I'm a PC user considering purchasing a Mac Powerbook G4, and I found the
speed ratings for the system bus and memory at odds with one another. The
system bus is 167MHz, while the memory is running at 333MHz. More
importantly, the maximum bandwidth a 167MHz 64-bit wide system bus could
reach would be 1.335GBps (167MHz * 8 bytes). The potential data throughput
of P2700 memory is 2.7GBps. Ergo, the memory is capable of twice the speeds
of the system bus, but is limited by that very system bus. Now, in the PC
world, the system bus and memory bandwidth are equal, or nearly so. For
instance, my AMD Athlon T-Bird has a potential throughput of 2.128GBps
second on its 64-bit wide system bus (266MHz * 8 bytes), and the memory is
PC2100, so the speeds are nearly matched. Why does Apple use PC2700 memory
in its notebooks if they cannot be properly utilized? Wouldn't it make more
sense to use a RAM technology that was a closer match in speed? It almost
seems to me that Apple is attemping to deceive people who don't know much
about computers, and assume the memory is really effectively used at 333MHz,
and do not realize the 167MHz system bus is the real factor involved. Is
Apple charging a false premium for a perceived benefit that doesn't exist?
Why not come clean on the issue? I have called Apple Care, and the data
wasn't "available to the public" regarding the Powerbook's architecture, so
I had to work out the figures from existing knowledge to the best of my
ability. Thanks for your input.

In article <UL%Ec.1527$R36.700@newsread2.news.pas.earthlink.n et>,
"MH" <removemstthiscrow5429spam@earthlink.net> wrote:

> I'm a PC user considering purchasing a Mac Powerbook G4, and I found the
> speed ratings for the system bus and memory at odds with one another. The
> system bus is 167MHz, while the memory is running at 333MHz. More
> importantly, the maximum bandwidth a 167MHz 64-bit wide system bus could
> reach would be 1.335GBps (167MHz * 8 bytes). The potential data throughput
> of P2700 memory is 2.7GBps. Ergo, the memory is capable of twice the speeds
> of the system bus, but is limited by that very system bus. Now, in the PC
> world, the system bus and memory bandwidth are equal, or nearly so. For
> instance, my AMD Athlon T-Bird has a potential throughput of 2.128GBps
> second on its 64-bit wide system bus (266MHz * 8 bytes), and the memory is
> PC2100, so the speeds are nearly matched. Why does Apple use PC2700 memory
> in its notebooks if they cannot be properly utilized? Wouldn't it make more
> sense to use a RAM technology that was a closer match in speed? It almost
> seems to me that Apple is attemping to deceive people who don't know much
> about computers, and assume the memory is really effectively used at 333MHz,
> and do not realize the 167MHz system bus is the real factor involved. Is
> Apple charging a false premium for a perceived benefit that doesn't exist?
> Why not come clean on the issue? I have called Apple Care, and the data
> wasn't "available to the public" regarding the Powerbook's architecture, so
> I had to work out the figures from existing knowledge to the best of my
> ability. Thanks for your input.

Well, I don't know exactly what you were told, but a lot of that kind of
information is available at www.apple.com/developer.

Which model PowerBook G4 were you looking at?

And moreover, a 167Mhz bus is what you're supposed to run 333MHz DDR ram
on...

<http://www.thetechzone.com/display.php?i=215&p=1>

"If your motherboard supports PC2700 DDR333 and you have a stick of
DDR333 RAM then your motherboard will set the CPU/RAM ratio to 4:5. This
ratio will run the CPU at 133MHz FSB and the RAM at 166MHz which is
333MHz DDR."

Or if you prefer:

<http://www.crucial.com/kb/answer.asp?qid=4032>

"PC2700 memory is DDR designed for use in systems with a 166-MHz
front-side bus (providing a 333 MT/s data transfer rate). The "2700"
refers to the module's bandwidth (the maximum amount of data it can
transfer each second), which is 2.7 GB."

Glad to help. <g>

--
Alan Baker
Vancouver, British Columbia
"If you raise the ceiling 4 feet, move the fireplace from that wall
to that wall, you'll still only get the full stereophonic effect
if you sit in the bottom of that cupboard."

