Corsair CM3X1024-1600C7DHX v3.1 & v4.1

BlowMeOneKenobi · 12 Nov 2008 at 18:44

I bought another pair(2x 1gb) DDR3 modules from OC a few months after the first pair.
I believe it was after installing the second pair that I started to get intermitent hard crashes(not sure this is the right terminology, no BSOD, just lockup). Sometimes this reset my bios settings which had to be changed back.
The only difference in the 2 pairs of modules is that they read v3.1 and v4.1 on the second pair. This info was not in the advert on OC when I bought it.

They are both Corsair CM3X1024-1600C7DHX.
Can someone confirm the two versions may not be compatible.
I have run both pairs seperately on their own without any problems.
I'm still waiting to hear off corsair.

---------------------
System: P5E3 Premium, QX9650 @ 3Ghz, 8800GTX 648/1566/2052, Xfi-extreme music, 2x500gb samsung SP f1 raid0, ocz gamextream 700w psu, Coolermaster stacker 832 case.

gordan · 12 Nov 2008 at 21:26

Most DDR3 setups don't work with 4 modules. See other similar threads. Your chances of getting 1600MHz stable with 4 DIMMs are somewhere between slim and none. There's a reason why kits come in pairs rather than fours. Clock it down to 1333MHz and you'll have better luck.

Cyber-Mav · 13 Nov 2008 at 00:25

i use the same 1600C7DHX--IN (IN part is some intel xmp crap) ram in my gigabyte p35t-dq6 board, the ram never stable for me at 7-7-7-20 timings, it worked fine though at 8-8-8-20 timings and i left it at that. the p35 chipset isnt the best chipset for ddr3 ram compatability at the end of the day, i could have tested timings such as 7-8-8-20 etc but i was too busy at the time to bother and left things as they were.

now i just picked up a kit of the same ram you got the 1600C7DHX non XMP (IN) model, the pc its in now is currently testing with all 4 sticks in the banks. so far i am also getting errors running at 400mhz fsb at the 8-8-8-20 timings with memtest. im gonna try stuff like upping the chipset voltage or something to see if it can get stable.

Cyber-Mav · 13 Nov 2008 at 02:05

just an update, iv had to drop the ram speed to 1280mhz using a 3.2 multiplier, for some reason the 3.33 multiplier wont work to give me 1333mhz ram speed so 3.20 multi it is for now. timings are at 7-7-7-20 and so far 20 passes of memtest are showing no errors for now. will run memtest for an hour more before i boot into windows and run intel burn test to make sure everything is still stable.

dropping from 1600mhz (800mhz) 8-8-8-20 ram speed to 1280mhz (640mhz) 7-7-7-20 ram speed showed a 200ish mb/s drop in speed in memtest, at 1600 it benched at 5600ish mb/s and at 1280ram speed it benches at 5414 mb/s.

any drop in speed though will be nothing compared to the gain in speed from having double the ram capacity anyways.

Cyber-Mav · 13 Nov 2008 at 04:30

just for your information and for something for you to go on here is a memory benchmark i ran in everest with the r 2gig ram at 1600mhz 7-7-7-20 timings, i will do a screenshot later on with the ram running at 1280mhz for you to see the difference.

edit: updated post with screenshot of 1280mhz 7-7-7-20 benchmark.

gordan · 13 Nov 2008 at 08:51

Just to make sure - you are running at 2T right? All the rated speeds/latencies will be for 2T command rate. The most you'll get stably at 1T is in the 1333MHz ball park.

Cyber-Mav · 13 Nov 2008 at 15:31

gordan said:
Just to make sure - you are running at 2T right? All the rated speeds/latencies will be for 2T command rate. The most you'll get stably at 1T is in the 1333MHz ball park.

yes its running at 2t, p35 chipset doesnt offer a 1t option, not on the board im using anyways (ga-p35t-dq6)

also to add, going to the 1280mhz ram speed lost me 60 points in vantage score, was nearly a 1000 point drop in the cpu benchtest.

gordan · 13 Nov 2008 at 18:18

Do you have a clock-twisting option? I've found that unless you're running a clean multiple, it needs to be set to "light" for 400 FSB -> 1333 RAM. Or at least that's the case on my X38 board.

Cyber-Mav · 13 Nov 2008 at 21:23

i didnt see anything going directly by the name of clock-twisting in the bios, if you could clarify some more i could give it a try. the board i use is a gigabyte ga-p35t-dq6 you could grab the manual from the gigabyte site and have a quick gander to see if that option is in there somethere.

Yellowbeard · 14 Nov 2008 at 22:28

BlowMeOneKenobi said:
I bought another pair(2x 1gb) DDR3 modules from OC a few months after the first pair.
I believe it was after installing the second pair that I started to get intermitent hard crashes(not sure this is the right terminology, no BSOD, just lockup). Sometimes this reset my bios settings which had to be changed back.
The only difference in the 2 pairs of modules is that they read v3.1 and v4.1 on the second pair. This info was not in the advert on OC when I bought it.

