ctk85 Posted December 27, 2015 Share Posted December 27, 2015 My Build: Intel i7 4790k, Gigabyte GA-Z97X-UD5H (rev 1.1) + Asus Z97-WS (in the process of migrating from the ud5h to the asus ws), Corsair RM750, MSI GTX970 4G, Corsair H110i GTX, 1 SSD Sandisk & 3x2TB HDD WD Black, MOTU PCIE-424, * RAM described below, * notes - Onboard audio disabled, IGFX disabled, CPU running stock. I bought the Corsair CMD8GX3M2A2400C10 Dominator Platinum 8GB (2x4GB) DDR3 2400 Mhz CL10 a year ago from amazon, put them in XMP profile 1 and never had a problem. 6 months later I bought the exact same pair from amazon to have a total of 4x4 (16gb), but my computer would never post and bios would report an overclocking failure. The highest setting I could use was 1600mhz XMP to even post. Next I tried putting only the new pair in and sure enough they ran at 2400mhz fine. This is the bit where it got weird - I tried putting one chip form the first pair and one from the new pair in and it couldn’t go higher than 1600mhz! Immediately I thought I had been sold the wrong product and decided to re-inspect the ram chips. I noticed the pairs had a different version number (ver4.13 & ver4.29) despite me simply clicked “buy again” on amazon assuming it would be an identical part. Obviously I didn’t check the label on the ram thoroughly before installing :( Corsair CMD8GX3M2A2400C10 Dominator Platinum 8GB (2x4GB) DDR3 2400 Mhz CL10 (ver4.13) Corsair CMD8GX3M2A2400C10 Dominator Platinum 8GB (2x4GB) DDR3 2400 Mhz CL10 (ver4.29) Next I ran memtest with all 4 chips in at 1600mhz XMP for 6 passes and got 0 errors. Is it worth running it o each chip individually at 2400? Finally I got a new motherboard (Asus Z97-WS) and I get the exact same behaviour. I’m past the time to return either pair to amazon so i’ll have to either sell a pair and try to rebuy the same ver # so they all have the same model and ver #, or contact corsair to see if I can RMA (or whatever the process is). Hoping I can get this sorted and stay with corsair. Any ideas what I should do next? Thanks! CPU-Z pics http://i.imgur.com/VLW6ElD.jpg Link to comment Share on other sites More sharing options...
ctk85 Posted December 28, 2015 Author Share Posted December 28, 2015 UPDATE I ran through memtest86 again but for every chip individually in slot 1. I loaded optimised defaults in bios then set XMP profile 1 on and checked the ram was running at 2400mhz 1.65V. CHIP 1 (ver4.13) - 86 errors. CHIP 2 (ver4.13) - 0 errors. CHIP 3 (ver4.29) - 10,000+ errors (I just stopped after the first pass on this one as it was finding errors constantly). CHIP 4 (ver4.29) - 0 errors. Glad that i've managed to make some progress diagnosing this but shame the two chips with errors are from different pairs. Chip 3 will have to be RMA'd I assume, but what about chip 1? Are 86 errors something I should be worried about? Link to comment Share on other sites More sharing options...
ctk85 Posted December 28, 2015 Author Share Posted December 28, 2015 System posts and is stable with all but chip 3 installed at 2400mhz :) progress! Link to comment Share on other sites More sharing options...
ctk85 Posted December 28, 2015 Author Share Posted December 28, 2015 nobody know if it's worth RMA'ing a RAM chip with over 80 errors on a 2 pass memtest? Link to comment Share on other sites More sharing options...
emissary42 Posted December 28, 2015 Share Posted December 28, 2015 If everything is working correctly you should not get any errors in memtest at all. There is one exception in the hammer subtest, since it is basically testing for a specific vulnerability to an exploit instead (errors in that subtest don't relate to problems with normal use). For mixing those two kits, make sure you use the secondary timings from the ver4.29 Kit. For example it might require a higher tRFC for stability than your Ver4.13 kit, because it is using higher density ICs. Link to comment Share on other sites More sharing options...
ctk85 Posted December 28, 2015 Author Share Posted December 28, 2015 If everything is working correctly you should not get any errors in memtest at all. There is one exception in the hammer subtest, since it is basically testing for a specific vulnerability to an exploit instead (errors in that subtest don't relate to problems with normal use). For mixing those two kits, make sure you use the secondary timings from the ver4.29 Kit. For example it might require a higher tRFC for stability than your Ver4.13 kit, because it is using higher density ICs. Thank you for the info. It was giving 5-10 errors per test (about 40 per pass). I'm not familiar with memtest86 so wasn't sure how many errors were deemed necessary for an RMA, since only the really bad chip won't post above 1600. Also thanks for the timing info, I'll post back when I get the units RMA'd. Enjoying a much more stable system with only the two good chips from different pairs. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.