• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

what's a multiplier anyway?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

emericanchaos

Member
Joined
Sep 13, 2002
Location
Williamsport, PA
everyone talks about dropping the multi to raise the FSB. i know what the FSB does and how it works. but what does the multiplier do? what happens when you change it? do you make the core of the processor weaker so you can raise the FSB or is it just a setting in the chip to keep a certain clock speed at the standard FSB speed?
 

PCphreak

Member
Joined
Jul 9, 2002
Location
Dayton, Ohio
The external processor bus (or Frontside bus) along with other busses including AGP and memory bus, PCI bus, etc...all cycle slower than the processor does internally. The back side bus, which is the L2 cache, is the only bus that cycles at the same rate of the processor. What I'm trying to get over here, is the relationship between each bus in terms of timing...

Now the mechanicals applied...

Located on all mainstream motherboards is a PLL, or clock synthesizer crystal, that produces a reference pulse. This pulse is used as the basis of timing for the busses in your system. Since the FSB is in timing to whatever the PLL is set for, all the other busses will either run synchronous or async in cycle time, but in the same rythym with the FSB. The busses cycling slower naturally have a divisor (PCI=1/4 of FSB, for instance) while busses cycling faster will have a multiple, like the processor (internal processor bus & backside bus). By increasing the multiplier on your processor, your simply doing nothing more but allowing your CPU to cycle at a multiple of the FSB. Even though your CPU may be cycling 12 times faster than the FSB, it's still in 'rythm', sending and recieving data on the same 'beats' so to speak.....Hope this helps some...

<edit>
To see a little more on the relation of cycling and frequency (This principle applies between any busses), I'll quote from my previous post.

------------
To get a full grasp on bandwidth and memory throughput, you need to understand it.

Right now DIMMs, whether SDR or DDR, are 64 bit's wide. This is important to note, because the CPU's external bus interface is 64-bits wide also. This is also why we are able to add memory a single DIMM at a time, whereas older SIMMS (32-bits wide) had to be added in pairs to match the CPU's external 64-bit bus interface.

The FSB and memory bus are two different busses, but must cooperatively operate together. They operate together more efficiently when their cycle time is lower and synchronous. As clock speed increases cycle time decreases. Example: a 2 GHz CPU cycles every .5 ns, while a DDR 400 bus cycles every 2.5 ns (ns = nanoseconds). ( 1 / MHz * 1000 = ns).
So what happens when a 2 GHz processor trys to directly access a DDR 400 bus? About 4.5 wait states (blank cycle CPU executes) till data is ready for the CPU on the DDR 400 bus's next event cycle. These wait states effectively slow down the CPU from 2 GHz to the 400 MHz of the bus. Same principle goes for any bus working together. Now you can see why it's important to run synchronous rather than async which introduces bottle necks.

I'm not going to get into the latency of memory, which is the time involved in a setting up a transfer. So I'll move on to bandwidth.

Memory bandwidth is figured by this:
(Bus Width) multiplied by (Clock Speed), and this product multiplied by (Data Cycles Per Clock).
Example: The Athlon's FSB is 64-bits wide and runs at clock speed 133 MHz and manages 2 data cycles per clock. This works out to be
( 8 bytes(64-bits) * 133 MHz * 2 = 2,128 MB/s ), or approx. 2.1 GB/s. To take full advantage of the FSB's bandwidth, you'd need to pair it with memory at the same bandwidth. DDR 2100 does this exactly. ( 8 * 133 MHz * 2 = 2,128 MB/s ). Or ( 8 * 266 MHz = 2,128 MB/s ); however you want to figure it.

Now you have it. Since Athlon is based on a DDR bus, it's better paired with DDR memory to directly match it. Remember, the higher the clock speed of the FSB the closer it's cycle time is to the CPU and less wait states occur in the event of a L2 cache miss.
As far as the P4 goes, it has a 128-byte line size and runs on a quad based bus (high bandwidth; 8 (64-bits) x 133 MHz x 4 = 4,256 MB/s ). So naturally it will benefit from a memory bus able to match. Right now, stock, Dual DDR 400 is right up it's ally.
---------------
</edit>


-PC
 
Last edited:

willlangford

Member
Joined
Aug 29, 2002
Location
Wash., USA, Earth, Milky Way
d--d i know how ya feel...here is what to do read all the stickys and then send a pm to a mod and ask them for help. they most always will know how to help ya.

~Will

u can also drop me a pm and ill help ya out.
 
OP
emericanchaos

emericanchaos

Member
Joined
Sep 13, 2002
Location
Williamsport, PA
thanks. but can we put it in a more general description.

my concern was, does dropping the multiplier make the processor weaker? or is it merely a setting. sounds like it's a setting as it wouldn't change the strength of the processor just the timing of the processor in relation to the other systems in teh machine.
 

repo man11

Member
Joined
Dec 28, 2001
I'm not sure what you mean by "making the processor weaker", but raising and lowering the CPU's multiplier is one of the ways to change the speed the CPU runs at.
Increasing the multiplier will increase the speed at which the CPU runs, and increase system speed and performance. Raising the FSB and leaving the multiplier static will increase the speed that the CPU runs at, and increase system performance. But increasing the FSB increases the speed at which the other system components run, most importantly the memory. So if the fastest you can run your processor is 1,500 MHz, you'll have better system performance by running it at 150MHz FSBx10 clock multiplier, than you will running it at 100MHz FSBx15 clockmultiplier.
 
OP
emericanchaos

emericanchaos

Member
Joined
Sep 13, 2002
Location
Williamsport, PA
by weaker i mean diminishing the point in buying a fast processor. if FSB is worth more than high multiplier than wouldn't buying AMD be pointless as intel's insane FSB speeds would surely asure better performance than the measley FSB speeds the VIA chipsets put ou tin comparison. what's the point of having a strong processor if FSB is where the performance is?
 

The Coolest

Senior Member
Joined
Jul 29, 2001
The FSB is NOT EVERYTHING, although its an important thing. Athlons do like high FSBs, but Intel P4s use a design which greatly differs from and AMD one, so although an Intel P4 does have higher FSBs it doesn't mean the if you take an AMD and a P4 at the same speeds and the P4 will beat it because it has more FSB. More FSB always helps to increase the performance of the system you are running. Intel P4's stronger point is it has wider bandwidth than AMD, but AMD has a much more efficient core. so its sort of a balance. But getting a really good RAM, unlocking the CPU and running as high as possible will most likely get you more performance than just by unlocking the chip and upping the multiplier, because the FSB is important, but its not everything.
 

smudged

Member
Joined
Mar 11, 2002
Location
Watford, South East england
Right this is the basics


Front Side bus X cpu multiplier = MHz

ie: 133 FSB x 11.5 =1529.5MHz (an xp1800)

if you wanted to over lock the whole PC you would increase the FSB

ie 166 x 11.5 = 1909 MHz

If you got to a point where things were not going any higher because the CPU has reached the maximium MHZ it will handle, then you could try unlocking the multiplier and lowering it.

ie: 166 x 10 = 1660 MHz

So you can say that upping FSB will overclock the whole PC and upping the multi will over clock just the CPU.

FSB X multi = MHz
133 x 11.5 =1529.5
133 x 12 =1596

hope this helps
 

The Coolest

Senior Member
Joined
Jul 29, 2001
yup as said above, depends on the benchmakr, a benchmark that only uses raw CPU power won't gain any perofrmance by having higher FSB at the same speed, but in ones that do depend on the FSB like 3DMark2001, you will notice a nice performance boost from just upping the FSB instead the multi
 
OP
emericanchaos

emericanchaos

Member
Joined
Sep 13, 2002
Location
Williamsport, PA
The Coolest said:
...depends on the benchmakr, a benchmark that only uses raw CPU power won't gain any perofrmance by having higher FSB at the same speed, but in ones that do depend on the FSB like 3DMark2001, you will notice a nice performance boost from just upping the FSB instead the multi

ok. so what we're doing by changing the multiplier is changing the rhythm or pattern of the CPU's.... let's call the term duty cycle. so dropping the multiplier would give us more room is this "duty cycle" to raise the FSB so as to give us the same over all result (clock speed) but with more time focused on FSB.

to give a crude example lets imagine a clock. (i'm using clock because of the way the term is used in this area) lets say 60% of the clock's time was red. leaving the other 40% blue. now if blue represents the FSB and red represents the core of the CPU than 60% of the time is dedicated to the processor. that means that the remaining 40 was for FSB. now if changing the multi lower was reducing the amount of time the CPU's core had to 50% of said clock than we would have a remaining 10%. this could be used to boost the amount of time the blue(FSB) side of the clock was used.

if that's the case then we're taking away from the core and weakening the "raw CPU power". right? or does the core not need so much time and is only given that time to make higher processor speeds?

another idea. say we ran a benchmark like SANDRA's burn in. and we ran FSB benchs as well as CPU benchs. now say we have a 1800Mhz processor set at 13.5*133. and we run the burn in and get our results. then we drop the multi to 10 and raise the FSB to 180. when we run the same exact benchs would we see a loss in raw CPU power in the second bench compared to the first?
 

Ishakir

Member
Joined
Sep 3, 2002
Location
Roselle, IL
PCphreak said:
The external processor bus (or Frontside bus) along with other busses including AGP and memory bus, PCI bus, etc...all cycle slower than the processor does internally. The back side bus, which is the L2 cache, is the only bus that cycles at the same rate of the processor. What I'm trying to get over here, is the relationship between each bus in terms of timing...

Now the mechanicals applied...

Located on all mainstream motherboards is a PLL, or clock synthesizer crystal, that produces a reference pulse. This pulse is used as the basis of timing for the busses in your system. Since the FSB is in timing to whatever the PLL is set for, all the other busses will either run synchronous or async in cycle time, but in the same rythym with the FSB. The busses cycling slower naturally have a divisor (PCI=1/4 of FSB, for instance) while busses cycling faster will have a multiple, like the processor (internal processor bus & backside bus). By increasing the multiplier on your processor, your simply doing nothing more but allowing your CPU to cycle at a multiple of the FSB. Even though your CPU may be cycling 12 times faster than the FSB, it's still in 'rythm'.....Hope this helps some...


