2500+ FSB at 200mhz, memory at 166mhz

[txxxx]

Hi,

Just wondering if anyone's tried running a cpu such as the AMD XP 2500+ at 200mhz whilst keeping the memory at an aggressive 166mhz?

In theory, it shouldnt make much difference but given the async nature, I expect to lose a little performance (not that I care/would notice it).

So just wondering whether anyone's tried that and had any success?

Thanks txxxx

 

[BlvdKing]

I hear that is suicide on the memory performance on just about any Athlon chipset.

 

[sugarkang]

let's use my aquamark benchies as an example of what happens during async:

BEFORE: KT266A + athlon xp 1700 + PC2100 = cpu score 4888
AFTER 1: NF2 + athlon xp 2500 + PC2100 = cpu score 5010
AFTER 2: NF2 + athlon xp 3200 + PC2100 = cpu score 5616
AFTER 3: NF2 + athlon xp 3200 + PC3200 = cpu score 6923

so, in the *before* scenario, i had an abit kt266a board with everything in sync.
notice when i upgraded to an asus a7n8x and barton 2500, my cpu score only went up by 100 points or so.
EVEN THOUGH i went from 1700 --> 2500.
in the next situation, i OCed the chip and gained 600 points.
but in the last situation when i *matched* the ram to the fsb, my score goes up a whopping 1300 points.

conclusion: DO NOT RUN ASYNCHRONIZED!

solution: go trade in your pc2700 for pc3200 *or* OC your ram to hit 200fsb levels. probably better to just upgrade to the 3200, and then take your total fsb up even higher.

 

[jjyiz28]

Originally posted by: sugarkang
let's use my aquamark benchies as an example of what happens during async:

BEFORE: KT266A + athlon xp 1700 + PC2100 = cpu score 4888
AFTER 1: NF2 + athlon xp 2500 + PC2100 = cpu score 5010
AFTER 2: NF2 + athlon xp 3200 + PC2100 = cpu score 5616
AFTER 3: NF2 + athlon xp 3200 + PC3200 = cpu score 6923

so, in the *before* scenario, i had an abit kt266a board with everything in sync.
notice when i upgraded to an asus a7n8x and barton 2500, my cpu score only went up by 100 points or so.
EVEN THOUGH i went from 1700 --> 2500.
in the next situation, i OCed the chip and gained 600 points.
but in the last situation when i *matched* the ram to the fsb, my score goes up a whopping 1300 points.

conclusion: DO NOT RUN ASYNCHRONIZED!

solution: go trade in your pc2700 for pc3200 *or* OC your ram to hit 200fsb levels. probably better to just upgrade to the 3200, and then take your total fsb up even higher.

you do know running only 1 benchmark is not definitive/????
a forum user in cpu did a benchmark testing exactly this. he found out , (with a lot of real world benchmarks), that 400fsb/333mem (tight timmings), wins majority of tests than 400fsb/400mem (relaxed timings).
his tests are ]here

also, read this

gamepc1

 

[Jeff7181]

Originally posted by: jjyiz28

Originally posted by: sugarkang
let's use my aquamark benchies as an example of what happens during async:

BEFORE: KT266A + athlon xp 1700 + PC2100 = cpu score 4888
AFTER 1: NF2 + athlon xp 2500 + PC2100 = cpu score 5010
AFTER 2: NF2 + athlon xp 3200 + PC2100 = cpu score 5616
AFTER 3: NF2 + athlon xp 3200 + PC3200 = cpu score 6923

so, in the *before* scenario, i had an abit kt266a board with everything in sync.
notice when i upgraded to an asus a7n8x and barton 2500, my cpu score only went up by 100 points or so.
EVEN THOUGH i went from 1700 --> 2500.
in the next situation, i OCed the chip and gained 600 points.
but in the last situation when i *matched* the ram to the fsb, my score goes up a whopping 1300 points.

conclusion: DO NOT RUN ASYNCHRONIZED!

solution: go trade in your pc2700 for pc3200 *or* OC your ram to hit 200fsb levels. probably better to just upgrade to the 3200, and then take your total fsb up even higher.

you do know running only 1 benchmark is not definitive/????
a forum user in cpu did a benchmark testing exactly this. he found out , (with a lot of real world benchmarks), that 400fsb/333mem (tight timmings), wins majority of tests than 400fsb/400mem (relaxed timings).
his tests are ]here

also, read this

gamepc1

Those tests Duvie did were using a Pentium 4 which is COMPLETELY different than an Athlon XP... so your argument means nothing.

