Speed Vs. Latency: Myths And Facts
There’s a myth that every new memory format brings with it a latency penalty. The myth is perpetuated by the method upon which latency labels are based: Clock cycles.
Consider the latency ratings of the three most recent memory formats: Upper-midrange DDR-333 was rated at CAS 2; similar-market DDR2-667 was rated at CAS 4 and today’s middle DDR3-1333 is often rated at CAS 8. Most people would be shocked to learn that these vastly different rated timings result in the same actual response time, which is specifically 12 nanoseconds.
Because cycle time is the inverse of clock speed (1/2 of DDR data rates), the DDR-333 reference clock cycled every six nanoseconds, DDR2-667 every three nanoseconds and DDR3-1333 every 1.5 nanoseconds. Latency is measured in clock cycles, and two 6ns cycles occur in the same time as four 3ns cycles or eight 1.5ns cycles. If you still have your doubts, do the math!
The problem perceived by many less-informed buyers is that faster memory responds more slowly, but it’s obvious from these examples that this simply isn’t often the case. The real problem isn’t that response times are getting slower, but instead that they’ve failed to get quicker! When we see astronomical "speeds," we hope that our entire systems will become "more responsive" as a result. Yet, memory latencies are one place where things really haven’t changed much.
We still hope to find some truly "quick" modules, so today’s tests will include both "highest stable speed" and "lowest stable latency" configurations.
But What Are All Those Numbers?
So latency is measured in clock cycles rather than time, but what do all its numbers refer to? Most buyers look at only the first four latency values, and these appear in order of importance with numbers such as 9-9-9-24 in the case of high-speed DDR3 modules. These are typically labelled CAS Latency (tCL), RAS to CAS Delay (tRCD), RAS Precharge Time (tRP) and Active Precharge Delay (tRAS). A full definition of these functions is found on Page 2 of our article "PC Memory: Just the Facts.
Latest Memory News
Latest Memory reviews
- 27/12 – Four High-End Quad-Channel DDR3 Memory Kits For X79, Reviewed
- 05/09 – Seven 8 GB DDR3 Memory Kits For Your AMD A75 Motherboard
- 18/04 – Eight 8 GB (2 x 4 GB) Memory Kits For P67 Express, Rounded Up
- 23/11 – Memory Upgrade: Is It Time To Add More RAM?
- 05/08 – A Tour Of The Kingston Memory Factory In Taiwan
One part on page 3 is a bit confusing to me:
"Because cycle time is the inverse of clock speed (1/2 of DDR data rates), the DDR-333 reference clock cycled every six nanoseconds, DDR2-667 every three nanoseconds and DDR3-1333 every 1.5 nanoseconds. Latency is measured in clock cycles, and two 6ns cycles occur in the same time as four 3ns cycles or eight 1.5ns cycles. If you still have your doubts, do the math!"
If I am reading this right, cycle time is the inverse of (1/2)*(DDR Data Rate or 333 for example). So 1/((1/2)*333)=.006006006 which rounds to 6 milli-seconds, not 6 nano-seconds.
Sorry for my confusion, but could someone please clarify for me?
I can not clarify...this is a good article, but im still anoyed at the spelling and punctuation in the articles...Im not one for utmost spelling and punction in informall chatting messaging etc (like this) but i am when its a published article, you wouldnt find a harry potter novel with spelling mistakes in it....
"and DDR3-1600 is called PC2-12800." Its surposed to say, PC3-12800 not...PC2-12800
On the first page mid way down ^^^
Therlian you got it wrong. Generally, period (cycle time) it's the inverse of frequency (clock speed). But as DDR stands for Dual Data Rate, its real frequency is 1/2 of the specified clock speed. What you did wrong it's that you didn't take in account that DDR-333 means a data rate of 333 MEGA-hertz. Because a hertz is the inverse of a second you get: 1/((1/2)*333 MHz)=.006 micro- seconds, which is equal to 6 nano-seconds. Hope I made it clear for you.
I see now danmari. Thanks for pointing that out. I don't know why I didn't think about the 333 being MHz as opposed to Hz.
"I can not clarify...this is a good article, but im still anoyed at the spelling and punctuation in the articles...Im not one for utmost spelling and punction in informall chatting messaging etc (like this) but i am when its a published article, you wouldnt find a harry potter novel with spelling mistakes in it...."
Sorry but when you make THAT many mistakes in your own post, you have no right to talk about spelling mistakes. The mistake you mention is a typo (hitting a wrong key) not a spelling error.
You however manage to spell "annoyed", "informal", and "supposed" wrongly. You also managed to spell "punctuation" wrongly (the second time.. not sure how you can spell it right the first time and wrong the second..)
If you don't know how to spell, then don't flame others for it.
anyone with notes PC2100,PC3200,PC4200,PC5300?
"Intel’s current fastest Front Side Bus (FSB) uses a 1333-MHz data rate"
No, that's not right!
It's 1600MHz (native) for a Socket 775 CPU and socket 771, and if you didn't notice, some people overclock, big time!
Although DDR3 is touted as using 1.50V, several modules use more!
The same is true of DDR2. In this article DDR2 is described as being 1.8V. In my experience most of the modules are 1.9V and some are higher.