Super User is a question and answer site for computer enthusiasts and power users. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

if an 8gb DDR3 SDRAM 1600MHz has a timing of 8-8-8-24, how can I calculate if it will have a better performance than a 2x4gb DDR3 SDRAM 1866MHz with a timing of 9-10-9-28?

Also, will a 2x4gb DDR3 SDRAM 1600MHz with a timing of 8-8-8-24 be better than both of the previously mentioned?

note: both of the 2x4gb have dual channel.

share|improve this question
up vote 8 down vote accepted

Before we get into calculating anything, I'd recommend that you double check that your motherboard and CPU support overclocking to use DDR3 1866, and if so, go with the 2x4GB DDR3 1866MHZ pack. You want two chips because the actual per-chip bandwidth of DDR modules is actually half the advertised rating, since they are made to be installed in pairs on dual BUSes or "channels", so always make sure you have an even number of chips of the same frequency installed if you want to get full bandwidth. The DDR3@1866 has the higher transfer rate (14928MB/s vs 12800MB/s), and as you will see, the chips both have pretty much equal advantage on each other in terms of timings.

Timings are given in clock ticks, so comparing the timings of a pair of chips of different frequency is an apples-to-oranges comparison, and requires some conversion. People can mean a lot of things when they say "performance" but I'm going to assume you mean "fast" and convert to seconds per rating. In general, as the frequency goes up, the time for a tick interval goes down (which is why faster RAM always seem to have higher timing values).

You can calculate the time interval between clock ticks with

1/(1/2 * frequency * 10e6) = T seconds

Since timings are in "ticks", you just multiply the rating by the interval in seconds.

For your RAM, we end up with (these figures are rounded to the hundredth):

800 8-8-8-24       1.25e-9 seconds per clock tick
CL = (1.25e-9 * 8)   =  1.0e-8 seconds
RtC = (1.25e-9 * 8)  =  1.0e-8 seconds
RtP = (1.25e-9 * 8)  =  1.0e-8 seconds
tRAS =(1.25e-9 * 24) =  3.0e-8 seconds

933 9-10-9-28      1.07e-9 seconds per clock tick
CL = (1.07e-9 * 9)   =  9.65e-9 seconds
RtC = (1.07e-9 * 10) =  1.07e-8 seconds
RtP = (1.07e-9 * 9)  =  9.65e-9 seconds
tRAS = (1.07e-9 * 28)=  3.0e-8  seconds (2.996e-8 actually)

So, by comparing the time, we can see that the 1866MHz chip:

  • Has a 0.00000000035 second advantage in CAS Latency and RAS-to-Precharge, (9@933 is faster than 8@800)

  • Is 0.0000000007 seconds slower than the 800MHz chip in RAS-to-CAS (8@800 is faster than 10@933)

  • Is 4e-10 faster at tRAS (this variance is seriously negligible but worth noting).

Wikipedia recommends more simply calculating the CL time in nanoseconds (ns) with

(CL/(freq-in-MHZ) * 1000) = timing in ns

rather than calculating all the values, which puts the 1866MHZ chip on top across the board. Remember, frequency is half the chip's label frequency.

share|improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .