E8600 on X48-DQ6

[Rivenson]

Hello to everybody

I am reading this forum about 6 months but i am posting for first time.
After so much reading i was under the impression that i can easily achieve mild overclock on my system for 24/7 use , but certain clues of instability confusing me and i start all over again to find the cause.
So i would be grateful for some help, and especially from Gigabyte knowledge monster Lsdmeasap...

My setup consists of

Motherboard : GA-X48-DQ6 F7

CPU : E8600 Vid 1.2500
CPU Cooler : Zalman 9500A
RAM 1150DDR: Kingston HyperX KHX9200D2K2/2G 555-15 2GB (2x1) at 2.3-2.35v
HDD : 2x 300 Velociraptors on ICH9R ide mode (planning to Raid0)
PSU : Corsair HX620
GPU : Radeon 3870
OS : XP SP3

Bios:

RGB >>>>>>>>>>>>>>>>>>>>>>>>> Auto
Cpu Clock Ratio >>>>>>>>>>>>>>>> 8X
Cpu Frequency >>>>>>>>>>>>>>>>> 3400.0
Cpu Host Clock Control >>>>>>>>>>> Enabled
CPU Host Frequency >>>>>>>>>>>>> 425
PCI Express >>>>>>>>>>>>>>>>>>> 100-102
C.I.A 2 >>>>>>>>>>>>>>>>>>>>>> Disabled
Performance >>>>>>>>>>>>>>>>>> Standard
System Mem Multi >>>>>>>>>>>>>> (I am trying D stap for 1133)
DRAM Timing >>>>>>>>>>>>>>>>>> Manual

Timings >>>>>>>>>>>>>>>>>>>>> 5-5-5-15

Advance Timings >>>>>>>>>>>>>>> All Auto for now ,except TRD which is set to 6 ( also trying 7)

CPU/PCIEX Clock Driving Control>>>>> 800mV
CPU Clock Skew Control >>>>>>>>>>> Normal
GMCH Clock Skew Control >>>>>>>>>> Normal
System Voltage >>>>>>>>>>>>>>>>> Manual
GTLR1 >>>>>>>>>>>>>>>>>>>>>>>> Auto
GTLR2 >>>>>>>>>>>>>>>>>>>>>>>> Normal (i don't know what this is)
DDR2 OverVoltage Control >>>>>>>>>> +0.50
PCI-E OverVoltage >>>>>>>>>>>>>>> Normal
FSB OverVoltage >>>>>>>>>>>>>>>> +0.05
GMCH OverVoltage >>>>>>>>>>>>>>> +0.100
LLC >>>>>>>>>>>>>>>>>>>>>>>>>> Disabled
CPU Voltage Control >>>>>>>>>>>>>> 1.2000 ( Win idle 1.168 - Prime95 LargeFFT load 1.152) which i think its enough for 3825 cause i have no errors for hours not even in 15 passes of IntelBurnTest)

I know that 4 Ghz for 8600 is piece of cake , but i was trying to check for stability first on mch-ram-fsb combo ,plus to check temperatures... (i have applied AC5 everywhere and NB is cooled actively by 40*40 fan)
NB temps are around 40-45 C

I'm on Prime Blend test for about 5 hours now with no errors
but if i set multi to 9x for 3825 Mhz , blend test fails (10 to 20 min) , and occasionally system crash (during the test)


Any advice??

[Lsdmeasap]

Re: E8600 on X48-DQ6

You may just need more Vcore, I'd bump it up a little bit just to be sure that is not the issue. Then if it still crashes you know that was not it, and you can lower it back down some.


I'd need to see your full exact settings you use when you try 3.8 to advise, you are correct you should not have many issues getting 4Ghz provided your cooling is ok are you are not overclocking your ram to hard.

You may have to use more MCH or use 8-9 tRD (Static tRead) for the 400 strap, sorry I can't remember for sure how X48 handles low tRD as I have not used in quite some time.

