It has to be done fairly
quickly, but with a bit of practice it becomes pretty easy. I then check the
join using the 100x and the close light to make sure there's no improper
joins, etc. Finally, for joining bridges, I use the pin to scrape the
laminate off the ends of the real bridges, woodglue to fill the gap, then
paint over the top of that using a similar method. I haven't managed to get
it to work nicely though, so there's probably a better way out there

For the 3rd L5, just fill it (gently!) with some conductive goop. The
standard automotive window defogger stuff should do fine, as would the
conductive ink pens (make sure you only join the one bridge though).
Supposedly pencil will work as well:
http://www.cpuheat.wz.cz/html/Photos/AXP_Multiplier_Photos.htm

I'm not 100% sure about how you rejoin the second L2, but I *think* you can
just fill it full of conductive goop as well. I'd personally recommend
joining rather than filling, though, as it's not been tried much AFAIK.

For the L6's you're sorta in luck. It's fairly clear above and below the
bridges, so the chance of hitting something is fairly small. I've tried
doing these with the pin and it's quite easy due to the amount of space you
have. However, a craft-knife does the job good as well, according to the
page above. Cut towards the middle of the bridge as opposed to cutting out,
as if you slip then you've got less chance of hitting comething important.

For the 1st L5, if you're planning on taking that route, you have to be a
little careful. There's a few traces just to the outside of the bridge, so
I'd recommend cutting towards the inside of the bridge. I scrape across the
bridge with a pin, though using a sharp knife to slice it also works quite
well apparently. Practice on the L6's before graduating to the 1st L5

The L8's are also quite tightly packed by traces, so make sure you're good
at doing the cuts before attempting these.

Do not, under ANY circumstances, use electricity to blow the bridges. This
has a very high failure rate on the newer chips, with either the bridge
being fused permanently to the "closed" state, or destruction of the chip.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Sun, 16 May 2004 00:19:45 +1200, "Michael Brown"
<see@signature.below> wrote:

>kevins_news2 wrote:
>[...]
>> Do you also have a link to a really good explanation of the best
>> current way join and how to cut the bridges?
>
>I haven't yet found any site that gives an easy way to do the mods, mainly
>because I think the "easiest" way varies depending on the equipment
>available and the patience of the modder. Personally, I do bridge cuts under
>a 100x microscope with a 100W lamp about 20cm from the chip, using a sharp
>needle to scrape across the bridges. For filling bridges, I tend to do a
>similar thing but move the lamp back a lot so that it doesn't dry the paint
>too much and the microscope at lower magnification. I use a similar needle,
>and squeeze a bit of the paint (I'm using a pen) onto a piece of plastic,
>then "scoop up" the paint with the tip of the pin. It has to be done fairly
>quickly, but with a bit of practice it becomes pretty easy. I then check the
>join using the 100x and the close light to make sure there's no improper
>joins, etc. Finally, for joining bridges, I use the pin to scrape the
>laminate off the ends of the real bridges, woodglue to fill the gap, then
>paint over the top of that using a similar method. I haven't managed to get
>it to work nicely though, so there's probably a better way out there
>
>For the 3rd L5, just fill it (gently!) with some conductive goop. The
>standard automotive window defogger stuff should do fine, as would the
>conductive ink pens (make sure you only join the one bridge though).
>Supposedly pencil will work as well:
>http://www.cpuheat.wz.cz/html/Photos/AXP_Multiplier_Photos.htm
>
>I'm not 100% sure about how you rejoin the second L2, but I *think* you can
>just fill it full of conductive goop as well. I'd personally recommend
>joining rather than filling, though, as it's not been tried much AFAIK.
>
>For the L6's you're sorta in luck. It's fairly clear above and below the
>bridges, so the chance of hitting something is fairly small. I've tried
>doing these with the pin and it's quite easy due to the amount of space you
>have. However, a craft-knife does the job good as well, according to the
>page above. Cut towards the middle of the bridge as opposed to cutting out,
>as if you slip then you've got less chance of hitting comething important.
>
>For the 1st L5, if you're planning on taking that route, you have to be a
>little careful. There's a few traces just to the outside of the bridge, so
>I'd recommend cutting towards the inside of the bridge. I scrape across the
>bridge with a pin, though using a sharp knife to slice it also works quite
>well apparently. Practice on the L6's before graduating to the 1st L5
>
>The L8's are also quite tightly packed by traces, so make sure you're good
>at doing the cuts before attempting these.
>
>Do not, under ANY circumstances, use electricity to blow the bridges. This
>has a very high failure rate on the newer chips, with either the bridge
>being fused permanently to the "closed" state, or destruction of the chip.