 

[jjyiz28]

Originally posted by: Jeff7181

Originally posted by: jjyiz28

Originally posted by: sugarkang
let's use my aquamark benchies as an example of what happens during async:

BEFORE: KT266A + athlon xp 1700 + PC2100 = cpu score 4888
AFTER 1: NF2 + athlon xp 2500 + PC2100 = cpu score 5010
AFTER 2: NF2 + athlon xp 3200 + PC2100 = cpu score 5616
AFTER 3: NF2 + athlon xp 3200 + PC3200 = cpu score 6923

so, in the *before* scenario, i had an abit kt266a board with everything in sync.
notice when i upgraded to an asus a7n8x and barton 2500, my cpu score only went up by 100 points or so.
EVEN THOUGH i went from 1700 --> 2500.
in the next situation, i OCed the chip and gained 600 points.
but in the last situation when i *matched* the ram to the fsb, my score goes up a whopping 1300 points.

conclusion: DO NOT RUN ASYNCHRONIZED!

solution: go trade in your pc2700 for pc3200 *or* OC your ram to hit 200fsb levels. probably better to just upgrade to the 3200, and then take your total fsb up even higher.

you do know running only 1 benchmark is not definitive/????
a forum user in cpu did a benchmark testing exactly this. he found out , (with a lot of real world benchmarks), that 400fsb/333mem (tight timmings), wins majority of tests than 400fsb/400mem (relaxed timings).
his tests are ]here

also, read this

gamepc1

Those tests Duvie did were using a Pentium 4 which is COMPLETELY different than an Athlon XP... so your argument means nothing.

read this and the later pages. amd setup dual channel.
they compared a 333fsb chip using 333mem, and 333fsb chip using 400mem. 333/400 is async just like 333/266 is async. rough average of 7% less in benchmarks.

 

[Jeff7181]

Originally posted by: jjyiz28

Originally posted by: Jeff7181

Originally posted by: jjyiz28

Originally posted by: sugarkang
let's use my aquamark benchies as an example of what happens during async:

BEFORE: KT266A + athlon xp 1700 + PC2100 = cpu score 4888
AFTER 1: NF2 + athlon xp 2500 + PC2100 = cpu score 5010
AFTER 2: NF2 + athlon xp 3200 + PC2100 = cpu score 5616
AFTER 3: NF2 + athlon xp 3200 + PC3200 = cpu score 6923

so, in the *before* scenario, i had an abit kt266a board with everything in sync.
notice when i upgraded to an asus a7n8x and barton 2500, my cpu score only went up by 100 points or so.
EVEN THOUGH i went from 1700 --> 2500.
in the next situation, i OCed the chip and gained 600 points.
but in the last situation when i *matched* the ram to the fsb, my score goes up a whopping 1300 points.

conclusion: DO NOT RUN ASYNCHRONIZED!

solution: go trade in your pc2700 for pc3200 *or* OC your ram to hit 200fsb levels. probably better to just upgrade to the 3200, and then take your total fsb up even higher.

you do know running only 1 benchmark is not definitive/????
a forum user in cpu did a benchmark testing exactly this. he found out , (with a lot of real world benchmarks), that 400fsb/333mem (tight timmings), wins majority of tests than 400fsb/400mem (relaxed timings).
his tests are ]here

also, read this

gamepc1

Those tests Duvie did were using a Pentium 4 which is COMPLETELY different than an Athlon XP... so your argument means nothing.

read this and the later pages. amd setup dual channel.
they compared a 333fsb chip using 333mem, and 333fsb chip using 400mem. 333/400 is async just like 333/266 is async. rough average of 7% less in benchmarks.

Thanks for proving my point for me... I didn't even have to dig for links

 

[jjyiz28]

Originally posted by: Jeff7181

Originally posted by: jjyiz28

Originally posted by: Jeff7181

Originally posted by: jjyiz28

Originally posted by: sugarkang
let's use my aquamark benchies as an example of what happens during async:

BEFORE: KT266A + athlon xp 1700 + PC2100 = cpu score 4888
AFTER 1: NF2 + athlon xp 2500 + PC2100 = cpu score 5010
AFTER 2: NF2 + athlon xp 3200 + PC2100 = cpu score 5616
AFTER 3: NF2 + athlon xp 3200 + PC3200 = cpu score 6923

so, in the *before* scenario, i had an abit kt266a board with everything in sync.
notice when i upgraded to an asus a7n8x and barton 2500, my cpu score only went up by 100 points or so.
EVEN THOUGH i went from 1700 --> 2500.
in the next situation, i OCed the chip and gained 600 points.
but in the last situation when i *matched* the ram to the fsb, my score goes up a whopping 1300 points.

conclusion: DO NOT RUN ASYNCHRONIZED!

solution: go trade in your pc2700 for pc3200 *or* OC your ram to hit 200fsb levels. probably better to just upgrade to the 3200, and then take your total fsb up even higher.

you do know running only 1 benchmark is not definitive/????
a forum user in cpu did a benchmark testing exactly this. he found out , (with a lot of real world benchmarks), that 400fsb/333mem (tight timmings), wins majority of tests than 400fsb/400mem (relaxed timings).
his tests are ]here

also, read this

gamepc1

Those tests Duvie did were using a Pentium 4 which is COMPLETELY different than an Athlon XP... so your argument means nothing.

read this and the later pages. amd setup dual channel.
they compared a 333fsb chip using 333mem, and 333fsb chip using 400mem. 333/400 is async just like 333/266 is async. rough average of 7% less in benchmarks.

Thanks for proving my point for me... I didn't even have to dig for links

7% increase rough ave. don't seem catatrosphic to me. im not willing to spend premium on faster ram, BUT, faster ram has been getting cheaper so......

 

[adams828]

Originally posted by: jjyiz28
7% increase rough ave. don't seem catatrosphic to me. im not willing to spend premium on faster ram, BUT, faster ram has been getting cheaper so......

definitely, if you shop around you can get some good deals. newegg's one day sale had kingston hyperx pc3500 for $107 the other day, cheaper than some of the comparable pc3200 sticks. not saying you need pc3500 or hyperX, just using this as an example showing you don't *have* to pay premiums for faster ram (hm of course, then i guess that means slower ram goes on sale sometimes for even cheaper too... d'oh! )

 

[sugarkang]

you don't have to spend much at all. i mean you don't need to keep your pc2100 right?
all i did was traded in my PC2100 for PC3200. total upgrade cost me $20.

 

[txxxx]

Originally posted by: adams828

Originally posted by: jjyiz28
7% increase rough ave. don't seem catatrosphic to me. im not willing to spend premium on faster ram, BUT, faster ram has been getting cheaper so......

definitely, if you shop around you can get some good deals. newegg's one day sale had kingston hyperx pc3500 for $107 the other day, cheaper than some of the comparable pc3200 sticks. not saying you need pc3500 or hyperX, just using this as an example showing you don't *have* to pay premiums for faster ram (hm of course, then i guess that means slower ram goes on sale sometimes for even cheaper too... d'oh! )

I refuse to buy new memory

 

[jjyiz28]

Originally posted by: sugarkang
you don't have to spend much at all. i mean you don't need to keep your pc2100 right?
all i did was traded in my PC2100 for PC3200. total upgrade cost me $20.

from where/ who??


holy shite!!!!! check out thte prices for sdr ram at newegg. i bought them like 70-80 bucks 512mb, and now that same exact one is for $115. im sure if you really wanted to, break even when i bought it to someone else that can't upgrade, or whatever. wow.

 

[sugarkang]

well i went to an open market swap meet kinda thing. there's a booth with a guy just selling ram every sunday. traded in my pc2100 for 3200. paid the difference.

even if you don't have that available, you can sell your old stick on ebay.

and regarding the ASYNC earlier.
if you've got a 200fsb but 233 ram, that kind of async shouldn't be as bad as having 200fsb and 166ram.

am i totally off on that?

 

[txxxx]

Originally posted by: sugarkang
well i went to an open market swap meet kinda thing. there's a booth with a guy just selling ram every sunday. traded in my pc2100 for 3200. paid the difference.

even if you don't have that available, you can sell your old stick on ebay.

and regarding the ASYNC earlier.
if you've got a 200fsb but 233 ram, that kind of async shouldn't be as bad as having 200fsb and 166ram.

am i totally off on that?

Hmmmm maybe worth considering...

 

[jjyiz28]

Originally posted by: sugarkang
well i went to an open market swap meet kinda thing. there's a booth with a guy just selling ram every sunday. traded in my pc2100 for 3200. paid the difference.

even if you don't have that available, you can sell your old stick on ebay.

and regarding the ASYNC earlier.
if you've got a 200fsb but 233 ram, that kind of async shouldn't be as bad as having 200fsb and 166ram.

am i totally off on that?

in my opnion, 200fsb/233mem, would be similar to 200fsb/166ram. but i think the former would be slightly better though since in some benchmarks it did better than 200/200.

 

[sugarkang]

that would make sense because you're always filling the 200mhz fsb in the first case, and not filling the fsb enough in the second case. i'd still want some decent info on async penalties.

still, there's an overall consensus that running SYNC is the best thing to do.

 

[sugarkang]

that would make sense because you're always filling the 200mhz fsb in the first case, and not filling the fsb enough in the second case. i'd still want some decent info on async penalties.

still, there's an overall consensus that running SYNC is the best thing to do.