[Rivenson]

Re: E8600 on X48-DQ6

Thank you for your response

First i'd like to tell you about memory i use that Kingston's KHX9200D2K2/2G are rated at DDR2-1150MHz 5-5-5-15 at 2.30-2.35v. Regardless of high voltage for ddr2 , i can assure you that touching them in my system feels comfortably warm -( probably my hand warms them ever more) plus they are in Gigabytes Qualified Vendors List.
?? Naturally?? any attempt to bring them to 4-4-4-12 at 960-1000 Mhz fail in Memtest (i understand they are not Mushkins). so for the time being i'm trying anything closest to their rated values.

I raised Vcore for E8600 to the default (1.2500v) to eliminate E8600 possible issues and then i change bios settings as follows:

M.I.T

RGB .................................................. . Auto
CPU Clock Ratio .................................... 8x
Fine Cpu clock ..................................... +0.0
CPU Frequency................................... 3.360 (420*8) Close to Cpu's Default
CPU Host Clock Control .......................... Enabled
CPU Host Frequency .............................. 420
PCI Express Freq. .................................. 100
C.I.A.2 ................................................ Disabled
Performance Enhance ............................. Standard

System Memory Multi .............(SPD)......... 2.66D
Memory Frequency .................800........... 1120
DRAM Timing Selectable........................... Manual

**** Standard Timing Control ****
(SPD)
CAS Latency Timing ................ 5 ........... 5
DRAM RAS# to CAS# Delay ....... 5 ........... 5
DRAM RAS# Precharge ............. 5 ........... 5
Precharge Delay (tRAS) ........... 15 .......... 15

**** Advance Timing Control ****
(SPD)
ACT to ACT Delay (tRRD) .......... 3 .......... Auto
Rank Write to Read Delay .......... 3 .......... Auto
Write to Precharge Delay .......... 6 .......... Auto
Refresh to ACT Delay ............... 42 .......... Auto
Read to Precharge Delay ........... 3 .......... Auto
Static tRead Value ................... 8 .......... ( Have try 8,7,6) Now 7
Static tRead Phaze Adj ............. 5 .......... Auto
Command Rate ........................ 2 .......... 2T

**** Clock Driving & Skew Control ****

CPU/PCIEx Clock Driving Control ................. 800mV
CPU Clock Skew Control ........................... Normal
GMCH Clock Skew Control ......................... Normal
System Voltage Control ............................ Manual
DDR2 Overvoltage Control ......................... +0.50v (as rated ... actual 2.32 at load)
PCI-E Overvoltage Control ........................ Normal
FSB Overvoltage Control ........................... +0.10v
GMCH Overvoltage Control ........................ +0.100
CPU GTLREF1 Voltage Ratio ....................... Auto
CPU GTLREF2 Voltage Ratio ....................... Normal
Load Line Calibration ................................ Disabled
CPU Voltage Control ................................. 1.25000

i have disable EIST and C1E


Prime's Blend test seems stable for 2.5 hours at tRD7, no issues (yesterday)

Orthos Blend test is stable so far ( 2 Hours ) With tRD6



screen:



BUT.....

memtest gives me errors and always at test7 , (tested with tRD 6,7)

screens:


and


is this an indication of faulty ram?




just to bring to your attention that X48
is capable of (for my understanding)
tRD of 6 at 425 Mhz at stock voltage
and that is one of the reasons that i bought it back then......

screen from anandtech

[Chike]

Re: E8600 on X48-DQ6

The same address failed twice (same bit) this may be indication to faulty RAM.
Try checking them individually at 1150 without oveclocking, maybe test 7 only to save time since it's likely to catch it.

[Rivenson]

Re: E8600 on X48-DQ6

i am on Orthos blend test for 10 hours with no error so far ,at 420 *9 (3780 Mhz) .
I will check sticks separately and i'll post back.....

[Lsdmeasap]

Re: E8600 on X48-DQ6

444-12 at 960 is asking for issues with about any brand really. Even at those high voltages you may never be able to run that low of a CAS at that high of a memory freq.

I'd suggest maybe 5-4-4-12 or 5-5-4-12, but you may just have to use 555-12 or 555-15

You are overclocking, so setting Vcore to default is not going to eliminate that as an issue. You may need to set it much higher, just to be sure. Just my thoughts on that part of your comments.

Ya I agree with Chike, an error in the same place more than once could mean a faulty module. Good things with that would be you do not need to find which really because if you RMA them you will need to send in the set all at once anyway.

