Archived from groups: alt.comp.hardware.pc-homebuilt,alt.comp.hardware.overclocking.amd,alt.comp.hardware.overclocking (
More info?)
Wes Newell wrote:
> On Fri, 01 Oct 2004 01:14:29 -0500, David Maynard wrote:
>
>
>>Wes Newell wrote:
>>
>>
>>>It's not my opinion. its clear fact. That's the way it has always been in
>>>computer science. ref:
>>>
>>>http://www.webopedia.com/TERM/d/data_transfer_rate.html
>>>
>>>
>>>
>>>>You seem to be under the false impression that the only thing 'Hz'
>>>>applies to is an electronic wave form.
>>>>
>>>
>>>No, I'm saying this is computer science, and it has it's definitions.
>>
>>You can 'say' whatever you want but that doesn't make it so.
>>
>
> I didn't say anything that the world of computer scuence has been saying
> for years. And I provide proof. You provide nothing but smoke and mirrors
> and I'm sure people will see that.
Yes, of course, a 'webopedia' definition is certainly the last word in
'computer science' whereas direct quotes out of AMD's data sheets are, as
you put it, "smoke and mirrors."
</end sarcasm mode>
>>>Bus speeds in computer science are measured in clock speeds, not data rates.
>>>ref:
>>>
>>>http://www.webopedia.com/TERM/b/bus.html
>>
>>Your definition says "Every bus has a clock speed measured in MHz."
>>
>>No one disputes that a bus has a "clock speed measured in MHz." The issue
>>is what other characteristics it has and one of those is how it transfers
>>data; such as one transfer per clock, two per clock, or 4 per clock.
>>
>
> Did you not read the above. Data rates are not measured in MHz, but
> BPs/bps. Now did you see anything about the bus speed being defined by
> data rate? No.
You also didn't see anything about synchronous and asynchronous busses and
the simple bus 'definition' didn't say any thing regarding data rate at
all, nor DDR or QDR, except for a meaningless comment about some busses
being 'fast', which makes them 'faster' (brilliant, and oh so 'scientific')
For you to suggest that simple description is an all inclusive 'last word'
on buses is supercilious.
As for your 'data rate' link, you *are* aware that "often" is not a synonym
of "always" or "exclusively," right?
>>Your definition 'proves' nothing because saying "Every bus has a clock
>>speed measured in MHz" is like saying "every internal combustion
>>engine's rotation is measured in RPM." That doesn't mean the horsepower
>>description is 'hype' nor that it 'violates' the RPM mention.
>>
>
> Wow are you dense. Just like bus speeds and data rates, they are 2
> different things, and as such are defined with 2 different measurements,
> RPM, and HP,
Horsepower and RPM are not unrelated things. Horsepower is the product of
the torque produced and RPM; much as data rate is the product of how many
transfers take place per clock cycle and the clock speed.
> not 1 like 200HP and then turn around and say 5000HP also.
Amusing, but completely inappropriate as that's not what's being done.
A more appropriate analogy is you keep insisting that the only 'real' way
to describe the power of a motor is maximum torque at the specified RPM and
I'm saying that horsepower is equally good as it is precisely the end
result of the exact same equation, and more intuitive to the user. But
somehow, to you in your 'marketing conspiracy' laden world, horsepower is
'hype'.
> It's exactly the same reason you don't use MHz for to seperate
> measurements in computer science, clock speed, and data rates.
Hz means cycles per second. It can be used with any cyclic phenomena.
> You've now
> proven my point. Thank you.
That you don't understand 400 is the same thing as 2 x 200? Yes. You're
welcome.
>>>>>Data rates deserve and do
>>>>>have their own measuring systems, actually there's two, bps and Bps,
>>>>>and it's not MHz.
>>>>
>>>>Obviously, by overwhelming usage it is as well.
>>>>
>>>>
>>>
>>>You will never see the data rate used as a bus speed in any technical
>>>documentation. AMD doesn't. It shows the actual clock speed, not the
>>>bogus numbers.
>>
>>AMD Athlon XP
>>Processor Model 10
>>Data Sheet
>>
>>Publication # 26237 Rev. C
>>Issue Date: May 2003
>>
>>Chapters:
>>
>>6 Advanced 333 Front-Side Bus AMD Athlon XP Processor Model 10
>>Specifications
>>
>>7 Advanced 400 Front-Side Bus AMD Athlon XP Processor Model 10
>>Specifications
>>
>>Those are 'data rate' descriptors for the FSB, which you claim AMD
>>"doesn't" do.
>>
>
> So tell me, where is MHz used above. And those are not data rate
> descriptors as you put it. They are simply names names assigned by AMD for
> a FSB system.
