Command Rate

[SelenaGomez]

I have G.SKill Ripjaws 2x4gb ddr3 1600mhz 8-8-8-24 running at Command rate 2T. What is this command rate? Is 1T better? OR should it be at 2T?

 

[corkyg]

The Command Per Clock (or Command Rate) setting determines the delay (in clocks) between the chip being selected and the controller beginning to send commands. 1T supposedly gives better performance, while 2T supposedly yields increased stability with older/cheaper RAM

Google it and you'll find many user opinions and examples.

https://www.google.com/search?sourc...T4GGLG_enUS329&q=What+is+memory+command+rate?

 

[DigDog]

command rate is a multiplier, essetially. you want the *number* to be low, lower = better.
1T is better, yes.

You cannot just run your ram at 1T however - it's kind of like overclocking. Most rams will run at 1T happily but you will need to "relax" (increase) the timings, so for example, 8-8-8-24 2T might need to become 10-10-14-38 1T.

overclocking ram is an art form i have never perfected, tbh. something liek this could help you:
http://www.ocfreaks.com/ram-overclocking-guide-tutorial/

 

[BonzaiDuck]

command rate is a multiplier, essetially. you want the *number* to be low, lower = better.
1T is better, yes.

You cannot just run your ram at 1T however - it's kind of like overclocking. Most rams will run at 1T happily but you will need to "relax" (increase) the timings, so for example, 8-8-8-24 2T might need to become 10-10-14-38 1T.

overclocking ram is an art form i have never perfected, tbh. something liek this could help you:
http://www.ocfreaks.com/ram-overclocking-guide-tutorial/

I became acquainted with 1T possibilities having read somewhere that you'd get a nice little boost in performance. It shows up in the latency result of benchmarks -- a few nano-seconds -- more or less.

With the G.SKILLs, the DDR3's I've had could run at 1T out of the box, and some motherboards like mine (and probably like SelenaGomez's) might actually default the CR in timings to 1 after choosing an XMP profile that specifies it as 2.

G.SKILL techs had informed me that this or that memory kit would run at 1T without bumping up the vDIMM voltage; other indications suggested it might take a little extra voltage to the IMC or in the vCCIO or vTT setting, and the G.SKILL tech would always say "you won't have to increase it by very much."

Of course, this was more certain with a 2x kit that left two DIMM slots vacant. With the RipJaws DDR3-1600 "-GBRL" kits, we could fill the slots with 4x4 and still run the modules at spec timings and spec vDIMM. But if your testing for stability doesn't respond to increases of a couple bumps at a time - I think the increment was 0.0625V -- if you were approaching 1.15V in the VCCIO -- it wouldn't work for you and there was nothing else you could do to get it. You could try increasing vDIMM itself, but I've seen where that didn't do any good either.

So if you used just a 2x4 or 2x8 kit of good G.SKILLs at the spec latency timings and vDIMM, you could likely pass a marathon memory-test with CR=1. If you got stop codes like 9C or 124, you might increase vCCIO by 30+ millivolts and that would get you there. But you should be able to run a G.SKILL Ripjaws 2x kit with vCCIO around 1.075V and CR =1.