Just because a board can do x tRD at x FSB does not mean that your particular ram will do that. Also even though it is on the QVL this also does not mean it will do that it only means that it has been tested to be compatible at stock settings per the CPU used and that ram.

Have you tested tRD 6-7 with more MCH at all, not often needed without 4GB or above, but your ram in particular may need it to run what you are trying.

I would guess from that ram needing 2.3+V to run stock specs that it may very well be pushing the boundaries of the IC's themselves since the voltage is already so high for DDR2. Since they are rated to allow such a high voltage, that means often that any less and they may not do the rated speed/timings, and that is a lot of voltage so they may actually be damaged from stock voltage settings.

I would never use above 2.2V for 24/7 use, and even then only with a fan on the ram, so to me a 2.3V+ stock voltage allowance would mean the ram needs very high voltage to run that speed and would not be good. That is how I see it, I mean I can do that speed with some Crucial's I have here with only 2.05, and the same with a few other brands with only 2.1-2.15. Not saying that your ram is bad, just that it would not be my choice for high speed ram due to the high voltages needed to push it there.

[Rivenson]

Re: E8600 on X48-DQ6

444-12 at 960 is asking for issues with about any brand really. Even at those high voltages you may never be able to run that low of a CAS at that high of a memory freq.

I'd suggest maybe 5-4-4-12 or 5-5-4-12, but you may just have to use 555-12 or 555-15

Ok man i get the picture

You are overclocking, so setting Vcore to default is not going to eliminate that as an issue. You may need to set it much higher, just to be sure. Just my thoughts on that part of your comments.

I know i'm OC thus with default Vcore also drop Cpu multiplier for stock Frequency (for my testing purposes) (CPU Frequency................................... 3.360 (420*8) Close to Cpu's Default)

Ya I agree with Chike, an error in the same place more than once could mean a faulty module. Good things with that would be you do not need to find which really because if you RMA them you will need to send in the set all at once anyway.

Thank you for clarification , and yes..
i've got a faulty one .....




Module checked in both channels individually.
Just wondering why Orthos and Prime didn't find errors.
Nevermind.

Just because a board can do x tRD at x FSB does not mean that your particular ram will do that. Also even though it is on the QVL this also does not mean it will do that it only means that it has been tested to be compatible at stock settings per the CPU used and that ram.

Perhaps my entire system does not mean anything...lol
Ok i can see where are you coming from.

Have you tested tRD 6-7 with more MCH at all, not often needed without 4GB or above, but your ram in particular may need it to run what you are trying.

All tests done with +0.100 and +0.150v MCH . Except Memtest errors (in test 7 only) ,in Prime and Orthos extensive tests (at least 10 Hours each) everything seems stable as rock with tRD6 at 420*9.

I would guess from that ram needing 2.3+V to run stock specs that it may very well be pushing the boundaries of the IC's themselves since the voltage is already so high for DDR2. Since they are rated to allow such a high voltage, that means often that any less and they may not do the rated speed/timings, and that is a lot of voltage so they may actually be damaged from stock voltage settings.

I would never use above 2.2V for 24/7 use, and even then only with a fan on the ram, so to me a 2.3V+ stock voltage allowance would mean the ram needs very high voltage to run that speed and would not be good. That is how I see it, I mean I can do that speed with some Crucial's I have here with only 2.05, and the same with a few other brands with only 2.1-2.15. Not saying that your ram is bad, just that it would not be my choice for high speed ram due to the high voltages needed to push it there.

I understand that -the cooler the better- and generally 2.2 and above, without fan is a non recommended situation for DDR2 , but perhaps there are some exceptions??
My ram runs without fan at 2.35v (only heat spreader ) at full load and i can touch them for like all day. Meaning, i dont feel that i kill my chip.
Unlike MCH that the temperature of varies from 40-45 (everest) degrees and i cannot touch the silent pipe above 10 sec.
Anyway i just wondering..
Thanks for your comment


Before i do RMA what do you think about Mushkins XP-9200

• 1150MHz (575MHz DDR)
• 5-5-4-12 latency
• supports EPP profiles
• Unbuffered
• 240 pin DIMM
• 2.2V-2.3V

Should i go for them??

[Chike]

Re: E8600 on X48-DQ6

Module checked in both channels individually.
Just wondering why Orthos and Prime didn't find errors.
Nevermind.