I overclocked my duron 800 by using a pencil to draw over one bridge.
But i've never tried cutting or painting yet. I might be a bit out of
my league here. But if i try to do it...

Judging by all the websites on mobilizing an xp2500 i think i only
have one thing to fill and three to cut.

- Fill the 2nd from the right L5 making it Mobile.

- Cut the middle L8 making the max voltage 1.8 (just in case) and then
setting it in windows with that CPUMSR utility.

- Cut the first and last L6 making the max multiplier 19x and setting
it in windows with the CPUMSR utility.

And that's it? No mention of the first L5 or anything with the L2s.
Should this be all i need to do my mod?

 

[minotaur]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

kevins_news2 wrote:

> On Sat, 15 May 2004 11:09:13 -0500, Ed <nobox@hotmail.com> wrote:
>
>
>>On Sat, 15 May 2004 15:39:45 GMT, kevins_news2
>><kevins_news2@rogers.n0spam.com> wrote:
>>
>>
>>>On Sat, 15 May 2004 06:55:49 GMT, Wes Newell
>>><w.newell@TAKEOUTverizon.net> wrote:
>>>
>>>
>>>>On Sat, 15 May 2004 06:08:24 +0000, kevins_news2 wrote:
>>>>
>>>>
>>>>>XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>>>>>or
>>>>>XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>>>>>clock higher.
>>>>>
>>>>
>>>>The 2100+ will give you the better performance. It's running as a 2700+
>>>>while the 2500+ is not even up to 2600+ speed.
>>>
>>>Strange my 3dmark 2003 scores with each are nearly identicle? I would
>>>have expected the CPU score for a 2158Mhz cpu to be better than that
>>>for a 1835Mhz (11x166 when i tested on the XP2500).
>>>
>>>Damn. Now i need to do more testing.
>>
>>The CPU isn't the bottle neck , that's why the scores are similar.
>
>
> No, i mean just the CPU score. I'm getting a cpu score of 299 with
> the XP2500 and 303 with the overclocked XP2100. All the video scores
> are equal (since i'm using the same vid card). In fact both those
> scores seem low for this range of processor. Now i'm even more
> confused. I didn't think much could screw up your CPU score.

Try running the XP2500 at 200FSB yet? is more than capable of it with
it's locked multipler of 11 for 2200MHz (XP3200+ speeds).
Might have to raise the VCore to 1.775 etc, and you do need to have
PC3200+ RAM.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

On Mon, 17 May 2004 22:40:51 +1000, Minotaur <antnel@hotmail.com>
wrote:

