Gigabyte GA X38 DQ6 Blog

[supershanks]

okay, so i'm back to 3.0GHz trying to figure out exactly what the problem is. so i currently have a ratio of 5 to 8. which seems to remain the same when i go to 355x9. is it better to have a ratio or 1 to 1 or do i want to go a lil higher than 1:1?

well 5:8 gives faster ram than 1:1 for same FSB so if it's stable than that's cool.

[Zenoth]

supershanks, so you say that increasing the PCI-E frequency and voltage should help for stability? But wouldn't it potentially damage my graphics card permanently? And why would that setting influence on the Memory for example? I wouldn't want to risk anything.

[supershanks]

No it won't damage your board, but it's up to you if you don't want to do it , it's fine.
It's a long used tip from when I 1st got my board back in october , there where 2 posts from the same guy, Alcibiades on Xtreme, who set me on the way with the board:-
Alcibiades Handy Hints #1
Alcibiades Handy Hints #2

luck

[lokeey]

since i can't seem to find a stable clock with 355x9, should i attempt maybe 400x8? do you think this would make a difference? or is this something that is not recommended?

[supershanks]

should i attempt maybe 400x8?

yep no reason why not

Luck

[Zenoth]

Alright supershanks look at this, I almost got it right for 3.6Ghz...

MB Intelligent Tweaker (MIT)
Robust Graphics Booster: Auto
CPU Clock Ratio: 9x
CPU Host Clock Control: Enabled
CPU Host Frequency(Mhz) : 400
PCI Express Frequency (Mhz) : 100
C.I.A.2: Disabled
Performance Enhance : Standard
System Memory Multiplier (SPD) : 2.66D (3:4)
Memory Frequency(Mhz): 1066
DRAM Timing Selectable : Manual

Standard Timing Control
CAS Latency Time: : 5
DRAM RAS# to CAS# : 5
DRAM RAS# Precharge : 5
Precharge Delay(tRAS) : 18

Advanced Timing Control
ACT to ACT Delay(tRRD) : 4
Rank Write to READ Delay : 4
Write To Precharge Delay : 8
Refresh to ACT Delay : 62
Read To Precharge Delay : 4
tRD : 8
tRD Phase Adjustment : Auto
Command to #CS Timing: : 2

Clock Driving and Skew Control
CPU/PCIEX Clock Driving Control : 800mv
CPU Clock Skew Control : Normal
(G)MCH Clock Skew Control : Normal

System Voltage Control : Manual
DDR2 OverVoltage Control : +0.15v (2.00v)
PCI-E OverVoltage Control : Normal
FSB OverVoltage Control : +0.20v
(G)MCH OverVoltage Control : +0.175v
MCH Reference Voltage Control : Normal
DDR Reference Voltage Control : Normal
DDR Termination Voltage Control : Normal
CPU GTLREF1 Voltage* : Normal
CPU GTLREF2 Voltage* : Normal
CPU Voltage Control : 1.30000v (1.280v in CPU-z)
Normal CPU Vcore : 1.22500v

With these settings my Orthos Blend test lasted 5 hours and 33 minutes before it failed, so far it has been the longest test time I've reached before a failure occurred, so I must be near stable settings. From my observations I've noticed that the Refresh to ACT Delay setting is greatly influencing on the overall stability. When I reduce that setting to less than 60 then none of my Prime95 or Orthos tests (Blend, with the Memory being tested of course) would last longer than 2 hours, or less. Above 61 it lasts for 3+ hours, at 62 like I've set it to, it lasted as I said a little more than 5 hours.

When I set everything to the factory defaults the BIOS tells me that Refresh to ACT Delay is set to 68, and that the four main timings tCL, tCRD, tRP and tRAS) are set to 5-7-7-24, however the Memory's frequency is properly set to 1066, but at default it doesn't run at a 1:1 ratio (usually it runs at 5:4). So I have no choices but to set everything manually. The more I play with the settings and the more I'm starting to believe that the actual Memory sticks themselves are physically in good shape and far from being faulty (although it still remains a possibility), and that my instability issues might just well come from the BIOS or just the Motherboard in general not being entirely compatible with the Crucial Ballistix modules. Because after all the actual Crucial brand is not listed at all in the officially Gigabyte's supported memory modules list for the GA-X38-DQ6 on their web-site.

As I type this I'm trying with the following settings, still for 3.6Ghz...

