What is memory cas latency: What Is CAS Latency in RAM? CL Timings Explained

What Is CAS Latency in RAM? CL Timings Explained

(Image credit: MaIII Themd/Shutterstock)

When buying RAM, you’ll see listings for their timings, such as CL16-18-18-38 or CL14-14-14-34 or CL 16-18-18-36. That number after the “CL” represents the RAM kit’s CAS latency, sometimes called CL. But what does CAS latency really mean? And what implications does a RAM kit’s CAS latency have for its performance?

The lower the CAS latency, the better

Why you can trust Tom’s Hardware
Our expert reviewers spend hours testing and comparing products and services so you can choose the best for you. Find out more about how we test.

A RAM module’s CAS (Column Address Strobe or Signal) latency is how many clock cycles in it takes for the RAM module to access a specific set of data in one of its columns (hence the name) and make that data available on its output pins, starting from when a memory controller tells it to. Another way to think of it is how many RAM clock cycles it takes for the RAM to output data called for by the CPU. A RAM kit with a CAS of 16 takes 16 RAM clock cycles to complete this task. The lower the CAS latency, the better.

CAS latency can be referred to in several different ways. A RAM kit with a CAS latency of 16, for example, can be described as CAS 16 or CL16 or as having CAS 16 timings.

A listing for a RAM kit on Newegg.com shows the CAS latency is 18. (Image credit: Newegg)

It’s important to note that two different RAM kits with the same data transfer rate, for example DDR4-3200, can have different CAS timings. Take the Team Group Delta Tuf Gaming RGB DDR4-3200 and G.Skill Trident Z Royal DDR4-3200. They both have a DDR4-3200 transfer rate. However, their CAS timings are different: CL16-18-18-38 (CAS 16) and CL14-14-14-34 (CAS 14), respectively.

This RAM kit’s label shows it’s a C18 kit. (Image credit: Tom’s Hardware)

RAM speed vs. RAM latency

While a RAM’s data transfer rate tells you how many mega transfers (1,000,000 data transfers) the RAM can do in one second (a DDR4-3200 RAM can conduct 3,200 mega transfers in 1 second), its CAS latency is also important for understanding RAM performance.

CAS latency tells you the total number of cycles it takes for the RAM to send data, but you should also consider the duration of each cycle to get a better idea of that RAM’s overall latency.

While DDR4 RAM is newer with better storage density and power efficiency than DDR3 RAM, it tends to have higher CAS latency. DDR3 RAM usually has a CAS latency of 9 or 10, while DDR4 will have a CAS latency of at least 15. However, because of its faster clock speeds, the newer standard has better performance overall.

This article is part of the Tom’s Hardware Glossary.

The Difference Between RAM Speed and CAS Latency

Memory (DRAM) performance is all about the relationship between speed and latency. While the two are closely related, they’re not connected in the way you might think. Here’s how speed and latency are related at a technical level – and how you can use this information to optimize your memory’s performance.

If you’re interested in how much memory your computer should have, read here.

The perception and the truth about latency

Perception

Many users believe CAS latency is an accurate indicator of real world latency performance
Many users also believe that because CAS latencies increase with speed gains, some of the speed gets nullified

Truth

Semiconductor engineers know that CAS latencies are an inaccurate indicator of performance
Latency is best measured in nanoseconds, which is a combination of speed and CAS latency
Both speed increases and latency decreases result in better system performance
- Example: because the latency in nanoseconds for DDR4-2400 CL17 and DDR4-2666 CL19 is roughly the same, the higher speed DDR4-2666 RAM will provide better performance
- Example: if the speed rating of a standard module and a gaming module is the same (i. e. DDR4-2666) but the CAS latencies are different (i.e. CL16 vs. CL19), then the lower CAS latency will provide better performance

The difference between the perception of latency and the truth of latency comes down to how latency is defined and measured.

The latency paradox

Latency is often misunderstood because on product flyers and spec comparisons, it’s noted in CAS Latency (CL), which is only half of the latency equation. Because CL ratings indicate only the total number of clock cycles, they don’t have anything to do with the duration of each clock cycle, and thus, they shouldn’t be extrapolated as the sole indicator of latency performance.

By looking at a module’s latency in terms of nanoseconds, you can best judge if one module is, in fact, more responsive than another. To calculate a module’s latency, multiply clock cycle duration by the total number of clock cycles. These numbers will be noted in official engineering documentation on a module’s data sheet. Here’s what these calculations look like.

What is latency and the latency equation?

At a basic level, latency refers to the time delay between when a command is entered and when the data is available. Latency is the gap between these two events. When the memory controller tells the memory to access a particular location, the data must go through a number of clock cycles in the Column Address Strobe (CAS) to get to its desired location and complete the command. With this in mind, there are two variables that determine a module’s latency:

The total number of clock cycles the data must go through (measured in CAS Latency, or CL, on data sheets)
The duration of each clock cycle (measured in nanoseconds)

Combining these two variables gives us the latency equation:

latency ^(ns) = clock cycle time ^(ns) x number of clock cycles

In the history of memory technology, as speeds have increased, (which means clock cycle times have decreased), the CAS latency values have also increased, however because of the faster clock cycle the true latency as measured in nanoseconds has remained roughly the same. By optimizing the balance between the maximum speed your processor is capable of and the lowest latency memory available within your budget, you are able to achieve a higher level of performance using newer, faster, and more efficient memory.