They are both Corsair CM3X1024-1600C7DHX.
Can someone confirm the two versions may not be compatible.
I have run both pairs seperately on their own without any problems.
I'm still waiting to hear off corsair.

---------------------
System: P5E3 Premium, QX9650 @ 3Ghz, 8800GTX 648/1566/2052, Xfi-extreme music, 2x500gb samsung SP f1 raid0, ocz gamextream 700w psu, Coolermaster stacker 832 case.

Try a single module of each in the proper slots for dual channel. Also, with 4 modules, you are likely going to need to raise the memory controller voltage.

Cyber-Mav · 15 Nov 2008 at 01:02

Yellowbeard said:
Try a single module of each in the proper slots for dual channel. Also, with 4 modules, you are likely going to need to raise the memory controller voltage.

does raising the mch voltage make much difference? i did put my mch voltage up 1 notch think it was an increase of 0.025v and i was still getting errors in memtest with the 4 sticks of ram running at 1600mhz.

Yellowbeard · 15 Nov 2008 at 02:00

Cyber-Mav said:
does raising the mch voltage make much difference? i did put my mch voltage up 1 notch think it was an increase of 0.025v and i was still getting errors in memtest with the 4 sticks of ram running at 1600mhz.

Yes, it is very important. 4 modules is twice the electrical load of 2 and MCH voltage is critical when trying to run 1600.

Cyber-Mav · 15 Nov 2008 at 02:06

hmm i wasnt aware that the electrical load would increase so much on the mch. im guessing the 0.025v was barely felt by the chipset. how much can i increase the mch voltage by without damage to the mobo? the board does use some gigabyte crazy cool heatsink stuff all over the chipset so i guess its gonna fairly cool.
have you known any other setups to run stable with 1600mhz ram speed on p35 chipset board with 3 sticks of ddr3 in them?

Cyber-Mav · 15 Nov 2008 at 12:29

hmm now this is annoying, i set the mch voltage to +0.050v in the bios and timings set as normal, managed to get 8 hours passed with 2 errors showing up before the 8 hour mark in test 5. not sure what more i can do to get it fully stable, back to the divider now and 1280mhz ram speed till i can test again tonight.

im running my manual 7-7-7-20 timings here, but looks to me like the board wants to run the ram at 8-8-8-20 if its put to auto.

better than before though, it used to error out straight away in memtest in the first pass of test 5.

Cyber-Mav · 15 Nov 2008 at 13:16

come to think of it, is the (G) MCH voltage option a setting that controls the northbridge voltage or the southbridge voltage?

whats the fsb overvoltage control option do?

gordan · 15 Nov 2008 at 16:27

The NB voltage may need to go up as much as 0.25V, and the DRAM voltage may need to go up by as much as 0.40V. If you dare run that high, that's fine, but the voltage boosts to get it stable at 1600MHz with quad DIMMs (if it works at all, which in my experience, it doesn't) are quite shocking. And if you're bumping up the VDIMM that much (also remember, DDR ref is vDDR/2, so more heat on the NB!) you may want to consider getting Corsair Dominator RAM fans.

Cyber-Mav · 15 Nov 2008 at 16:31

criky, and here is me planning on doing a +0.050v bump on the northbridge and running memtest overnight.

gordon what are your results like with 4 sticks of ram in memtest? do you get as many errors as i did in an 8 hour run? i may just bump up northbridge volts to +0.10 tonight and then let memtest run to see how it is.

Yellowbeard · 15 Nov 2008 at 16:51

Just for the testing, try the RAM timings at 9-9-9-24.

Cyber-Mav · 15 Nov 2008 at 17:11

sure thing yellowbeard, testing takes a long time though since i want to get atleast a minimum of 12 hours in memtest passed. tonight i will run the northbridge voltage increased and at the stock 7-7-7-20 timings, if it fails that i will try the stock northbridge voltage the next day and run at your suggested timings of 9-9-9-24.

im guessing im close to stability if i just get 2 errors at around 8 hours of memtest with no increase on the northbridge voltage. after some reading up i found out that the (g)mch voltage is for the southbridge and increasing that wont really make a difference, its the fsb overvoltage setting which is the northbridge voltage control that needs to be increased and then tested again.

will try it tonight at 7-7-7-20 with around a +0.050v increase on the northbridge.

Yellowbeard · 15 Nov 2008 at 17:24

Cyber-Mav said:
sure thing yellowbeard, testing takes a long time though since i want to get atleast a minimum of 12 hours in memtest passed. tonight i will run the northbridge voltage increased and at the stock 7-7-7-20 timings, if it fails that i will try the stock northbridge voltage the next day and run at your suggested timings of 9-9-9-24.

If it fails at CAS7, try CAS9 at the same voltage as CAS7.