MB Intelligent Tweaker (MIT)
Robust Graphics Booster: Auto
CPU Clock Ratio: 9x
CPU Host Clock Control: Enabled
CPU Host Frequency(Mhz) : 400
PCI Express Frequency (Mhz) : 100
C.I.A.2: Disabled
Performance Enhance : Standard
System Memory Multiplier (SPD) : 2.66D (3:4)
Memory Frequency(Mhz): 1066
DRAM Timing Selectable : Manual

Standard Timing Control
CAS Latency Time: : 5
DRAM RAS# to CAS# : 5
DRAM RAS# Precharge : 5
Precharge Delay(tRAS) : 18

Advanced Timing Control
ACT to ACT Delay(tRRD) : 4
Rank Write to READ Delay : 4, now set to 10
Write To Precharge Delay : 8, now set to 10
Refresh to ACT Delay : 62, now set to 68
Read To Precharge Delay : 4
tRD : 8
tRD Phase Adjustment : Auto
Command to #CS Timing: : 2

Clock Driving and Skew Control
CPU/PCIEX Clock Driving Control : 800mv
CPU Clock Skew Control : Normal
(G)MCH Clock Skew Control : Normal

System Voltage Control : Manual
DDR2 OverVoltage Control : +0.15v (2.00v), now set to +0.25v (2.10v)
PCI-E OverVoltage Control : Normal
FSB OverVoltage Control : +0.20v
(G)MCH OverVoltage Control : +0.175v
MCH Reference Voltage Control : Normal
DDR Reference Voltage Control : Normal
DDR Termination Voltage Control : Normal
CPU GTLREF1 Voltage* : Normal
CPU GTLREF2 Voltage* : Normal
CPU Voltage Control : 1.30000v (1.280v in CPU-z)
Normal CPU Vcore : 1.22500v

I hope it remains stable forever and not crash at all.

But you know what's even more weird in all this? It's the fact that while it was stress testing in Orthos Blend yesterday I could launch Steam and play Team Fortress 2 without a single crash (and TF2 usually uses about 75% to 90% of a single CPU Core). I also let my P2P program run in the background. Later I could play Diablo II and then later I even played Age of Empires III. Not a single one of the games I've tried ever crashed once... yet Orthos failed. I am aware that no games out there stresses Memory and CPU like Orthos or Prime95 do, but still... I'm not sure about what could be wrong but if it ain't the Memory only then I don't know what it is.

[supershanks]

What size fft did it fail on ??
The dram voltage looks liker the most likely suspect. the 8500 is rated at 2.2v the jded ddr2 std is 1.80v so DDR2 OverVoltage Control : = 2.2-1.8=0.4v

Robust Graphics Booster: Auto

= Turbo

Precharge Delay(tRAS) : 18

12 should be ok probably 10

Refresh to ACT Delay : 62

30 ~ 35 should be ok

Write To Precharge Delay : 8

8 as now ~6

CPU/PCIEX Clock Driving Control : 800mv

1000mv

Keep up the good work your getting there

[Zenoth]

Ok, the second settings in my previous post gave me 3 hours and 29 seconds before it failed.

I'm now testing the following:

MB Intelligent Tweaker (MIT)
Robust Graphics Booster: Turbo
CPU Clock Ratio: 9x
CPU Host Clock Control: Enabled
CPU Host Frequency(Mhz) : 400
PCI Express Frequency (Mhz) : 100
C.I.A.2: Disabled
Performance Enhance : Standard
System Memory Multiplier (SPD) : 2.66D (3:4)
Memory Frequency(Mhz): 1066
DRAM Timing Selectable : Manual

Standard Timing Control
CAS Latency Time: : 5
DRAM RAS# to CAS# : 5
DRAM RAS# Precharge : 5
Precharge Delay(tRAS) : 18

Advanced Timing Control
ACT to ACT Delay(tRRD) : 4
Rank Write to READ Delay : 10
Write To Precharge Delay : 8
Refresh to ACT Delay : 68
Read To Precharge Delay : 4
tRD : 8
tRD Phase Adjustment : Auto
Command to #CS Timing: : 2

Clock Driving and Skew Control
CPU/PCIEX Clock Driving Control : 1000mv
CPU Clock Skew Control : Normal
(G)MCH Clock Skew Control : Normal

System Voltage Control : Manual
DDR2 OverVoltage Control : +0.40v (2.25v)
PCI-E OverVoltage Control : Normal
FSB OverVoltage Control : +0.20v
(G)MCH OverVoltage Control : +0.175v
MCH Reference Voltage Control : Normal
DDR Reference Voltage Control : Normal
DDR Termination Voltage Control : Normal
CPU GTLREF1 Voltage* : Normal
CPU GTLREF2 Voltage* : Normal
CPU Voltage Control : 1.30000v (1.280v in CPU-z)
Normal CPU Vcore : 1.22500v

[Zenoth]

Well, bad news, it failed once more, and it took 1 hour and 37 minutes.

http://img521.imageshack.us/img521/646/orthostc0.png

I don't know what to do anymore. I think I'll just leave it as is, because my games don't crash, and eventually I will go buy new memory modules from another brand, perhaps Corsair or OCZ, but I won't touch Crucial anymore I don't even want to RMA them. I've increased the vCore from 1.30000v to 1.33500v, giving me 1.312v in CPU-z, hoping that the Orthos failure was related to a lack of voltage (although I highly doubt it, considering some people can each 3.6Ghz on stock voltage.

My last theory is that the Power Supply just can't do it (Antec NeoHE 550).

[Vbs]

Hi all!

This is my first post. I've bought this week my new cpu/mobo/ram and after throughly reading most of this huge thread, I'd like to share with you guys my o/c so far. Many thanks to supershanks and dynapl (we have the same memory kit ) for their informative and helpful posts!

CPU: Intel Core 2 Quad Q6600 G0 @ 3.6 GHz (BX80562Q6600SLACR, Version #: E10342-001)
Cooler: Thermalright Ultra-120 eXtreme + Scythe "S-FLEX™" S-FDB 120mm (SFF21F, 1600rpm) [review]
Motherboard: The one and only GA-X38-DQ6
RAM: G.Skill 4GB (2x2GB) PC2-8000 (F2-8000CL5D-4GBPQ), CL5-5-5-15 2.0~2.1V
PSU: Enermax Liberty 500W (ELT500AWT) (has 2 +12V rails of 22A each)

My overclock so far has been 7+ hours stable in Prime 25.6 (on Vista x64) and 5+ hours Memtest x86+ 2.0!
Temps: 29ºC idle / 50ºC full load (Prime 4 cores, In-place large FFTs)

------------- BIOS Settings -----------------------------
MB Intelligent Tweaker (MIT) BIOS F5h
Robust Graphics Booster: [Turbo]
CPU Clock Ratio: [9X]
CPU Host Clock Control: [Enabled]
CPU Host Frequency(Mhz) : [400]
PCI Express Frequency (Mhz) : [100]
C.I.A.2: [Disabled]
Performance Enhance : [Extreme]
System Memory Multiplier (SPD) [2.66D]
Memory Frequency(Mhz): 800 1064
DRAM Timing Selectable : [Manual]
********Standard Timing Control ***********
CAS Latency Time: 5 [5]
DRAM RAS# to CAS# : 5 [5]
DRAM RAS# Precharge : 5 [5]
Precharge Delay(tRAS) : 15 [12]
******** Advanced Timing Control ***************
ACT to ACT Delay(tRRD) : 3 [1]
Rank Write to READ Delay : 3 [3]
Write To Precharge Delay : 6 [3]
Refresh to ACT Delay : 52 [45]
Read To Precharge Delay : 3 [1]
Static tRead Value : 8 [6]
Static tRead Phase Adjust : 5 [Auto]
Command Rate (CMD) : 2 [2T]
******** Clock Driving and Skew Control **********
CPU/PCIEX Clock Driving Control : [1000mV]
CPU Clock Skew Control : [Normal]
(G)MCH Clock Skew Control : [Normal]
System Voltage Control: [Manual]
DDR2 OverVoltage Control : [+0.25V]
PCI-E OverVoltage Control : [+0.10V]
FSB OverVoltage Control : [+0.10V]
(G)MCH OverVoltage Control : [+0.025]
MCH Reference Voltage Control : [Normal]
DDR Reference Voltage Control : [Normal]
DDR Termination Voltage Control : [Normal]
CPU Voltage Control : [1.41250V]
Normal CPU Vcore: 1.26250

Memory timings are the most tight possible without crashes on boot or errors in Memtest

Things of note:
1) Disabling CPU Smart FAN Control is helpful on trying to get the right Vcore, to make sure the CPU is getting 100% cooled (you can Enable the fan control afterwards and use EasyTune to adjust the fan speeds)
2) BIOS F5h seems to be using "Loadline Calibration" Enabled by default (this setting doesn't even appear in this BIOS, only in BIOS F7). This rather "controversial" setting helps reduce Vdroop, and could be the reason why I had a stable o/c at this Vcore with BIOS F5h (I couldn't get stable in F7 with the same settings above and with Loadline Calibration on Disabled).
3) If you're gaming on Vista x64, get 4GB of RAM, the speed increase was noticeable in 2006, imagine now...

Cheers,
Vbs