More Nv-30 speculation !!
- Thread starter Ascended Saiyan
- Start date
arjan de lumens
Veteran
<nitpick>nAo said:Assuming data movements of one bit per cycle per pin both the unit measures are the same thingarjan de lumens said:
ciao,
Marco
Yes, but isn't the point of DDR that you move two bits per cycle per pin?
</nitpick>
Yeah, but in this case the cycle is not the memory module interface cycle but the cycle associated to the data change rate on the pin. My fault that I didn't make it clear before.<nitpick>
Yes, but isn't the point of DDR that you move two bits per cycle per pin?
</nitpick>
ciao,
Marco
arjan de lumens
Veteran
ok .. just wondering what unit measures you referred to with "both" - one of them presumably being the Mbps/pin - I assumed that the other one was MHz?
nAo said:
As the bus is not a single-bit one that would result in 0Hz actually
Anyway, if you're going to try to make a Hz measurement of the data signals the only thing that makes sense is possibly to do some kind of statistical analysis to find out how often the data at a certain pin changes on average or something like that.
Hmm...Hertz refers to a cycle. Regardless of where in the cycle data is sent, Hertz continues to reference the rate of cycles per second. In this case, the cycle is the clock used to drive data transfer. If 4 bits of data is sent in one cycle, so be it, but the Hertz is in reference to the clock cycle.
Data is measured in bits, nybbles (still used?), bytes, etc. If twice the data is sent per Hertz, this should be reflected in the data rate number, not the Hertz number.
It seems to me anything else is just the result of marketing.
Forgive me for, I think, repeating the sentiment of many, but I'm just putting it out in a way that is clearer to myself.
Data is measured in bits, nybbles (still used?), bytes, etc. If twice the data is sent per Hertz, this should be reflected in the data rate number, not the Hertz number.
It seems to me anything else is just the result of marketing.
Forgive me for, I think, repeating the sentiment of many, but I'm just putting it out in a way that is clearer to myself.
Sorry you didn't catch the idea. Anyway, alternates 0xaaaa and 0x5555 on a per total-bus-width basis and you're doneAs the bus is not a single-bit one that would result in 0Hz actually
NO
You don't need any statistics, the data itself DOES change every half clock. The fact it could be potentially the same all the time doesn't mean anything.
ciao,
Marco
Since I'm quite sure that you all agree that the unit Hz is used commonly (popular and technically) for things that aren't related to the number of cycles of a waveform per second (be it square, sine or whatever), I don't understand why you would go so far as to say that it's not allowed for datarate. (As long as the signals aren't more than bi-level and transmit more than one bit per symbol.)
I agree that when refering to the 2x number, it should be explicitly stated that it's the datarate. But using the unit Hz isn't wrong.
Best thing is of course to use the neutral standard that luckily is consistent through SDR/DDR/DDRII, where the new Samsung memory would be called DDRII1000.
I agree that when refering to the 2x number, it should be explicitly stated that it's the datarate. But using the unit Hz isn't wrong.
Best thing is of course to use the neutral standard that luckily is consistent through SDR/DDR/DDRII, where the new Samsung memory would be called DDRII1000.
arjan de lumens
Veteran
Hmmm ..
The SI unit 'Hertz' is nothing more special than just the inverse of a second (no need for a specific cycling waveform or anything like that) so saying "N times per second" means the same as "N Hertz" - so e.g. a DDR-II module that transfers data 1,000,000,000 times per second runs at 1 GHz data transfer rate.
You could make a point out of the fact that you cannot represent a frequency higher than 500 MHz using a data pin running at 1GHz transfer rate - that's similar to how a CD with a sample rate of 44100 Hz cannot represent frequencies above 22050 Hz, etc., but that's really another discussion ....
The SI unit 'Hertz' is nothing more special than just the inverse of a second (no need for a specific cycling waveform or anything like that) so saying "N times per second" means the same as "N Hertz" - so e.g. a DDR-II module that transfers data 1,000,000,000 times per second runs at 1 GHz data transfer rate.
You could make a point out of the fact that you cannot represent a frequency higher than 500 MHz using a data pin running at 1GHz transfer rate - that's similar to how a CD with a sample rate of 44100 Hz cannot represent frequencies above 22050 Hz, etc., but that's really another discussion ....
Ahhh, Music. It's a world of hertz.
From Samsungs info:
Burst Length: 4 (Read/Write Interrupt Prohibited but only Read interrupted by Read & Write interrupted by Write are allowed), 8(Interleave/nibble sequential
This synchronous device achieve high speed double-data-rate transfer rates of up to 533Mb/sec/pin (DDR533) for general applications.
Top wack DDR is running at 133MHz (x2 = 266Mhz giving 266Mbits/sec/pin) whilst DDRII is running at 267MHz (x2 = 533MHz giving 533Mbits/sec/pin) with a CAS of 4. Ouch. So they both transfer 1 bit per cycle and given the rising edge and falling edge it still equates to 2bits per cycle with internal burst cycles allowing 4 and 8 bit transfers. Either that or I read the paper wrong.
From Samsungs info:
Burst Length: 4 (Read/Write Interrupt Prohibited but only Read interrupted by Read & Write interrupted by Write are allowed), 8(Interleave/nibble sequential
This synchronous device achieve high speed double-data-rate transfer rates of up to 533Mb/sec/pin (DDR533) for general applications.
Top wack DDR is running at 133MHz (x2 = 266Mhz giving 266Mbits/sec/pin) whilst DDRII is running at 267MHz (x2 = 533MHz giving 533Mbits/sec/pin) with a CAS of 4. Ouch. So they both transfer 1 bit per cycle and given the rising edge and falling edge it still equates to 2bits per cycle with internal burst cycles allowing 4 and 8 bit transfers. Either that or I read the paper wrong.
nAo said:NO
You don't need any statistics, the data itself DOES change every half clock. The fact it could be potentially the same all the time doesn't mean anything.
If it's the same, has it really changed?
If it has changed, how come it's still the same?
Anyway, if we start to use the real definitions as such "bla bla / second" then one could equally much argue we should count the bus width too. If it's changing on 128 places, why count that as one? It'll still be 1 / s.
In computer hardware Hz has always referred to the clock, I can't see why DDR memory should be treated differently, except that it looks better on marketing slides.
After all, what we're really interested in is how much it can transfer. Instead of Hz I think it's more interesting to see the number of bytes / second instead.
arjan de lumens said:Hmmm ..
The SI unit 'Hertz' is nothing more special than just the inverse of a second (no need for a specific cycling waveform or anything like that) so saying "N times per second" means the same as "N Hertz" - so e.g. a DDR-II module that transfers data 1,000,000,000 times per second runs at 1 GHz data transfer rate.
You could make a point out of the fact that you cannot represent a frequency higher than 500 MHz using a data pin running at 1GHz transfer rate - that's similar to how a CD with a sample rate of 44100 Hz cannot represent frequencies above 22050 Hz, etc., but that's really another discussion ....
The unit for Hertz is inverse second.
The definition of Hertz is cycle per second. That's because cycle isn't a unit, but an idea, so that's why we use "Hertz" for cycle per second and not inverse second notation.
An example:
The unit for Barns is 10^-24 cm^2.
The definition for Barns is the probability of an interaction at the atomic level. Again, dropping the idea behind the specific unit name confuses the issue.
I did learn this specific definition of Hertz, I did find this specific definition of Hertz when I searched the web, and I've never encounteered another definition of Hertz, so I'm confused why you think otherwise.
The Hertz in memory speed is the number of times a clock cycle goes through the cycle sufficiently to then subsequently repeat the same behavior again, aka a "cycle". Distorting the meaning by focusing only on the units and substituting something not clearly defined as a "cycle" equates "Hertz" to "bits per second" or any other arbitrary "per second" value, and this strikes me as incorrect.
arjan de lumens
Veteran
OK then ... a little more web search finds this:
hertz (Hz)
the SI unit of frequency, equal to one cycle per second. The hertz is used to measure the rates of events that happen periodically in a fixed and definite cycle; the becquerel, also equal to one "event" per second, is used to measure the rates of things which happen randomly or unpredictably.
I would still say that the individual data transfers of that DDR-II @ 500 MHz input clock happen "periodically in a fixed and definite cycle" at 1 GHz data rate - exactly once every nanosecond a valid piece of data appears on the data bus - which I'd say is close enough to being a 'cycle' to warrant the use of the unit Hz for it.
hertz (Hz)
the SI unit of frequency, equal to one cycle per second. The hertz is used to measure the rates of events that happen periodically in a fixed and definite cycle; the becquerel, also equal to one "event" per second, is used to measure the rates of things which happen randomly or unpredictably.
I would still say that the individual data transfers of that DDR-II @ 500 MHz input clock happen "periodically in a fixed and definite cycle" at 1 GHz data rate - exactly once every nanosecond a valid piece of data appears on the data bus - which I'd say is close enough to being a 'cycle' to warrant the use of the unit Hz for it.
demalion said:
Well, saying that the DDR memory in my computer is currently running at 266MHz is not incorrect, as it's 266 million data cycles per second, but it isn't 266 million waveforms per second.
Coming from a physics perspective, Hz is just the number of something per second. It could be the rate of particles striking a surface, or the number of flies killed around the world every second.
That said, the most common use for the term Hz is as related to waves. That doesn't mean it can't be used for other things that happen at a nearly constant rate.
arjan de lumens said:
Now you're just abusing the word "cycle". Why was the unit "becquerel" invented if the significance of a cycle wasn't important?
Discrete events are not arbitrarily equal to the term cycle, and what you describe are discrete events. The data transmission occurs twice per cycle, or is the term "Double Data Rate" now insignificant?
Chalnoth said:
What is this "data cycle" term in this context? Currently, I think you just threw the two terms together, but if you have a specific mention of a "data cycle" and the term Hertz being a proper unit of measure for it, please introduce me to it.
I'm pretty sure there are things that could be called "data cycles", though I'd think they are "data transfer cycles" and refer to an abstract concept that would map more accurately to data transfer rate.
Ok, could you please simply then provide an example, as I've asked, that is "something per second" where the "something" is not a cycle/reference to a repeating waveform?
Why did' arjan's quote above mention the "becquerel" then? Seems there is more distinction intended than you credit.
Yeah, and the proper use of English doesn't mean we can't have l33t sp35k....that comment is just circular. Please provide, as I've asked, a more concrete example, preferrably one there aren't people out there criticizing as "improper usage" (since we're discussing such an "improper usage" at the moment). Note that arjan's quote and the existence of the term "becquerel" seem, to me, to contradict the principle of what you put forth.
arjan de lumens
Veteran
demalion said:
Hmmm .. It's still discrete events taking place at perfectly constant intervals, which would appear to match 'Hertz' better than 'becquerel' ..? Or are there other 1/s units that are more appropriate for this purpose?
arjan de lumens said:
The discussion started about whether DDR should be referred to by its clock rate, or its effective clock rate compared to "SDR".
Whatever has been made popular, and I think due to marketing, I would say:
1) I'm under the impression that it actually isn't exactly twice the data rate.
2) We have all sorts of things we can put before "per second", like bits, bytes, and other things that are actually meaningful and useful.
Therefore, your use of this reasoning as a justification for the use of "Hz" as a unit for this does not make sense because:
What is this "cycle" you refer to in regards to this? You're either ignoring the use of the word "cycle" in your definition as if it is not there, or failing to provide a clear link to the term "cycle" and the way in which DDR achieves its data transfer. I say again, isn't DDR specifically because it can achieve data transfer twice per clock cycle? If so, doesn't it make sense that the proper unit of measurement including a unit of data in it? The amount of data transferred does not change the clock cycle, and that is what the Hz in the specific case of memory makes sense for.
Again, if you have examples counter to this, please provide, but what it looks like to me is you're working backwards from the fact that people (when I say that, I really think "marketing") have used "Hz" to refer to this, and attempting to justify it, and therefore the logical support so far (to me) looks bass ackwards.
Note that the vigorous attack of your position is just my own personal values in action and not related to my personal opinion of you, a disclaimer that might be important with some of the atmosphere that has been present here in the past few months.
Similar threads
