Skip navigation
This discussion is locked

RAM compatible with 3930K and ASUS P9X79 PRO + more

Jun 4, 2012 10:26 AM

Tags: #cs6 #3930k #first_build #p9x79 #32gb
  Latest reply: Jeff Bellune, Aug 20, 2012 1:21 PM
Replies 1 2 Previous Next
  • Currently Being Moderated
    Jun 12, 2012 9:50 AM   in reply to Colin Grigson

    There were several ways we were able to determine stability.

     

    1. restart the system several times and see if it posted correctly and without kicking on the watch dog.

    2. Memtest host programs over an extended period of time

    3. Load windows cleanly without corruption errors.

    4. Run burn in with applications such as Intel CPU Diag tool and Sisoft Sandra for an extended period of time without BSOD's.

    5. Run Premiere or Cubase with test projects including encoding without errors for a long period of time without errors.

     

    One of the major benefits the Cuda based applications such as Premiere CS5/6 and Pro Audio applications with low latency Asio drivers is the intensive load on the ram. The ram is constantly buffered to and updated besides the usual read write operations. This often finds issues with latency timings on the ram because if the read, write, erase, and refresh timings are off, then you are going to have corruption and application crashes. With audio the data corruption comes across as distorted audio from the audio engine. You can also duplicate some of this by Overclocking the CPU some to see if the stability remains with certain applications because the CPU handles the memory management. We were able to detect errors consistently and without failure with these methods especially the latency timing compatibility issues. The biggest challenge the new Memory controller CPU architecture brought to the table is syncing the host controller to the modules across a bus effected by heat, signal attenuation, and Voltage regulation. This requires the controller and the board to re-detect memory timings every time the board boots up from a non-powered or powered state along with maintaining voltage levels with a very fine margin of fluctuation. The Board manufacturers have to adjust that detection based on modules in the market and considering the scope of modules, that is no small challenge. Now add to that Quad channel and changes/version differences in the host controllers ie Memory controllers and you have the current problems you see today in the memory compatibility.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 12, 2012 1:28 PM   in reply to ECBowen

    Eric - wow!  Thanks!  You certainly know your stuff.  Great you can share this knowledge with us.

     

    I've downloaded and prepared a boot cd from http://www.memtest.org and will use those other tools as well.

     

    Unfortunately I'm probably in for a world of frustration as most of my parts are already ordered and will be with me in a couple of days.  If I weren't downunder, I would certainly be using you to build my machine.

     

    I have coming a ASUS P9X79 PRO with an INTEL 3960x processor

     

    The ram is 32GB of G.SKILL Ripjaws Z Series 1866 (PC3 14900)

    http://www.newegg.com/Product/Product.aspx?Item=N82E16820231499

     

    According to all the good advice above, it's not going to be fun getting them working.

     

    Combination was recommended by VideoGuys in their DIY9 build.  Will let everyone know how the testing goes...

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 13, 2012 4:08 PM   in reply to Colin Grigson

    The 3960X has far better memory compatibility than the 3930K. I always suspected the 2 CPU's had different memory controllers like many of the I7 Gen 1 CPU's. However when I asked the CPU Engineer's at the ISS conference they stated that was not the case with the new SB-E CPU's. Now I dont know if they were just not aware of what was happening in the East or if that is really the case. Either way in all of the ram testing I did here and other system configurators I talk to have done, the 3960X was able to run most of the Ram modules at 1600 that did not work with the 3930K. This was also the Case with the Xeon CPU's as well. Unless these current G-Skill modules have issues with timings due to new IC's I have not tested then they should be ok. With the Ram parts bidding market, you never know what is going to be bought by the manufacturers for production. You definitely have a far better chance with the 3960X than you do the 3930K of getting those to run at 1600.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 20, 2012 7:37 AM   in reply to ECBowen

    Hi folks!

     

    First of all, thank you for all your input here, been a great help! Here`s a little update. Today the I got 32gb of the http://www.crucial.com/store/partspecs.aspx?imodule=CT2KIT51264BD160B kit delivered, which I discussed earlier in this thread with you. I just installed 2x4gb modules to test it and it seems to be working fine. The mobo I use is a gigabyte ga x79 ud5, with of course the 3930k.

    When I look in the bios the cpu runs at 3500mhz, the memory at1600mhz. However, the memory seems to be running at 1.5v, at least the bios is saying so. Should I adjust this to 1.35v, or is it ok to leave it at the automatic setting of 1.5v?

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 20, 2012 12:05 PM   in reply to Tamas

    That is the Base Profile for those modules that I spoke of before. If you load the XMP Profile then the voltage and timings will change.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 4:41 AM   in reply to ECBowen

    Hi Eric,

     

    When I go in the bios of the x79 ud5, the enable xmp profile option is greyed out and set to disabled. I have the feeling that these modules do'nt support xmp at all. Do I have to set some things manually now? And if so, which settings do you recommend?

    I'm also thinking to overclock the 3930k to 4,37ghz. In that case, do I have to use different settings for the ram as well?

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 6:05 AM   in reply to Harm Millaard

    Hi Harm,

    for your RAM information and the further RAM discussions/recommenadtions.

    I run ASUS P9X79 WS with 64 GB RAM, i3930K

    G.Skill RipJawsZ DIMM Kit 64GB PC3-12800U CL10-10-10-30 (DDR3-1600) (F3-12800CL10Q2-64GBZL)

    EUR 445 at www.geizhals.at

    no problems, After Effects CS6 with "extra Global Perf. Cache SSD" works great.

     

    regards

    Klaus klfi

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 7:24 AM   in reply to MatthewsM7

    I can report that the G-Skill Ripjaws Z DDR3 PC3-17000 9-11-10-28 2133MHz in a 8x4GB setup at 1.5v 1600MHz is working with the following build:

     

    Gigabyte X79-UD5 mobo

    Intel 3930K i7 @ 4.2GHz OC

    Rosewill Thor V2-White Edition (love this case)

    EVGA GeForce GTX 580 3GB

    2x Seagate Barracuda 7200 rpm 1TB in RAID 0 (onboard)

    2x Seagate Barracuda 7200 rpm 2TB (one for exports, one for storage)

    1x Samsung 830 SSD 128GB for OS & Progs

    Pioneer Internal BD writer (BDR-207DBK)

    Corsair AX-1200 watt Gold PSU

    Noctua NH-D14

    Windows 7 Pro SP1

    And CS6 of course.

     

    I did have issues when I tried to run the ram in XMP mode, which I guess it should since the 3930K doesn't work with clock speeds over 1600. So I left it alone and this machine screams. I do think I need to tweak some things as the PPBM5.5 came in at #47, and TrevorD with a very similar build is up at #6. He does have a 10K rpm drive for projects, a 4.4 OC which I think is making a huge difference. He also is running a GTX 570 instead of my 580. I wonder if that makes a difference? Either way, when I have time I am going to try to get into the top 10.

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 7:58 AM   in reply to ChrisLeeDallas

    Chris,

     

    TrevorD is running CS5 which runs the PPBM5 test faster than CS5.5 / CS6 on the same hardware (Harm and Bill say it is something having to do with how Adobe changed the caching in the different versions). He also has a 5xRAID 0 that will do better at matching the powerful cpu you both are running.

     

    You will also need to do lots of system tweaking (mostly Windows, not CS itself) to get CS performing at its best.

     

    Your build is a nice one - enjoy it!

     

    Regards,

     

    Jim

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 8:18 AM   in reply to JEShort01

    Jim, Thanks for the reply. I didn't realize he was running CS5 or the 5 HDD Raid. That will definitely make a difference.

     

    I have tweaked Windows quite a bit. I didn't mean tweak CS, but Windows and hardware to get the most out of CS.

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 10:51 AM   in reply to Tamas

    Tamas, If the base Profile is 1600 then I would not change anything since they did not include an XMP profile on these modules. Run with the base since it is working.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 22, 2012 11:08 AM   in reply to ChrisLeeDallas

    When looking at our PPBM5 test results the first thing that should be done is filter it to look at the correct CS version as each are different as far as scoring goes

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 26, 2012 4:36 AM   in reply to Bill Gehrke

    Hi All.

    Can anybody tell me why, in this thread, there is a big down on Macron?

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 26, 2012 6:16 AM   in reply to iammykyl

    If you mean Micron, it is because they were not delivering low voltage chips as required by the LGA 2011 CPU's.  Maybe by now they have "seen the light".

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 26, 2012 11:03 AM   in reply to iammykyl

    Compatibility issues are the reason Micron is such a problem lately and PSC is also a problem. Keep in mind most of this has only revolved around the Socket 2011 platform

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 26, 2012 4:58 PM   in reply to Bill Gehrke

    Thanks guys for your response.

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 27, 2012 3:11 PM   in reply to ChrisLeeDallas

    Chris, we have pretty similar builds, mine as set up for the PPBM5 benchmark is:

     

    Asus P9X79WS

    i7-3930K

    Noctua NH-D14 cooler

    2 kits of 4x4 F3-12800CL9Q-16GBXL GSkill Ripjaws (DDR3-1600, CL9-9-9-24)

    GTX 570 HD (2.5GB GDDR5)

    120GB SSD for my OS

    3 x 1Tb Caviar Blacks for everything,including Pagefile

    A very sparse Windows7 Pro installation

    All of Harm's Windows tweaks - no anti-virus, disconnected from internet after downloads

    Premiere Pro (6.0.1) and Adobe Media Encoder  (6.0.1.31) from Adobe's cloud

     

    I ran a few PPBM5 runs with 16Gb of RAM and then put in the second set - started up without a hitch even tho the new kit was way newer and even has a different physical setup. I was even able to squeak the new sticks in underneath the Noctua cooler by taking out one of the fans temporarily.

     

    The PPBM scores went from 44, 119, 64, 46/5 to 44, 75, 52, 4 without any overclocking.

     

    I've been trying out various overclocking modes - nothing I can do will let me operate with a 48 multiplier, yet I am quite stable at a 4.7 OC with the RAM set to 1600 and with my CPU Vcore Voltage in Offset mode set at +0.200

     

    That gets my scores to 37,71,44,60/4 which I am very happy with! And I am enormously grateful to Harm, Bill and all of those who take the time to answer questions in this Forum - and to those who ask the questions that many of the rest of us have as well.

     

    My guess is that you will have to bump up your overclocking to get your times down any further on the benchmark.

     

    My next step is to quit playing with the machine and put it to work on converting my HD DSLR footage (Canon 5DMarkII) to short videos (5min or less).

    This means I will have to come to grips with deciding how to set up my operating disk system for actual use vs. benchmark testing. I'm posting that issue on a new thread.

     

    Dorothy

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 27, 2012 4:09 PM   in reply to DeePeeBee

    what are the - Harm's Windows tweaks - is there a link where i could read about?

     

    and why no anti-virus? is it important? i could not live with out one!

     
    |
    Mark as:
  • Currently Being Moderated
    Jun 27, 2012 5:45 PM   in reply to anat_usa

    Anat - this is a machine that I have just finished building and I intend to use it ONLY for my photography hobby using Premiere Pro and Photoshop, so after I installed Windows 7 and downloaded my other programs - and UPDATED them - I disconnected the machine from the internet. I will only reconnect it for updates, during those I will activate the Windows firewall but do not plan to have any anti-virus at all. I have more than enough other computers, tablets, etc. for surfing the web, e-mail, etc. and ALL of them have some kind of protection, such as anti-virus. My goal with this new machine is to make it as fast as possible, and I know that anti-virus, spam filters, etc. are constantly inspecting what is happening in the system - which has to be slowing things down, so I am trying to work without them.  I have also stopped Sidebar, indexing, etc as is described in "the Guide for installing and tuning a Vista 64 pc". If you click the header above at discussions, you will get a box for searching, just type in "guide for installing and tuning" and you will get to the article written by Harm Millaard which is excellent and which still applies to Windows 7. You can also search for other discussions of how to tweak Windows, especially note those comments by Harm Millaard, Bill Gehrke and others listed as "top participants". There is an enormous wealth of information to be found in their postings and they provided inestimable help to my build.

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 3, 2012 4:05 PM   in reply to DeePeeBee

    Out of interest Dorothy, what case and PSU are you using?

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 4, 2012 10:30 AM   in reply to CaillinAU

    I'm  not at home to check for sure, but I think the PSU is an OCZ, Gold and I'm  sure it's 1000watts. The case is a Silverstone and I can't remember the model number but I will post it next week.

    Dorothy

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 9, 2012 9:52 AM   in reply to DeePeeBee

    So I am planning out a new system and it will have a 3930K and the P9X79 WS. I want to have 64GB of ram so I was looking at the G.Skill Ripjaws Z Series. I was wondering if there is a huge diffrence between buying the 8 x 8GB kit of buying two of the 4 x 8GB kit, because they are normally about the same price but the 4 x 8 is currently on sale on newegg for $30 less each so it would be $60 cheeper to get the two kits. Is that worth it or do you still think it's better to buy the 8 x 8 kit.

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 9, 2012 10:01 AM   in reply to Zanzaben

    Difficult question to answer, but a very valid one. If the XMP profiles are the same (hard to tell from the packaging, though) and given the fact that the two sets are installed in different banks, also electrically, you will likely have no problems with two sets of four, but is it worth the risk and potential hassle if you run into problems? I can't answer it for myself at the moment, let alone advise you. Maybe Eric can bow in with his reactions.

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 9, 2012 10:32 AM   in reply to Harm Millaard

    On newegg the 4 x 8 kit says Intel XMP support for 1st, 2nd and 3rd Generation Intel Core Processors. For the 8 x 8 it says Intel XMP 1.3 ready. I don't know if this helps.

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 9, 2012 11:16 AM   in reply to Zanzaben

    You normally want to avoid using multiple kits incase the ram is from different builds/revisions. If you feel the benefit out ways the possible problems and test the ram well then it may be worth the savings. Keep in mind though that the likely chance of problems grow in this scenario.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 10, 2012 8:06 AM   in reply to ChrisLeeDallas

    How important is it to buy an Nvidia card with 4 Gb of video memory ?


    I read a comment on one of the threads, that you may run out of VRAM if you have too many things happening on the time line in PP,

    which is a bit concerning to me.

     

    Just how much stuff can we load up on the time line without running out of memory ( with 2 Gb )  ?

     

    Do the majority of you, who bought any of the new video cards ( 670, 680 ) have any issues with "just" 2 Gb of VRAM ?

     

    At the moment, cards with 4 Gb of VRAM are hard to find, and I'm guessing that most of you have no problems with cards

    with 2 GB of VRAM.

     

    Any thoughts ?

     

    Dave.

     
    |
    Mark as:
  • Currently Being Moderated
    Jul 10, 2012 9:03 AM   in reply to David Zeno

    Unfortunately the amount of Vram used is a Fluid data set. This is decided by allot of factors including effects used, codecs of material, frame resolution and frame rate. Last I checked which has been a while ago Premiere was not really using up to the 1.5GB of ram on allot of the 500 series cards. I would have to check again with CS6. Others can check by downloading and running GPU-Z. Keep in mind though this info is going to be workflow and project specific. The 4GB cards though give far more headroom for future improvements and changes to Adobe. I always recommend the 4GB cards for future proofing if you have the budget.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 12:25 AM   in reply to ECBowen

    Since 1.35v RAM is being recommended here and I can't seem to find the Samsung option in a kit with the memory size I'm looking for, what do you all think about the Mushkin blackline http://www.newegg.com/Product/Product.aspx?Item=N82E16820226329?

     

    I don't plan to overclock. I just want things to work well. I don't care about the fancy heat spreaders. I'd prefer not having them so I can use a better cpu cooler.

     

    I'm planning an Asus PX79 LE with a 3930k processor. The link above is for 32GB set but maybe I can live with 4x4 (16GB). I haven't picked the video card yet. I'm considering the Asus GTX 560 or 570.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 6:14 AM   in reply to jetboy10

    jetboy, I am using two sets of GSkill ripjaws 4x4 GB 1600 ram sticks for a total of 32 GB. The two sets were purchased separately over a year apart and they are running just fine. My box is currently 4th on the CS6 ranking.  I do however overclock. Going from 16 to 32 GB made a considerable difference in PPBM5 speed (before overclocking, I haven't run 16 vs 32 since I overclocked).  Newegg has a 32GB set on sale for $159 if you use code EMCNBNC43, G.SKILL Ripjaws Z Series 32GB (4 x 8GB) DDR3 1600 (PC3 12800) Sorry, but the code expires TODAY.

     

    Note that the latency and timing are different from the Mushkin set.

     

    Good luck, Dorothy

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 9:27 AM   in reply to DeePeeBee

    Thanks Dorothy, (sorry to you and everyone else for the jetboy name - I try to be anonymous on the web if that's even possible). The g.skill ripjaws seem to be recommended a lot around the web and it's good to know they work when you have all 8 slots filled. that's a good sign. The only downside to me is the fancy heat spreaders (I know. The mushkins I linked to has them too).

     

    I wonder why several people in this thread seems focused on LV RAM. Is this really better for consistent stability. Besides individual experiences, are there any links to where this was tested or where Intel discusses the 1.35V benefits?

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 3:44 PM   in reply to jetboy10

    We tested it extensively when the platform first released and only went to 1.35V ram when the 1.65 and 1.5V was showing data issues(heat and voltage levels) or compatibility. Some of the manuals with the boards use to have a warning recommendation for LV for Full population on the X79. That has since changed in the manuals. Intel is not going to release information or a position that will alienate half of the ram partners. They Officially don't support any ram above DDR3 1600 on that platform. Some 1.5V ram is working now fully populated and has been linked in the Adobe forum. Look for those if you have concerns besides the known good 1.35.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 13, 2012 8:18 PM   in reply to ECBowen

    Thanks Eric. I decided to go with the samsung 1.35V 4GB memory sticks (sold in pairs). I can get the 32GB if I populate all 8 slots or maybe I only need 16GB. I currently use 12GB now with an X58 board so 16GB of RAM (plus an P9X79 LE board and a 3930k processor) should be a noticable improvement. I'm also going to get an ASUS GTX-570 GPU card for this build. 

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 19, 2012 10:34 PM   in reply to MatthewsM7

    I wish the question of RAM for a X79 - i73930K - LGA 2011 rig were simple, and that there were more 8 stick/64GB kits available in stores.

     

    By what I I've read in these forums, the X79's sound difficult to get running at first, because of the RAM.

     

    The more I read, the less confident I am about building a LGA 2011 platform PC. Especially considering it will be my first ever build.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 20, 2012 6:13 AM   in reply to jetboy10

    jetboy10 wrote:

     

    Thanks Dorothy, (sorry to you and everyone else for the jetboy name - I try to be anonymous on the web if that's even possible). The g.skill ripjaws seem to be recommended a lot around the web and it's good to know they work when you have all 8 slots filled. that's a good sign. The only downside to me is the fancy heat spreaders (I know. The mushkins I linked to has them too).

     

    I wonder why several people in this thread seems focused on LV RAM. Is this really better for consistent stability. Besides individual experiences, are there any links to where this was tested or where Intel discusses the 1.35V benefits?

    It runs colder.  That's about it, lol.  I've had nothing but reported problems with low voltage ram since some boards say they support it but then can't seem to detect it properly and send in the wrong voltage.  You often times have to set the voltage and the speed and timing manually, which isn't the end of the world but it's annoying.

     

    I would stay far away from low voltage memory even if you're packing every single slot.  Basically if you're not using a top notch board and power supply, the voltage could drop too low on the circuit that powers the RAM for a tiny amount of time and your RAM won't work properly.  Low voltage memory is extremely sensitive to that because they already had to jump through hoops to get it to run at such a low voltage.  Basically they're more sensitive to slight voltage dips.  If you think heat is a problem if you've got 8 sticks in a row, it's not.  They're like 2-3W each or something and most come with excessive aluminum coolers attached anyway.

     

    I've never had a problem with any RAM at standard voltage in any system ever but I also only buy from 2 vendors, both that give lifetime warranties and are known for being the absolute best.

     

    Also, someone back a ways in this thread said just the speed, not the timings are important and you should just ignore the timings.  WHAT?!?!?!?!  The speed is the number of cycles per second the RAM runs at and the timings are how long they take to do an operation.  The lower the timings, the faster the memory.  You should always get the lower timings that seem cost effective.  It will get much, much higher WEI ratings and benchmark numbers if it has lower timings.

     

    People have had problems with nonstandard timing memory though not booting or not being recognized in their BIOS.  Well, that's simple.  Regardless, turn on XMP profiles on the motherboard or yeah, it won't work so well.  Otherwise:

    1. buy really good, high rated, no reported compatibility problems RAM like for example GSkill.  Remember, 7 out of the 10 top rated RAM sets available for sale right now are from GSkill.

    2. before you buy, look up the motherboard's QVL (qualified vendors list) which contains a list of many specific memory modules that were thoroughly tested by the manufacturer and certified to work with your board.  That eliminated 100% of the problem.  If your perfectly matching RAM doesn't work, it's you doing something wrong in that case.  If your motherboard doesn't have QVL, you're buying the wrong brand

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 20, 2012 6:31 AM   in reply to VHC-CO-IT

    1) QVL seriously?  please do tell me how the motherboard manufacturers test? and with how many sticks.. cant wait for this answer...

     

    2) so is 1.5v /1.6v good for all 2011? are the memory controllers the same on all 2011 processors?

     

    3) using the default XMP just works huh? so do you think that ram is running @ 1600? sigh....

     

    4) do you even know what chips are bieing used by gskill?

     

    EL Plates.. make sure you buy Samsung original 1.35V ONLY.. this will ensure compatibilty.. socket 1155 is not near as problematic as 2011..

     

    Scott

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 20, 2012 7:45 AM   in reply to Scott Chichelli

    I'm not sure I follow much of that.  They run an extensive custom RAM test for hours plus their own special back end diagnostics before certifying that memory works.  Every manufacturer does it differently but respectable ones do it that way.  Asus for example not only ran an extensive test but re-ran it 2 months later to add additional RAM that had been released since then.  Check it out:

     

    http://www.asus.com/Motherboards/Intel_Socket_1155/P8Z77M_PRO/#MSL

     

    Not really sure what you mean by the XMP thing either.  Since JEDEC SPD standards only officially go up to 1333 9-9-9-24, if your memory is 1600, then you better have XMP turned on.  And if you do turn it on, it will be 1600.  In fact, look in the RAM timing section and verify it.  Or run CPU-Z from Windows and verify it.  It really, sincerely is running at 1600 if it's rated for 1600.

     

    And what chips are inside?  I don't care.  That's an awfully roundabout way to go about things. I'm pretty sure they're made by GSkill themselves but I don't go with theoretical brand problems.  I know Hynix and Elpida and several other actual chip makers suck so I avoid them for example but it's a lot faster to just see if the other thousand people that bought the RAM you're looking at had any problems with it.  GSkill ram has lifetime warranties, a perfect track record, and well, what's the difference between these customer ratings:

    http://www.newegg.com/Product/Product.aspx?Item=N82E16820231277

     

    and these customer ratings

    http://www.newegg.com/Product/Product.aspx?Item=N82E16820178238

     

    Ooooh, good job, PNY lol

     

    If people have any DOA sticks, any problem configuring them, any system stability, or any problems at all, they do not hesitate to go give it a low rating on newegg.  So since GSkill DDR3 is the highest rated memory on the entire site and almost every individual speed category, there's the one to buy.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 20, 2012 7:55 AM   in reply to VHC-CO-IT

    “””I'm not sure I follow much of that”””

     

     

     

    Exactly my point..

     

     

     

    What about the memory controllers on socket 2011?  Still waiting on an answer to that…

     

     

     

    “”””Not really sure what you mean by the XMP thing either. “””

     

    Exactly my point..

     

     

     

    Eric?

     

     

     

    Care to educate the boy?

     

     

     

     

     

    Ahh newegg again… you are way toooo funny..

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 20, 2012 9:34 AM   in reply to VHC-CO-IT

    VHC-CO-IT wrote:

     

    jetboy10 wrote:

     

    Thanks Dorothy, (sorry to you and everyone else for the jetboy name - I try to be anonymous on the web if that's even possible). The g.skill ripjaws seem to be recommended a lot around the web and it's good to know they work when you have all 8 slots filled. that's a good sign. The only downside to me is the fancy heat spreaders (I know. The mushkins I linked to has them too).

     

    I wonder why several people in this thread seems focused on LV RAM. Is this really better for consistent stability. Besides individual experiences, are there any links to where this was tested or where Intel discusses the 1.35V benefits?

    It runs colder.  That's about it, lol.  I've had nothing but reported problems with low voltage ram since some boards say they support it but then can't seem to detect it properly and send in the wrong voltage.  You often times have to set the voltage and the speed and timing manually, which isn't the end of the world but it's annoying.

     

    I would stay far away from low voltage memory even if you're packing every single slot.  Basically if you're not using a top notch board and power supply, the voltage could drop too low on the circuit that powers the RAM for a tiny amount of time and your RAM won't work properly.  Low voltage memory is extremely sensitive to that because they already had to jump through hoops to get it to run at such a low voltage.  Basically they're more sensitive to slight voltage dips.  If you think heat is a problem if you've got 8 sticks in a row, it's not.  They're like 2-3W each or something and most come with excessive aluminum coolers attached anyway.

     

    I've never had a problem with any RAM at standard voltage in any system ever but I also only buy from 2 vendors, both that give lifetime warranties and are known for being the absolute best.

     

    Also, someone back a ways in this thread said just the speed, not the timings are important and you should just ignore the timings.  WHAT?!?!?!?!  The speed is the number of cycles per second the RAM runs at and the timings are how long they take to do an operation.  The lower the timings, the faster the memory.  You should always get the lower timings that seem cost effective.  It will get much, much higher WEI ratings and benchmark numbers if it has lower timings.

     

    People have had problems with nonstandard timing memory though not booting or not being recognized in their BIOS.  Well, that's simple.  Regardless, turn on XMP profiles on the motherboard or yeah, it won't work so well.  Otherwise:

    1. buy really good, high rated, no reported compatibility problems RAM like for example GSkill.  Remember, 7 out of the 10 top rated RAM sets available for sale right now are from GSkill.

    2. before you buy, look up the motherboard's QVL (qualified vendors list) which contains a list of many specific memory modules that were thoroughly tested by the manufacturer and certified to work with your board.  That eliminated 100% of the problem.  If your perfectly matching RAM doesn't work, it's you doing something wrong in that case.  If your motherboard doesn't have QVL, you're buying the wrong brand

     

    There so many incorrect points or assumptions here it's hard to find a place to begin.

     

    1. The 1.35V ram was spec'ed and designed for the Quad Channel Socket 2011 platform. This was not spec'ed for the 1155 platform which you are referencing issues. The issue has nothing to do with the ram and everything to do with the XMP Profiles/voltage settings the board supports with the bios. If the board does not have charts for the low voltage ram, then you cannot turn on the XMP profile for the low voltage memory. That does not mean however you cannot run the ram on those 1155 boards at all. You just have to run the ram at the base 1.5V profile that almost all of the Low Voltage ram comes with. Those detect fine. Many of the ram Manufacturers are using DDR3 1600 profiles as their base profile such as Kingston, Samsung, and others.

     

    2. The comments on Low voltage having issues with voltage fluctuation are completely backwards. The Low Voltage ram was designed to resolve this issue with signal attenuation and voltage level issues on the bus between the memory controller in the CPU and Ram in the slots. The manufacturers use higher grade chips such as those used for DDR3 2000+ to make the low voltage chips. These are the highest grade chips in the market which also give the best stability with varying voltage levels. This information however is not common knowledge outside the memory manufacturing engineers. You have to talk to one of the engineers to know. I talk to them in Taiwan all the time. Do you?

     

    3.  Some of the Memory manufacturers are using a Hybrid model when manufacturing low voltage. They are using higher grade chips that are not as high as the DDR3 2000+ and then will raise latency timings some to ensure stability. This is a completely acceptable production method since Latency timings have little to no impact on ram/system performance in the DDR3 with the Intel Platforms. The current Dual channel performance on the Intel platforms is around 20GB/s and the Quad channel is 35GB/s. This bandwidth is so high and above what the current system requires, that the marginal difference in CL latency has little to no impact anymore. Latency was a concern in the DDR1 and DDR2 days when the ram bandwidth was under 5GB/s or 2.5GB/s. However all the testing done that I have seen by us or others shows little to no impact on high performance applications such as videoediting applications. The information you posted here is simply wrong and unfounded today.

