UNOFFICIAL tREF TABLE THREAD

[Dracula]

Hello,

 

Looking at the table,

 

How does 7.8us relate to 200Mhz, 0016 and 0064.

In my opening post of this thread,

here

, under the

GLOSSARY

section:

 

tREF = Click

HERE

 

For 7.8us memory choose either 0016 or 0064 in the BIOS.

 

Or am I reading the table wrong?

You are reading it correctly.

Cheery Bye...

[Fight Game]

If this is only for 200mhz memory, then why do you say good luck overclocking, and also in the OC section, since 200mhz is not oc'ed at all?

 

Doesn't Angry Games (and others) say that auto is a good setting, and it will find the best setting here?

[red930]

If this is only for 200mhz memory, then why do you say good luck overclocking, and also in the OC section, since 200mhz is not oc'ed at all?

 

Doesn't Angry Games (and others) say that auto is a good setting, and it will find the best setting here?

lol yeah and for the last year DIDNT HE SAY DO NOT USE CORSAIR RAM IN THESE BOARDS and the first two pic we see of his pc what do you see ? yeap CORSAIR RAM lol is he not learning too?

[Dracula]

If this is only for 200mhz memory, then why do you say good luck overclocking, and also in the OC section, since 200mhz is not oc'ed at all?

This is a good question. Firstly, do you want me to wish people bad luck?? And secondly, no where in my original post to this thread which is

here

, did I ever state, “...this is

only

for 200mhz memory...”. If you had read carefully, you would have seen straight away and stated obviously, in the whole second sentence of the first paragraph, that it says, and I quote, “It is

based

upon

a

200MHz

Front Side Bus with

no

divider in 1T mode, which equates to a

5ns

memCLK.” What would you have a comprehensive tREF Table based upon, a

455MHz

FSB or maybe a

1MHz

FSB?? Any table, like mine, needs to be based upon the

standard

,

stock

settings in order to form the proper foundation on which

O

ver

C

locking will be predicated. My tREF Table is a guide and a tool to facilitate help for those wishing to OverClock. So yes, not only does it belong in the OverClocking section of DFI-Street, it would be inappropriate to have it anywhere else.

 

Very appropriately, I included a few illustrations of OverClocked FSB’s that are clearly displayed right in the contents of my tREF Table. As you can plainly read at the top of the

YELLOW

section labeled, “

DELAY BETWEEN REFRESHES MEASURED IN CLOCK-CYCLES/TICKS

”, it shows exceptionally lucid examples of OverClocked FSB’s: “...i.e. - a 200MHz(FSB) = 5ns(memCLK),

250MHz

= 4ns

,

300MHz

= 31/3ns

, etc...}”. Anything beyond that, you will need to do your own math for your own custom settings.

 

Doesn't Angry Games (and others) say that auto is a good setting, and it will find the best setting here?

Travis’ official recommendation was “

3120

” or “auto”. Both he and I have also said that individual results will vary because of widely differing hardware, software and applicably contrasting personal settings.

 

lol yeah and for the last year DIDNT HE SAY DO NOT USE CORSAIR RAM IN THESE BOARDS and the first two pic we see of his pc what do you see ? yeap CORSAIR RAM lol is he not learning too?

I totally agree, “DO NOT USE CORSAIR RAM IN THESE BOARDS”. When I purchased my LANPARTY NF4 SLI-DR board and the Corsair memory, that particular motherboard had

just

been released and the Corsair was the most highly rated RAM on most review sites at that time. Many, many things (like newly available hardware) have changed since then and WE ALL have learned a lot. I got rid of that Corsair RAM ages ago now.

Cheery Bye...

[chennhui]

Hi Dracula, could you please explain in details how to deduct delay (in clock cycles) for a overclocked memory. Say, for example my PDP XBLK TCCD which have native refresh rate of 200MHz 3.9µs, overclocked to 237MHz?

[Sharp]

Hello,

 

If I am reading it correctly and you are saying that for a 7.8us rated module at 200Mhz to choose 0016 and 0064 then something is wrong.

 

tREF values

Auto = I get 2436MB/s in memtest

1560 = 2436MB/s in memtest

 

0016 = 1605MB/s in memtest

0064 = 2354MB/s in memtest

 

 

What I was asking is, how did you work this out?

3120 = 200MHz(15.6µs)

Did you work this out or copy it from somewhere?

 

 

You said this, (I agree with the statement)

The tREF is defined as the number of clock cycles between refreshes, or the duration of the pause between each actual refresh

But in the BIOS the tREF values are in clock cycles so how does 0016 clock cycles and 0064 clock cycles = 7.8us for a frequency of 200Mhz?

[chennhui]

I suspect 7.8us_refresh_rate_RAM (as listed in Everest) is related to tREF 1536/3072 (in 3.9us column). That is why A64 tweaker would show the tREF as 200, 3.9us for 7.8us_refresh_rate_RAM. Now what I wonder is this 780Hz is related to 7.8us_refresh_rate_RAM?

 

...If I am reading it correctly and you are saying that for a 7.8us rated module at 200Mhz to choose 0016 and 0064 then something is wrong.

[Dracula]

I am bringing this post from another area because it is pertinent to this thread and is worthy of debate and intelligent discussion. The comments and/or questions quoted below are legitimate and demonstrate a worthy level of contemplation. If you have some useful thoughts or data that would be applicable here and you would like to contribute to this thread, please feel free to do so respectfully.

 

AND THE CONFUSION ESCALATES

 

Not trying to stir the coals anymore but tref 128 is not really "faster". In ram faster pretty much means "more often". In general you want everything to happen more often such as reads and writes but tref - time refresh - is not really one of them. You want the ram refreshed often enough to keep the charge from dissipating and producing errors but refresh too much and you throw uneeded cycles in which prevent reads and writes from happening "faster" or "more often" which when it comes to getting more out of the ram, reads and writes, too many refreshes is actually slower.

 

Those sandra tests are so close as to probably be within the margin of error anyway. A better test would probably be 3dmark o1 or superpi to see best performance.

 

 

I have a feeling that a certain tref might be optimal at a certain voltage for a given ram but once you change the voltage or the clock another tref might be better as you've changed the voltage of the stored charge and also how often it gets reread and rewritten. A fresh write doesn't need a refresh.

 

this is from the overclcking sticky....

 

Paraphrased From Adrian Wong’s site: http://www.rojakpot.com/ ”This BIOS feature allows you to set the refresh interval of the memory chips. There are (several) different settings as well as an Auto option. If the Auto option is selected, the BIOS will query the memory modules' SPD chips and use the lowest setting found for maximum compatibility. For better performance, you should consider increasing the Refresh Interval from the default values (15.6 µsec for 128Mbit or smaller memory chips and 7.8 µsec for 256Mbit or larger memory chips) up to 128 µsec. Please note that if you increase the Refresh Interval too much, the memory cells may lose their contents. Therefore, you should start with small increases in the Refresh Interval and test your system after each hike before increasing it further. If you face stability problems upon increasing the refresh interval, reduce the refresh interval step by step until the system is stable.

 

From Sierra at ABXzone: The information below is taken from an old RAM guide. In a nutshell a memory module is made up of electrical cells. The refresh process recharges these cells, which are arranged on the chips in rows. The refresh cycle refers to the number of rows that must be refreshed. "Periodically the charge stored in each bit must be refreshed or the charge will decay and the value of the bit of data will be lost. DRAM (Dynamic Random Access Memory) is really just a bunch of capacitors that can store energy in an array of bits. The array of bits can be accessed randomly. However, the capacitors can only store this energy for a short time before it discharges it. Therefore DRAM must be refreshed (re-energizing of the capacitors) every 15.6µs (a microsecond equals 10-6 seconds) per row. Each time the capacitors are refreshed the memory is re-written. For this reason DRAM is also called volatile memory. Using the RAS-ONLY refresh (ROR) method, the refresh is done is a systematic manner, each column is refreshed row by row in sequence. In a typical EDO module each row takes 15.6µs to refresh. Therefore in a 2K module the refresh time per column would be 15.6µs x 2048 rows = 32ms (1 millisecond equals 10-6 seconds). This value is called the tREF. It refers to the refresh interval of the entire array." Here is an interesting discussion of tREF on the DFI forum: http://www.dfi-street.com/forum/showthread.php?t=10411

 

Also the number that you see in the bios aren't really a linear setting. They start low and work up but they only correlate when you in the same grouping of nano seconds like the 3.9's are linear but not a higher 3.9 vs a 15.6. There was a thread where some people tried to actually do the math but I don't think anybody knew wtf they were talking about and I haven't seen anybody explain it correctly.

 

Here is some testing I did with tref. The ram was way overclocked so it gave a lot of errors no matter what. You can see that numbers that are close together numerically don't scale with errors. Proving the point of the nanosecond times changing the calculation but probably only oscar wu knows what it means.

 

http://www.dfi-street.com/forum/showpost.p...48&postcount=27

 

Here is the data sorted by the value in the bios.

 

0016 - would reset after post or memtest wouldn't run

0032 - 508

0064 - 206

0338 - 118

0516 - 117

0648 - 82

0776 - 120

0780 - 80

0908 - 80

1032 - 126

1032 - 86

1168 - 120

1296 - 97

1536 - 106

1560 - 98

1816 - 74

2048 - 90

2064 - 70

2064 - 80 or 94 (I hit escape and couldn't remember exactly)

2336 - 108

2560 - 104

2592 - 93

3072 - 98

3120 - 77

3632 - 88

4128 - 80

4672 - 89

 

 

Anyway this is one of the better threads I've seen in a while and despite the four letter words in post or two it was just healthy debate. High drama from flying insults and it's about something interesting which is good for a change. After this book of a post I'm still not sure about 2.5-2-2-x. Nobody does it and thats probably for a reason, on the other hand 0128 may not be such and extreme tref after all since nobody knows what it means anyway. The best way to find the best tref is just to test. I admire you for at least trying something different despite the vicious headwind. Trying stuff is the only way to find out whats good.