-PC

PCphreak is god! I've been here a while now, your posts have by far been the best and most accurate. Heh, just wanted to tell you that :)

Won't waste your time anymore lol...

-Ish-
 

PCphreak

Member
Joined
Jul 9, 2002
Location
Dayton, Ohio
Ishakir said:

PCphreak is god! I've been here a while now, your posts have by far been the best and most accurate. Heh, just wanted to tell you that :)
Won't waste your time anymore lol...
-Ish-

Sheesh:) ...I am merely a mortal...LoL:D ....Anyhow, I really do appreciate the kind comment. Makes me feel good, as I try to put most of my free time into these Forums here, either helping other's or learning myself. Overclockers' forums is a well rounded place, full of great individuals!

-PC
 

The Coolest

Senior Member
Joined
Jul 29, 2001
emericanchaos said:


ok. so what we're doing by changing the multiplier is changing the rhythm or pattern of the CPU's.... let's call the term duty cycle. so dropping the multiplier would give us more room is this "duty cycle" to raise the FSB so as to give us the same over all result (clock speed) but with more time focused on FSB.

to give a crude example lets imagine a clock. (i'm using clock because of the way the term is used in this area) lets say 60% of the clock's time was red. leaving the other 40% blue. now if blue represents the FSB and red represents the core of the CPU than 60% of the time is dedicated to the processor. that means that the remaining 40 was for FSB. now if changing the multi lower was reducing the amount of time the CPU's core had to 50% of said clock than we would have a remaining 10%. this could be used to boost the amount of time the blue(FSB) side of the clock was used.

if that's the case then we're taking away from the core and weakening the "raw CPU power". right? or does the core not need so much time and is only given that time to make higher processor speeds?

another idea. say we ran a benchmark like SANDRA's burn in. and we ran FSB benchs as well as CPU benchs. now say we have a 1800Mhz processor set at 13.5*133. and we run the burn in and get our results. then we drop the multi to 10 and raise the FSB to 180. when we run the same exact benchs would we see a loss in raw CPU power in the second bench compared to the first?

No. the raw CPU power is represented in MHz, doesn't matter what the FSB is. so it wouldn't matter running 100x18 or 10x180, the raw CPU power WILL stay the same, but the overall system performance will be greater with the "10x180" setting, because of the higher speed that the memory is running. so always look to get the FSB as high as possible and only then start upping the multi till you reach the fastest stable speed.

on a side note, also running Prime95 as a test for stability is far more efficient than running sandra, one of the reasons is that if Prime95 encounters an error which is a stability issue, it stops, while Sandra can continue, until it hangs on some calculation that it was unable to solve due to high speeds. so prime is much more sensitive to instabilities, while sandra can run as long as it wants, but you still may see stability issues, I highly recommend getting Prime95
 

PCphreak

Member
Joined
Jul 9, 2002
Location
Dayton, Ohio
emericanchaos said:

ok. so what we're doing by changing the multiplier is changing the rhythm or pattern of the CPU's....

Ok. Stop there. Yes, your changing the rhythm of the CPU in terms of MHz, or more accurately called frequency. Frequency can also be converted to cycle time which is expressed in nano seconds.
Example:
133 MHz = 7.5 ns. --( 1 / MHz * 1000 = ns )--

The lower the ns the faster the cycle.

The rule: As frequency increases, cycle time decreases.

The front-side bus frequency is used as a reference for the CPU. The CPU's frequency will always be derived from it's multiplier, multiplied by the FSB's frequency.


emericanchaos said:

another idea. say we ran a benchmark like SANDRA's burn in. and we ran FSB benchs as well as CPU benchs. now say we have a 1800Mhz processor set at 13.5*133. and we run the burn in and get our results. then we drop the multi to 10 and raise the FSB to 180. when we run the same exact benchs would we see a loss in raw CPU power in the second bench compared to the first?

IPC. Stands for: Instruction Per Clock Cycle. The amount of instruction capable of being completed per clock will always remain the same for that CPU.
The only variable in: (Instructions * Clock Cycle), is the clock cycle. Remember cycle time can also be expressed in frequency: MHz. So, the higher the frequency of the CPU, the more instructions that are capable of being completed in a given time frame. Naturally, the lower the frequency, the less instructions capable of being completed. The amount of Instruction processed in a given time, can be directly translated into "CPU power".
Example: Athlon XP
1,800 MHz * 9 (IPC) = 16.2 Billion Instructions capable of being completed per second (CPU power). MHz = Millions of times per second (which is what it's cycling).

emericanchaos said:

if that's the case then we're taking away from the core and weakening the "raw CPU power". right? or does the core not need so much time and is only given that time to make higher processor speeds?

So yes- from my above explanation, you are weaking the "CPU power" by changing the multiplier- since the multiplier controls the frequency of the CPU.


-PC
 
Last edited: