A computer components & hardware forum. HardwareBanter

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » HardwareBanter forum » Processors » Overclocking
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

fsb speed - why does it matter?



 
 
Thread Tools Display Modes
  #51  
Old November 5th 04, 05:35 PM
Richard Hopkins
external usenet poster
 
Posts: n/a
Default

"James Hanley" wrote in message...
Course Neither Dual inline nor DDR increase the actual speed.


You have to be careful with your terminology here, as you're in danger of
confusing/transposing different factors. In particular it looks like you're
a little too keen to mix bus speeds and bandwidths.

However, DDR is considered to increase the effective speed, even
though it does not increase the speed in cycles per second.


Double data rate increases bandwidth, not speed. ;-)

It just writes twice as much per cycle, which has the same effect
as working at twice the frequency.


Correct.

according to the table in pcguide.com article "Memory Banks
and Package Bit Width", it says Pentiums have a 64-bit data
bus.


Yes, eight bytes/64 bits. I mentioned this in my previous post.

So a Pentium with DDR RAM DIMMS has a 64-bit data bus and
64-bit memory bus. If the effective speeds are the same, - for
example - the actual FSB is the same, the FSB is dual pumped
and so is the memory bus,


No. The front side bus on the Pentium 4 CPU's is *quad* pumped, so four data
transfers per clock, not two as you say above.

If the memory bus can throw around twice as much data as the
FSB, then it is not an efficient set up.


That comment is incorrect, because your underlying assumption about the P4
processor bus being double pumped is wrong.

It is only efficient if the DDR RAM is half the speed of the FSB.


See above.

Since in one memory bus clock cycle when reading, there are 2
servings for the FSB, it would require 2 FSB cycles to pick up
the data. Similarly when writing.


No. See above.

No, that's wrong. You're ass-u-ming that the geometry of the processor
and memory buses are the same. In most cases they're not.


yeah, I was assuming that, but based on the tables I saw at
pcguide.com and scott mueller's article. Collectively they said that
Pentiums have 64-bit data buses and DDR RAM (at least PC66-
PC4300) has 64-bit memory bus.


They do.

To me, that means that the geometry is the same.


No. You're forgetting that one is DDR and the other QDR.

Snip RH's explanation of P4 and DDR bus geometry

but isn't that a beautiful real world illustration of what I mean.


Lol, depends what you're talking about, as you've just spent the first half
of this post claiming that the memory bus runs twice as quick as the
processor bus, so do you mind telling us exactly what you do mean?

In that example, the effective speed of the FSB is 800=4*200.
And the effective speed of the memory bus is 400=2*200. It's
efficient, because the bandwidth is equal.


Correct.

Although that required the actual speeds to be the same, it means
that the FSB's effective speed is half that of the memory bus.


Eh? You seem to be confusing yourself with all this talk of "effective
speeds". The Pentium 4 bus's "effective speed" (as you call it) is four
times faster than the FSB clock, as there are four transfers per clock
(QDR), so the FSB's effective speed is twice that of the DDR memory bus,
but, to compensate, a dual channel DDR memory bus is twice as wide as the
processor's.

When you say "If you keep your memory synchronous, you also
raise the bandwidth of the memory bus by 33%." You mean
keeping the actual memory speed equal to the processor speed,
right?


Yes.

The word synchronous is a funny one to use, because SDRAM is
always synchronous, that's what the S stands for.


The "S" in SDRAM is there partly for marketing reasons, partly because it
refers to the way the memory works inside each DIMM. You need to separate
the term "Synchronous" (capital S) as it applies to SDRAM memory modules
from the word "synchronous".

If the actual speeds are different, it would stiull be synchronous,


No. You have to understand that "synchronous" (small s) is just a verb that
could be used in any scenario. Could even say that my girlfriend's heart is
synchronous to mine when they start beating together, but that's another
story. ;-)

It literally refers to things running at the same speed. In this scenario,
when I say that the FSB and memory buses are synchronous, it means they're
running at a 1:1 multiplier. No more or less. If I say the memory and
processor buses are asynchronous, it means that the multiplier is something
other than 1:1.

If, OTOH, you keep the FSB the same, but raise the CPU and
memory multipliers by 33%, you get the faster memory and CPU
clocks, but you don't raise the processor bus, so the resulting
bottleneck - assuming that the bandwidths matched in the first
place - will curtail your performance gain.


And in your earlier example
P4 800MHz FSB(200*4). DDR-SDRAM PC3200 400MHz=(200*2)
- a 2:1 multiplier. There's no bottleneck because the bandwidths of the
fsb and memory bus are equal.


Yep, sure, assuming that the memory was dual channel, the bandwidths would
match. What I was trying to say though, was that, if you assumed that the
CPU in the system above was an engineering sample, your "FSB doesn't matter"
theory would be comprehensively disproved by the following test:

Test configuration A:
CPU multiplier: 12x
FSB: 266MHz
FSB:Memory bus multiplier: 1:1

Test configuration B:
CPU multiplier: 18x
FSB: 200MHz
FSB:Memory bus multiplier: 3:4

In the configurations above, CPU core speed A=CPU core speed B, while memory
bus speed A=memory bus speed B, give or take any floating point differences.
However, in configuration A, the processor bus bandwidth would be 8.5GB/sec,
theoretically matched to the memory bandwidth. In configuration B, the
processor bus bandwidth would be reduced to 6.4GB/sec, creating the
bottleneck. Even allowing for inefficiency and latency in the memory bus,
configuration A should still have a tidy performance advantage, despite,
processor and memory speeds being the same.

yeah, and those calculations i did with the 33% increase all agree
with what you're saying.

thanks


No worries. :-)
--


Richard Hopkins
Cardiff, Wales, United Kingdom
(replace .nospam with .com in reply address)

The UK's leading technology reseller www.dabs.com
Get the most out of your digital photos www.dabsxpose.com


  #52  
Old November 5th 04, 05:53 PM
Richard Hopkins
external usenet poster
 
Posts: n/a
Default

"James Hanley" wrote in message...
Course Neither Dual inline nor DDR increase the actual speed.


You have to be careful with your terminology here, as you're in danger of
confusing/transposing different factors. In particular it looks like you're
a little too keen to mix bus speeds and bandwidths.

However, DDR is considered to increase the effective speed, even
though it does not increase the speed in cycles per second.


Double data rate increases bandwidth, not speed. ;-)

It just writes twice as much per cycle, which has the same effect
as working at twice the frequency.


Correct.

according to the table in pcguide.com article "Memory Banks
and Package Bit Width", it says Pentiums have a 64-bit data
bus.


Yes, eight bytes/64 bits. I mentioned this in my previous post.

So a Pentium with DDR RAM DIMMS has a 64-bit data bus and
64-bit memory bus. If the effective speeds are the same, - for
example - the actual FSB is the same, the FSB is dual pumped
and so is the memory bus,


No. The front side bus on the Pentium 4 CPU's is *quad* pumped, so four data
transfers per clock, not two as you say above.

If the memory bus can throw around twice as much data as the
FSB, then it is not an efficient set up.


That comment is incorrect, because your underlying assumption about the P4
processor bus being double pumped is wrong.

It is only efficient if the DDR RAM is half the speed of the FSB.


See above.

Since in one memory bus clock cycle when reading, there are 2
servings for the FSB, it would require 2 FSB cycles to pick up
the data. Similarly when writing.


No. See above.

No, that's wrong. You're ass-u-ming that the geometry of the processor
and memory buses are the same. In most cases they're not.


yeah, I was assuming that, but based on the tables I saw at
pcguide.com and scott mueller's article. Collectively they said that
Pentiums have 64-bit data buses and DDR RAM (at least PC66-
PC4300) has 64-bit memory bus.


They do.

To me, that means that the geometry is the same.


No. You're forgetting that one is DDR and the other QDR.

Snip RH's explanation of P4 and DDR bus geometry

but isn't that a beautiful real world illustration of what I mean.


Lol, depends what you're talking about, as you've just spent the first half
of this post claiming that the memory bus runs twice as quick as the
processor bus, so do you mind telling us exactly what you do mean?

In that example, the effective speed of the FSB is 800=4*200.
And the effective speed of the memory bus is 400=2*200. It's
efficient, because the bandwidth is equal.


Correct.

Although that required the actual speeds to be the same, it means
that the FSB's effective speed is half that of the memory bus.


Eh? You seem to be confusing yourself with all this talk of "effective
speeds". The Pentium 4 bus's "effective speed" (as you call it) is four
times faster than the FSB clock, as there are four transfers per clock
(QDR), so the FSB's effective speed is twice that of the DDR memory bus,
but, to compensate, a dual channel DDR memory bus is twice as wide as the
processor's.

When you say "If you keep your memory synchronous, you also
raise the bandwidth of the memory bus by 33%." You mean
keeping the actual memory speed equal to the processor speed,
right?


Yes.

The word synchronous is a funny one to use, because SDRAM is
always synchronous, that's what the S stands for.


The "S" in SDRAM is there partly for marketing reasons, partly because it
refers to the way the memory works inside each DIMM. You need to separate
the term "Synchronous" (capital S) as it applies to SDRAM memory modules
from the word "synchronous".

If the actual speeds are different, it would stiull be synchronous,


No. You have to understand that "synchronous" (small s) is just a verb that
could be used in any scenario. Could even say that my girlfriend's heart is
synchronous with mine when they start beating together, but I'd never use
such a clinical word to describe such a lovely thing. ;-)

Synchronous literally refers to things running at the same speed. In this
scenario, when I say that the FSB and memory buses are synchronous, it means
they're running at a 1:1 multiplier. No more or less. If I say the memory
and processor buses are asynchronous, it means that the multiplier is
something other than 1:1.

If, OTOH, you keep the FSB the same, but raise the CPU and
memory multipliers by 33%, you get the faster memory and CPU
clocks, but you don't raise the processor bus, so the resulting
bottleneck - assuming that the bandwidths matched in the first
place - will curtail your performance gain.


And in your earlier example
P4 800MHz FSB(200*4). DDR-SDRAM PC3200 400MHz=(200*2)
- a 2:1 multiplier. There's no bottleneck because the bandwidths of the
fsb and memory bus are equal.


Yep, sure, assuming that the memory was dual channel, the bandwidths would
match. What I was trying to say though, was that, if you assumed that the
CPU in the system above was an engineering sample, your "FSB doesn't matter"
theory would be comprehensively disproved by the following test:

Test configuration A:
CPU multiplier: 12x
FSB: 266MHz
FSB:Memory bus multiplier: 1:1

Test configuration B:
CPU multiplier: 18x
FSB: 200MHz
FSB:Memory bus multiplier: 3:4

In the configurations above, CPU core speed A=CPU core speed B, while memory
bus speed A=memory bus speed B, give or take any floating point differences.
However, in configuration A, the processor bus bandwidth would be 8.5GB/sec,
theoretically matched to the memory bandwidth. In configuration B, the
processor bus bandwidth would be reduced to 6.4GB/sec, creating the
bottleneck. Even allowing for inefficiency and latency in the memory bus,
configuration A should still have a tidy performance advantage, despite,
processor and memory speeds being the same.

yeah, and those calculations i did with the 33% increase all agree
with what you're saying.

thanks


No worries. :-)
--


Richard Hopkins
Cardiff, Wales, United Kingdom
(replace .nospam with .com in reply address)

The UK's leading technology reseller www.dabs.com
Get the most out of your digital photos www.dabsxpose.com


  #53  
Old November 7th 04, 01:04 PM
James Hanley
external usenet poster
 
Posts: n/a
Default

"Richard Hopkins" wrote in message ...
"James Hanley" wrote in message...
Course Neither Dual inline nor DDR increase the actual speed.


You have to be careful with your terminology here, as you're in danger of
confusing/transposing different factors. In particular it looks like you're
a little too keen to mix bus speeds and bandwidths.


I slipped up before with the dual punped, quad pomped thing
What I meant was that with the p4 running dual inline memory modules,
is efficient, no bottleneck. Also, the P4's effective speed of
800(200*4) compared to the memory's effective speed of 400(200*2). is
double. To compensate, the memory serves double, since its width is
doubled, and the system is efficient.

However, DDR is considered to increase the effective speed, even
though it does not increase the speed in cycles per second.


Double data rate increases bandwidth, not speed. ;-)


but since the concept of 'effective speed' was invented, memory that
is DDR with an actual clock of 200MHz, is said to have an effective
speed of 400MHz, it's called DDR 400. Even though it does not increase
speed or cycles per second, it just increases bandwidth.
Dual inline also only increases bandwidth, but oddly, it's considered
not to increase 'effective speed'.
Perhaps the term/concept of 'effective speed' should not be used when
discussing what's actually happening. It's just useful to some for
calculating bandwidth, so the FSB's bandwidth of 800*8 comes up the
same as 200*(4*8).
 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
How to speed up my CPU? MC General 11 December 12th 04 09:11 PM
AthlonXP 2000 on MSI KT4AV with (VIA KT400A) chipset Mainboard has Speed ÎÔ»¢²ØÁúCrouching Tiger Hidden Dragon Overclocking AMD Processors 18 May 6th 04 12:14 AM
AthlonXP 2000 on MSI KT4AV with (VIA KT400A) chipset Mainboard has Speed Complexity LongBow Overclocking AMD Processors 7 May 2nd 04 12:23 AM
D865GLC + CPU Fan Speed HELP Ron Reaugh General 1 December 16th 03 03:28 PM
CD burning speed determines read speed? David K General 4 July 22nd 03 09:31 AM


All times are GMT +1. The time now is 10:52 AM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 HardwareBanter.
The comments are property of their posters.