Alan Baker <alangbaker@telus.net> writes:

> And moreover, a 167Mhz bus is what you're supposed to run 333MHz DDR ram
> on...

Yes. That's the first "D" in DDR (Double Data Rate). DDR basically
means that you get 2 data transfers per cycle (one on the up-tick
and one on the down-tick).

I think you are just seeing differencees in how people quote
data rates. The ram manufacturers like to quote the bigger
number because it sounds better to say that this is a 333MHz ram
than to say it is a 167Mhz one with double data rate. Too many
people just stop after seeing the number.

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain | experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

MH wrote:

A while back (and I've heard nothing new lately that would contradict
this) the G4 processors were not capable of -really- using the DDR
RAM, due to Motorola's failure to make a better chip.

The chip could easily be starved of data because it the internals of the
processor (especially the altivec unit) can process data faster
than it could be supplied by the remainder of the chip.

This is not to say the benefits of DDR RAM are entirely wasted - other
parts of the system are able to make use of it (such as the ethernet
& ATA controllers which can dump data into memory very quickly).

Please correct me if any of this has changed recently...

Dale

--
dstanbro@spam.o.matic.bigpond.net.au

In article <MrNoSpam-065309.09562902072004@news-server.bigpond.net.au>,
Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:

> MH wrote:
>
> A while back (and I've heard nothing new lately that would contradict
> this) the G4 processors were not capable of -really- using the DDR
> RAM, due to Motorola's failure to make a better chip.

You heard this, huh? Where?

>
> The chip could easily be starved of data because it the internals of the
> processor (especially the altivec unit) can process data faster
> than it could be supplied by the remainder of the chip.

Again: a reference?

>
> This is not to say the benefits of DDR RAM are entirely wasted - other
> parts of the system are able to make use of it (such as the ethernet
> & ATA controllers which can dump data into memory very quickly).
>
> Please correct me if any of this has changed recently...


After you've shown that it was true in the first place.

--
Alan Baker
Vancouver, British Columbia
"If you raise the ceiling 4 feet, move the fireplace from that wall
to that wall, you'll still only get the full stereophonic effect
if you sit in the bottom of that cupboard."

In article <alangbaker-4DDA71.18372701072004@news.telus.net>,
Alan Baker <alangbaker@telus.net> wrote:

> In article <MrNoSpam-065309.09562902072004@news-server.bigpond.net.au>,
> Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:
>
> > MH wrote:
> >
> > A while back (and I've heard nothing new lately that would contradict
> > this) the G4 processors were not capable of -really- using the DDR
> > RAM, due to Motorola's failure to make a better chip.
>
> You heard this, huh? Where?

I've read this too, Alan. I think he's right.

--
George Graves
------------------

"This election is shaping up great. Our choices
are a guy who has a lot of second thoughts, or
a guy who has never had a first thought."
-- Jay Leno

In article <gmgravesnos-E1FAEB.19015501072004@news.sf.sbcglobal.net>,
George Graves <gmgravesnos@pacbell.net> wrote:

> In article <alangbaker-4DDA71.18372701072004@news.telus.net>,
> Alan Baker <alangbaker@telus.net> wrote:
>
> > In article <MrNoSpam-065309.09562902072004@news-server.bigpond.net.au>,
> > Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:
> >
> > > MH wrote:
> > >
> > > A while back (and I've heard nothing new lately that would contradict
> > > this) the G4 processors were not capable of -really- using the DDR
> > > RAM, due to Motorola's failure to make a better chip.
> >
> > You heard this, huh? Where?
>
> I've read this too, Alan. I think he's right.

OK. Where?

--
Alan Baker
Vancouver, British Columbia
"If you raise the ceiling 4 feet, move the fireplace from that wall
to that wall, you'll still only get the full stereophonic effect
if you sit in the bottom of that cupboard."

> And moreover, a 167Mhz bus is what you're supposed to run 333MHz DDR ram
> on...

I think there's some confusion in the way different people talk about
these specs...

