Jump to content
HWBOT Community Forums

redux

Members
 View Profile  See their activity

  • Posts

    153

  • Joined

  • Last visited

  • Days Won

    4

 Content Type 

Posts posted by redux

    Asrock Z97 OC Formula-PSC

    in Memory Heaven (air/extreme)

    Posted

    5 hours ago, K404 said:

    On the subject of PSC.... how much variation is there in Haswell IMC? None of my PSC does anything good- tested on MSI, Gigabyte and Asus, but I wonder if im a bit rusty or being too soft with vSA and vI/O

     

    :)

    (First note I have been binning IMCs with 8M, not 32M.)

    I have had multiple CPUs that were not able to do 2666 8-12-8 tight, though this was only in quick profile testing and maybe some of those could do it with some tweaking. I use IMC volts +0.3/+0.25/+0.25 and very lazy +0.05 to each if the chip fails. It depends on the sticks and how many channels as well, I have had multiple chips fail 2600c8 with dual channel sticks that do 2666 all day, and then do 2750 8-12-8 in single on a better stick.

    For harders numbers, in my most recent round of IMC testing (2666 8-12-8 tight and 2750 8-12-8 tight) I had 23% fail 2666, 77% pass 2666 (~5% of passing chips needed raised IMC), and 41% pass 2750 (~25% of passing chips needed raised IMC). This was across 39 G3258/4670K/4770K.

    IMC also matters for VDIMM. I did some mem binning on a crap 4670K that struggled with 2666 8-12-8 last year. When I rested with a very good IMC (now dead :() some of the same sticks ran settings at 50mV lower volts.

    And of course there's more. Some IMCs hate tight tRCD or tRP, some don't like extreme VDIMM, etc.

     

    5 hours ago, Noxinite said:

    Honestly I've only had one CPU ever that seemed unable to do 2666C8, but it was pretty early on with me OCing Haswell.

    Most CPUs will work with typical volts (+0.35/+0.15/+0.2) or "Bullant" (+0.08/+0.06/+0.180).

    Or, like my 4770K, they need stupid IOD just to get 2666C8 to post (+0.4) but then will do 2770C8 fine on air. With a little cold on SS, it works fine like any other CPU though. Samsung OC was also totally unaffected compared to other CPUs.

    Binning IMC for Samsung (Rev.D) is much much easier. Probably 90% of my chips I tested (all G3258?) passed 2800c9 no problem so I decided it wasn't worth doing.

    • Thanks 1

    POLL: CPUZ Feelz, let your opinion be known.

    in Benchmark software

    Posted · Edited by redux
    by core count

      

    49 minutes ago, buildzoid said:

    I agree that if you want a all core valid ranking it should just be separate from the existing CPU-z valids.

    This seems like the simplest solution. I don't care how many cores people use to validate but I don't like the idea of restricting the option of disabling cores because some people don't like it. It doesn't really make sense when CPU-Z isn't ranked by core count and modern hardware and overclocking is moving away from fixed all-core clocks anyway.

    Asrock Z97 OC Formula-PSC

    in Memory Heaven (air/extreme)

    Posted · Edited by redux

    After testing about a dozen quad cores I have found an L310B 4770K with an IMC good enough for my purposes and I've been doing a lot of PSC binning on this board lately. Tested only ~a dozen sticks on this setup so far but some nice results :D

    I'm not sure if it's because of this IMC or what, but I have had crazy scaling on most sticks, 2V+ easy on all so far and with the right settings even scaling 2.1V+, 2.2V+, and at least booting even 2.3V on some sticks (a mediocre 2000C9 Trident booted 2.33V!), with normal ambients and auto VTTDDR. Helps me not to feel bad about average sticks by MHz/volt terms with scaling like this ? :

    memec6.thumb.jpg.5b4eff6ae88faa46ed338a57f71cda23.jpg

    Maybe even 2666C7 tCWL6 or 2666C6 could be done with cold ambients and stupid volts?

     

    tRDRDs can go a little tighter here (not sure about 4-5-4), but just a bit of fun:

    meme3.thumb.jpg.c1dea055f6e37aefc5f6f1ee01991718.jpg

     

    For point of reference this stick does 1200 7-11-7-28 tCWL8 1.7V, 1300 8-12-7-28 tCWL6 1.84V, 1333 8-12-8-28 tCWL6 1.9V, 1350 8-12-8-28 tCWL6 2.01V, 1400 8-13-8-28 tCWL6 2.08V all tight subs. Again, far from golden lovo. +0.3/+0.25/+0.25 used for all testing except the 2950 screen where a bump to +0.28 IOD helped boot consistency. An even better IMC I suspect will improve tRCD12 and tRCD13 ceilings for this kind of testing, 1375 8-12-8 is at least bootable yet 8-13-8 stops at only 1414 32M...

    • Like 2
    • Thanks 1
×
×
  • Create New...