AND

 

Ummm, I don't get it either. I still don't think anyone knows precisely what this is. I used memtest to to try them all and found 1560 game a lot of bandwidth ad stability.

 

It was stated earlier that "200mhz is the amount of data bandwidth that will be refreshed". Bandwidth is a measure of capacity in a bus or circuit, it's a measurement not a thing and can't be refreshed. MHZ is a measurement of time and not a measure of capacity such as bytes so this makes no sense either.

AND AGAIN

 

I still believe in what I said at the beginning.

 

780 for 3.9us memory @200Mhz

1560 for 7.8us memory @200Mhz

3120 for 15.6us memory @200Mhz

• “Bandwidth: how much information can be transferred over a connection in a given period of time.
  It's usually measured in bits per second or bytes per second
  , e.g.: 1.5 Mbit/s, (1.5 million bits per second) or 150 MB/s. (150 million bytes per second)” – Reference.com
  
  
   
  
  

http://www.reference.com/browse/wiki/Compa...y_and_bandwidth

 

• “1) In electronic communication,
  bandwidth is the width of the range (or band) of frequencies that an electronic signal uses on a given transmission medium
  . In this usage, bandwidth is expressed in terms of the difference between the highest-frequency signal component and the lowest-frequency signal component.
  Since the frequency of a signal is measured in hertz (the number of cycles of change per second), a given bandwidth is the difference in hertz
  between the highest frequency the signal uses and the lowest frequency it uses. A typical voice signal has a bandwidth of approximately three kilohertz (3 kHz); an analog television (TV) broadcast video signal has a bandwidth of six megahertz (6 MHz) -- some 2,000 times as wide as the voice signal.
  
  
   
  
  
  2) In computer networks,
  bandwidth
  is
  often
  used
  as
  a
  synonym
  for
  data transfer rate
  - the amount of data that can be carried from one point to another in a given time period (usually a second). This kind of bandwidth is usually expressed in bits (of data) per second (bps). Occasionally, it's expressed as bytes per second (Bps). A modem that works at 57,600 bps has twice the bandwidth of a modem that works at 28,800 bps. In general, a link with a high bandwidth is one that may be able to carry enough information to sustain the succession of images in a video presentation.
  
  
   
  
  
  It should be remembered that a real communications path usually consists of a succession of links, each with its own bandwidth. If one of these is much slower than the rest, it is said to be a bandwidth bottleneck.” - searchNetworking.com
  
  
   
  
  

http://searchnetworking.techtarget.com/sDe...i211634,00.html

 

Just so everyone knows, this is a

VERY

ambiguous topic to which even the "EXPERTS" are confused or speak with great conflict, contradicting their own words in the same breath. Here are some links on some definitions of "bandwidth":

 

http://www.answers.com/bandwidth

 

AND

 

http://dictionary.reference.com/search?q=bandwidth

 

AND

 

http://en.wikipedia.org/wiki/Bandwidth

 

• "
  2.
  The
  amount
  of
  data
  that can be passed along a communications channel in a given
  period of time
  ." - Answers.com
  
  

 

• "
  1.
  Capacity
  or
  time
  . Computer people sometimes use the term
  very
  broadly
  ..." - Answers.com
  
  

 

• "
  Meaning #1:
  a data transmission rate; the maximum
  amount of information
  (bits/second) that can be transmitted along a channel" - Answers.com
  
  

 

• n. 1. [common] Used by hackers (in a generalization
  
  
  of its technical meaning) as the
  volume of information per unit time
  
  
  that a computer, person, or transmission medium can handle..." - Dictionary.com
  
  

 

• "Bandwidth is a measure of frequency range...measured in hertz...In a digital communication system, bandwidth has a
  dual
  meaning. Technically, it is a synonym for baud rate, the rate at which symbols may be transmitted through the system. It is also used colloquially to describe channel capacity, the rate at which bits may be transmitted through the system. Hence a 66 MHz digital data bus with 32 separate data lines may properly be said to have a bandwidth of 66 MHz and a capacity of 2.1 Gbit/s — but it would not be surprising to hear such a bus described as a having a "bandwidth of 2.1 Gbit/s". Similar confusion exists for analog modems, where each symbol carries multiple bits of information so that an modem may transmit 56 kbit/s of information over a phone line with a bandwidth of only 12 kHz...In telecommunications and computing, bit rate (sometimes written bitrate or Rbit) is the frequency at which bits are passing a given (physical or metaphorical) “point”. It is quantified using hertz, the SI unit of frequency, or the bit per second (bit/s) unit..." - Wikipedia
  
  

 

MHz

is

bandwidth and is a measure of speed (not conventional velocity) and time and

does

correlate directly to data. These below, are

not

algorithms:

 

780 for 3.9us memory @200Mhz

1560 for 7.8us memory @200Mhz

3120 for 15.6us memory @200Mhz

 

A tREF of

780 T

is in the

200MHz (15.6µs)

category,

1560 T

is in the

200MHz (3.9 µs)

category and

3120 T

is in the

200MHz (15.6 µs)

category. There are many sources on the net that back this up, including the program

A64 Tweaker

. And WHAT AUTHORITY (if there really is such a thing)

ever

stated absolutely that "bandwidth"

cannot

be refreshed? Additionally, there's even a simple table displayed about this (that is not mine and preceded mine) here at DFI-Street, provided by Angry_Games himself.

 

Cheery Bye...

[Sharp]

My questions are based on results.

 

You are implying that for a 7.8us refresh to choose 0016 or 0064 but memtest shows a decrease in performance.

 

0016 = 1605MB/s in memtest

0064 = 2354MB/s in memtest

 

Therefore your table cannot be correct.

Can you explain this?

 

I am also aware of A64 tweaker and I believe the program is wrong.

Prove me wrong and I will go away.

[chennhui]

mm, all we need to know how we can make use of the table on a overclocked system, such as for a given memory frequency, how we determine the tREF. Less people actually care what it mean at stock since leaving it Auto will just work fine. Therefore, could you please given a short tutorial on how you obtaining the tREF value for a given memory frequency, i.e. in you case at DDR400.

 

Hi Dracula, could you please explain in details how to deduct delay (in clock cycles) for a overclocked memory. Say, for example my PDP XBLK TCCD which have native refresh rate of 200MHz 3.9µs, overclocked to 237MHz?

[gbomb944]

MHz is bandwidth and is a measure of speed (not conventional velocity) and time and does correlate directly to data. Not precisely true. MHz is how many cycles between on and off, charge and no charge or charge and negative charge in a second. It only relates to data if something reads the cycles and turns it into 1's and 0's or data.

 

These below, are not algorithms: 780 for 3.9us memory @200Mhz 1560 for 7.8us memory @200Mhz 3120 for 15.6us memory @200Mhz A tREF of 780 T is in the 200MHz (15.6µs) category, 1560 T is in the 200MHz (3.9 µs) category and 3120 T is in the 200MHz (15.6 µs) category.

 

I'm pretty sure you are correct about the above. The us such as 3.9 is how long each bank is refreshed, the mhz is how often (not sure about this, especially how it would change when using a 100mhz setting while your ram is clocked at 260 or whatever). I think that they perform an algorithm on these two separate variables and come up with "tfactor" or tref. I think there are only 100, 133, 200, etc because these are the only supported speeds for ddr.

 

There are many sources on the net that back this up, including the program A64 Tweaker. And WHAT AUTHORITY (if there really is such a thing) ever stated absolutely that "bandwidth" cannot be refreshed?

 

 

I don't have an issue with your findings on tref, ( I sort of get it but still don't understand the math), I just think you are using the the term "data bandwidth" in regard to "being refreshed" incorrectly. I'm certainly not an "authority" but I did work in satcom and rf for awhile so I know my way around a spectrum analyzer and an oscilliscope.

 

Bandwidth refers to the amount of data that can pass through a given circuit, (usually called a "carrier"), in a given medium, be it air or copper in any moment in time. A circuit can continuously pass data so it doesn't get refreshed unless you change some parameter of the circuit like the frequency.

 

If you look at a "band", (a frequency range such as FM for instance 88 - 108 Mhz), on a spectrum analyzer you'll see at flat line at the bottom then when you see you're favorite radio station it will protrude up a certain amount from the bottom, (power of the signal over background noise), and it will be so wide, (it's bandwidth or how much of the frequency range it occupies. this is the precise meaning of bandwidth and it has no other). Wider signals encompass more of a given spectrum, they are "broadband" and can cary more data per second or whatever time measure you like. They are continuous and never get refreshed. They transmit either a 1 or a 0 nonstop for ever, the wider they are the more 1's and 0's they can transmit at a time so they have more "bandwidth". Data being transmitted down the tiny wires of your mobo is the same thing it's just going through copper, same with ethernet, or the data that comes through your cable modem.

 

You can look here at the last couple of images to get an idea what a spectrum plot looks like. http://www.sencore.com/newsletter/Oct02/Sp...umAnalyzer1.htm

 

When talking about tref we are refreshing the charge of the pages in ram that are holding 1's and 0's. Everything in computers and is done with electricity 0= no charge 1 being has a charge. See here mainly for the pic of the banks on a stick. http://arstechnica.com/paedia/r/ram_guide/...de.part3-1.html

 

It is a physical capacity of storage therefore we are not refreshing bandwidth we are refreshing a certain number of banks on the sticks themselves. These are capacitors on the sticks. They as a group have a capacity for storage but they do not transmit anything thus they do not have a bandwidth and shouldn't be described as such. Your hard drive is the same thing. It can store so many gigabytes which is it's capacity. When you write some data to it you didn't do anything to it's bandwidth. If you test how fast you can write to it you are testing it's "bandwidth".

 

When we buy ram it isn't measured in bandwidth it's in megabytes, 512 or whatever which refers to how many 1's and 0's it can hold. If it runs at a high clock speed it can read and write at high speed and output, (actually be read by but close enough), data to the memcontroller quickly so it could be said to have "high bandwidth" compared to a lesser stick. Bandwidth being how much data it can read/write at any given moment in time.

 

Hopefully this makes sense. I think you may understand the math of 200mhhz 15.6us (I think I do but don't want to go there), but not precisely what is happening when ram stores data or how it's written to or read, and moreso the use of bandwidth to describe it. Bandwidth should be used to describe throughput or output but not refreshing the charge of capacitive storage. I'm not an electrical engineer and don't even have a degree so I could be wrong. Also please don't take this as an affront as I'm not trying to attack you. I just like talking about advanced stuff like this and also having to post and detail things makes me think about what I actually do know and if it's right or not. I appreciate the discourse.

[Dracula]

AN ANALOGY

(and yes I do know that it is not the exact same thing):

 

“Imagine you are an engineer who has complete control of a water system (computer) that contains water-pipes (circuits designed for the sole purpose of carrying data) that are 20’ in diameter and are under continuous pressure from multiple 1000HP submersible pumps; drawing water from a major water source. Let’s say for the sake of argument that each pipe, working together with the pumps, can push through a maximum of 200 Megaliters (200Mb) of water (data) per second (MHz). The 200 Megaliters per second (MHz) thereby represents the maximum

potential

capacity of each pipeline. Whether the FULL capacity of each pipeline is used depends DIRECTLY upon the amount of water present in each pipe. No mater what amount of water is in a pipe, the pipe’s capacity remains constant. But, just because the pipe’s capacity is constant does not mean that the “pipeline” cannot be “flushed” (refreshed) periodically. The

WATER

that each pipe carries is the sole reason why all the pipes were built in the first place.

 

Now, let’s factor in timed valves. Each pipe has a valve on it that opens up 100%, once every 3,072 system cycles (tREF delay). The 3,072 system cycles, each cycle being equal to 5 nanoseconds, is the length of time a valve remains

closed

. When a valve opens, it remains completely open for exactly 3.9 microseconds and then closes again. It remains closed again for another 3,072 system cycles at which end it opens for yet another 3.9 microseconds and so on, and so on, continually...Each time that valve opens for the 3.9 microseconds, it has a maximum

potential

capacity, during that period, of 780 Liters that can be “

put through

” a valve to reach its destination. Whether the full capacity of a 3.9 microseconds pipeline is used or not is irrelevant because the

potential

of 780 Liters remains

constant

. When water passes through a valve from a pipe (data throughput) to reach its destination, the water that remains in the pipe is “

freshened

” with new water from the original source from which it was drawn. The contents of the pipeline has been exchanged and freshened; the pipeline has, ever so slightly, been flushed (refreshed).”

 

I don't have an issue with your findings on tref, ( I sort of get it but still don't understand the math), I just think you are using the the term "data bandwidth" in regard to "being refreshed" incorrectly. I'm certainly not an "authority" but I did work in satcom and rf for awhile so I know my way around a spectrum analyzer and an oscilliscope.

There is NO question in my mind that you are a knowledgeable person and know what it is that you are talking about when a subject of interest to you arises. This post of yours that I am replying to here is not only the most cogent and thoughtful post in this thread but it also shows intelligence and civility, unlike the tripe in the

“

Tref problem, by

abracadabra

”

post

,

which is all about “thread-crapping”, “flaming” and acting like a bunch of childish little snot-nosed brats

.

What you said was well spoken and correct. Yes, I admitted that what you said was correct (imagine that). So that means that I must now eat crow and tell the whole world how misinformed and thick-headed I am because I am wrong

...well, ...um, ...actually, ...uh, ...NOPE! You aren’t wrong but neither am I.

 

Your post was remarkable and very informative, accurate and totally on target, except for one teeny-tiny thing; the target you were aiming at was NOT the same target presented in this thread. I will explain this in great detail, as I progress in my replies to each section of your post.

 

Bandwidth can be capacity when referring to radio frequencies and circuits, and yes, I am also very familiar with oscilloscopes, VU meters, spectrum analyzers and things like AM / FM radio, etc... In this thread, we are specifically talking about computers

. The facts and figures that you listed in an earlier post on RAM were also correct, but are based on very old technology from the late 1990’s (Kingston’s Ultimate Memory Guide). That data is not wrong, but it is not up to date either.

