What determines your FSB setting?

G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I think
PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated for what
determines what you set your front-side bus speed to (in this case 166
MHZ)?
 

Jim

Distinguished
Mar 31, 2004
2,444
0
19,780
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

You're on the right track, but there is a more precise, accurate way to
describe it. To keep it simple, let's assume no overclocking.

You typically configure the CPU (FSB) on the motherboard to match the clock
speed of the processor. Assuming this Barton is 333MHz DDR (166MHz actual,
or "clock"), that's what you use. Usually you buy *matching* memory so you
can run synchronously (1:1), which simply means it has the same "clock"
speed as the CPU FSB. PC2700 is 166MHz actual/clock (333MHz DDR), so it is
a perfect match in this case.

So in the end, you arrived at the same/right conclusion, but I didn't want
to leave a misimpression. You don't typically buy memory at some given
speed "X", then set your CPU FSB to match it. That's literally how you
described it. You never let the memory speed *drive* the CPU FSB, it's the
other way around. You buy the best CPU you can afford, they buy memory to
match it. The cart pulls the horse, not the other way around!

In the simplest case, you match CPU FSB "clock" speed to memory "clock"
speed. But nothing prevents you from running FASTER memory if you prefer,
just so long as your motherboard supports it. So it may also be possible to
run say, PC3200 (which is 200MHz "clock"). In this case, you'd have the CPU
FSB running asynchronously to the memory (166MHz vs. 200MHz). It would
require the motherboard to support a 5:6 CPU/DRAM ratio too (166 / 5 * 6 =
200), at least to run at spec. This isn't always more efficient,
asynchronous behavior sometimes undermines any benefits you expected by
increasing the memory speed, in this example, by 33MHz. But some people do
anyway, and claim more performance.

Some people also buy faster clocked memory because they intend to overclock
the CPU FSB (run above it's spec of 166MHz in this example). By doing so,
their attempts to get more clock speed out of the FSB and continue to run
synchronously (CPU clock = Memory clock) is not hindered by the memory, the
*extra* clock speed afforded by the memory at 200MHz provides *headroom* for
the overclocking attempt on the CPU FSB. IOW, on the march to reach 200MHz
w/ the CPU FSB, the memory is never overclocked itself, indeed, it always
remains underclocked.

HTH

Jim


"Jim Caldwell" <Jamesw@grandecom.net> wrote in message
news:c5jhol$2g73b$1@ID-153828.news.uni-berlin.de...
> I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I
think
> PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated for
what
> determines what you set your front-side bus speed to (in this case 166
> MHZ)?
>
>
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

PC2700 = 266MHz. The DDR (Double Data Rate) runs at Double the FSB of
133Mhz, thus at 266MHz.

--
DaveW



"Jim Caldwell" <Jamesw@grandecom.net> wrote in message
news:c5jhol$2g73b$1@ID-153828.news.uni-berlin.de...
> I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I
think
> PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated for
what
> determines what you set your front-side bus speed to (in this case 166
> MHZ)?
>
>
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

Hi Jim
you can have perfect synchronizing with your computer
your mother board support up to 400 Mhz FSB (provided that you have
BIOS version 2, see Asus website)
and your AMD XP2600 support 333Mhz FSB. At the same time your memory
speed is 2700/8=337.5
SO you can set the FSB in your BIOS to be 333Mhz. I think the mother
board decided the speed of the FSB.
HTH

"Jim Caldwell" <Jamesw@grandecom.net> wrote in message news:<c5jhol$2g73b$1@ID-153828.news.uni-berlin.de>...
> I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I think
> PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated for what
> determines what you set your front-side bus speed to (in this case 166
> MHZ)?
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

Jim you said
<quote>... In this case, you'd have the CPU FSB running asynchronously
to the memory (166MHz vs. 200MHz).</quote>

But I thought AMD XP2600 has FSB = 333Mhz. So in this case even with
PC3200 he can get good synchronization! Am I right??

