Announcement

Collapse
No announcement yet.

4GBRL Timing Question

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • 4GBRL Timing Question

    I recently upgraded to a windows 7 64-bit computer with a 4gb kit of g-skill 4GBRL ram in it. I have tweaked everything in my system for efficiency, last of which was this ram. The stock timings are 9-9-9-24-40 2T @ 1.50 volts..

    I have been able to tighten them to 8-9-8-24-40 1T completely stable @ 1.561 volts, 12 hours+ prime95 64 bit on 4 cores and 7+ passes in memtest86+.

    Going to 8-8-8-24-33 1T is a different story, I have not got this stable even @ 1.63 volts and I refuse to go any higher because I value the lifespan of my ram more than any marginal gains it would bring. Actual tests at this speed showed performance degradation in some areas actually, not to mention being unstable.

    My question here is what I should be running the 'row cycle time' any lower @ 8-9-8-24 1T? Is there an optimal setting for 'row cycle time' at this timing, and furthermore is it not optimal to run the first three settings @ 8-9-8 instead of having them even across the board? (8-8-8 or 9-9-9) I have seen in AIDA64 tests that the memory read, write & copy times benefit from these timings, but these are synthetic benchmarks to the best of my knowledge, and I would like to know if the 8-9-8 timings would work best with a certain "row cycle time"?

    Is ram best run with the TRP, TRCD & TCL (8-9-8 currently) @ equivalent numbers (8-8-8, or 9-9-9) and what should the 'row cycle time' be set to with 8-9-8-24-1T timings, optimally?


    I am a novice overclocker with very favorable results recently.


    System Specs:

    AMD Phenom x2 555 BE 3.2ghz unlocked to x4 3.6GHZ (200x 18) 50 degrees load air cooled in a centurion 5 II case w/ 5 fans (cpu, rear, top, HDD, side panel) w/stock CPU voltage.

    4gb G-Skill 2x2 4GBRL Ram stock 9-9-9-24-40-2T @ 1.50 volt-- running @ 8-9-8-24-40-1T 1.56 volt, ddr3-1600 (800mhz)

    ASRock 870 Extreme3 mobo, NB 2400mhz (x12), ht 2000mhz

    Thank you g-skill for making ram that I trust in any system! I know the voltage I have the ram running at is only slighly above stock 1.5 voltage, but I would like for it to last a while, and be reasonably tightened to it's potential-- that's why I'm asking. I was very happy to see it run flawlessly with 1T settings and would just like to match the row cycle time to this efficiency level.

    Thank You!
    Last edited by MordecaiWalfish; 02-20-2011, 10:08 AM. Reason: clarity

  • #2
    You can try lowering down to 33, that is typically the lowest feasible. Although, the performance increase will not be dramatic, if any.

    If you raise DRAM Voltage just a little more, 1.60V-1.65V, you should be able to even those timings and achieve 8-8-8-24.
    The over voltage will be perfectly fine.

    Thank you
    GSKILL TECH

    Comment


    • #3
      Thank you for the reply, I was more so wondering if it was beneficial for the ram to be run at even timings, I will try going to 1.65 volts with the settings you described but if the performance gains after testing are not enough to warrant jumping from 1.56 volts I would rather run at that voltage with the settings I currently have..

      I have heard that certain scenarios can actually make performance worse in some aspects with overclocking ram and wanted to make sure one of those scenarios is not whether all the timings are even or not..

      Thank You!

      Comment


      • #4
        SS

        DRAM normally has a sweet spot, where maximum performance is attained and can really only be found by testing, normally the increase is barely noticible except with benchmark programs, so it's up to the individual the time they want to put into it

        Comment


        • #5
          Found it. ( 4GBRL ASRock 870 Extreme3 )

          For me the "sweet spot" with my rig is actually the settings I had in the original post. I tested with AIDA64 each time a change was made to the timings, then did stability testing.

          I tested the ram @ 8-8-8-24-40-1T over-volted to both 1.645 & 1.659v settings in my bios (there was no *exact* 1.65 volt setting, these were the closest). Benchmarking @ both voltages via AIDA64 reported back *almost exact* stats as the ones reported @ 8-9-8-24-40-1T w/ 1.561 volts. Stability testing failed at both voltages with the 8-8-8-24 timings as well, with prime64 reporting errors within 5 minutes of the tests starting @ 1.645v & completely locking windows up during the test @ 1.659v, No blue screen or anything, just completely froze.

          I found that the 8-9-8-24-40-1T @ 1.561v timings I had originally, tested in AIDA64 pretty much exactly as the more tightened timings (only real difference being half a nanosecond faster in latency, big deal). Fully stable tested with 10+ passes in memtest, 3 passes in windows 7's "memory diagnostic tool" (on the "extended" setting) and 12 hours+ in prime95 64-bit version.


          The motherboard I'm using is a bit on the lower end so perhaps this is why it could not maintain stability with the tightened settings @ the higher voltages, but either way, I'm happy with the performance gained with the slight overvolt (1.561) I'm using now.

          For reference of anyone who's potentially using the same motherboard I am with this ram, the model is "ASRock 870 Extreme3" (bios revision 1.60).

          =)

          Comment


          • #6
            Well done indeed! Now you can enjoy the computer knowing it is at max performance. =)

            Thank you
            GSKILL TECH

            Comment

            Working...
            X