 

I want to clear something up now. I

NEVER

stated that any

set

“bandwidth” would be changed during a “refresh”; I never said that bandwidth did not have EVERYTHING to do with capacity in a circuit. What I DID SAY, was that “bandwidth” has to do with oscillations (data transmitted in the form an electrical signal - send and receive) measured in Hertz as regarding the rising and falling edge of the memory’s clock. Those “oscillations” pertain directly to data sent (rise) and data received (fall), or “looping” as it is sometimes called. The

bandwidth

demonstrates the

full capability

,

maximum capacity

and

speed

of

data throughput

in a given period of time, in a channel or circuit. What do you think the whole point of such capacity is??

Each

time

that data throughput is

interrupted

then

released

, interrupted then released, you have, in essence, “refreshed” the “data” in each of those releases. The amount of that data is directly proportional to the “capacity”/“bandwidth” of the channel or circuit in which it is being transmitted through.

 

Example

- 200MHz is 200MHz is 200MHz, yes? This fact is NOT in dispute. 200MHz represents the maximum capacity of “potential” data throughput in any given

one

second

period of time in a circuit or channel. Each one Hz has a total capacity of one bit of potential data throughput. Therefore, if you have a maximum “bandwidth” capacity of 200 Million Hertz (or oscillations - send / receive) every second, you will also have a

potential

data throughput that is capable of a maximum of 200 Million Bits of data every one second as well. And yes, I am aware that 1’s and 0’s are charge and no charge or can be compared analogously to turning a light switch ON and OFF.

 

Bandwidth refers to the amount of data that can pass through a given circuit, (usually called a "carrier"), in a given medium, be it air or copper in any moment in time. A circuit can continuously (the appropriate term here is “continually”) pass data so it doesn't get refreshed unless you change some parameter of the circuit like the frequency.

 

If you look at a "band", (a frequency range such as FM for instance 88 - 108 Mhz), on a spectrum analyzer you'll see at flat line at the bottom then when you see you're favorite radio station it will protrude up a certain amount from the bottom, (power of the signal over background noise), and it will be so wide, (it's bandwidth or how much of the frequency range it occupies. this is the precise meaning of bandwidth and it has no other). Wider signals encompass more of a given spectrum, they are "broadband" and can cary more data per second or whatever time measure you like. They are continuous and never get refreshed. They transmit either a 1 or a 0 nonstop for ever, the wider they are the more 1's and 0's they can transmit at a time so they have more "bandwidth". Data being transmitted down the tiny wires of your mobo is the same thing it's just going through copper, same with ethernet, or the data that comes through your cable modem.

I am very familiar with bandwidth and "carrier circuits" and to tell you the truth, just like ALL the

multiple

sources

I have shown you in my previous posts, the “medium” in which the term is used can make a drastic difference in the term’s meaning; “BANDWIDTH”

IS

A VERY BROADLY USED TERM. I will give you yet another example:

 

“

I used to have a really good old 19” CRT Monitor that was capable of very high resolutions. I ALWAYS kept the refresh rate of that monitor at 100Hz, WHICH IS ITS REFRESH DATA BANDWIDTH RATE. The

data

on the screen is re-created or replenished or “refreshed” at a rate of 100 cycles per whatever unit of time...The advantage of higher refresh rates on CRT monitors is that you get less of a flicker effect and it is easier on the eyes. It can also be hard on a monitor that is not designed for high level refreshes. Frequency is ALWAYS measured in HERTZ (Hz); HERTZ and BANDWIDTH and DATA are all connected in monitor technology. The refresh rate of a monitor is measured in Hertz (frequency) and has EVERYTHING to do with replenishing the OLD DATA on the screen with NEW UPDATED DATA. As you can clearly see, REFRESH and FREQUENCY and HERTZ are all used to describe the SAME THING - BANDWIDTH!

”

You can look here at the last couple of images to get an idea what a spectrum plot looks like. http://www.sencore.com/newsletter/Oct02/Sp...umAnalyzer1.htm

 

When talking about tref we are refreshing the charge of the pages in ram that are holding 1's and 0's. Everything in computers and is done with electricity 0= no charge 1 being has a charge. See here mainly for the pic of the banks on a stick. http://arstechnica.com/paedia/r/ram_guide/...de.part3-1.html

 

It is a physical capacity of storage therefore we are not refreshing bandwidth we are refreshing a certain number of banks on the sticks themselves.

First of all, "

Electricity

"

is a Field of Study in Science

and is not the same thing as "

Electrical Current

". And no, bandwidth IS NOT a

set

capacity of “physical storage”. I'll say that again, BANDWIDTH IS NOT “PHYSICAL” STORAGE. The “storage” aspect of RAM is achieved “dynamically”,

inside

the CELLS of the chips on the PCB. The circuits that run to and from those “cells” act as data-paths or passageways that facilitate a certain amount of bandwidth, in the form of rapid electrical pulses, specifically for data. That

data-bandwidth

can be defined as

the potential maximum capacity of DATA THROUGHPUT in a set period of time, which is measured in units of frequency called Hertz

(please leave the Rental Car company out of this :eek2: ).