I suppose it's just 'coincidence' that they're the data rate numbers: 333,
for 2 times a 166.6 MHz Clocked FSB, and 400 for 2 times a 200 MHz clock
FSB. LOL
> Now go back and read the table under these systems and tell
> me what speed in MHz they define them. Well, what do you know, it's clock
> speed is 166MHz, and 200MHz.
Yes, the "clock speed" is 166.6 Mhz and 200 Mhz, respectively. That is not
the data rate because they are DDR busses and is why the title lists them
as 333 and 400 FSBs: the DATA RATE, not the 'clock'.
> Nowhere will you find the bus defined as
> 333Mhz or 400MHz. And nowhere will you even find the data rate as 333MHz
> or 400MHz.
Except in the large, bold, title of the chapters and every time the
processor 'name' is described or the bus is described, as in "Advanced 333
FSB technology available" or as in
Table 9. Interface Signal Groupings
Signal Group
AMD Athlon™
System Bus
See “Advanced 333 FSB AMD Athlon™ System Bus DC
Characteristics” on page 24, See “Advanced 400 FSB
AMD Athlon™ System Bus DC Characteristics” on page
28, Table 3, “Advanced 333 FSB AMD Athlon™ System
Bus AC Characteristics,” on page 23, Table 7, “Advanced
400 FSB AMD Athlon™ System Bus AC Characteristics,”
on page 27, and “CLKFWDRST Pin” on page 72.
> You will find a note that says the "system" operates at twice
> the clock speed. Now what does that imply? DDR.
Yes. And what is double 166.6 and 200?
Surprise, there's more than one way to say something.
>>>Actually, it's not even the data rate of the bus. It's only the data
>>>rate for a single line of the "bus".
>>
>>That statement is a nonsensical absurdity.
>>
>>
>>>The actual data rate of the bus, more commonly called it's bandwidth is
>>>about 3.2GBps.
>>
>>You could use Bps but that's not 'really true' either as it's a
>>theoretical capability unsustainable with any known synchronous RAM.
>>
>
> You're reaching now. The ram bus is totally different than the FSB.
That doesn't alter the fact that your 'bandwidth' numbers are theoretical
maximums.
>>However, the issue isn't whether you, or I, or anyone else, might think
>>Bps is a 'better' means to describe the bus; the issue is whether using
>>the DDR rate is 'hype'. And it's not 'hype' because it does, in fact,
>>describe something real and tangible. A '400', double data rate 200 Mhz
>>clocked, bus is faster than a single data rate 200 MHz clocked bus.
>>
>
> The base bus is not faster.
Of course it is and, contrary to your implication, 'the clock' is not 'the
bus'.
> The only thing faster is the data rate.
Which, frankly, is what really matters. The clock is simply a means of
timing on the bus and how the clock is USED varies with the type of bus.
> And
> this is because of the 2 devices the bus connects to supports DDR.
Well, that's a useless waste of words. Of course the devices 'support' DDR.
They *define* the bus. The empty wires have no 'clock', or any other signals.
>>And, as I pointed out before, the data rate expressed in 'bus speed' is
>>consistent with the previous single data rate description whereas Bps is
>>not.
>>
>
> Prior to multi-data per clock there was never a problem. A 66MHz bus was
> just that and everyone knew that the clockspeed of that bus was 66MHz. The
> data rate, or bandwidth, is determied by the width of the bus, the data
> per bits clock, not the clockspeed.
That is because earlier busses were single data rate transferring once per
clock cycle. A DDR bus transfers twice per clock cycle, which means looking
at simply the clock speed no longer represents the capability of the bus.
There are a number of ways to 'describe' that. One could say, as some
prefer, 200MHz DDR, but that presumes the reader has a clue what the heck
'DDR' means. One could state the 'bandwidth', as you have suggested, but
that provides no common reference with the SDR busses that are described
simply by 'speed'.
Using the data rate, however, is consistent with the previous 'speed'
description. With SDR, the data rate IS the 'clock speed', so saying "100
Mhz" specifies BOTH the 'clock' and the 'data rate'. With DDR the data rate
is, by definition, 2 times the clock speed and the equivalency of data rate
vs clock no longer holds. So saying '400', for a DDR bus, vs the old SDR
'200' accurately encapsulates the difference for the thing that MATTERS:
the data rate, which is transferred twice per clock with DDR.
>>>And my conclusion is that you write a lot of BS for some reason.
>>
>>You're the one writing 'B.S.' by suggesting there is no meaning to a bus
>>being DDR vs single data rate. I.E. that it's just 'hype'.
>>
>
> No, nowhere have you found me "suggesting" that.
It has been the gist of your argument from the beginning.
>>It does, however, describe something real and tangible and so is not
>>'hype', regardless of how much you dislike the usage.
>>
>
> No it doesn't.
Yes, it does, and I've explained it over and over. The burst data rate on a
200 Mhz clocked DDR bus is 400 Mhz. Pure and simple.
> All it does is confuse people.
Lots of things confuse people but that doesn't mean it's 'hype'. And people
wouldn't be so confused if it weren't for people, like you, saying it's
'just marketing hype'.
> And even you stated (in this
> same post) that the speeds are not obtainable.
I said not sustainable. It IS "obtainable," for bursts.
The 400, and indeed the SDR 'speed', are burst data rates. And, as I've
already pointed out, the data rate description is entirely consistent with
the SDR 'speed' description: the same 'burst rate', means of operation, and
other limiting factors, apply in the same manner.
> You keep hanging yourself. when are you going to give up with the BS or
> provide some documented facts.
You remind me of the cowboy who, laying there pounded into the dirt with
bovine hoof marks up and down his body, kept saying "what stampede?"
>>>>>Now before you reply, consider that by definition, bus speeds are
>>>>>measured by their clock speeds, not data rates.
>>>>
>>>>Any by who's definition would that be?
>>>>
>>>>
>>>
>>>Already answered above. And if you don't like that one, welll, just you
>>>can look at the processor specs from AMD. And I said specs, not ads.
>>
>>You failed at both, as I've shown.
>>
>
> You must have a lot more free time than me to keep your stupid arguement
> up. It obvious that you will not embrace the truth, so I will just let you
> live in your own little world and believe what you want to.
And I suppose you could stay in your own little world ignorant of what
double data rate means to a bus and believing that 2 times 200 isn't 400.
I would have preferred you finally understood it but it's ultimately your
choice.
>>>The fact of the matter is that you write a lot but have absolutely
>>>nothing technical to back up any of your claims.
>>
>>Of course I do, as shown, but it really isn't necessary because you, as
>>the one levying the 'accusation' of the term being 'hype', have the
>>obligation to prove the charge.
>>
>>So ante up. Explain how the P4 quad pumped 100MHz clocked FSB is 'hype'
>>and that it makes no difference whether it's a P4 'hyped' 400Mhz FSB or
>>a P3 100Mhz FSB bus because they both operate at the 'same real' 100 MHz
>>clock speed.
>>
>
> It's not the quad pumped bus that's hype, and I've never said it was. It's
> the definition of the bus that is not correct. It is a 100MHz QDR bus. And
> QDR is short for quad pumped.
So your position is that there's only ONE way to ever say anything? A
patently absurd argument.
Btw, what does quad pumped 'mean', in the context of what USE is it? Hint:
it transfers data 4 times per clock, so it's burst transfer rate is 4 times
faster. And if I want to give someone the gist of it, saying 400 vs 100
makes a heck of a lot more intuitive sense than '100 QDR'.
And, btw, that a '400', 200Mhz clocked DDR bus, isn't 'twice as fast' as a
SDR 200 Mhz clocked bus in overall performance is no different than a 133.3
Mhz (SDR) bus not being twice as fast in overall performance as a 66.6 Mhz
SDR bus.
There is nothing magical about 'the clock' nor is it 'the real thing'. It
is simply a means to an end, the transferring of data, and the DATA RATE is
the thing of importance to a "data processing" machine.
>>>And since you didn't answer the
>>>question,
>>
>>I most certainly did answer the question.
>>
>>
>>>i can only assume that you don't know of any logical use for the 400
>>>number. And I can fully understand that since there is none unless you
>>>divide it by 2 to get to the base clockspeed.
>>
>>I repeat, it's apparent you have no understanding of what double data
>>rate means, or your feigning ignorance.
>>
>
> You have no answer, so you fall back on insultsd that you know are lies.
I said it's "apparent" from what you say, as opposed to flat out saying
something like "wow, are you dense." You could, of course, demonstrate some
understanding of it in contrast.
>>When streaming, I.E. when doing what your '3.2 GBps bandwidth' number
>>describes, there are two data transfer cycles per clock cycle or, in
>>other words, the data rate, for a double data rate bus clocked at 200
>>MHz, is 400 million data transfer cycles per second (burst rate). e.g.
>>400 Mhz.
>
>
> It's kind of comical that you are explaining this to me when I have
> designed memory controllers for computer systems, and I've explained the
> FSB and the data rate on my website. Maybe you should have a look. It's
> the link below.
>
I read it, and all it does is repeat the same nonsense you've been saying
here. Not surprising since it's you.