Very simple, it's just one byte out of 2GB they have probably never used it or it may have only trashed a string or data that have never been used.

[Rivenson]

Re: E8600 on X48-DQ6

Thank you Chike

I'm on 4.037 MHz CPU at 1.232v now (load) , 425 fsb ,tRD 6, memory at 1133 at 555-15
and still no errors on LargeFFTs (12 hours).

so when i complete all tests i'll post my final settings
to anyone that might interesting.

[Rivenson]

Re: E8600 on X48-DQ6 final

Well it took me some time to post back after RMA memory modules but store run out of stock thus i needed to buy Kingstons from other store.
Also i took the chance to update my bios with F8e for the benefit of newer AHCI rom and build my raid 0 array

Now i dont know if f8e has to do with succesfull and easy overclock, but i had the possibility of pushing the modules to the "prohibited" area....

here's the settings

M.I.T

RGB .................................................. . Auto
CPU Clock Ratio .................................... 10x
Fine Cpu clock ..................................... +0.0
CPU Frequency...................................... 4000
CPU Host Clock Control .......................... Enabled
CPU Host Frequency .............................. 400
PCI Express Freq. .................................. 100
C.I.A.2 ................................................ Disabled
Performance Enhance ............................. Standard

System Memory Multi .............(SPD)......... 2.40B
Memory Frequency .................800........... 960
DRAM Timing Selectable........................... Manual

**** Standard Timing Control ****
(SPD)
CAS Latency Timing ................ 5 ........... 4
DRAM RAS# to CAS# Delay ....... 5 ........... 4
DRAM RAS# Precharge ............. 5 ........... 4
Precharge Delay (tRAS) ........... 15 .......... 12

**** Advance Timing Control ****
(SPD)
ACT to ACT Delay (tRRD) .......... 3 .......... 3
Rank Write to Read Delay .......... 3 .......... 3
Write to Precharge Delay .......... 6 .......... 4
Refresh to ACT Delay ............... 42 .......... 38
Read to Precharge Delay ........... 3 .......... 3
Static tRead Value ................... 6 .......... 6
Static tRead Phaze Adj ............. 0 .......... Auto
Command Rate ........................ 2 .......... 2T

**** Clock Driving & Skew Control ****

CPU/PCIEx Clock Driving Control ................. 800mV
CPU Clock Skew Control ........................... Normal
GMCH Clock Skew Control ......................... Normal
System Voltage Control ............................ Manual
DDR2 Overvoltage Control ......................... +0.25
PCI-E Overvoltage Control ........................ Normal
FSB Overvoltage Control ........................... Normal
GMCH Overvoltage Control ........................ +0.050
CPU GTLREF1 Voltage Ratio ....................... Auto
CPU GTLREF2 Voltage Ratio ....................... Normal
Load Line Calibration ................................ Disabled
CPU Voltage Control ................................. 1.29375




and





i was able to push further at 425 fsb , and +0.30 vDimm and those crazy timings....




Well this is it...
Thank you guys for your help...

[Chike]

Re: E8600 on X48-DQ6

Yhea pretty tight timings, tRD 6 is nice, what are memory performance?

[Rivenson]

Re: E8600 on X48-DQ6

Here

400fsb




and 425fsb




I know i can push further like 450 fsb 1080 ddr same settings (adding a little more vDimm) but i'm happy as it is...
Bare in mind that results came with GMCH stock voltage , i just put +0.050 for the little extra.

[Lsdmeasap]

Re: E8600 on X48-DQ6

Looks good!! So it was the ram before then that was faulty right?

[Rivenson]

Re: E8600 on X48-DQ6

Probably ... plus that i didn't count correctly motherboard's vDrop/Vdroop for E8600 and i had put little vcore (per your suggestion) for the needs of 3.8-4 Ghz.
Also as i mentioned earlier i don't know what changes came with F8e (besides newer AHCI) but surely it's a stable one......

[Lsdmeasap]

Re: E8600 on X48-DQ6

Well I am glad you like the BIOS, newer is good I think! I always run the latest beta and never the finals

So I suggested not enough Vcore? If so, sorry about that, I musta overlooked something and not told you that you may have needed more.