- Joined
- Nov 11, 2003
- Location
- Metropolitan Area
I hear people over clockering cpus d they got like 200fsbs, so i'm wondering..., isn't fsb more important than clock speed?
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
Isn't FSB important?
- Thread starter Keikan
- Start date
- Joined
- Oct 18, 2003
- Location
- Leawood, KS
correct me if im wrong, but clock speed is what the CPU is runnign @ i.e. 2.0ghz. FSB has something to do /w the mobo.
also "over clockering"?
also "over clockering"?
- Joined
- Nov 2, 2003
- Location
- South of Boston
err you should find a basics of computers page
cpu speed = fsb x multiplier
higher fsb has the ram running faster (faster system)
so, if you have a high fsb, and a low multi, but the clock speed is still high, it's best performance
cpu speed = fsb x multiplier
higher fsb has the ram running faster (faster system)
so, if you have a high fsb, and a low multi, but the clock speed is still high, it's best performance
- Joined
- Feb 19, 2003
- Location
- Northants England
crimedog said:
not always
while the fsb might be high u need a good multiplier for the cpu to handle all that bandwidth or all that fsb goes to waste
In the tests I have done with my system I have found that *processor* speed and performance remains fairly constant regardless of FSB - raw MHz seems more important (although a high FSB will tip the scales a few points).
However, a higher FSB means faster RAM speed as well (timings etc. play a factor too though) - and higher benchmarks as a direct result.
FSB ain't "all-that" - but it does affect performance a noticeably.
However, a higher FSB means faster RAM speed as well (timings etc. play a factor too though) - and higher benchmarks as a direct result.
FSB ain't "all-that" - but it does affect performance a noticeably.
Very good example stamasd - but, I'll bet that in Sandra, Prime95, and PCMark 02 (all CPU benchmarks) it performs roughy the same all the way through - when you remove RAM as a factor FSB makes little difference towards processor performance.
And how much of a difference does the higher FSB make to your 3DMark benches? Probably not more than 100 points by my guess. FSB makes a difference - but not a *huge* one. That is with the same Ram timings, and whatnot all the way through
And how much of a difference does the higher FSB make to your 3DMark benches? Probably not more than 100 points by my guess. FSB makes a difference - but not a *huge* one. That is with the same Ram timings, and whatnot all the way through
- Joined
- Jul 24, 2002
- Location
- Seattle, WA
More general system performance probably, yes. In terms of games though, I bet a 10mhz clockspeed increase on your video card's RAM is going to be a lot more beneficial than even a 25mhz increase in FSB.
For a given system, one can approximate the performance or benchmark number like this:
benchmark = A FSB + B CPU
where A and B are some constants to be determined,
FSB is the FSB frequency in MHz,
CPU is the CPU frequency in MHz.
One can look at A as the benchmark gain per FSB MHz, and B as the benchmark gain per CPU MHz.
By running the same benchmark twice with two different settings of FSB and CPU, one can solve A and B (two equations, two unknowns).
For programs that are highly dependent on FSB and less dependent on CPU, the number A is relatively large, e.g. memory benchmark, ....
For programs that are highly dependent on CPU and less dependent on FSB, the number B is relatively large, e.g. Prime95, Sandra CPU MIPS and FLOPS, ....
For programs that are dependent closely on CPU and FSB, both A and B are important, e.g. 3DMark01, ....
benchmark = A FSB + B CPU
where A and B are some constants to be determined,
FSB is the FSB frequency in MHz,
CPU is the CPU frequency in MHz.
One can look at A as the benchmark gain per FSB MHz, and B as the benchmark gain per CPU MHz.
By running the same benchmark twice with two different settings of FSB and CPU, one can solve A and B (two equations, two unknowns).
For programs that are highly dependent on FSB and less dependent on CPU, the number A is relatively large, e.g. memory benchmark, ....
For programs that are highly dependent on CPU and less dependent on FSB, the number B is relatively large, e.g. Prime95, Sandra CPU MIPS and FLOPS, ....
For programs that are dependent closely on CPU and FSB, both A and B are important, e.g. 3DMark01, ....
Last edited:
- Joined
- Feb 13, 2001
- Location
- Twin Cities
You give with one hand and you take with the other. Sometimes you need to retard your memory timings to run at higher FSBs and that takes its toll on overall performance including gaming. Here is a partial quote from another thread:
Don't get sucked into that trap where fsb speed is everything. As I said before, running Synchronously, Dual Channel, my 3D Mark 2001se score for 11x200 @ 2-2-2-11 is 13,639, whereas 10x220 2.5-3-3-9 (needed for 220 FSB) yields a 3D MArk 2001se score of 13670. Well less than 1% improvement...
Whoo, big difference!
Hoot
Don't get sucked into that trap where fsb speed is everything. As I said before, running Synchronously, Dual Channel, my 3D Mark 2001se score for 11x200 @ 2-2-2-11 is 13,639, whereas 10x220 2.5-3-3-9 (needed for 220 FSB) yields a 3D MArk 2001se score of 13670. Well less than 1% improvement...
Whoo, big difference!
Hoot
- Joined
- Jul 24, 2002
- Location
- Seattle, WA
Well, duh. Higher overclock means higher performance...but try some other benches. At high settings (what you'd usually play games at) framerates start to become a lot more dependent on video card bandwidth. I'm saying that if you have a constant CPU speed, then the extra memory bandwidth ALONE you get from a 25mhz FSB increase isn't going to make as much of a difference in games as a 10mhz video card RAM increase, at least in the majority of games.
- Joined
- Jul 20, 2003
thats such bull... oc your vid cards CORE 10 mhz.. i bet u dont even get 2fps diff.. you get performance from the vid cards memory more than you do the core...
the only games/benchies that are out now that are limited by the vid card so much where 10mhz would make a diff is DX9 games.. .. and when those dx9 games start to flow out... then i'll bother ocin my vid card. its pointless to say there is no point in having a high fsb.. why dont u just say there is no point in having a high anything.. go back to 100mhz sdr then
the only games/benchies that are out now that are limited by the vid card so much where 10mhz would make a diff is DX9 games.. .. and when those dx9 games start to flow out... then i'll bother ocin my vid card. its pointless to say there is no point in having a high fsb.. why dont u just say there is no point in having a high anything.. go back to 100mhz sdr then
- Joined
- Jul 24, 2002
- Location
- Seattle, WA
I in no way even implied that there's no point in having a high FSB.
There's definitely a point in having a high everything...that's why the proc in my sig runs around 2.6GHz, at as high an FSB as I can get. If you can point out where I said "there's no point to having a high FSB" then I'd be very happy to edit it. The higher the FSB the better. It just isn't as important to most games these days as a faster video card is.
If you bothered to read my post, you'd see that I said:
By RAM I meant memory.
There's definitely a point in having a high everything...that's why the proc in my sig runs around 2.6GHz, at as high an FSB as I can get. If you can point out where I said "there's no point to having a high FSB" then I'd be very happy to edit it. The higher the FSB the better. It just isn't as important to most games these days as a faster video card is.
Similar threads