These are capacitors on the sticks. They as a group have a capacity for storage but they do not transmit anything thus they do not have a bandwidth and shouldn't be described as such. Your hard drive is the same thing. It can store so many gigabytes which is it's capacity. When you write some data to it you didn't do anything to it's bandwidth. If you test how fast you can write to it you are testing it's "bandwidth".

A

capacitor

is designed to act like a FAST battery, “charging” and “discharging” a set

amount

of electrical capacitance for a VERY, VERY BRIEF period of time. Charge, discharge, millions of times per second, producing a “strobe-like” effect...creating, renewing, “REFRESHING” cells of data, line by line immensely fast. So, what do you think “bandwidth” is exactly? If it is “capacity” as you say and I agree that it IS, regarding data throughput, for what do you think that “capacity” is being used, STORAGE? Yes, the cells in the chips that are soldered onto the PCB (Printed Circuit Board) hold data, one bit each. Also, you yourself stated in a previous post, “the bits of data stored in the cells must be periodically refreshed or else the data (

The MINNOW, lol

) would be lost”. So, yes the “CELLS” store the data temporarily as dynamic data, but also they MUST

ALWAYS

BE ELECTRICALLY “REFRESHED” IN SET INTERVALS OF TIME. Why in the world would you think that bandwidth / capacity is not

directly

related to data throughput? The data is replenished BECAUSE the “bandwidth” has been “interrupted”. The “bandwidth” and the data throughput are SYMBIOTIC; you CANNOT have one without the other - they facilitate each other, always.

 

When we buy ram it isn't measured in bandwidth it's in megabytes, 512 or whatever which refers to how many 1's and 0's it can hold...

 

True but that ain’t a full description either. When I or you or anyone in Taiwan purchases DDR SDRAM, I or you or they buy a certain size, yes and additionally what other information is there about that RAM?? DDR 600MHz, PC - 4800. What on Earth do you think those two numbers are referring to? The DDR (Double

Data

Rate) is the measure of the “effective” clock in DDR mode, based on a 300MHz FSB which is measured as 600MHz...Hmmm, there’s that bothersome frequency abbreviation “MHz”, referring to “bandwidth” yet again, and it is used in the same breath as Double

Data

Rate

. The 4800 in “PC - ...” is also referring to the “bandwidth” and total data throughput on an 8 bit scale. Once you go up to 16 bit you are referring to RDRAM, but that would be a topic for another thread.

 

...If it runs at a high clock speed it can read and write at high speed and output, (actually be read by but close enough), data to the memcontroller quickly so it could be said to have "high bandwidth" compared to a lesser stick. Bandwidth being how much data it can read/write (or send / receive) at any given moment in time.

I could not have said it better myself. G-Bomb, dude, I HAVE been saying exactly

that

this whole time. You’re getting way too hung-up on the explicit definition of a term that has a very broad usage and implicit meaning. But here, you just said, and I quote, “

Bandwidth

being

how

much

data

it can read/write (or send / receive) at

any

given

moment

in

time

”.

“HOW

REFRESHING

!” THAT IS WHAT I HAVE BEEN SAYING ALL ALONG!

 

Hopefully this makes sense. I think you may understand the math of 200mhhz 15.6us (I think I do but don't want to go there), but not precisely what is happening when ram stores data or how it's written to or read, and moreso the use of bandwidth to describe it. Bandwidth should be used to describe throughput or output but not refreshing the charge of capacitive storage.

When a ROW or COLUMN Address is “refreshed” (RAS and CAS), it receives “strobes” (Row Address Strobe and Column Address Strobe) of electrical pulses that update each cell, a ROW and/or COLUMN at a time - very, very rapidly of course. The “bandwidth” IS regarding the “

data

throughput” and is done so electrically.

 

I'm not an electrical engineer...

Nor am I, but one day perhaps. And here, appropriately, is still another example:

 

“Every time you flip a light switch to the “ON” position, a set amount of electrical current, measured in amps, flows through the wires to the light bulb. The amount of current those wires can handle can be measured in a set amount of amperes. If I turn that light OFF, the electrical current in the wires will dissipate rapidly. Though there is NO current in the disengaged circuits and wires, the capacity

potential

of amps has NOT been changed. Now, if I turn that light switch ON and OFF extremely rapidly, in varying patterns of complex logic, the current in the wires to the light bulb could now be interpreted as rapid electrical signals or pulses. Those “pulses” radiating through the wires to the light bulb would be expresses externally as flashes. The flashes could also be interpreted as data code and the amount of those electrical “pulses” per second, up to the amperage capacity of the wires, could be considered the “BANDWIDTH”. Every time that “data code” needs to be updated with NEW SIGNAL PATTERN STRUCTURES, there would be just an ever so brief pause and then the electrical signal of current in the wires would commence again but in a changed pattern. The wire, that carries the pulses of current has been updated or “refreshed” with a NEW signal pattern. An update in a signal pattern IS THE SAME AS A REFRESH. The pattern of the signal conveys the data as pulsed code and can expand through the wires to the extent of those wire’s amperage limit.

 

THE AMOUNT OF FREQUENCY OF ELECTRICAL CURRENT IN A CIRCUIT, CONDUCTED AS A “PULSED” PATTERN OF CODED DATA TRANSMISSION IN AND AS THAT EXACT SAME ELECTRICAL CURRENT, ARE BOTH “DATA BANDWIDTH AND DATA THROUGHPUT”, ACTING SYMBIOTICALLY TOGETHER AS ONE AND THE SAME

.

”

 

------------------------------------------------------------------------------------------------------------------------------------

------------------------------------------------------------------------------------------------------------------------------------

------------------------------------------------------------------------------------------------------------------------------------

 

With ALL this said, it is time to get back to

tREF

and WHAT IT IS

exactly

. The term “

tREF

” by its very name and definition, fundamentally means “

REFRESH

”! Now, like our elementary school teachers used to say, “It’s time to put on our ‘Thinking Caps’!”

 

Let’s use the BIOS tREF setting of

4708 T

with a

200MHz FSB

at a

1:1

in

1T mode

for our example, on this occasion. The setting of

4708 T

falls directly into the

166MHz

,

1.95µs

category of the BIOS tREF “refresh” bracket. So, let us break down what

each

of these numbers

must

mean, by

fact

, and through the Age-Old “process of elimination.”

 

We already know that

4708 T

is the number of memClock Cycles of

delay

between

each

“REFRESH”. We also know, for fact, that a

200MHz FSB

with a

1:1

in

1T mode

WILL produce a

5ns

“Clock” [(1 ÷ 200MHz FSB, Memory Frequency) x 1000]. In addition, we know that

1.95µs

WILL BE the duration, in microseconds (one millionth of a second), of the actual “REFRESH” (and no,

A64 Tweaker

is

not

“wrong”. THAT IS a ridiculous and laughable statement with absolutely

no

proof

,

whatsoever

, to back it up...but we will cover

that

rubbish

in a different post). But what then, you ask, is it that is BEING “refreshed”, exactly? Well, since the

4708 T

IS the

delay

and the

1.95µs

IS the actual “refresh” interval and tREF means “REFRESH”, that leaves us with ONLY

one

number left to look at,

166MHz

. Since we have unequivocally established that tREF is a “REFRESH” setting in the DRAM Configuration Settings in the BIOS on your motherboard, it HAS TO refer to the

166MHz

figure. Since

166MHz

is in terms of “Hertz (Hz)”, which IS a measurement of frequency, and since frequency CAN BE defined as “bandwidth” and since “

BANDWIDTH

” CAN BE defined as, “

the full capability, maximum capacity and speed of data-throughput in a given period of time, in a channel or circuit

”, and since IT IS THE ONLY REMAINING FIGURE OF THE ORIGINAL 3 LEFT, and since

tREF

means “

refresh

” and SOME--THING IS BEING “REFRESHED”, it MUST BE the “

data-bandwidth

”, which IS the

166MHz

figure because it IS the

ONLY

figure left! THAT “data-bandwidth” IS

EXACTLY

WHAT

IS

BEING “REFRESHED”! Repetition, repetition, repetition...! You see, a

painfully

simple

and

completely

logical

, “process of elimination!” :nod:

 

...and don't even have a degree so I could be wrong. Also please don't take this as an affront as I'm not trying to attack you. I just like talking about advanced stuff like this and also having to post and detail things makes me think about what I actually do know and if it's right or not. I appreciate the discourse.

I AGREE! This helps us ALL to think these things through thoroughly, giving us ALL a much better understanding of what is going on here. My reply to you is also not meant as an attack, but I cannot go against that which I know to be true, now knowing what your understanding of data bandwidth is. Your’s is a very scientific, as in physics, type of definition of the term “bandwidth”. While not incorrect, it is not precisely the “target topic” of this thread. “Bandwidth” is a VERY BROADLY USED TERM and cannot be pigeonholed into one

narrow

definition. It not only refers to “speed and

potential

capacity”, but it CAN also refer to data and data throughput. For these reasons, I will not cede my tREF Table to your definition of bandwidth nor will I change its premise. You are very knowledgeable and a very intelligent person, but the word “bandwidth” IS NOT confined merely to the “light-spectrum”, “radio-wave” and “oscilloscope” definitions to which you desire to imprison it.

 

Cheery Bye...

 

P.S. - Sharp and Chennhui, I WILL be responding to the two of you very soon... Mwahahaha! :eek: