View Full Version : Whats faster, more FSB or a higher multi?
ShadowFold
05-16-2008, 09:48 PM
Im just wondering if 400 x 8 or 337 x 9.5 is a faster 3.2ghz or are they the same?
Solaris17
05-16-2008, 09:50 PM
higher fsb
erocker
05-16-2008, 09:50 PM
A higher FSB would equal more performance.
niko084
05-16-2008, 09:51 PM
Higher FSB and because it affects more than your processor.
imperialreign
05-16-2008, 09:53 PM
agreed, better overall system performance from a higher BUS.
unless you do a lot of task and applications that are CPU-only intensive, higher CPU clock isn't as effective of a performance gain.
spearman914
05-16-2008, 09:56 PM
Higher FSB. Because the FSB affects other components too but multi only affects the cpu.
MOAR FSB
even if you had settings that had everything running at stock except for the CPU and FSB, same speed on the CPU produced by both settings... moar fsb wins
ShadowFold
05-16-2008, 10:08 PM
Thanks
Thanks
Unless you have very good RAM, a higher FSB will mean higher timings for the RAM which will actually slow you down.
spearman914
05-16-2008, 10:14 PM
Unless you have very good RAM, a higher FSB will mean higher timings for the RAM which will actually slow you down.
Also higher fsb will need more voltages which will decreasing liftime which is a bad thing....
Higher speed doesn't always warrant higher voltages.
Also higher fsb will need more voltages which will decreasing liftime which is a bad thing....
Totally right, dude!
That's why there aren't many x * 7 multi overclocks out there! Even with good RAM, those wouldn't last very long (unless they were low OCs).
SirKeldon
05-16-2008, 10:19 PM
It is the same thing for Intel and AMD? Cause i know their architechtures are different in the FSB implementation, a higher FSB is always better?
Ty in advance :toast:
zCexVe
05-16-2008, 10:21 PM
In AMD they always win in Bandwidth with their more superior Hyper Transport.
Solaris17
05-16-2008, 10:21 PM
yes higher FSB is better regardless of architecture..because its over all system bandwidth not just ram or proc...the diff you may be thinking of is timings...from what i here AMD likes tigher timings were as intel it doesnt matter so much
farlex85
05-16-2008, 10:23 PM
Intel's FSB is the chokepoint for the whole system. All the information from ram to cpu and cpu to everything must pass through the fsb. Getting that as high as possible is more important than tight timings. Of course, RAM is a key factor and if it is working very ineffectively, then sacrificing a little fsb for RAM performance is good. Its a balance.
AMD does not have a FSB sirkeldon. Their procs communicate directly w/ the ram via HT i believe, however I am not as familiar w/ their architecture.
niko084
05-16-2008, 10:26 PM
Your correct Farlex85!
This is part of the reason AMD you want tight timings and don't need such massive bandwidth for equal performance.
Intel's FSB is the chokepoint for the whole system. All the information from ram to cpu and cpu to everything must pass through the fsb. Getting that as high as possible is more important than tight timings. Of course, RAM is a key factor and if it is working very ineffectively, then sacrificing a little fsb for RAM performance is good. Its a balance.
AMD does not have a FSB sirkeldon. Their procs communicate directly w/ the ram via HT i believe, however I am not as familiar w/ their architecture.
That's why, in my case, with RAM @ 5-5-5-15 @ 1062 MHz is actually slower then RAM @ 4-4-4-10 @ 884 MHz, even though Everest says otherwise (difference of about 400 in memory read). In Nero Recode, it's slightly faster with 884 then 1062: about 12 seconds.
jbruneau
05-17-2008, 02:05 AM
I agree, a higher FSB would benefit a system more than simply using a higher multiplier. Now, both on the other hand would be even better!
HAT, I'm with you, an overclock does not always involve added voltage. Though, with sufficient cooling, the additional voltage will not affect shelf-life in any way.
As for those worried about memory, that is why they invented dividers. You can run your memory faster than the FSB, the same speed, or slower. Run the crap out of them, drop some more voltage, active cooling and get those timings down. Either that or run them stock. Whichever is your preference.
Ravenas
05-17-2008, 02:07 AM
I agree, a higher FSB would benefit a system more than simply using a higher multiplier. Now, both on the other hand would be even better!
HAT, I'm with you, an overclock does not always involve added voltage. Though, with sufficient cooling, the additional voltage will not affect shelf-life in any way.
As for those worried about memory, that is why they invented dividers. You can run your memory faster than the FSB, the same speed, or slower. Run the crap out of them, drop some more voltage, active cooling and get those timings down. Either that or run them stock. Whichever is your preference.
This is the truth. Both are the best for the greatest overall performance boost.
trog100
05-17-2008, 03:24 AM
the important thing is to get the cpu speed up.. useing a higher multiplier is just an easier way to do this without stressing the rest of the less important bits or adding more instability problems than u have to..
but in theory (in a perfect world) everything should be as fast as possible but higher cpu speed is the best performance gain.. the rest is secondary and might cause more problems than its worth for any small additional gains..
trog
Wile E
05-17-2008, 03:30 AM
With his board, 400fsb is a non-issue. So 400x8 is better either way.
I agree, a higher FSB would benefit a system more than simply using a higher multiplier. Now, both on the other hand would be even better!
HAT, I'm with you, an overclock does not always involve added voltage. Though, with sufficient cooling, the additional voltage will not affect shelf-life in any way.
As for those worried about memory, that is why they invented dividers. You can run your memory faster than the FSB, the same speed, or slower. Run the crap out of them, drop some more voltage, active cooling and get those timings down. Either that or run them stock. Whichever is your preference.
excessive voltage will shorten the life no matter what the cooling is. If I had one of those uber leet xigmatek coolers and I lived in the north pole, I could still run 1.6v through my processor all the time and knock some life off it.
Solaris17
05-17-2008, 04:21 AM
thats absolutely true hat..but with how fast most of this community change parts it usually isnt an issue.
vBulletin® v3.7.0, Copyright ©2000-2008, Jelsoft Enterprises Ltd.