F3-14900CL9D-8GBSR timings for 1333 and 1600MHz

[alexan_e]

Hi

I bought a 2x4GB set F3-14900CL9D-8GBSR (1866MHz DDR3) , I intend to use them with Asus CrossHair IV Formula + AMD Phenom II X6 Black 1090T.

The timings for 1866MHz are 9-10-9-28-2N, my question is what are the timings for this memory used @1333MHz and @1600MHz?

I would expect the timings to be less for the lower frequencies but the SPD I found using google (mine hasn't arrived yet) show worse timings:



The reason for asking is that although buying a 1866 memory I intend to use it at a lower frequency and take advantage of lower timings.

Thank you
Alex

[Tradesman]

1600 8-9-8-27 1333 7-8-7-26 is a good starting point - can prob get them tighter, but I normally 1 step on each timing for each freq range you drop

[alexan_e]

Thank you for your reply.

The RAM voltage spec is 1.5v, I know from other threads that increasing the voltage can help them overclock in the frequency domain (I guess up to 1.65v it is safe), does that apply to tighter timings too or they are irrelevant of voltages?

Is the timing calculation based on strict math?
I mean the CAS latency of 9 clocks in 1866MHz is 9 * (1/1866Mhz)=4.82ns
so should I expect the same memory to work with 4.82ns/(1/1333MHz)= 6.42 clocks CAS latency @1333MHz?
Of course the clock result has to be rounded to an integer so it should be 7 clocks but if voltage helps maybe it can go down to 6 clocks.

One more thing, it's about the command rate which is 2T for 1866Mhz , can this be lowered to 1T for 1333Mhz or it stays 2T for any frequency? (I mean should I try 1T or it is a lost cause)

Best regards
Alex

[Tradesman]

At lower freqs can prob drop the voltage if you want to 1.45 or so and experiment from there, you can often undervolt some when underclocking. If OCing 1.65 would be safe though I try and stay to no more than + 0.10 over when possible. Timings are laid out as the optimal for the DRAM and math can play in - in fact the norm used to be add the first 4 bbase timings and that would be the fourth - i.e. 5-5-5-15, 6-6-6-18,etc. As I mentioned if under-clocking or OCing I normally go up +1 or -1 per freq step as a starting point and fine tune from there.

[alexan_e]

Sorry for asking again but I'm not clear on the RAM voltage effect on timings, does overvoltage help to get better timings or it only help to get higher frequencies?

Can you please comment on the command rate, it is 2T for 1866MHz , the SPD has no info for 1333MHz so I'm not sure if 1T can be used or if it remains constant at 2T.

Thank you for your help
Alex

[Tradesman]

Both on the voltage question, it can help OCing to higher freqs and on tightening timings (basically it help the sticks pull higher or tighter than the rated spec.

Running 1T depends on a lot of things - some sticks are a little stronger than others and are capable of running the 1T at spec while others fall into the 'norm' of running at 2T - same with running at lower freqs, it's often a combo of the timings and voltage, often you can move up to 1T with higher voltage or if you stay at spec voltage by raising the CL, say from 8 to 9 or 9 to 10, this mehod however can be counter productive as raising the CL can counter the effects of tightening the CR to 1T i.e. you might get to 1T but the higher CL degrades performance from spec CL and 2T - it's a matter of experimentation with voltage and timings and running BenchMarks at each setting for best throughput

[alexan_e]

I have currently set the memory to 7-8-7-24 1T @1447Mhz and seems to be working fine.
I have increased the voltage to 1.55v in case it helps.

I have run multiple passes of Memtest test #5 and #8 (I read they are good to reveal faults ) and I have run three passes of all the Memetest sequence (took about 2 1/2 hours I think).
I have also run a few simultaneous super Pi 32M tests.

I would appreciate your insight of what is considered a good RAM stability test.
What do you recommend?

Thank you
Alex

[Tradesman]

Looking good, for both DRAM and your overall system Prime95 is a good test