I have an Abit motherboard in my PC, with a stated capability of 266,
333, or 400 MHz FSB. In it I use an Athlon XP 3000+ with 400 MHz FSB.
When I put them together and boot - it autoconfigures to 200 MHz
"external clock" and a 10.5 multiplier (internal frequency 2100 MHz).

So... I guess a 200 MHz "external clock" is equal to 400 MHz FSB.

I have PC3200 RAM installed. It clocks at 400 MHz but in the BIOS is
set to 200 MHz. I just assumed it meant 200 MHz on one side of the cycle.

Now, with the Mac, it seems like their language is slightly different.
My specs say "Bus Speed: 167 MHz" on my powerbook, but is that the FSB
or the external clock? If it's FSB, then can you see how it doesn't
match the PC specs?

"MH" <removemstthiscrow5429spam@earthlink.net> wrote in message news:<UL%Ec.1527$R36.700@newsread2.news.pas.earthl ink.net>...
> I'm a PC user considering purchasing a Mac Powerbook G4, and I found the
> speed ratings for the system bus and memory at odds with one another. The
> system bus is 167MHz, while the memory is running at 333MHz. More
> importantly, the maximum bandwidth a 167MHz 64-bit wide system bus could
> reach would be 1.335GBps (167MHz * 8 bytes). The potential data throughput
> of P2700 memory is 2.7GBps. Ergo, the memory is capable of twice the speeds
> of the system bus, but is limited by that very system bus.

Yes, I for one believe this is correct. The G4 has suffered from a
lame-ass FSB wrt Intel since Intel first started quad-pumping. This
exact same issue came up with the 2002 MDD G4'sm which had the same
FSB and the same DDR-333 memory. It is also one of the reasons I'm
still happy on my 800Mhz PBG4 w/ 133Mhz FSB, I know moving up to a
1.5Ghz G4 isn't going to be that different an experience.

> Now, in the PC
> world, the system bus and memory bandwidth are equal, or nearly so. For
> instance, my AMD Athlon T-Bird has a potential throughput of 2.128GBps
> second on its 64-bit wide system bus (266MHz * 8 bytes), and the memory is
> PC2100, so the speeds are nearly matched. Why does Apple use PC2700 memory
> in its notebooks if they cannot be properly utilized?

It's probably the best match Apple can get. Slower DDR memory isn't
going to be any less expensive than PC-2700. Plus with Apple's
"Intrepid" memory/IO controller DMA accessess can go directly into
memory without CPU intervention. Plus AGP transfers can too.

> Wouldn't it make more
> sense to use a RAM technology that was a closer match in speed?

There is no SDR 167Mhz RAM, so Apple chose to use the next best thing,
DDR-333.

> It almost
> seems to me that Apple is attemping to deceive people who don't know much
> about computers, and assume the memory is really effectively used at 333MHz,
> and do not realize the 167MHz system bus is the real factor involved.

Undoubtedly Apple benefits from this confusion. Perhaps you have the
basis for a class action lawsuit. Lord knows they've been sued on less
bs issues.

> Is Apple charging a false premium for a perceived benefit that doesn't exist?

One could view it this way, though one would have to see the marketing
materials for any specific misleading claim. NB: Intel's quad-pumped
200x4 FSB does not result in an effective 800Mhz throughput, more like
500-600Mhz.

> Why not come clean on the issue?

What issue?

>I have called Apple Care, and the data
> wasn't "available to the public" regarding the Powerbook's architecture, so
> I had to work out the figures from existing knowledge to the best of my
> ability. Thanks for your input.

http://developer.apple.com/documenta...kG4/index.html

knock yourself out.

In article <cc2ia1$21g9$1@netnews.upenn.edu>,
David Besack <daveREMOVEbesack@mac.com> wrote:

> > And moreover, a 167Mhz bus is what you're supposed to run 333MHz DDR ram
> > on...
>
> I think there's some confusion in the way different people talk about
> these specs...
>
> I have an Abit motherboard in my PC, with a stated capability of 266,
> 333, or 400 MHz FSB. In it I use an Athlon XP 3000+ with 400 MHz FSB.
> When I put them together and boot - it autoconfigures to 200 MHz
> "external clock" and a 10.5 multiplier (internal frequency 2100 MHz).
>
> So... I guess a 200 MHz "external clock" is equal to 400 MHz FSB.
>
> I have PC3200 RAM installed. It clocks at 400 MHz but in the BIOS is
> set to 200 MHz. I just assumed it meant 200 MHz on one side of the cycle.

No. It clocks at 400MHz, but the RAM operates on both the rising *and*
falling edge of the clock cycle.

>
> Now, with the Mac, it seems like their language is slightly different.
> My specs say "Bus Speed: 167 MHz" on my powerbook, but is that the FSB
> or the external clock? If it's FSB, then can you see how it doesn't
> match the PC specs?

It's both the FSB *and* the clock. And the RAM operates as expected, on
the rising and falling edge: i.e. at 333MHz.

--
Alan Baker
Vancouver, British Columbia
"If you raise the ceiling 4 feet, move the fireplace from that wall
to that wall, you'll still only get the full stereophonic effect
if you sit in the bottom of that cupboard."

In article <dd5de929.0407011909.1d95ada6@posting.google.com >,
imouttahere@mac.com (Heywood Mogroot) wrote:

> "MH" <removemstthiscrow5429spam@earthlink.net> wrote in message
> news:<UL%Ec.1527$R36.700@newsread2.news.pas.earthl ink.net>...
> > I'm a PC user considering purchasing a Mac Powerbook G4, and I found the
> > speed ratings for the system bus and memory at odds with one another. The
> > system bus is 167MHz, while the memory is running at 333MHz. More
> > importantly, the maximum bandwidth a 167MHz 64-bit wide system bus could
> > reach would be 1.335GBps (167MHz * 8 bytes). The potential data throughput
> > of P2700 memory is 2.7GBps. Ergo, the memory is capable of twice the
> > speeds
> > of the system bus, but is limited by that very system bus.
>
> Yes, I for one believe this is correct. The G4 has suffered from a
> lame-ass FSB wrt Intel since Intel first started quad-pumping. This
> exact same issue came up with the 2002 MDD G4'sm which had the same
> FSB and the same DDR-333 memory. It is also one of the reasons I'm
> still happy on my 800Mhz PBG4 w/ 133Mhz FSB, I know moving up to a
> 1.5Ghz G4 isn't going to be that different an experience.
>
> > Now, in the PC
> > world, the system bus and memory bandwidth are equal, or nearly so. For
> > instance, my AMD Athlon T-Bird has a potential throughput of 2.128GBps
> > second on its 64-bit wide system bus (266MHz * 8 bytes), and the memory is
> > PC2100, so the speeds are nearly matched. Why does Apple use PC2700 memory
> > in its notebooks if they cannot be properly utilized?
>
> It's probably the best match Apple can get. Slower DDR memory isn't
> going to be any less expensive than PC-2700. Plus with Apple's
> "Intrepid" memory/IO controller DMA accessess can go directly into
> memory without CPU intervention. Plus AGP transfers can too.
>
> > Wouldn't it make more
> > sense to use a RAM technology that was a closer match in speed?
>
> There is no SDR 167Mhz RAM, so Apple chose to use the next best thing,
> DDR-333.

You've got it wrong.

333MHz DDR RAM is supposed to run on a 167MHz front side bus. If you're
thinking that it's running on a 333MHz bus, consider:

The "2700" in the name "PC2700" for this RAM indicates the maximum
bandwidth of the memory system that uses it.

167MHz

x 2 (for DDR)

x 64 (bits)

/ 8 (for bytes)

= 2,666MB/s

or rounded up to 2700 for marketing purposes.

<http://www.crucial.com/kb/answer.asp?qid=4032>

"PC2700 memory is DDR designed for use in systems with a 166-MHz
front-side bus (providing a 333 MT/s data transfer rate). The "2700"
refers to the module's bandwidth (the maximum amount of data it can
transfer each second), which is 2.7 GB. "

I think you owe Apple an apology.

>
> > It almost
> > seems to me that Apple is attemping to deceive people who don't know much
> > about computers, and assume the memory is really effectively used at
> > 333MHz,
> > and do not realize the 167MHz system bus is the real factor involved.
>
> Undoubtedly Apple benefits from this confusion. Perhaps you have the
> basis for a class action lawsuit. Lord knows they've been sued on less
> bs issues.
>
> > Is Apple charging a false premium for a perceived benefit that doesn't
> > exist?
>
> One could view it this way, though one would have to see the marketing
> materials for any specific misleading claim. NB: Intel's quad-pumped
> 200x4 FSB does not result in an effective 800Mhz throughput, more like
> 500-600Mhz.

>
> > Why not come clean on the issue?
>
> What issue?
>
> >I have called Apple Care, and the data
> > wasn't "available to the public" regarding the Powerbook's architecture, so
> > I had to work out the figures from existing knowledge to the best of my
> > ability. Thanks for your input.
>
> http://developer.apple.com/documenta...s/Macintosh_CP
> Us-G4/15inchPowerBookG4/index.html
>
> knock yourself out.

--
Alan Baker
Vancouver, British Columbia
"If you raise the ceiling 4 feet, move the fireplace from that wall
to that wall, you'll still only get the full stereophonic effect
if you sit in the bottom of that cupboard."

Alan Baker <alangbaker@telus.net> wrote:
> Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:
>> MH wrote:
>> A while back (and I've heard nothing new lately that would contradict
>> this) the G4 processors were not capable of -really- using the DDR
>> RAM, due to Motorola's failure to make a better chip.
>
>You heard this, huh? Where?

This is absolutely true, and can easily be verified by reading Motorola's
documentation on the PPC745x series. There is no member of that family
capable of using a double-pumped memory interface.

>> The chip could easily be starved of data because it the internals of the
>> processor (especially the altivec unit) can process data faster
>> than it could be supplied by the remainder of the chip.
>
>Again: a reference?

What is so unusual about this? I can't think of a single modern system
that doesn't have a memory bus slower than its theoretical computational
bandwidth. CPU speed has generally outstripped memory speed for quite
some time now.

>> This is not to say the benefits of DDR RAM are entirely wasted - other
>> parts of the system are able to make use of it (such as the ethernet
>> & ATA controllers which can dump data into memory very quickly).
>>
>> Please correct me if any of this has changed recently...
>
>After you've shown that it was true in the first place.

It's something Apple itself claimed when they released the original Xserve
(the first Mac to use DDR RAM): even though the G4 processor was not able
to access memory at full bandwidth, the system controller was designed to
allow other components to make use of the remaining bandwidth, creating a
"balanced I/O architecture". Why are you being so defensive?

-- Erick

In article <cc2qt0$180$1@morgoth.sfu.ca>,
erick@sfu.ca (Erick Bryce Wong) wrote:

> Alan Baker <alangbaker@telus.net> wrote:
> > Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:
> >> MH wrote:
> >> A while back (and I've heard nothing new lately that would contradict
> >> this) the G4 processors were not capable of -really- using the DDR
> >> RAM, due to Motorola's failure to make a better chip.
> >
> >You heard this, huh? Where?
>
> This is absolutely true, and can easily be verified by reading Motorola's
> documentation on the PPC745x series. There is no member of that family
> capable of using a double-pumped memory interface.
>
> >> The chip could easily be starved of data because it the internals of the
> >> processor (especially the altivec unit) can process data faster
> >> than it could be supplied by the remainder of the chip.
> >
> >Again: a reference?
>
> What is so unusual about this? I can't think of a single modern system
> that doesn't have a memory bus slower than its theoretical computational
> bandwidth. CPU speed has generally outstripped memory speed for quite
> some time now.
>
> >> This is not to say the benefits of DDR RAM are entirely wasted - other
> >> parts of the system are able to make use of it (such as the ethernet
> >> & ATA controllers which can dump data into memory very quickly).
> >>
> >> Please correct me if any of this has changed recently...
> >
> >After you've shown that it was true in the first place.
>
> It's something Apple itself claimed when they released the original Xserve
> (the first Mac to use DDR RAM): even though the G4 processor was not able
> to access memory at full bandwidth, the system controller was designed to
> allow other components to make use of the remaining bandwidth, creating a
> "balanced I/O architecture". Why are you being so defensive?

I'm not defensive about it at all.

I'd just like specifics.

--
Alan Baker
Vancouver, British Columbia
"If you raise the ceiling 4 feet, move the fireplace from that wall
to that wall, you'll still only get the full stereophonic effect
if you sit in the bottom of that cupboard."

On Fri, 02 Jul 2004 06:15:32 GMT, Alan Baker <alangbaker@telus.net>
chose to bless us with the following wisdom:

>In article <cc2qt0$180$1@morgoth.sfu.ca>,
> erick@sfu.ca (Erick Bryce Wong) wrote:
>
>> Alan Baker <alangbaker@telus.net> wrote:
>> > Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:
>> >> MH wrote:
>> >> A while back (and I've heard nothing new lately that would contradict
>> >> this) the G4 processors were not capable of -really- using the DDR
>> >> RAM, due to Motorola's failure to make a better chip.
>> >
>> >You heard this, huh? Where?
>>
>> This is absolutely true, and can easily be verified by reading Motorola's
>> documentation on the PPC745x series. There is no member of that family
>> capable of using a double-pumped memory interface.
>>
>> >> The chip could easily be starved of data because it the internals of the
>> >> processor (especially the altivec unit) can process data faster
>> >> than it could be supplied by the remainder of the chip.
>> >
>> >Again: a reference?
>>
>> What is so unusual about this? I can't think of a single modern system
>> that doesn't have a memory bus slower than its theoretical computational
>> bandwidth. CPU speed has generally outstripped memory speed for quite
>> some time now.
>>
>> >> This is not to say the benefits of DDR RAM are entirely wasted - other
>> >> parts of the system are able to make use of it (such as the ethernet
>> >> & ATA controllers which can dump data into memory very quickly).
>> >>
>> >> Please correct me if any of this has changed recently...
>> >
>> >After you've shown that it was true in the first place.
>>
>> It's something Apple itself claimed when they released the original Xserve
>> (the first Mac to use DDR RAM): even though the G4 processor was not able
>> to access memory at full bandwidth, the system controller was designed to
>> allow other components to make use of the remaining bandwidth, creating a
>> "balanced I/O architecture". Why are you being so defensive?
>
>I'm not defensive about it at all.
>
>I'd just like specifics.

Hey! We've got something in common. I like specifics too! Specifically
I'd like to know when you're going to come up with one of those 'too
numerous to mention' times I said Apple was going to go out of
business. Failing that you could just be a man and admit that you were
wrong. I expect that's asking waaay too much of you though, Alan 'I'll
cry like a little girl until you give me a URL but I'll never back up
anything I say with one' Baker.


--
Why settle for the lesser evil?
Cthulhu for President 2004

In article <alangbaker-35EE11.19123901072004@news.telus.net>,
Alan Baker <alangbaker@telus.net> wrote:

> In article <gmgravesnos-E1FAEB.19015501072004@news.sf.sbcglobal.net>,
> George Graves <gmgravesnos@pacbell.net> wrote:
>
> > In article <alangbaker-4DDA71.18372701072004@news.telus.net>,
> > Alan Baker <alangbaker@telus.net> wrote:
> >
> > > In article <MrNoSpam-065309.09562902072004@news-server.bigpond.net.au>,
> > > Dale Stanbrough <MrNoSpam@bigpoop.net.au> wrote:
> > >
> > > > MH wrote:
> > > >
> > > > A while back (and I've heard nothing new lately that would contradict
> > > > this) the G4 processors were not capable of -really- using the DDR
> > > > RAM, due to Motorola's failure to make a better chip.
> > >
> > > You heard this, huh? Where?
> >
> > I've read this too, Alan. I think he's right.
>
> OK. Where?

Seems to me in a review of the first G5s. They said that the machine was
the first Mac to fully utilize the DDR RAM, because the G4 processor
doesn't support it even though the last generation G4 machines came with
it. I read so many links referred to on this group, it's hard to say
exactly where I saw it.

--
George Graves
------------------

"This election is shaping up great. Our choices
are a guy who has a lot of second thoughts, or
a guy who has never had a first thought."
-- Jay Leno