Which is more important: speed or latency?

Based on in-depth engineering analysis and extensive testing in the Crucial Performance Lab, the answer to this classic question is BOTH! Speed and latency both play a critical role in system performance, so when looking to upgrade we recommend:

Step 1: Identify the highest memory speed supported by both your processor and motherboard (including overclocking profiles).
Step 2: Select the lowest latency memory that fits within your budget at that speed, remembering that a superior (i.e. lower) latency means superior system performance.

©2018 Micron Technology, Inc. All rights reserved. Information, products, and/or specifications are subject to change without notice. Neither Crucial nor Micron Technology, Inc. is responsible for omissions or errors in typography or photography. Micron, the Micron logo, Crucial, and the Crucial logo are trademarks or registered trademarks of Micron Technology, Inc. All other trademarks and service marks are the property of their respective owners.

what is it and what does it mean?

Greetings dear guests! Today we’ll talk about the latency of RAM — what it is, what happens, how timings affect it, and what it means in practical terms.

How RAM works

Structurally, any RAM bar is a kind of matrix, divided into rows and columns. Each cell can have a value of 1 (fully charged) or 0 (fully discharged). In fact, any such bar is a kind of table consisting of many microscopic capacitors. nine0003

Each element in the RAM has its own unique address, which is directly accessed by the processor or peripheral devices.

In addition, the capacitors are grouped into banks, the number of which depends on the cell density. It takes longer to open a line in one bank than if you go to another bank, because the line in use must first be closed. The principle of row alternation is applied when a new row is opened in a new bank. nine0003

What does memory module latency mean

The literal definition of this RAM parameter is “latency”, that is, the time required to read, write and copy data.

Despite the high speed of modern computers, all actions are not performed instantly. In another way, such delays are called timings and, for the convenience of users, they put such characteristics on the nameplate, which must be pasted in accordance with the standards (although this is not always the case). For example, like this: 4-4-4-6. Each digit in this marking indicates the time in milliseconds that elapses before the start of the command. nine0003

Four types of latency go in a row here:

CL — delay between giving a command to the RAM and starting writing or reading;
Trcd – time required to activate a bank row, that is, the time between the selection of a column and a row;
Trp – row activity time, i. e. its opening and giving a command to overwrite;
Tras — the time required to fully charge the capacitor with a specific address.

The higher the clock frequency of the RAM, the more timings it will have. nine0003

Therefore, modern memory may have values of cl 11, cl15, cl 16 and even cl19. For example, for a DDR3 memory bar with a clock frequency of 1333 MHz, CL 9 is considered the optimal value.

DDR4 with a frequency of 2800 MHz has average latency values of 14-15. By the way, in this case we are talking about the so-called CAS latency, that is, the delay between sending the data column address to RAM and the start of data transfer — the time required to read the first bit.

Read more about latency values in RAM and which ones are better to read soon on the blog. nine0003

Latency setting

It is this number (CL) that is most interesting in practical terms, so often only it is indicated in the marking.

In order to find out more detailed information about the delay intervals of the RAM, you have to look for its full specification on the manufacturer’s website. Latency information is recorded in the RAM bar itself in the SPD chip, which is in any RAM module. As a rule, when assembling a computer, you do not need to make additional settings in the BIOS: the latency provided by the manufacturer is set automatically. nine0003

All the user needs to do is to properly mount the RAM modules in the correct slots.

However, if necessary, the timings can still be configured, for which the BIOS provides a number of options for the user. For this, the DRAM Timings function is used, with which you can set the values of the four main latency values.

When set to AUTO, the default settings will be used — those on which the bar works with optimal performance. nine0003

You may need to set the timings yourself when overclocking the memory module: since there is an inverse relationship, lower latency leads to faster RAM.

In addition, this may be necessary when trying to make friends with a pair of slightly different RAM sticks in order to make them work in dual-channel mode. On the other hand, increasing latency slightly slows down the memory, but makes it more stable.

Such «dancing with a tambourine» is recommended for users who know exactly what they are doing and why. nine0003

If you are not yet familiar with computer hardware and will not be well versed in it soon, I recommend setting the default latency. I also advise you to read the publications “What affects the frequency of RAM” and “What is it — support for ECC RAM”.

I remind you that by sharing the articles of this blog on social networks, you contribute to its promotion, which will allow me to publish even more useful instructions. In order not to miss the news, subscribe to the email notifications. And that’s all for today. See you all tomorrow! nine0003

Best regards, Andrey Andreev.

RAM Latency — How Speed and Latency Matter

While this is perhaps one of the easiest components to install, understanding what makes your RAM work is another game altogether.

These innocent-looking memory sticks are more complicated than they look. Let’s try to simplify one aspect today — memory latency .

We’ll take a look at what this means, how it can impact your workloads, and help you decide whether you should be chasing those lightning speeds or looking for low latency memory kits. nine0003

RAM speed and latency

While memory speed (or data transfer rate) determines how fast your memory controller can access or write data to memory, RAM latency focuses on how soon it can start a process .

Memory speed is measured in MT/s (mega transfers per second), and latency is measured in nanoseconds.

Before we delve into calculating memory latency or RAM latency, there are a few terms you should know:

Memory latency

Simply put, latency is latency.

This delay can be measured in nanoseconds (real time) . However, when it comes to digital electronics, we often use the clock cycle because that way we get comparative numbers that are independent of frequency or data rate.

Memory timings

Unlike latency, memory timings (that sequence of numbers you see on your memory module) are measured in clock cycles. nine0003

Thus, each number in memory timings, such as 16-19-19-39, indicates the number of ticks or cycles required to complete a particular task.

Here’s a quick overview of what these timings mean, from first to last (everything is measured in clock cycles).

To simplify things, think of memory space as a giant spreadsheet with rows and columns, where each cell can contain binary data (0 or 1).

CAS Latency (tCL) — The first memory clock is called the Column Access Strobe (CAS) latency. While the term «strobe» is a bit obsolete today as it’s a holdover from the days of asynchronous DRAM, the term CAS is still used in the industry. The CAS Latency of RAM specifies the number of cycles it takes to get a response from memory after the memory controller sends the column it needs to access (think of the spreadsheet analogy I mentioned above). Unlike all other timings below, tCL is an exact number, not a maximum/minimum. nine0024
Row Address Delay Column Address (tRCD) — The second number is the minimum number of ticks it takes to open a row (again, in this giant spreadsheet) and access the required column. Remember that unlike tCL, tRCD is not an exact number, but the maximum delay.
Row Precharge Time (tRP) — The third number in this 4-digit sequence indicates the minimum clock cycle delay to access another row in the same selected column. nine0024
Row active time (tRAS) — The last number in this memory time sequence indicates the minimum number of clock cycles that the row must remain open for data access. This is usually the biggest delay.

Calculate RAM latency or CAS latency

CAS Latency is the time it takes for your memory to respond to a request from the memory controller. Here is a table that will make this process easier for you:

In this section, we will learn how to calculate RAM latency.

Of course, you can also use the RAM latency calculator if you want to skip the math.

However, since advertised CAS latency is measured in clock cycles, we need to account for memory speed to get real CAS latency in nanoseconds.

Here is the formula to use: (CL×2000)/Frequency

Let’s say we have a DDR4-3200 CL16 memory kit. You get latency in nanoseconds (16×2000)/3200 = 10 ns .

First Word Latency

Now if you’re wondering, what about other memory timings? Don’t they affect latency?

Yes, it is.

However, CAS latency is still the most commonly used metric to compare memory latency because it more directly affects (immediately) how fast your RAM module responds to a request.

However, some engineers talk about the benefits of using First Word delay when it comes to memory. This latency takes into account the timing of the primary memory, along with the burst length, to get a latency that essentially tells you how long it takes to read a word from memory. nine0003

Low latency RAM or high speed RAM

Which is faster?

Consider the following three sets:

32 GB (2×16) DDR4-3200 CL16
32 GB (2×16) DDR4-3600 CL18
32 GB (2×16) DDR4-4000 CL20

Which one do you think is «the fastest»?

Well, this is where things get a little more complicated (or more interesting).

For 3D and active workloads

We tested it ourselves. And it turns out that the task itself is so severely limited by IPC and the clock speed of one processor core that it doesn’t matter what latency you choose as long as the data transfer rate or memory timings don’t actively hinder processor performance.

Ryzen 9 5950X	Benchmark (average over 10 runs)
DDR4-3600 CL 20	1230.75
DDR4-3600 CL 16	1237.50
DDR4-2666 CL 20	1230.88

As you can see, with Ryzen processors, both lower timings and lower data rates affect the performance of the processor itself (even single-core).

However, this opens up an interesting discussion.

Should you go for the cheaper DDR4-2666 CL20 kit (or DDR4-2666 CL16 kit) instead of the slightly more expensive DDR4-3600 CL20 kit if you’re going to get similar performance? nine0003

I would say no. You’ll see better performance with faster memory gains for any task that puts a heavy load directly on the CPU, spread across multiple cores.

One example of this is fast scrolling through a Premiere Pro timeline with compressed source material.

For CPU/GPU rendering workloads

The short answer is not a big deal. If these are your primary workloads, we recommend that you prioritize system stability rather than pursuing minor gains. nine0003

Frequently asked questions about memory latency

Is mixing RAM with different latency a bad idea?

Yes . Mixed memory is generally not ideal. This is true even if you have different memory timings.

As far as what behavior to expect, there are two possibilities:

If you’re lucky, the memory modules can be configured to run at slower memory timings.
if you’re not lucky, you may not see stability until you revert memory overclocking to JEDEC specs, which are usually much slower. nine0024

How can I check the CAS latency of my RAM?

You can find out the current memory latency in no time! First download CPU-Z from CPUID official website.