View Full Version : 8GB (4x2GB) G.Skill @ 1068MHz
I have 4x1GB sticks of corsair Corsair Dominator PC2 8500 that failed memtest v3.4 at 1068MHz, but the G.Skill (4x2GB) passed memtest v3.4.
How hard is it to run 4 sticks of DDR2 memory at 1068MHz at the stock NB voltage of a IP35 MB?
you may have a faulty module if the corsair doesnt pass memtest(id use memtest86+ to test them).
nb voltage is essentially dependant on the mb and voltage the mem needs imo(the higher the voltage needed for the mem the higher the nb may need to go).some need more voltage to run smooth with all four dimms poulated some dont.youll have to test yours to find out what works best.
Careface
02-22-08, 01:47 AM
Holy hell, surely you mean 2.4V in your original post? 3.4V in DDR2.. surprised they lived.
Im not sure about the nb being too stressed running 4 sticks, but to test you could always just up the volts on the nb a little and see if it passes memtest or not. If not, then you could very well have a faulty stick
With 4 modules of memory, the northbridge chip will always be under more stress than when using 2 modules. You will have to increase northbridge voltage for better stability. How much voltage varies depending on NB chipset and memory used.
JamesXP
02-23-08, 01:51 PM
He means memtest 3.4 I think...
Memtest version 3.4, the G.Skill F2-8500CL5D-4GBPK is at 2.1 volts. As soon as I found Memtest86+ v2.01 I ran that and my system passed 6 times with no errors, it is much faster than the version 3.4 I was using. The MCH is still at the stock 1.25 volts, I was just wondering how lucky I was to get 1068MHz on 4x2GB configuration.:attn:
I bought 4x1GB sticks of corsair dominator CM2X1024-8500C5D and they would not all run together at 1068MHz, I didn't try to go over ~3.5volts on the mch.
Memtest version 3.4, the G.Skill F2-8500CL5D-4GBPK is at 2.1 volts. As soon as I found Memtest86+ v2.01 I ran that and my system passed 6 times with no errors, it is much faster than the version 3.4 I was using. The MCH is still at the stock 1.25 volts, I was just wondering how lucky I was to get 1068MHz on 4x2GB configuration.:attn:
I bought 4x1GB sticks of corsair dominator CM2X1024-8500C5D and they would not all run together at 1068MHz, I didn't try to go over ~3.5volts on the mch.
I would recommend you lower the voltage on the G.Skill modules to 2.0v. More voltage is not better with those modules. I know they are rated 2.0-2.1v, but I've had mine up to 1080MHz at 2.0v and increasing the voltage only increases temperatures and causes them to get unstable. Start at 2.0v and see if they still run the same speed, then increase in .025v increments until they stabilize, but I'm still willing to bet they will run just fine at 2.0v.
Blazing fire
02-23-08, 06:30 PM
Why not test 1 stick at a time? Also, does your mobo support 4 sticks of your corsair memory? I don't use abit so I don't know. Check the manual or something.
Unless I'm misreading something, it seems to me that the G.Skill memory is working just fine and the Corsair is not. Why are we trying to diagnose his Corsair problem? I say keep the G.Skill memory and be happy.
I would recommend you lower the voltage on the G.Skill modules to 2.0v. More voltage is not better with those modules. I know they are rated 2.0-2.1v, but I've had mine up to 1080MHz at 2.0v and increasing the voltage only increases temperatures and causes them to get unstable. Start at 2.0v and see if they still run the same speed, then increase in .025v increments until they stabilize, but I'm still willing to bet they will run just fine at 2.0v.
I tried 2.0 volts and all 4 passed, I'm going to let memtest run......but wow only 2.0 volts needed for 4 sticks!:drool:
I'm happy with 8GB of G.Skill, it works at lower volts and I didn't have to OC the NB. I thought it was just strange that 4x2GB runs fine at 1068MHz, but the 4x1GB corsair would not.
I tried 2.0 volts and all 4 passed, I'm going to let memtest run......but wow only 2.0 volts needed for 4 sticks!:drool:
I'm happy with 8GB of G.Skill, it works at lower volts and I didn't have to OC the NB. I thought it was just strange that 4x2GB runs fine at 1068MHz, but the 4x1GB corsair would not.
Those G.SKill modules are awesome performers. They run over their rated speed at very low voltage(2.0v). I can't recommend them enough.
Passed 11 times at 2.0 volts!
http://webpages.charter.net/stg556/memtest.jpg
Blazing fire
02-25-08, 04:43 AM
Nice.
redduc900
02-25-08, 12:14 PM
What are the rest of your voltages and sub-timings at 1068?...
CPU Core Voltage... 1.32v
DDR2 Voltage... 2.0v
CPU VTT 1.2V Voltage
MCH 1.25V Voltage... 1.25v
ICH 1.05V Voltage
ICHIO 1.5V Voltage
DDR2 Reference Voltage
CPU GTLREF 0+2
CPU GTLREF 1+4
Refresh Cycle Time (tRFC)
Write Recovery Time (tWR)
Write to Read Delay (tWTR)
ACT to ACT Time (tRRD)
Read to Precharge (tRTP)
Command Rate
All of my sub timings are set to auto in the BIOS. The only manual settings I use are 5-5-5-15. 2.0v Vdimm and 1080MHz and I'm using the older 1000MHz 2GB modules, which still run fine well above their rated speed.
Here is a shoot of Prime95 after 24+ hours.
http://webpages.charter.net/stg556/3.jpg
I'll take some of the BIOS so you can see what I'm working with...............
leftheaded
02-25-08, 06:51 PM
just an fyi... i had errors after 25 passes! i'd let memtest86 run for 24 hours or 50 passes, whichever comes first.. yeah, both of those numbers are arbitrary, but in my experience 18 hours and 25 passes wasn't sufficient
http://www.xtremesystems.org/forums/showthread.php?t=174179
just an fyi... i had errors after 25 passes! i'd let memtest86 run for 24 hours or 50 passes, whichever comes first.. yeah, both of those numbers are arbitrary, but in my experience 18 hours and 25 passes wasn't sufficient
http://www.xtremesystems.org/forums/showthread.php?t=174179
Later in the week I'll let my rig run memtest for 24+ hours, I'm still running prime95 right now. Each memtest pass takes about 1 hour with 8GB:mad:.
Is 30 hours of prime95 good, or should I let it run more?
Here are some BIOS pics......
http://webpages.charter.net/stg556/bios1.jpg
http://webpages.charter.net/stg556/bios2.jpg
http://webpages.charter.net/stg556/bios3.jpg
CPU Core Voltage... 1.3275v
DDR2 Voltage... 2.000v
CPU VTT 1.2V Voltage... 1.20v
MCH 1.25V Voltage... 1.25v
ICH 1.05V Voltage... 1.05v
ICHIO 1.5V Voltage... 1.50v
DDR2 Reference Voltage... 0%
CPU GTLREF 0+2... 67%
CPU GTLREF 1+4... 67%
Refresh Cycle Time (tRFC)... Auto
Write Recovery Time (tWR)... Auto
Write to Read Delay (tWTR)... Auto
ACT to ACT Time (tRRD)... Auto
Read to Precharge (tRTP)... Auto
Command Rate... Auto
redduc900
02-26-08, 03:09 AM
Thanks for the pics and the timings / voltages... that's some impressive RAM you've got there.
Do you have any extra cooling on the RAM? I'd imagine those sticks heat up a bit.
I measured some temps with a Raytek noncontact thermometer:
The hottest spot on the side of the G.Skill heatsink is 110F.
The NB/MCH heatsink is 112F.
The hottest spot on the back of the pcb on my EVGA 8800 ultra is 106F.
(temps were taken after 2 passes of memtest, with a room temp of 75F)
I want to put a small fan on the NB/MCH & I may try to put some fans on the ram.
http://webpages.charter.net/stg556/ram.jpg
Also, I forgot to add that the EIST function is turned on. At 3204MHz(2136MHz) 356x9(356x6) the EIST seems to work fine, anything higher was not 100% stable with EIST turned on.
http://webpages.charter.net/stg556/bios4.jpg
Do you have any extra cooling on the RAM? I'd imagine those sticks heat up a bit.
They don't really get that hot. In fact, these G.Skill modules run cooler than any Corsair or Crucial memory I've ever owned.
Shiggity
02-26-08, 06:59 PM
My gskill ram actually responds negatively to increased voltage. The rated is 1.8v-2.0v, but at 2.0v theres almost no stability.
My gskill ram actually responds negatively to increased voltage. The rated is 1.8v-2.0v, but at 2.0v theres almost no stability.
That's been my experience as well with all the G.Skill modules I've owned. I've had the best results running them at their minimum rated voltage.
Evilsizer
02-26-08, 09:41 PM
what ever the manufacture of the ram is doesnt play a role in how the oc. the IC used on the ram is more of role then anything. any ic that is not Micron based responds to looser timings then voltage. micron based ram responds more to voltage then timings, changing both on microns. high clock yeilds are better on micron base ram then non microns. at that point your talking the extreme of speeds. the newer powerchip and promos high density ic's have shown promise. starting with the ddr2-1000 4gig kit from gskill, showing how non micron ram can do. though it has taken some time for them "2gig dimms" to get to this point. even 1gig dimms about a year ago had issues as well, as my corsair v2.1's with promos ic's i was able to get them to 570mhz/ddr2-1140 5-6-6 @2.3v. no one else at the time made it that high with non-micron ram, in terms of ram speed.
http://img.photobucket.com/albums/v733/Evilsizer/ocforums/superpi%201m%20runs/34g1140g.jpg
this is back when they had EPP settings in the ram for ddr2-1066. getting the same ram now they dont have those settings.
Any idea what IC the G.Skill F2-8500CL5D-4GBPK uses? Powerchip??:confused:
Can you tell what IC the F2-8500CL5D-4GBPK uses without removing the heatsink?
Evilsizer
02-26-08, 10:11 PM
yes, from what i can gather it is powerchip. as the only thing different between the gskill 4gig kits ddr2-1000 and ddr2-1066 is just spec speed. when the ddr2-1000 kits came out many people reported being able to do ddr2-1066 on stock-v and timings.
yes, from what i can gather it is powerchip. as the only thing different between the gskill 4gig kits ddr2-1000 and ddr2-1066 is just spec speed. when the ddr2-1000 kits came out many people reported being able to do ddr2-1066 on stock-v and timings.
I bought a set of the G.Skill F2-8000CL5D-4GBPQ and they would not run at 1068MHz @ 2.1volts, the 2x2GB pair failed memtest(but passed at the rated 1000MHz). That is why I bought the new F2-8500CL5D-4GBPK when I saw it at newegg finally.:santa:
Evilsizer
02-26-08, 10:31 PM
at some point they prolly started to take the ones running at the higher speed, to make the model you own. this does happen alot with ram though, first round of them oc great. then later ones oc less or dont at all. it is the way or ram binning...
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.