     

    4. Heat was most definitely an issue with the X79 platform and 1.65V ram. We tested this extensively when the platform released with many different methods. The method that displayed this issue the quickest was Pro Audio testing. The data would corrupt in the ram when playing a heavy Pro Audio plugin test that is very memory intensive. The result was complete audio distortion out of the audio interface. This showed up in all our testing with the 1.65V ram at 1600 until we cooled the memory. At that point all of the distortion went away and the test ran perfectly. This did not show up with 1.5V or 1.35V ram however. So your statement on memory and heat once again is completely inaccurate and shows a complete lack of any real world testing. Once again this was the X79/socket 2011 platform only. This had nothing to do with the 1155 platform. It also makes perfect sense if you have an understanding of electronic principles. Heat lowers resistance which lowers voltage. This is why you get signal attenuation.

     

    5. The last comments on Non standard timings, XMP, and QVL show a text book answer without any real knowledge or experience with the problems people are running into with ram. It also shows a lack in real understanding of the ram production and component bidding market. The QVL lists were a large part of the problem. The motherboard manufacturers vary in testing method. However most test with just populating 1 or half the banks which is not completely qualifying the ram. There are many sub timing and compatibility issues that dont show up when testing with 1 stick or half the banks populated. You find this out often to late when calling the motherboard manufacturer for support and they ask you how many sticks are you populating. Then their support will say remove half or down to 1 and test again. If the problem goes away, they say contact the memory manufacturer  since the issue is theirs. However when you contact the ram manufacturer they say the issue is the Motherboard manufacturer since the board wont accept the ram fully populated. They didn't test it fully populated either. You have no idea how many times I fought with different memory manufacturers I was working with to test the ram with the boards we were going to use fully populated. They have to adjust sub timings to handle that. Once again if you talked to ram engineers you would know this. Now we finally have a manufacturer who does all their testing fully populated and our issues have evaporated since then.

     

    The other issue with QVL's have to do with how often they are updated compared to how often the components in those same model ram change. The Ram bidding market is completely volatile. This means Manufacturer A may use PSC chips in 1 production run and Hynix in the next. Well the motherboard manufacturer originally QVl'ed with the PSC versions. The bios was updated for timing detection for those models based on that. However the Hynix modules require different sub timings and therefore the compatibility.detection charts will no longer work for those sticks even though their still on the QVL. I see this all of the time and contrary to your belief, most often with Gskill. Your predictions on eliminating 100% of the issues with memory timings by enabling the XMP profiles and following the QVL is once again completely false. I often do paid support and the ram manufacturer I see the most is Gskill. That is also the ram I most often have to set the timings manually because the XMP profile does not work at all and is far to aggressive. Once I set the timings manually, many times they will work. atleast at 1333 if needed. You may want to actually go through the motherboard manufacturer forums and actually see all of the XMP ram problems that are showing up. That will give a far more accurate view of the current QVL and ram issues.

     

    Posting information that is neither accurate or has zero real world testing behind it helps no one. Instead that is setting up someone who is coming to the forums for help to experience serious problems that can impact their business, lives, and reputations. This is nothing to take lightly since they are relying on people who post to share information they just dont have and really dont have time to learn. They also dont have the time outside of the profession that supports their Families or lives to learn a second. They simply hope those who have the knowledge will tell them the best way to accomplish this. I have already had several paid support clients who have been disastrously misled by miss information in this forum which has cost them thousands in paid repair costs replacing hardware, rebuilding and lost production time not to mention the impact on their business reputation. This is not something anyone should play around with and shows a complete lack of responsibility or professionalism. Those Forum readers who are following this thread, also please take heed of this. There is allot of miss information currently going on in this forum. Be very careful what you believe right now. Please verify with other sources before you act.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 20, 2012 1:21 PM   in reply to ECBowen

    [Thread locked]

     
    |
    Mark as:
1 2 Previous Next
Actions

More Like This

  • Retrieving data ...

Bookmarked By (1)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points