"Jim" <null@null.com> wrote in message news:<r4dfc.2650$Yf6.587@fed1read07>...
> You're on the right track, but there is a more precise, accurate way to
> describe it. To keep it simple, let's assume no overclocking.
>
> You typically configure the CPU (FSB) on the motherboard to match the clock
> speed of the processor. Assuming this Barton is 333MHz DDR (166MHz actual,
> or "clock"), that's what you use. Usually you buy *matching* memory so you
> can run synchronously (1:1), which simply means it has the same "clock"
> speed as the CPU FSB. PC2700 is 166MHz actual/clock (333MHz DDR), so it is
> a perfect match in this case.
>
> So in the end, you arrived at the same/right conclusion, but I didn't want
> to leave a misimpression. You don't typically buy memory at some given
> speed "X", then set your CPU FSB to match it. That's literally how you
> described it. You never let the memory speed *drive* the CPU FSB, it's the
> other way around. You buy the best CPU you can afford, they buy memory to
> match it. The cart pulls the horse, not the other way around!
>
> In the simplest case, you match CPU FSB "clock" speed to memory "clock"
> speed. But nothing prevents you from running FASTER memory if you prefer,
> just so long as your motherboard supports it. So it may also be possible to
> run say, PC3200 (which is 200MHz "clock"). In this case, you'd have the CPU
> FSB running asynchronously to the memory (166MHz vs. 200MHz). It would
> require the motherboard to support a 5:6 CPU/DRAM ratio too (166 / 5 * 6 =
> 200), at least to run at spec. This isn't always more efficient,
> asynchronous behavior sometimes undermines any benefits you expected by
> increasing the memory speed, in this example, by 33MHz. But some people do
> anyway, and claim more performance.
>
> Some people also buy faster clocked memory because they intend to overclock
> the CPU FSB (run above it's spec of 166MHz in this example). By doing so,
> their attempts to get more clock speed out of the FSB and continue to run
> synchronously (CPU clock = Memory clock) is not hindered by the memory, the
> *extra* clock speed afforded by the memory at 200MHz provides *headroom* for
> the overclocking attempt on the CPU FSB. IOW, on the march to reach 200MHz
> w/ the CPU FSB, the memory is never overclocked itself, indeed, it always
> remains underclocked.
>
> HTH
>
> Jim
>
>
> "Jim Caldwell" <Jamesw@grandecom.net> wrote in message
> news:c5jhol$2g73b$1@ID-153828.news.uni-berlin.de...
> > I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I
> think
> > PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated for
> what
> > determines what you set your front-side bus speed to (in this case 166
> > MHZ)?
> >
> >
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

How you got 266Mhz??
pC2700 = 2700/8 = 337.5 MBit/Sec (DDR) = 168.75 MBit/sec

"DaveW" <none@zero.org> wrote in message news:<YFjfc.37951$xn4.95902@attbi_s51>...
> PC2700 = 266MHz. The DDR (Double Data Rate) runs at Double the FSB of
> 133Mhz, thus at 266MHz.
>
> --
> DaveW
>
>
>
> "Jim Caldwell" <Jamesw@grandecom.net> wrote in message
> news:c5jhol$2g73b$1@ID-153828.news.uni-berlin.de...
> > I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I
> think
> > PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated for
> what
> > determines what you set your front-side bus speed to (in this case 166
> > MHZ)?
> >
> >
 

Jim

Distinguished
Mar 31, 2004
2,444
0
19,780
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

"esara" <esara123@hotmail.com> wrote in message
news:fd7d27e7.0404150759.28d49a3d@posting.google.com...
> Jim you said
> <quote>... In this case, you'd have the CPU FSB running asynchronously
> to the memory (166MHz vs. 200MHz).</quote>
>
> But I thought AMD XP2600 has FSB = 333Mhz. So in this case even with
> PC3200 he can get good synchronization! Am I right??

This is where the confusion comes in. We have to differentiate between the
actual "clock" speed, vs. *effective* speed (for lack of a better term).
When you install that AMD XP2600 w/ FSB 333 in your mobo, the motherboard
will generate a 166MHz "clock" to drive the FSB. That AMD CPU supports an
actual clock of 166MHz (spec'd, assuming no OC'ing), but because it
implements DDR, it's doing twice (2x) the amount of work as a non-DDR
processor w/ the same 166MHz clock. It does this by sending data on BOTH
the up and down side of the clock cycle. IOW, it's more efficient. AMD
sells and markets the CPU as 333MHz FSB (2 x 166MHz clock) to make clear
this is the peformance level you can expect, but it's not the actual "clock"
speed that it runs at.

Same holds true for PC3200. This memory "clocks" at 200MHz (spec'd,
assuming no OC'ing). The PC3200 label is simply a reflection of its
performance, which is 200MHz (clock) * 2 (for DDR) * 8 (bits wide) = 3200!
IOW, the speed of the memory is 3.2Gb/sec (theoretically).

Finally, for synchronization purposes, it's the clock that matters. Since
that AMD is 166MHz clock, and the PC3200 memory is 200MHz clock, these are
NOT synnchronized. Using that AMD processor (w/ 166MHz clock), the *ideal*,
synchronized memory would be PC2700 (2700 / 2 (for DDR) / 8 (bit wide) =
166MHz clock).

Of course, some people try to *force* synchronization by OC'ing one or the
other, but that's another story for another time.

HTH

Jim

>
> "Jim" <null@null.com> wrote in message
news:<r4dfc.2650$Yf6.587@fed1read07>...
> > You're on the right track, but there is a more precise, accurate way to
> > describe it. To keep it simple, let's assume no overclocking.
> >
> > You typically configure the CPU (FSB) on the motherboard to match the
clock
> > speed of the processor. Assuming this Barton is 333MHz DDR (166MHz
actual,
> > or "clock"), that's what you use. Usually you buy *matching* memory so
you
> > can run synchronously (1:1), which simply means it has the same "clock"
> > speed as the CPU FSB. PC2700 is 166MHz actual/clock (333MHz DDR), so it
is
> > a perfect match in this case.
> >
> > So in the end, you arrived at the same/right conclusion, but I didn't
want
> > to leave a misimpression. You don't typically buy memory at some given
> > speed "X", then set your CPU FSB to match it. That's literally how you
> > described it. You never let the memory speed *drive* the CPU FSB, it's
the
> > other way around. You buy the best CPU you can afford, they buy memory
to
> > match it. The cart pulls the horse, not the other way around!
> >
> > In the simplest case, you match CPU FSB "clock" speed to memory "clock"
> > speed. But nothing prevents you from running FASTER memory if you
prefer,
> > just so long as your motherboard supports it. So it may also be
possible to
> > run say, PC3200 (which is 200MHz "clock"). In this case, you'd have the
CPU
> > FSB running asynchronously to the memory (166MHz vs. 200MHz). It would
> > require the motherboard to support a 5:6 CPU/DRAM ratio too (166 / 5 * 6
=
> > 200), at least to run at spec. This isn't always more efficient,
> > asynchronous behavior sometimes undermines any benefits you expected by
> > increasing the memory speed, in this example, by 33MHz. But some people
do
> > anyway, and claim more performance.
> >
> > Some people also buy faster clocked memory because they intend to
overclock
> > the CPU FSB (run above it's spec of 166MHz in this example). By doing
so,
> > their attempts to get more clock speed out of the FSB and continue to
run
> > synchronously (CPU clock = Memory clock) is not hindered by the memory,
the
> > *extra* clock speed afforded by the memory at 200MHz provides *headroom*
for
> > the overclocking attempt on the CPU FSB. IOW, on the march to reach
200MHz
> > w/ the CPU FSB, the memory is never overclocked itself, indeed, it
always
> > remains underclocked.
> >
> > HTH
> >
> > Jim
> >
> >
> > "Jim Caldwell" <Jamesw@grandecom.net> wrote in message
> > news:c5jhol$2g73b$1@ID-153828.news.uni-berlin.de...
> > > I have a Asus A7N8X with a XP 2600 Barton, 1 gb of PC 2700 DDR ram. I
> > think
> > > PC 2700 is equivalent to 166 Mhz., so is whatever your RAM is rated
for
> > what
> > > determines what you set your front-side bus speed to (in this case
166
> > > MHZ)?
> > >
> > >
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Wed, 14 Apr 2004 23:19:20 +0000, DaveW wrote:

> PC2700 = 266MHz. The DDR (Double Data Rate) runs at Double the FSB of
> 133Mhz, thus at 266MHz.

PC2700 = 2700MBps, not 266MHz, There's nothing in it running at 266MHz.
The Bus speed is 166MHz, DDR (333Mbps per line, NOT MHz). And the FSB and
memory bus are 2 seperate buses.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

Wes
which bus speed you are talking about.. do you mean the memory bus??
in this case yes it is 166 but not the FSB which is 333Mhz

Wes Newell <w.newell@TAKEOUTverizon.net> wrote in message news:<pan.2004.04.15.10.37.29.714403@TAKEOUTverizon.net>...
> On Wed, 14 Apr 2004 23:19:20 +0000, DaveW wrote:
>
> > PC2700 = 266MHz. The DDR (Double Data Rate) runs at Double the FSB of
> > 133Mhz, thus at 266MHz.
>
> PC2700 = 2700MBps, not 266MHz, There's nothing in it running at 266MHz.
> The Bus speed is 166MHz, DDR (333Mbps per line, NOT MHz). And the FSB and
> memory bus are 2 seperate buses.
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Thu, 15 Apr 2004 09:05:54 -0700, esara wrote:

> Wes Newell <w.newell@TAKEOUTverizon.net> wrote in message
> news:<pan.2004.04.15.10.37.29.714403@TAKEOUTverizon.net>...
>> On Wed, 14 Apr 2004 23:19:20 +0000, DaveW wrote:
>>
>> > PC2700 = 266MHz. The DDR (Double Data Rate) runs at Double the FSB
>> > of 133Mhz, thus at 266MHz.
>>
>> PC2700 = 2700MBps, not 266MHz, There's nothing in it running at 266MHz.
>> The Bus speed is 166MHz, DDR (333Mbps per line, NOT MHz). And the FSB
>> and memory bus are 2 seperate buses.
>
> Wes
> which bus speed you are talking about.. do you mean the memory bus?? in
> this case yes it is 166 but not the FSB which is 333Mhz
>
Fixed top posted message.
AMD (or Intel for that matter) doesn't make a cpu with a FSB speed of more
than 200MHz. Actually, they don't make cpu's with any FSB speed. FSB speed
is determined by the MB and can be set to any value the MB supports. Bus
speeds are measured by the clock speed in Hz. This clock speed along with
the multiplier define the cpu internal clock. Why am I typing this
again... See The Real Front Side Bus in link below.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On 15 Apr 2004 09:05:54 -0700, esara123@hotmail.com (esara) wrote:

>Wes
>which bus speed you are talking about.. do you mean the memory bus??
>in this case yes it is 166 but not the FSB which is 333Mhz

333MHz is the *effective FSB speed* on an AMD FSB at a *FSB clock
speed* of 166MHz.

Just "bus speed" is a good way to confuse things. ;-)

The frequency of data transfers occurring on the bus is 333MHz. The
width of the bus is 8 Bytes, so we get a max data rate of 8B X 333MHz
= 2667x10^6 Bps

(And here's a question for Wes, if he reads this:
Is 2667x10^6 Bps equal to 2667 MBps or 2543 MBps?)

The transfers are synched on both rising and falling flanks of the bus
clock, so the *FSB clock speed* is only 166MHz.

There is one clock. This clock controls both cpu clock speed and the
FSB. And this clock is in fact identical to 'FSB clock speed'.

There's some risk of confusing 'effective FSB speed' - introduced by
the cpu makers - and 'FSB clock speed'. So for completeness sake, a
little table:

'effective FSB' 'FSB clock speed'

Intel P4 at 800FSB 200MHz
Intel P4 at 533FSB 133MHz
Athlon at 400FSB 200MHz
Athlon at 333FSB 166MHz
Athlon at 266FSB 133MHz

And this doesn't involve 'memory bus' in any way. But on the memory
bus, you have the corresponding confusion with *DDR* rating, Like
DDR333 and the memory bus frequency,166MHz for DDR333. This is analog
to *effective FSB speed* and *FSB clock speed*, but concerns another
bus.

ancra
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Thu, 15 Apr 2004 23:40:00 +0200, somebody wrote:

> The frequency of data transfers occurring on the bus is 333MHz.

For clearity, that's 333Mbps per line.

> The width of the bus is 8 Bytes, so we get a max data rate of 8B X
> 333MHz = 2667x10^6 Bps
>
Don't know how you come up with these numbers. I get 64x333M = 21312Mbps
or 2664MBps.

> (And here's a question for Wes, if he reads this: Is 2667x10^6 Bps equal
> to 2667 MBps or 2543 MBps?)
>
2667x((10^6)x8)=21336Mbps or 2667MBps. :)
Calculators work wonders. I still don't know where you came up with 2667.
2664 maybe?

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

ancra
How did you know the "Effective FSB" and "FSB Clock Speed" for Intel
and AMD?? Does the table below that you wrote come with the spec of
each CPU? If not how can I know the "effective FSB" and the "FSB Clock
speed or the actual FSB speed"
Thanks a lot.

Intel P4 at 800FSB 200MHz
Intel P4 at 533FSB 133MHz
Athlon at 400FSB 200MHz
Athlon at 333FSB 166MHz
Athlon at 266FSB 133MHz

somebody@some.domain wrote in message news:<t50u70dnsanhpt3dp3b3r6c6u0nevn18el@4ax.com>...
> On 15 Apr 2004 09:05:54 -0700, esara123@hotmail.com (esara) wrote:
>
> >Wes
> >which bus speed you are talking about.. do you mean the memory bus??
> >in this case yes it is 166 but not the FSB which is 333Mhz
>
> 333MHz is the *effective FSB speed* on an AMD FSB at a *FSB clock
> speed* of 166MHz.
>
> Just "bus speed" is a good way to confuse things. ;-)
>
> The frequency of data transfers occurring on the bus is 333MHz. The
> width of the bus is 8 Bytes, so we get a max data rate of 8B X 333MHz
> = 2667x10^6 Bps
>
> (And here's a question for Wes, if he reads this:
> Is 2667x10^6 Bps equal to 2667 MBps or 2543 MBps?)
>
> The transfers are synched on both rising and falling flanks of the bus
> clock, so the *FSB clock speed* is only 166MHz.
>
> There is one clock. This clock controls both cpu clock speed and the
> FSB. And this clock is in fact identical to 'FSB clock speed'.
>
> There's some risk of confusing 'effective FSB speed' - introduced by
> the cpu makers - and 'FSB clock speed'. So for completeness sake, a
> little table:
>
> 'effective FSB' 'FSB clock speed'
>
> Intel P4 at 800FSB 200MHz
> Intel P4 at 533FSB 133MHz
> Athlon at 400FSB 200MHz
> Athlon at 333FSB 166MHz
> Athlon at 266FSB 133MHz
>
> And this doesn't involve 'memory bus' in any way. But on the memory
> bus, you have the corresponding confusion with *DDR* rating, Like
> DDR333 and the memory bus frequency,166MHz for DDR333. This is analog
> to *effective FSB speed* and *FSB clock speed*, but concerns another
> bus.
>
> ancra
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

ancra
Sorry I also forgot to ask about the DDR memory .. can we say that the
"actual speed or clock speed of DDR memory is 1/2 the effective
speed". I mean if the DDR is 333Mhz then the actual speed is 333/2 and
if DDR is 266 (which is the effective speed) then the actual speed is
266/2. Can I say so??
Thanks..

somebody@some.domain wrote in message news:<t50u70dnsanhpt3dp3b3r6c6u0nevn18el@4ax.com>...
> On 15 Apr 2004 09:05:54 -0700, esara123@hotmail.com (esara) wrote:
>
> >Wes
> >which bus speed you are talking about.. do you mean the memory bus??
> >in this case yes it is 166 but not the FSB which is 333Mhz
>
> 333MHz is the *effective FSB speed* on an AMD FSB at a *FSB clock
> speed* of 166MHz.
>
> Just "bus speed" is a good way to confuse things. ;-)
>
> The frequency of data transfers occurring on the bus is 333MHz. The
> width of the bus is 8 Bytes, so we get a max data rate of 8B X 333MHz
> = 2667x10^6 Bps
>
> (And here's a question for Wes, if he reads this:
> Is 2667x10^6 Bps equal to 2667 MBps or 2543 MBps?)
>
> The transfers are synched on both rising and falling flanks of the bus
> clock, so the *FSB clock speed* is only 166MHz.
>
> There is one clock. This clock controls both cpu clock speed and the
> FSB. And this clock is in fact identical to 'FSB clock speed'.
>
> There's some risk of confusing 'effective FSB speed' - introduced by
> the cpu makers - and 'FSB clock speed'. So for completeness sake, a
> little table:
>
> 'effective FSB' 'FSB clock speed'
>
> Intel P4 at 800FSB 200MHz
> Intel P4 at 533FSB 133MHz
> Athlon at 400FSB 200MHz
> Athlon at 333FSB 166MHz
> Athlon at 266FSB 133MHz
>
> And this doesn't involve 'memory bus' in any way. But on the memory
> bus, you have the corresponding confusion with *DDR* rating, Like
> DDR333 and the memory bus frequency,166MHz for DDR333. This is analog
> to *effective FSB speed* and *FSB clock speed*, but concerns another
> bus.
>
> ancra
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Sat, 17 Apr 2004 10:38:24 -0700, esara wrote:

> Sorry I also forgot to ask about the DDR memory .. can we say that the
> "actual speed or clock speed of DDR memory is 1/2 the effective
> speed". I mean if the DDR is 333Mhz then the actual speed is 333/2 and
> if DDR is 266 (which is the effective speed) then the actual speed is
> 266/2. Can I say so??
> Thanks..
>
Why even bother with "effective speed". It's bascally a worthless
statement without knowing what you are comparing the efectiveness to.:)
Some examples:
A P4/800 only has an effective speed of 400MHz when compared to an Athlon
200MHz (or is that 400MHz effective) FSB cpu.:)
So I could say an Athlon/400 has an effective FSB speed of 400MHz and the
P4/800 also has an effective FSB speed of 400MHz and be completely
correct.
And that just one more reason the CPU FSB speed is measured in clock
cycles, not data rates.
And until you admit to ones self that the numbers are just bogus BS, you
will continue to have this confusion.
Bus clocks are measured in Hz, data rates aren't, and never should the two
be intertwined.


--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On 17 Apr 2004 10:30:54 -0700, esara123@hotmail.com (esara) wrote:

>ancra
>How did you know the "Effective FSB" and "FSB Clock Speed" for Intel
>and AMD?? Does the table below that you wrote come with the spec of
>each CPU? If not how can I know the "effective FSB" and the "FSB Clock
>speed or the actual FSB speed"
>Thanks a lot.
>
>Intel P4 at 800FSB 200MHz
>Intel P4 at 533FSB 133MHz
>Athlon at 400FSB 200MHz
>Athlon at 333FSB 166MHz
>Athlon at 266FSB 133MHz

The cpus are marketed with their 'effective FSB speed'. That's how you
buy them. The 'FSB clock speed' is what you need to set the mobo FSB
clock to.
(There's other older P4s and Athlons too, other than those I listed.)

"actual FSB speed" I don't understand. Please keep to "effective" or
*clock* respectively. Both are actual and real.
FSB Clockspeed is the frequency of the clock that syncs transfers.
Effective FSB speed is the frequency of transfers.

The Intel P4 varieties (P4B, P4C, P4E and P4EE) with 533 and 800 MHz
FSB, have what Intel call a "quad pumped" bus. This means the
transfers can occur 4 times per clock cycle. So 4 * 200 = 800 and 4 *
133 = 533.

AMD is using DEC's old Alpha bus. This syncs on both rise and fall of
clock. So effective speed is simply twice that of the clock.

ancra
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On 17 Apr 2004 10:38:24 -0700, esara123@hotmail.com (esara) wrote:

>ancra
>Sorry I also forgot to ask about the DDR memory .. can we say that the
>"actual speed or clock speed of DDR memory is 1/2 the effective
>speed". I mean if the DDR is 333Mhz then the actual speed is 333/2 and
>if DDR is 266 (which is the effective speed) then the actual speed is
>266/2. Can I say so??
>Thanks..

This (333/2, 266/2 etc) is what the memory bus should be set to, for
DDR ram. So, for instance, DDR333 should run on a 166MHz memory bus.

ancra
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Sun, 18 Apr 2004 03:10:18 +0200, somebody wrote:

> On Sat, 17 Apr 2004 06:40:53 GMT, Wes Newell
> <w.newell@TAKEOUTverizon.net> wrote:
>
>>On Fri, 16 Apr 2004 23:33:49 +0200, somebody wrote:
>>
>>> On Fri, 16 Apr 2004 08:44:25 GMT, Wes Newell
>>> <w.newell@TAKEOUTverizon.net> wrote:
>>>
>>>>On Thu, 15 Apr 2004 23:40:00 +0200, somebody wrote:
>>>>
>>>>> The frequency of data transfers occurring on the bus is 333MHz.
>>>>
>>>>For clearity, that's 333Mbps per line.
>>>
>>> The nomenclature in wide use is "333MHz FSB effective speed", so it's
>>> not objective that Mbpspb is clearer. But my question was rather if it
>>> really is the same thing. Superficially, one might be tempted by:
>>>
>>> 333 Mbps per bitline = 333 Mb ps pb = 333 M * b * s^-1 * b^-1 =
>>> = 333 M * s^-1 = 333MHz
>>>
>>> Here's the problem: MHz is 10^6 * Hz.
>>> But what does the M stand for in Mb?
>>>
>>Since it originally equaled 1 million (in 333MHz or 333Mbps however you
>>want to put it), then it carries that value through the equation, and not
>>1024K. It started as a decimal value, not hex, so it can't be changed
>>during the equation. And throughput is always defined in decimal afaik.
>
> Thank you. I thought I remembered having seen that (decimal) stated
> before somewhere, but I wasn't sure.
>
> That means 'MHz' and 'Mbps per bit' are exactly the same unit for
> frequency. (I prefer MHz ;) And Mbps is something different,
> troughput, rather than frequency)
>
No it doesn't mean anything of the sort. Hz is used for a clock cycle. A
clock cycle is constant. Bps means bits per second. This could be constant
or not. It doesn't have a consant cycle from low to high as does the clock.

> But it also means things like:
>
> "MB per second" is not the same as "MBps". Interesting observation, I
> think.
>
Huh! Both mean MegaBytes per second. In data throughput M is always 1
million. Actaully, it probably properly MiliBytes, but everyone uses the
former.


--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Sun, 18 Apr 2004 03:08:39 +0200, somebody wrote:

> The cpus are marketed with their 'effective FSB speed'. That's how you
> buy them. The 'FSB clock speed' is what you need to set the mobo FSB
> clock to.
> (There's other older P4s and Athlons too, other than those I listed.)
>
So this kinda makes the so called efective speed pretty much worthless.:)

> "actual FSB speed" I don't understand. Please keep to "effective" or
> *clock* respectively. Both are actual and real. FSB Clockspeed is the
> frequency of the clock that syncs transfers. Effective FSB speed is the
> frequency of transfers.
>
Let me help you. Actual means real. FSB means Front Side Bus. And speed
means the clock speed. Now you know.:)

> The Intel P4 varieties (P4B, P4C, P4E and P4EE) with 533 and 800 MHz
> FSB, have what Intel call a "quad pumped" bus. This means the transfers
> can occur 4 times per clock cycle. So 4 * 200 = 800 and 4 * 133 = 533.
>
I think everyone that reads this uderstand the Intel bus. I prefer QDR for
Quad Data Rate, which is more precise than quad pumped, which sounds like
you're at a gas station pumping gas instead of data.:)

> AMD is using DEC's old Alpha bus. This syncs on both rise and fall of
> clock. So effective speed is simply twice that of the clock.
>
DDR would have done here. :)

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Sun, 18 Apr 2004 06:20:15 GMT, Wes Newell
<w.newell@TAKEOUTverizon.net> wrote:

>On Sun, 18 Apr 2004 03:10:18 +0200, somebody wrote:

>> That means 'MHz' and 'Mbps per bit' are exactly the same unit for
>> frequency. (I prefer MHz ;) And Mbps is something different,
>> troughput, rather than frequency)
>>
>No it doesn't mean anything of the sort. Hz is used for a clock cycle. A
>clock cycle is constant. Bps means bits per second. This could be constant
>or not. It doesn't have a consant cycle from low to high as does the clock.

(Bps means bytes per second, ...and I know you were starting a new
sentence :))
I didn't say 'bps' I said 'bps p b', which was your construct, I
believe. Personally, I find that *datawidth per datawidth* a somewhat
contorted effort to hide the fact that we're really dealing with s^-1.

As for MHz:
The nomenclature used by AMD and Intel in their specs and in their
technical documentation, is: "800MHz bus", "800MHz speed", "800MHz
FSB" etc, " _off_ a 200MHz system clock", " - 200MHz bus clock" etc.
This is also the nomenclature used by chip set and mobo manufacturers,
when referring to supported features.
It's also the nomenclature used by distributors, retailers and media,
and so on.
"Effective speed" is (again) _not_ data rate. And it is always
measured in MHz, whether you think it's appropriate or not. And MHz is
a unit for frequency. SI definition being 10^6 * s^-1. (Which does
seem appropriate enough to me, apples being apples.)

