Official GIGABYTE Forum
Questions about GIGABYTE products => Motherboards with Intel processors => Topic started by: birket on November 14, 2011, 11:18:27 am
-
We have 3 newly built workstations with the sanme hardware:-
Z68XP-UD3-iSSD boards
16GB (4X4) Corsair Vengeance 1600MHZ LP ram (1.35V)
120GB Corsair Force GT SSD main disk
Asus 480GTX GPU
Corsair AX850 PSU
Intel i7 2700k CPU
Western Digital SATA3 500GB secondary drives
We are having many random crashes where the system just shuts down unexpectedly. We have fresh installs of Windows 7 Ultimate x64. I have tried a number of things to try and figure out the root of the problem.
1) diabling onboard SSD drive (which we were hoping to use to accelerate the secondary SATA3 drive)
2) leaving clocking of CPU at default (35x)
3) updating BIOS to latest F8 revision
4) uninstalling all Gigabyte utilities
5) renaming 'gdrv.sys' to 'gdrv.sy_' (which a few postings had suggested could cause issues)
None of the above have made any difference.
The only thing which so far seems to have helped is disabling the XMP profile of the memory in BIOS and have it running at 1333MHz rather than its default 1600MHz. Using CPU-Z, the memory details are as follows in this non-XMP mode:-
DRAM Frequency - 668.7MHz
FSB DRAM - 1:5
CL - 9.0
tRCD - 9
tRP - 9
tRAS - 24
CR - 2T
This memory is good quality RAM from Corsair, but it should be running at 1.35V and at 1600MHz (not sure why CPU-Z is telling me it is 668.7MHz as I am pretty sure BIOS tells me 1333MHz). We are running Prime95 torture tests to stress the systems and monitoring the temps using RealTemp 3.60. The massive Thermalright coolers are preventing the CPU core temps from going above about 55 degrees C at full load, so temps aren't an issue. Crashes (when XMP mode is enabled) happened randomly, at full load or not at full load.
(The way I have come to the conclusion of XMP being the problem is that in an attempt to get stability, I loaded failsafe BIOS settings which removes the XMP profile)
-
Welcome birket .....top notch editing machines ;)
You would be probably be better off with Optimized default... then disable XMP manually...... load all memory timings and voltage manually and claw back total control.....this has helped countless people in system stability.....as you have 3 systems displaying the same problem....yes it's most certainly a con-fig issue .... was the memory purchased as matched kits ....and two.....out of curiosity....where they on the compatibility table on the gigabyte or Corsair website ?
Post back if you have anymore problems.....If you get some BSOD codes to work with ....all the better!
Aussie Allan
-
Hi friends,
It seems that when doing the oc, you must release the memory timing.
Use timing JDEC. Check with CPUID Cpu-z . SPD Tab. You find a more timing less than 9-9-924
EDIT01 : or with this timing but different voltage, like 1.5volt.
And not use effectively XMP profile, if you prefer. Or else if you whant keep XMP Profil, Ajust votage the ram to 1.5 in bios.
I modify my post after read Allan Aussie's comments. Thanks AA. :P
Gloup_Gloup
-
Hi All,
I found this article see if this helps to shed some light on the matter:
http://www.tomshardware.co.uk/forum/256351-11-problems-profile
Hope this helps
Teknology9
-
Welcome birket .....top notch editing machines ;)
You would be probably be better off with Optimized default... then disable XMP manually...... load all memory timings and voltage manually and claw back total control.....this has helped countless people in system stability.....as you have 3 systems displaying the same problem....yes it's most certainly a con-fig issue .... was the memory purchased as matched kits ....and two.....out of curiosity....where they on the compatibility table on the gigabyte or Corsair website ?
Post back if you have anymore problems.....If you get some BSOD codes to work with ....all the better!
Aussie Allan
I had a go at putting the memory timings in manually, but I got another system freeze. Power to the keyboard and mouse just stopped and system locked up, so I had to do a reset.
I'll confirm what I entered, but essentially I set it so the multiplier (or whatever it is called) on the memory was 16 (so that gave me the 1600MHz), the timings even on auto seemed correct at 9,9,9,24. I manually set the voltage to 1.35v (though there did seem to be a whole lot of other voltage items, but I left all these at AUTO). It ran through Prime95 for about 10-15minutes, but then I got the system freeze. Now I've put it back to 'Load Failsafe Defaults' (so 1333MHz).
Memory was bought in pairs from Scan (in the UK), so each pack was an 8GB (2x4GB) pack. I made sure I put sticks from the same pack in the same coloured DIM slots. Pretty certain they are on the compatability table, but will double check. Code is CML8GX3M2A1600C9W.
-
Hi friends,
It seems that when doing the oc, you must release the memory timing.
Use timing JDEC. Check with CPUID Cpu-z . SPD Tab. You find a more timing less than 9-9-924.
And not effectively XMP profile.
Gloup_Gloup
Hi. TBH, I'm not even really trying to overclock these systems yet - just get them stable. The CPU multiplier is 35 and thevoltage is Auto which I think is the default.
Not really sure what JEDEC means, but I did see in the timings table the 3 columns of JEDEC timings as follows:-
JEDEC #2 518MHz, 7,7,7,19,27, 1.5V
JEDEC #3 592MHz, 8,8,8,22,30, 1.5V
JEDEC #4 666MHz, 9,9,9,24,34, 1.5V
XMP-1600, 800MHz, 9,9,9,24,41,2T 1.35V
The XMP listing is I presume what CPU-Z is reading from the dims. Are the Jedec settings tested setups?
-
OK....I'm assuming you had 9,9,9,24,34,2T and set to (which are loose timings for testing) 1.35V......do this again but bump the Voltage (Ram) a notch or 2 .......even as high as 1.45v-1.5v and at 1333
the next thing that can give conflicts in the "Let's try this easy stuff first".... is ......if no one us using any USB3 toys.....disable USB3 in BIOS and delete the driver package.....it will default back to USB2 and will still be usable.....this has also given problems at the forum.
Aussie Allan
-
Hi birket
Hope I'm notbeing naive but won't memory usually want more Volts to run faster? You show a drop from 1.5V at 1333MHz to 1.35 at 1600MHz. The spec for my corsair ballistix memory is 1.5V at 1333MHz but for XMP mode at 1600MHz requires 1.65V. Worth trying double checking rated voltage on maufacturers website and uping the Volts?
jkjgb
-
Quite correct, you normally require a bit more voltage to get that extra bit of speed otherwise it isn't stable even if you can get it to run. Often too even though it comes with an XMP attached it won't work properly on auto and requires the timings and voltage to be manually input into the BIOS.
-
Hi birket
Hope I'm notbeing naive but won't memory usually want more Volts to run faster? You show a drop from 1.5V at 1333MHz to 1.35 at 1600MHz. The spec for my corsair ballistix memory is 1.5V at 1333MHz but for XMP mode at 1600MHz requires 1.65V. Worth trying double checking rated voltage on maufacturers website and uping the Volts?
jkjgb
Hi there. This RAM is designed to be run at 1.35v at 1600MHz, so I am really just trying to get it to run at this setting. I am not even trying to overclock the memory at all.
Here's a link to the RAM we are using:-
http://www.corsair.com/memory-by-product-family/vengeance/vengeance-low-profile-8gb-dual-channel-ddr3-memory-kit-cml8gx3m2a1600c9w.html
-
Sorry to harp on but.....".if no one us using any USB3 toys.....disable USB3 in BIOS and delete the driver package."
If you have not tried this....it's worth a try.....it very well could be a conflict ;)
Aussie Allan
-
Sorry to harp on but.....".if no one us using any USB3 toys.....disable USB3 in BIOS and delete the driver package."
If you have not tried this....it's worth a try.....it very well could be a conflict ;)
Aussie Allan
I will try, its just that I'm working on the workstations as we speak and it normally takes around half an hour of Prime95 to knock it over, so I need to find half an hour of freedom to check this.
-
disable USB3 in BIOS and delete the driver package."[/b]
I have now disabled both instances of USB3 that I can find in BIOS, but am not sure how to delete the driver package.
I am, nevertheless, running Prime95.
-
My Prime95 tests failed I'm afraid. System completely froze again. Doesn't look like a USB3 issue.
I have re-enabled USB3 in BIOS and have manually upped the voltage to 1.45v but left the clock speed at 1600MHz.
So, to sum up:-
- setting XMP correctly sets speed to 1600MHz and voltage to 1.35v, but this causes instability
- setting to 'failsafe' with 1333Mhz and 1.5v seems stable
- setting to XMP with USB disabled causes instability
- am now trying 1600MHz with 1.45v
...UPDATE! I've been running Prime95 with 1600MHz/1.45v now for 40 minutes. This is longer than I would expect - previous tests have crashed the system after 20-30 minutes max. Could it be that the RAM shouldn't really be run as low as 1.35v, even though that is what it is advertised to do by Corsair themselves? I will leave it overnight and see how it goes.
-
Hi birket, ;)
As already observed by Aussie Allan, me and Dark Mantis, you can set the BIOS settings for the ram to 1.5Volt.
I am talking about an increase of 0.15.
But if all goes well with 1.45volt, keep it.
In passing, I appreciate the active participation of Aussie Allan and Dark Mantis.
Teknology9 and jkgb, they have given helpful comments.
Nice work team. ;)
Gloup_Gloup
-
birket
How did prime go at the lower speed at the high volts ?.......40mins is a good sign.......might require 1.5 for long term stability if it crashes overnight!
Aussie Allan
-
birket
How did prime go at the lower speed at the high volts ?.......40mins is a good sign.......might require 1.5 for long term stability if it crashes overnight!
Aussie Allan
Well, its been running overnight without a problem, so that's fantastic news! Thanks for your tips guys. However, I am slightly frustrated by the fact that this Corsair ram is supposed to be set to 1.35v. I have had lots of issues getting these machines stable, and it is all because of this RAM! I will be writing an email to Corsair to highlight my concern.
-
Well that really good news if only semi-good news for you personally......you seem to have a belief of the virtues of memory that runs at 1.35v
Could you share your thoughts behind this ?
All I could add if you really want the lowest voltage possible is to reduce the volts .04/.05 at a time and test accordingly......as you are probably be aware.....tighter memory timings will play a far bigger part re performance then brute Mhz speed
Aussie Allan
-
you seem to have a belief of the virtues of memory that runs at 1.35v
Could you share your thoughts behind this ?
It has nothing to do with a belief or virtues of memory that runs at 1.35v: these ram modules are designed to run at 1.35v, but as I have shown, they don't. I don't care if they are 1.35v, 1.5v, 1.65v or 13v as long as they work as the manufacturer states!
Here's a review of the RAM if you are interested:- http://www.overclock3d.net/reviews/memory/corsair_vengeance_lp_8gb_1_35v_review/1
-
You know....no matter how much I read and people I ask.....there seems to be no real reason behind ultra-low rated memory except a higher max voltage potential ...... some company's push the power saving angle but the difference between 1.5 and 1.35v is about 2Watt/h per week running 24/7 :D :D :D :D :D :D
Here's an interesting snip from a link I'll include at the bottom if you find it interesting....
DDR3 was specified by the JEDEC Solid State Technology Association to run between 1.5 V and 1.575 V (a 30% drop from DDR2's 1.8 V), and to withstand voltages of up to 1.975 V before taking damage.
http://www.tomshardware.com/reviews/lovo-ddr3-power,2650.html
Let us know if you need any further help.
Aussie Allan
-
You know....no matter how much I read and people I ask.....there seems to be no real reason behind ultra-low rated memory except a higher max voltage potential ...... some company's push the power saving angle but the difference between 1.5 and 1.35v is about 2Watt/h per week running 24/7 :D :D :D :D :D :D
Here's an interesting snip from a link I'll include at the bottom if you find it interesting....
DDR3 was specified by the JEDEC Solid State Technology Association to run between 1.5 V and 1.575 V (a 30% drop from DDR2's 1.8 V), and to withstand voltages of up to 1.975 V before taking damage.
http://www.tomshardware.com/reviews/lovo-ddr3-power,2650.html
Let us know if you need any further help.
Aussie Allan
Interesting. Its bloody annoying though. You wouldn't believe how much swearing has been flying around our office over the last few days, all because of a bloody voltage specified by the manufacturer that isn't appropriate! The money saving side of it didn't play a part in my decision to buy the memory. It was available from our supplier and ran at the right speed and seemed a decent price. I then found the above review and it gave it a good write-up. This was really what made me buy it. Nothing at all to do with the voltage.
Thanks for your help.
-
It is good to hear that you have now got the issue sorted and I too can't understand why they try and make low voltage memory when most of the time it needs increasing to get stability. I think a lot of it is simply trying to appeal to the "save the earth" group, even though it is all a false economy.