When CAS Latency (latency timing) for a 200 MHz system bus (FSB) is set to CL 5.0, the column access time (tCAC) takes 50 nanoseconds and the clock cycle time (tCLK) 10 nanoseconds (this follows the well-known formula tCLK * CAS Latency >= tCAC). Other items are explained in the following table:
| RAS | Row Access Strobe |
| CAS | Column Access Strobe |
| tRCD | The time interval between RAS and CAS access |
| tRP | The time interval to switch between memory banks |
| tAC | The time interval to ready output transmission |
| tCAC | The column access time |

Latency information from CPU-Z on a PC running Windows XP.
A typical memory configuration looks something like this: DDR2-533 with CL 4.0 or DDR-667 with CL 5.0. Only rarely can you find DDR2 memory with nominal latency timings of CL 3.0, and when you do, it's pretty expensive because short reaction times in connection with high clock rates are difficult to achieve.

A complete set of parameters is stored in this SPD chip; the motherboard can read these directly from the memory module itself.

The CPU-Z program also reads SPD values directly from your memory modules.

Increasing input voltage improves signal stability.
Another concern surrounds input voltage for DDR2 memory modules. The standard level is 1.80 volts, with a tolerance of ± 0.05 volts. Some manufacturers permit their memory modules to be operated at input voltages of up to 2.2 volts. You can set input voltage levels this high for your RAM without risk, because DDR2 demands only modest power intake and thus heats up only slightly during operation. Even in our most extreme experiments we sustained no damage to our RAM modules resulting from over-voltage.
- Optimal DRAM For Overclocking
- DDR2 Parts Details
- Two DRAM Sides Are Better Than One
- BIOS Settings To Boost Performance
- CL4 Or CL3: Boosts Performance By Up To 5%
- More Speed By Tightening Latency Timings
- Boosting Memory Clock Speed: 23% Performance Gains
- Here's How We Tested At THG's Munich Labs
- Wstream And Everest Benchmarks
- Overclocking Tests And Motherboards Used
- Default Settings Test: DDR2-667
- Default Settings Test: DDR2-667, Continued
- Overclocking By Tightening Timings
- Overclocking By Tightening Timings, Continued
- Overclocking To The Max
- Overclocking To The Max, Continued
- Only Asus Boards Convince
- An Overview Of The 16 Test Candidates
- Aeneon DDR2-533: Unbeatable Price/Performance
- Buffalo's DDR2-667 Is Unspectacular, But Solid
- Corsair DDR2-1000: Absolutely The Fastest
- Crucial/Micron DDR2-667 (Ballistix) Is Geared For Overclocking
- GeIL's DDR2-533 Fails To Impress
- G.Skill's DDR2-675 Confusing Label
- Kingmax DDR2-667: Choose The Chip's Colour
- Kingston DDR2-900 Offers Minimal Overclocking
- Mushkin DDR2-667's Double-Sided Module With A Need For Speed
- OCZ DDR2-800: The Third-Fastest Test Candidate
- Patriot Memory DDR2-1000 Offers Top Performance Reads At 8.6 GB/sec
- PQI's DDR2-667 Is Fast, With No Overclocking Headroom
- Samsung DDR2-667: Conservative RAM Sans Overclocking
- Twinmos' DDR2-667 Is Slow, But Overclocks Well
- Wintec's (AMP) DDR2-667: 475 MHz When Overclocked
- Conclusion