>> "MB per second" is not the same as "MBps". Interesting observation, I
>> think.
>>
>Huh! Both mean MegaBytes per second. In data throughput M is always 1
>million. Actaully, it probably properly MiliBytes, but everyone uses the
>former.

Well, thank you again, I take that to mean that I can usually depend
on M being decimal in any troughput figure.

ancra
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Sun, 18 Apr 2004 06:37:05 GMT, Wes Newell
<w.newell@TAKEOUTverizon.net> wrote:

>On Sun, 18 Apr 2004 03:08:39 +0200, somebody wrote:

>> The Intel P4 varieties (P4B, P4C, P4E and P4EE) with 533 and 800 MHz
>> FSB, have what Intel call a "quad pumped" bus. This means the transfers
>> can occur 4 times per clock cycle. So 4 * 200 = 800 and 4 * 133 = 533.
>>
>I think everyone that reads this uderstand the Intel bus.

(?) ...I assumed esara asked.

> I prefer QDR for
>Quad Data Rate, which is more precise than quad pumped, which sounds like
>you're at a gas station pumping gas instead of data.:)

Ok, the nomenclature Intel is using, in their technical documentation
is "quad pumped", so I'll just stick to that, for the sake of
consistency.

>> AMD is using DEC's old Alpha bus. This syncs on both rise and fall of
>> clock. So effective speed is simply twice that of the clock.
>>
>DDR would have done here. :)

Would it? Then I'd have to be clear about not meaning DDR ram etc.
....No Wes, there was a smilie on that, but reign your horses.

ancra
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Sun, 18 Apr 2004 01:22:55 GMT, Wes Newell
<w.newell@TAKEOUTverizon.net> wrote:

>On Sat, 17 Apr 2004 10:38:24 -0700, esara wrote:
>
>> Sorry I also forgot to ask about the DDR memory .. can we say that the
>> "actual speed or clock speed of DDR memory is 1/2 the effective
>> speed". I mean if the DDR is 333Mhz then the actual speed is 333/2 and
>> if DDR is 266 (which is the effective speed) then the actual speed is
>> 266/2. Can I say so??
>> Thanks..
>>
>Why even bother with "effective speed".

Because it's all over the place. In sales specs, in technical
documentation, in media, on cpu markings. But usually nobody even
bothers much about "effective". Even Intel and AMD are mostly just
simply stating "bus speed" or "FSB speed". Even in their technical
manuals. - And no, it's not the clock speed they're referring to.

>It's bascally a worthless
>statement without knowing what you are comparing the efectiveness to.:)

Fundamentally, everything is worthless if you don't have a purpose for
it. Fundamentally, every purpose requires additional knowledge.
Anyone trying to set his FSB clock to AMD/Intel advertised "FSB speed"
has been misled.
Anyone not understanding why dual channel DDR400 makes sense for his
P4C, since it has only got a "200MHz bus clock" has also been misled.

>Some examples:
>A P4/800 only has an effective speed of 400MHz when compared to an Athlon
>200MHz (or is that 400MHz effective) FSB cpu.:)
>So I could say an Athlon/400 has an effective FSB speed of 400MHz and the
>P4/800 also has an effective FSB speed of 400MHz and be completely
>correct.

No, that's only you fooling around with what you think should or could
be the meaning of "effective bus speed". I think that AMD and Intel
intend it to mean frequency of transfers.

>And that just one more reason the CPU FSB speed is measured in clock
>cycles, not data rates.
>And until you admit to ones self that the numbers are just bogus BS, you
>will continue to have this confusion.
>Bus clocks are measured in Hz, data rates aren't, and never should the two
>be intertwined.

(But effective bus speed is clearly _not_ exactly data rate. It's a
different property (a frequency, - virtual or real), even if it's
closely related to data rate.)

In your opinion, if I got you right, you feel AMD and Intel should
have spec'ed their FSBs with max data rate instead? Fair enough, that
argument has some merit IMO, but that's not the way things are now.

ancra
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Mon, 19 Apr 2004 17:17:00 +0200, somebody wrote:

> As for MHz:
> The nomenclature used by AMD and Intel in their specs and in their
> technical documentation, is: "800MHz bus", "800MHz speed", "800MHz
> FSB" etc, " _off_ a 200MHz system clock", " - 200MHz bus clock" etc.

Can't say what Intel says, but AMD's data sheets use the clock speed to
describe the system bus, not the so called effective speed either. IOW's,
nowhere in the doc do they call 333FSB, 333MHz FSB. They do use 166MHz for
it though with a note that bus operrates at twice the clock speed. No
other details are given. And even if they did, it wouldn't matter, it
would just be wrong.:)

> This is also the nomenclature used by chip set and mobo manufacturers,

Hmmm.... There's no FSB setting anywhere close to 333, much less 400 in
the bios of any board I've used. They use the real speed too. My latest
board is an Athlon 64 board. Top FSB speed in it is 233Mhz.

> when referring to supported features. It's also the nomenclature used by
> distributors, retailers and media, and so on. "Effective speed" is
> (again) _not_ data rate. And it is always measured in MHz, whether you
> think it's appropriate or not. And MHz is a unit for frequency. SI
> definition being 10^6 * s^-1. (Which does seem appropriate enough to me,
> apples being apples.)
>
You can say the moon is made of green cheese too.:)

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

"Wes Newell" <w.newell@TAKEOUTverizon.net> wrote in message
news:pan.2004.04.18.06.38.02.372769@TAKEOUTverizon.net...
> On Sun, 18 Apr 2004 03:08:39 +0200, somebody wrote:
>
> > The cpus are marketed with their 'effective FSB speed'. That's how you
> > buy them. The 'FSB clock speed' is what you need to set the mobo FSB
> > clock to.
> > (There's other older P4s and Athlons too, other than those I listed.)
> >
> So this kinda makes the so called efective speed pretty much worthless.:)
>
> > "actual FSB speed" I don't understand. Please keep to "effective" or
> > *clock* respectively. Both are actual and real. FSB Clockspeed is the
> > frequency of the clock that syncs transfers. Effective FSB speed is the
> > frequency of transfers.
> >
> Let me help you. Actual means real. FSB means Front Side Bus. And speed
> means the clock speed. Now you know.:)
>
> > The Intel P4 varieties (P4B, P4C, P4E and P4EE) with 533 and 800 MHz
> > FSB, have what Intel call a "quad pumped" bus. This means the transfers
> > can occur 4 times per clock cycle. So 4 * 200 = 800 and 4 * 133 = 533.
> >
> I think everyone that reads this uderstand the Intel bus. I prefer QDR for
> Quad Data Rate, which is more precise than quad pumped, which sounds like
> you're at a gas station pumping gas instead of data.:)
>
> > AMD is using DEC's old Alpha bus. This syncs on both rise and fall of
> > clock. So effective speed is simply twice that of the clock.
> >
> DDR would have done here. :)
>
> --
> Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
> http://mysite.verizon.net/res0exft/cpu.htm

9,600 bps modems had 2400 baud signalling with 4 bit-wide "word", this is
analogous to "QDR" sending data 4 times fold the signalling rate. Using a
MegaHertz unit is meaningfull when you state what is it that completing a
"cycle" or "phase".
 

somebody

Distinguished
Apr 2, 2004
154
0
18,680
Archived from groups: alt.comp.hardware.pc-homebuilt (More info?)

On Mon, 19 Apr 2004 23:43:18 GMT, Wes Newell
<w.newell@TAKEOUTverizon.net> wrote:

>On Mon, 19 Apr 2004 17:17:00 +0200, somebody wrote:
>
>> As for MHz:
>> The nomenclature used by AMD and Intel in their specs and in their
>> technical documentation, is: "800MHz bus", "800MHz speed", "800MHz
>> FSB" etc, " _off_ a 200MHz system clock", " - 200MHz bus clock" etc.
>
>Can't say what Intel says, but AMD's data sheets use the clock speed to
>describe the system bus, not the so called effective speed either. IOW's,
>nowhere in the doc do they call 333FSB, 333MHz FSB.

I only checked Intel, but I didn't have to check AMD (...or so I
thought):
AMD Athlon XP processor
Page 1 model 6 data sheet, Overview, 6 lines from bottom.

However, just as I was about to to rub your nose in this ;-), I
noticed to my horror that in _later_ model data sheets, the MHz unit
is gone:
Page 2 model 8 data sheet, Overview, 3, 4 and 26 lines from top.
Page 2 model 10 data sheet, Overview, 3 to 5 lines from top.
Here it's just "333 FSB" and "400 FSB".
(So, this is most frustrating and annoying :))

But still, not only do these data sheets consistently use "266/333/400
FSB" , riffling through, I can't see the clock ever being referred to
as "bus speed". It's always called system clock or bus clock.
In tables like in chapter 6, page 22 and chapter 7 page 26, for
instance, we see that "maximum clock frequency" is indeed 166MHz and
200MHz for "333 FSB sysclk and sysclk# AC characteristics" and "400
FSB" respectively.

>> This is also the nomenclature used by chip set and mobo manufacturers,
>
>Hmmm.... There's no FSB setting anywhere close to 333, much less 400 in
>the bios of any board I've used.

....Since it's the clock you set. However, mobo manuals state things
like 266, 333, 400 MHz FSB in specs etc. However, you know that too,
so you're just being argumentative ;-).

> They use the real speed too. My latest
>board is an Athlon 64 board. Top FSB speed in it is 233Mhz.

That's neat, since the A64 doesn't have any FSB. ;-)
(I bet it has an external clock though.)

ancra