>kevins_news2 wrote:
>
>> On Sat, 15 May 2004 11:09:13 -0500, Ed <nobox@hotmail.com> wrote:
>>
>>
>>>On Sat, 15 May 2004 15:39:45 GMT, kevins_news2
>>><kevins_news2@rogers.n0spam.com> wrote:
>>>
>>>
>>>>On Sat, 15 May 2004 06:55:49 GMT, Wes Newell
>>>><w.newell@TAKEOUTverizon.net> wrote:
>>>>
>>>>
>>>>>On Sat, 15 May 2004 06:08:24 +0000, kevins_news2 wrote:
>>>>>
>>>>>
>>>>>>XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>>>>>>or
>>>>>>XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>>>>>>clock higher.
>>>>>>
>>>>>
>>>>>The 2100+ will give you the better performance. It's running as a 2700+
>>>>>while the 2500+ is not even up to 2600+ speed.
>>>>
>>>>Strange my 3dmark 2003 scores with each are nearly identicle? I would
>>>>have expected the CPU score for a 2158Mhz cpu to be better than that
>>>>for a 1835Mhz (11x166 when i tested on the XP2500).
>>>>
>>>>Damn. Now i need to do more testing.
>>>
>>>The CPU isn't the bottle neck , that's why the scores are similar.
>>
>>
>> No, i mean just the CPU score. I'm getting a cpu score of 299 with
>> the XP2500 and 303 with the overclocked XP2100. All the video scores
>> are equal (since i'm using the same vid card). In fact both those
>> scores seem low for this range of processor. Now i'm even more
>> confused. I didn't think much could screw up your CPU score.
>
>Try running the XP2500 at 200FSB yet? is more than capable of it with
>it's locked multipler of 11 for 2200MHz (XP3200+ speeds).
>Might have to raise the VCore to 1.775 etc, and you do need to have
>PC3200+ RAM.

My motherboard doesn't have a 1/6 divider. so i'd be majorly
overclocking the PCI/AGP busses. And i don't have PC3200 ram either.

In the end i kept the XP2500. I did a few benchmark tests. Sisoft
Sandra, Unreal 2004 demo benchmark, Quake 3 demo, 3dMark 2003, and a
software 3D rendering benchmark called Kwikbench or something.
Here's the results in case anyone was interested. The numbers don't
really mean anything other than to compare the processors. I won't
even mention what graphics card i have.

Sisoft Sandra
XP2500 at 11x171 = 1881Mhz
CPu arithmetic 6955/2885
CPU Multimedia 17284/18376

XP2100 at 14x140 = 1960 Mhz
CPu arithmetic 7061/2971
CPU Multimedia 18451/19710

XP2100 at 13x170 = 2210Mhz
CPU Arithmetic 8199/3411
CPU Multimedia 20434/21794

Performance scales with clockspeed.
But games were another issue.

XP2100 at 14x140
Ut2k4 25 fps
Q3 199 fps
3dMark cpu score of 290
Kwikbench 25 fps

XP2100 at 13x170
Ut2k4 26 fps
Q3 210 fps
3dMark cpu score of 301
Kwikbench 29 fps

XP2500 at 11x171
Ut2k4 28 fps
Q3 221 fps
3dMark cpu score of 309
Kwikbench 34 fps

So i kept the XP2500 even though it's running at a lower clock speed.
I assume it's the extra L2 cache that makes the difference. Hopefully
i'll get my hands on a mobo that does 200Mhz fsb and some PC3200 ram
sometime in the future. OR maybe i'll even try messing with the CPU
to turn it into a mobile and therefore unlock the other multipliers.

Thanks for all the responses.

 

[minotaur]

Archived from groups: alt.comp.hardware.overclocking.amd (More info?)

kevins_news2 wrote:

> On Sat, 15 May 2004 16:29:29 GMT, "Augustus" <tiberius@weeik.com>
> wrote:
>
>
>>>XP2500 (hard locked) at 11x170 = 1870Mhz. 512KB L2 cache
>>>or
>>>XP2100 (unlocked) at 13x166 = 2158Mhz. 256Kb L2 cache. Possible could
>>>clock higher.
>>
>>You must have the only Barton 2500 in the entire world that can only be
>>overclocked to a maximum of 4Mhz above stock 166Mhz. I've never heard of
>>this.
>>
>
>
> Could be my memory. It was cheap stuff when i got it a year and a
> half ago.
>
>

Try reducing your RAM timings. Best idea find out what type of RAM it is
and goto the Manufactures web site for specs. You would definantly have
to change the timings between 166FSB and 200FSB, see what they suggest
for 200FSB?.

Minotaur (8*