ram timings

[MRSMIFFY]

can someone explain to me about ram timings i did a google search but that was way to complicated for me? i have 8gb of crucial ballstix 1333mhz i think but i have it at 1866mhz would i benefit from changing ram timings? 

[AlwaysFSX]

The timings are the latency for the ram to send or receive information from the cpu, even if they're lower than what they're already at you won't notice much of a difference.

[UnknownEngineer]

I think you are referring to CAS latency, the number usually states the time (in milliseconds) it take for the RAM to read data, it could also mean write but i cannot remember. To put it short it is the time it takes for your ram to receive and/or send the data. I modern applications and usage scenarios it doesn't really make a whole lot of a difference.

Also, when overclocking RAM you usually have to increase the CAS latency, i am not absolutely sure why but it is what i remember. I would recommend not messing with the latency's, there is no detectable difference for reducing or increasing them. You will most likely just cause instability. 

[Wats]

can someone explain to me about ram timings i did a google search but that was way to complicated for me? i have 8gb of crucial ballstix 1333mhz i think but i have it at 1866mhz would i benefit from changing ram timings? 

uh, wow. no way that can be stable if all you did was adjust the frequency... have you at least stress tested it?

 

Yes, changing timings can actually make your ram faster. on my 1600cl9 kit of corsair, running with cl10 actually improves my latency by several nano seconds (milliseconds delay on ram would be crazy, i think Finland above me got them mixed up). But you cannot just take someone else's word or work, you have to try it all out for your self.

[UnknownEngineer]

uh, wow. no way that can be stable if all you did was adjust the frequency... have you at least stress tested it?

 

Yes, changing timings can actually make your ram faster. on my 1600cl9 kit of corsair, running with cl10 actually improves my latency by several nano seconds (milliseconds delay on ram would be crazy, i think Finland above me got them mixed up). But you cannot just take someone else's word or work, you have to try it all out for your self.

Could have been nano, IDR. but regardless the faster timings don't make for more performance, at least i haven't heard of anyone noticing a difference with a PC using CL 9 Vs CL 11.

[Wats]

Could have been nano, IDR. but regardless the faster timings don't make for more performance, at least i haven't heard of anyone noticing a difference with a PC using CL 9 Vs CL 11.

the only way you will notice a difference, is if you run benchmarks that give numbers.. in fact i have some benchmark screenies... i think ima put some thumbnails here for OP. EDIT: WHATS THE POINT OF THUMBNAILS IF THEY'RE THE SIZE OF THE ORIGINAL?!

 

4.0 -- 1.024

4.1 -- 1.040

4.2 -- 1.072

4.3 -- 1.112

4.4 -- 1.144

4.5 -- 1.200

4.6 -- 1.232

4.7 -- 1.280

4.8 -- 1.344

4.9 -- 1.400

5.0 -- 1.440

5.1 -- 1.504

 

(tested with some Pi benchmark)

1600 9-9-9-24 1.50XMP -- 21.772s

1866 9-10-9-24 1.54 -- 20.980s

2000 13-14-13-31 1.65 -- 21.096s

2000 12-14-13-31 1.65 -- 20.956

2k 12-13-12-31 --------- 20.886

 2k 12-12-12-31--------- 21.080?

2GHz 11-13-12-31------- 20.912

2      11-13-11-31-------- 20.858

2      11-12-11-31-------- 20.191

11-13-12-31

9.64

1.2310

！介ッシツツじシツシツツッジシッ

▬▬▬▬▬▬▬▬▬▬ஜ۩۞۩ஜ▬▬▬▬▬▬▬▬▬▬

ＤＡＭＮ ＴＨＩＳ ＣＯＭＭＥＮＴ ＩＳ ＦＡＮＣY

▬▬▬▬▬▬▬▬▬▬ஜ۩۞۩ஜ▬▬▬▬▬▬▬▬▬▬

 

Good luck, i've got to go. Yes, I know I have way too many things running in the background while benching...

 

1.4V and peaked at 70c lol... on an ivy bridge

[MRSMIFFY]

so my ram shouldnt be stable after just changing the frequency ? i stressed it using p95 never been able to get aida64 to work 

[Wats]

so my ram shouldnt be stable after just changing the frequency ? i stressed it using p95 never been able to get aida64 to work 

P95 is a last resort, it does work though. Get HCI Memtest, run a couple windows of that at 250MB. let run for a few thousand % coverage. much faster to use a dedicated memory test.