[ This information was compiled from the FAQ at www.crucial.com on November 29th 2000.]

The bottom line...

...for the average consumer when buying RAM is as follows.  (These are intended to be guidelines; write to me if you have a question.)

  • Get DIMM RAM if you've got an Intel Celeron, Pentium II or newer (Fast Page and EDO SIMMs are for older systems).
  • Get 3.3 volt (all DIMM RAM that I've installed when building and upgrading PCs has been 3.3 volts).
  • Get unbufferred (not registered, not bufferred).
  • Get pc100 or pc133 (100 or 133 MHz) or faster, and you can mix speeds in most cases.
  • Don't worry about the nanosecond speed (for DIMMs, commonly 10ns, 8ns, 7.5ns)... that is accounted for in the MHz rating.
  • Don't worry about CAS Latency (CL2 vs CL3), and you can mix latencies.
  • 16x64 vs 16x72... this has to do with whether the RAM is parity/ECC or non-parity.  Either is OK, and except for mission-critical systems, parity/ECC is not necessary.  But generally you cannot mix the two types.

Read on for more detailed descriptions of all of these specifications.

What is the difference between buffered and registered memory?

DDR and SDRAM can be unbuffered or registered. EDO and FPM can be buffered or unbuffered. Buffered modules contain a buffer to help the chipset cope with the large electrical load required when the system has a lot of memory. Registered modules are unbuffered modules that contain a register that delays all information transferred to the module by one clock cycle. Buffered and registered modules are typically used only in servers and other mission-critical systems where it is extremely important that the data is properly handled.

What does "registered" SDRAM mean?

The term "registered" refers to how the memory module processes signals. Registered modules contain a register that delays all information transferred to the module by one clock cycle. This type of memory is primarily used in servers and was designed for modules with 32 or more chips on them to help ensure that data is properly handled.

While most PCs will only accept unbuffered SDRAM, there are some that accept registered SDRAM. Keep in mind that when you install registered SDRAM, all of the modules installed in your PC must be registered because unbuffered and registered modules are not interchangeable.

What is the performance difference between CL2 and CL3?

CL2 parts process data a little quicker than CL3 parts in that you have to wait one less clock cycle for the initial data. However, after the first piece of data is processed, the rest of the data is processed at equal speeds. Latency only affects the initial burst of data. Once data starts flowing, there is no effect. Bear in mind, a clock cycle for a PC100 module is 10 nanoseconds so you probably won't notice a significant performance difference. Most systems will accept either latency part. However, there are some systems that require either CL2 or CL3 parts. These requirements are built into our Memory Selector.

What is CAS Latency?

CAS Latency (also referred to as latency) is the amount of time it takes for your memory to respond to a command. Specifically, it is the length of time between memory receiving a command to read data, and the first piece of data being output from memory. Latency is measured in terms of clock cycles and is often noted as CL2 (two clock cycles) or CL3 (three clock cycles).

Can I mix CAS2 and CAS3 parts? If so, what settings should I use in my motherboard's BIOS for the CL timing?

CAS2 and CAS3 parts can be mixed. BIOS settings should reflect the CAS3 setting, as the CAS2 module will run at the CAS3 setting when it is installed with a CAS3 module.

What is the difference between ECC and non-parity memory? Is there a performance difference?

ECC is indicated by a 36 or 72 in our part number. Non-parity is designated with a 32 or 64. If you already have a PC and are unsure which type you have, count the number of small, black, IC chips mounted on one of your existing DIMMs. If the number of chips is evenly divisible by three, then you need ECC. If the number of chips is NOT evenly divisible by three, you have non-parity memory.

If you are building a PC and deciding which type to use, the following guidelines should help. If you plan to use your system as a server or a similar mission critical type machine, it is to your advantage to use ECC. If you plan to use your PC for regular home, office, or gaming applications, you are better off with non-parity.

ECC (Error Checking and Correcting) performs "double bit detection and single bit correction". This means that if you have a single bit memory error, the chipset and memory will find and repair the error on the fly without you knowing that it happened. If you have a double bit memory error, it will detect and report it. Using ECC decreases your PC's performance by about 2%. Current technology DRAM is very stable and memory errors are rare, so unless you have a need for ECC, you are better served with non-parity SDRAM.

How do I know if I need parity, non-parity or ECC memory?

The general rule of thumb in deciding what type of memory you need is to look at what's already installed in your system. To find out if you have ECC, parity or non-parity memory, count the number of chips on the module. Divide the total number of chips by three. If you can evenly divide the number of chips by three, the module is ECC or parity, if not, then it is a non-parity module.

So what if your system does have ECC or parity memory (the chips are evenly divisible by three), how do you know which one you have? One way is to look at the part numbers on the chips of your module. If each chip has the same part number, you have ECC. If one chip is different, you have parity. An even easier way is use the Crucial Memory Selector. Just select your system make and model and the Memory Selector will tell you which type of memory works in your system.

If you are building a PC and deciding which type to use, the following guidelines should help. If you plan to use your system as a server or a similar mission critical type machine, it is to your advantage to use ECC. If you plan to use your PC for regular home, office, or gaming applications, you are better off with non-parity.

Using ECC decreases your PC's performance by about 2%. Current technology DRAM is very stable and memory errors are rare, so unless you have a need for ECC, you are better served with non-parity SDRAM.

Can I use memory with ECC on a board that does not support ECC?

Most motherboards that do not have an ECC function within the BIOS are still able to use a module with ECC, but the module will run in non-ECC mode. Keep in mind, there are some cases where the motherboard will not accept an ECC module, depending on the BIOS programming. The only sure-fire way to test this is to place the module in the motherboard and see if the BIOS will recognize the memory addition.

What do 16x64, 16x72, etc. represent when describing memory, and how do they affect what I should buy?

You may see the size of a memory module written in the format 16Meg x 64, 16Meg x 72, etc. In this format, the second number gives the width of the data path in bits. If this number is divisible by nine, it is a parity or ECC module; otherwise, it is a non-parity module.

You can determine the size of your module from this format by multiplying the two numbers together, then dividing by eight or nine, whichever results in an even number. This gives you the size of your module in megabytes (MB).

Can I use a 128MB PC100 SDRAM DIMM with a 128MB PC133 SDRAM DIMM?

Generally you can mix PC100 and PC133 memory in the same system. The faster memory will run at the slower memory's speed (in this case PC100). However, there are some systems that will not correctly run this memory configuration. It is rare, but we suggest you contact your computer manufacturer or motherboard manual for specifics.

How do I tell the difference between 2 clock and 4 clock SDRAM?

Physically looking at a memory module, there really is not a distinguishing mark to determine if the module is 2 clock SDRAM or 4 clock SDRAM. The only way to distinguish the two parts is through the part number markings or data sheets supplied by the manufacturer.

My motherboard supports SPD and non-SPD memory. What's the difference and which is better?

SPD (serial presence detect) is a small non-volatile RAM chip attached to SDRAM modules that contains information about the memory. This information includes the number of row addresses, number of column addresses, error detection/correction, refresh rates, data width, and the interface standard. It also contains less important information such as the module serial number and manufacturer code. When your computer powers up, it sets the row and column settings and the timings for the module based on the information in the SPD.

SPD is required in SDRAM that is 66MHz, PC100 and PC133 compliant. Standards set by Intel and JEDEC ensure that data is entered in appropriate locations so the motherboard BIOS can understand what this data means. The SPD standard allows greater flexibility for incorporating identification of new features and technologies on memory modules.