Q9450 + Asus Striker II Extreme + Raid0 OC

[Shanara Red]

Hi everyone.

I have been an avid PC enthisiast for the last 20 odd years. After following the OC scene with a bit of awe and fear, I decided, that I will delve into the OC waters with the new setup I just bought. After two months of info gathering and manual/review/Forum reading, I decided for the following:

Intel Q9450 @ 2.66GHz / Asus STRIKER II EXTREME / 2x2Gb Mushkin Ascent Extreme DDR3 - 1600 / XFX 260GTX, 896 / Thermalright 120 TRUE + Scythe Slipstream 1200 / 2x 130 Gb WD Raid! / 1x WD 640 Gb / 750 TX Corsair PSU / Lite-On Optical Drive / Lian Li PC A-77

It seemed like a solid set-up, with the expensive memory and case being cherries on top.

I decided on the Q9450 because of it's supposed overclockability and the 12mb of cache.

Now. After reading a wondrous amount of forums, guides, reviews and some more forums. I decided to give it a go. And found Clunk's Forums to be the best reference by far. Especially CrapCrasher's thread seemed to concern my setup.

And so my OC ode begun. Sadly with very dissapointing result thus far.

After putting everything togheter, I first flashed the BIOS to the newest available version and installed the RAID drivers for the 2x 130 Gb WD's (I wanted RAID0 for the faster HDD performace, as I often have to save huge CAD, .ai and .psd files often while I work (and then copy them to the larger 'backup' drive at the end of the work session)). After a bit of hassle because of the missing floppy drive, I managed it just fine, and installed Win XP (SP 1) and all other necessary drivers (GPU, etc ...). I followed p with cpu testing software (CPUZ, RealTemp, Prime 95, Sandra, etc ...)

Then off I go.

The goal was 3.2 GHz. And stable. And I failed. Miserably.

After reading CrapCrasher's posts (and MANY other posts and guides allover the web), I figured a 3,2 GHz is not overally ambitious and should be stable, without many problems.

So I decided to OC my cpu first. I started off, by testing my stock settings, and everything works fine. I should add, that I left everything on auto, except the memory (1600, 7/7/7/20/1T @ 1.80V). It all tested as rock solid for 12 hours.

Ok ... I figured. Let's fire her up. But first, I need to find all the Voltage values I require and set the MANUALLY. Here's what I got mostly from the MB's [Auto] settings, but I then entered the values manually. At these settings, everything's peachy after 6 hours of prime 95.



BIOS: 0901

-= Extreme Tweaker =-

CPU Level Up: [Auto]
Memory Level UP: [Auto]

Ai Overclock Tuner: [Manual]
CPU Multiplier: [8.0]
FSB - Memory Clock Mode: [Unlinked]
FSB (QDR), MHz: [1333]
MEM (DDR), MHz: [1600]

LDT Frequency: [x5]
PCIE Bus, Slot 1&2, MHz: [100]
PCIE Bus, Slot6 3, MHz: [100]
SPP <-> MCP Ref Clock, MHz: [Auto]
SLI Ready Memory: [Disabled]


-Memory Timing Setting-

P1: [Auto]
P2: [Auto]

tCL (CAS Latency): [7] -> 7
tRCD: [7] -> 7
tRP: [7] -> 7
tRAS: [7] -> 20
Command Per Clock: [1T] -> 1T

Advanced Memory Settings

tRRD: [Auto] -> 4
tRC: [Auto] -> 27
tWR: [Auto] -> 10
tWTR: [Auto] -> 14
tFAW: [Auto] -> 21
tRD: [Auto] -> 9
tRTP: [Auto] -> 4
tRFC: [Auto] -> 59
tREF: [Auto] -> 7,8uS


-Over Voltage-

CPU Voltage: [1.25V]
Loadline Calibration: [Disabled]
CPU PLL Voltage: [1.53V]
CPU VTT Voltage: [1.14V]
Memory Volatge: [1.80V]
NB Core Voltage: [1.32V]
SB Core Voltage: [1.5V]

CPU GTL_REF0 Ratio: [Auto] (when tweaking any of these values, I use generally +/- 67% of CPU Vtt value)
CPU GTL_REF1 Ratio: [Auto]
CPU GTL_REF2 Ratio: [Auto]
CPU GTL_REF3 Ratio: [Auto]

NB GTL_REF Ratio: [Auto]

DDR3 CHA1 Ref Voltage: [Auto]
DDR3 CHB1 Ref Voltage: [Auto]
DDR3 CHA2 Ref Voltage: [Auto]
DDR3 CHB2 Ref Voltage: [Auto]


-CPU Configuration-

CPU Type: Intel (R) Core(TM)2 Quad CPU
Q9450 @ 2.66 GHz

CPU Speed: 2.66
Cache: 6144K x2

CPU Internal Thermal Control: [Auto]
Limit CPUID MaxVal: [Disabled]
Enhanced C1 (C1E): [Diasbled]
Execute Disable Bit: [Enabled]
Virtualisation Technology: [Disabled]
Enhanced Intel SpeedStep(tm) Tech. : [Disabled]

CPU Core 2: [Enabled]
CPU Core 3: [Enabled]
CPU Core 4: [Enabled]


-Spread Spectrum Control-

CPU Spread Spectrum: [Disabled]
PCIE Spread Spectrum: [Disabled]
SATA Spread Spectum: [Disabled]
LDT Spread Spectrum: [Disabled]

(+++ NOTE: Please feel free to use as a template for future Striker II Extreme OC's +++)

...

I then upped the FSB to 1600 (meaning a Linked and Synced CPU:Memory and 3.2 GHz Core speed). As I expected, I did not work. So after playing around with the settings and some CMOS resets, I entered CrapCrashers values. And I used these values as 'ideal' guidelines and went on from there. Basically, I did 27 iterations of the V values using the afore mentioned numbers as guidelines. I tried to achieve 3.2 GHz by changing/using the following values:

V Core: 1.25V up to 1.337V
PLL: 1.52V
VTT: 1.13V up to 1.28V
NB: 1.32V up to 1.44VV
(I always changed the GTL_REF's according to the VTT)

The rest of the setting were as above in the template post.

Not once could I get the system to boot. I actually got past the POST test, only to get the 'NTLDR not found, press any key to reboot' This was at the last two sets of Voltage values I OC'd to:

Vcore: 1.331V and 1.337V
PLL: 1.52V
VTT: 1.28V
NB 1.44V

After raising the NB to 1.44 (I figured the NB which is the link to the SB was underpowered due to RAID), and still no result, I gave up on this methodology. It just wasn't working, and the voltages were getting WAY past CrapCrasher's values for 3.2 GHz, and some exceeded his 3.4GHz OC. (For Reference: CrapCrasher Voltages @3.4 stable: Vcore: 1.337V , PLL:1.510V , VTT:1.250V , NB:1.367V )

So I unplug my RAID hdd's and install XP on the single 640 Gb hdd. Boots without problems at stock and is stable for 12 hours.

I then figured I would give the 'CPU level Up' MB function a go. It allowed a 3.2 GHz CPU overclock with everything on [auto] except the memory.It was stable for 15 mins under Prime, then the CPU crashed. Here are the MB's own settings for the 3.2GHz OC (found in BIOS under 'Hardware Monitor' / 'Voltage Monitor'):

Vcore: 1.31V
PLL:1.52V
VTT: 1.24V
NB: 1.32V

All the values are LOWER than what I had already tried! Ok, I figure. I'm over-volting some components. I go back to BIOS, set those same 'CPU Level-up' values manually (even try Vcore at 0.3V higher because of Vdroop) for the 3.2 OC. Nothing. No boot.

Figuring I may be too ambitious, I then repeat the whole 'manually-alter-Voltages-from-low-to-high'' process again for FSB 1400 (2.8 GHz). The whole shabang (Highest Vcore: 1.337V). Nothing. No boot.

I then go back to Clunk's P5K and OC for beginners guides. I leave EVERYTHING on [auto] (even the otherwise stable OC'd memory in one attempt) except the FSB. I set it to 1400. Which is 77MHz over stock. No boot. I raise the Vcore, PLL and VTT (and GTL accordingly) by single increments. I raise each of the values (first V Core then PLL and then VTT) incrementally 5 times. And still nothing. No boot.

I am completely lost. I have tried to OC agresivelly, by the book, and by-the-beginners way. Please, if someone has bothered to read this wall o' text, let me know if I have screwed up anywhere, and which steps I should retrace. I uninstalled the RAID0, and am now working on a single boot hdd of XP. Just to eliminate any excess 'random-error' factors.

I truly do not know where I have gone wrong, and what to try next. If any one of you has any advice, I would be VERY happy to read it, as after a few days work I'm still nowhere, and am beginning to lose hope.

Thank you,

-SR

P.S.- The system would and still ALWAYS boots at stock values (either [Auto] or entered manually), that you see on the BIOS template-post above.

[Clunk]

Welcome to the forums

Unfortunately, you have picked some of the most difficult components to overclock, at least for a beginner anyway.

The 790i chipset is a known pig with many issues, so that is something to be aware of.

The thread you linked to of crapcrashers is for the DFI X48 T2R which is an Intel chipset based board and is DDR2, so it's no wonder you aren't having much luck there!

There are no guides on this site that reference the 790i, simply because it wasn't worth doing...but that's not what you want to hear

What I would suggest is starting a fresh.

Clear the CMOS and start up with one stick of memory, If you have optimised defaults, load those and then change the memory voltage and timings to whatever they are rated for and put a fan over them., load memtest86+ v2.xx and run test 5 on each stick individually for about an hour each at stock speeds.

If you see errors, put that stick of memory to one side and test the next.

Let us know how that goes first and we can go from there.

[Shanara Red]

Mr Clunk, (had to say it )... thank you for a swift reply. Both sticks check out without errors @ 1600 MHZ, 7-7-7-20-1T, 1.78V ... with the rest of the settings on stock.

I realise there are differences between ddr2 and ddr3 boards, but how and why would that influence cpu OC'ing? Well, I suppose the answer is not really neccesary, as I realise the whole MB-NB-etc setup is in fact interconnected and at some level it's bound to influence the whole thing. I'm not and electro-current engineer :P.

Anyway. The sticks are ok. Where do I go from here? Follow the basics guide again?

EDIT: Note: There is no test 5 in memtest86 v3.4. Under 'select tests' there are only options 1-4 for testing. Options: 1-Default tests, 2-Skip current test, 3-Skip Test, 4-Select Bit Fade test, 0-Cancel. I therefore ran the sticks under option: 1-Default tests.

[Clunk]

It's good that the memory checks out ok

It's not not just the fact that the DDR2 and 3 boards are different, your's is an Nvidia chipset which works in a completely different way to Intel chipsets.

As I say, there are no guides on here for your board, so se can help you, but it's a bit of a "test it and see" kind of scenario because I don;t have a 790i board here.

Also, you have the wrong version of memtest. I said, memtest86+ v2.xx - you have memtest86 v3.4 - You need the one I said and to run test 5.

Let us know when you've done that bit

[Shanara Red]

Ok. I need to repost the above post.

I ran a second set of tests, just while waiting for some futher replies. Please ignore the above post. The results I'm getting are not looking promising anymore. Memtest seems to freeze up (the Pass and Test values stay at a fixed value), although some numbers are still being run by memtest, and I figured for the first two tests in the posrs above, that that was ok, so I reported no errors. I then saw, that the screen had frozen up, except some running numbers. The WallTime timer seems stuck at 0:02:36. There are no errors reported on this (the topmost) table. I am however getting a continous count of errors (from 0-32767 or 0-32766) on the 'fourth' test line under the first table: Max Contigouous Errors: 1

my remarks are in (...)


Memtest-86 v 3.4 : Pass 6% ##
Pentium III 2667 MHz (+pentium 3???+) : Test 27% ###########
L1 cache: 64K 43716 MB/s : Test #4 [Moving inversions, random pattern]
L2 cache: Unknown : Testing 120k - 2047M 2047M
Memory: 2047M 17429 MB/s :Pattern d554da23
Chipset: (+nothing written here+)

Wall Time Cached RsvdMem MemMap Cache ECC Test Pass Errors ECC errors
--------- ------- --------- -------- ------ ---- ---- ----- ----- ------
0:02:36 2047M 276M e820-Std on off Std 0 0

-----------------------------------------------------------------------------------------------
Error Confidence Value: Test Errors
Lowest Error Adress: 00000000009 - 0.0 MB 0 0
Highest Error Adress: 000ffffffffb - 4095.9 MB 1 0
Bits in Error Mask: ffffffff 2 0
Bits in Error - Total: 32 Min:2 Max:32 Avg: 16 3 0
Max Contigous errors: 1 4 0-32766
ECC Correctable errors: 5 0
Errors per Memory Slot: 0 : 23143567 (+and counting up!+)


Well yes it's messy, but it's the best I can do ... I've found no Memtest template.

So it would seem, that the stick gets stuck at the 4th test (when the bunch of values and numbers appear below the top table), and although the program seems to be running, the % of the test remain the same.

I have also insterted the same stick into the B1 slot. Exactly the same results.

I upped the voltage to 1.86. Same results.

I then tried the same stick under 1.9V which is the max ref voltage of the manufacturer for this speed (1600 / 7-7-7-20 @ 1.9V). In the A1 slot again. Same results.

I then went on to the second stick.

Again I first tried 1600 / 7-7-7-20 @ 1.86V.


Both Stick were run under these conditions in A1 and B1 slots at all times, the cpu ran at stock and unlinked to the memory:

1st: Manually set vcore:1.25v /PLL: 1.52v /VTT: 1.18v /nb: 1.32v (the last of which could be a problem I suppose? Even tough Prime ran on it without probs for 12 hours at stock?!?!) DDR3 @ 1.86V. +Errors only.

2nd: Manually set vcore/PLL/VTT/nb DDR3 @ 1.90 V. Errors only.

3rd: [Auto] set everything except the memory overlclocks (1600 7-7-7-20 @ [Auto]V) Same results. One of the sticks (Stick 01) errored after 2:30 mins at the exact same point as all the other tests in both slots. Stick 02, actually made it to the past the 5th test and into the 6th! It froze up at 0:07:51 (Test 6) with NO numbers appearing below the first table. I re-ran the test and it made it to 0:09:27, froze up at 45% of test #6.

So this is what I have so far. When I come back home, I will re-run all the tests at the 'fourth experiment' values:

4th: Everything left to [Auto] - Deafult speeds that MB reads for RAM's - 1066 x-x-x-xx @ xxv. I figured ... this has to work, otherwise the Prime blend test would have found these errors. I will post in about 2hrs time.

[Shanara Red]

lol ... I had the post screen open for about two hours now, and I didn't see you post. I feel like such a beginner. Ok. When I return home, and re-do all the test in memtest 2.xx like you instructed I will post again. :/

[Clunk]

It's a common mistake, don't worry

Don't forget to cool that RAM too.

[Shanara Red]

Ok. I will be restarting testing in a few minutes.

But how is it possible that the sticks failed so many on the memtest 86 v 3.4 tests? It wouldn't have somehing to do with the fact, that Memtest's v3.4 tests are NOT optimised for ddr3?

Plus ... is the methodology I'm using correct? Am I on the right track?

[Clunk]

The V2.xx is optimised for all the latest memory, so use that.

No, your methodology is wrong because you were on the wrong version of the program

Let us know how you get on with the proper version.

[Shanara Red]

Originally Posted by Clunk

The V2.xx is optimised for all the latest memory, so use that.

No, your methodology is wrong because you were on the wrong version of the program

Let us know how you get on with the proper version.

Ok. Went back to sqaure 1. Manual (stock)settings with boot capability. It's looking great atm. Using Memtest86+ v2.01. I have ran test 5 for an hour and blend tests are currently being run at 1600 7-7-7-20-1T @ 1.86v. No errors.

CPU manual settings are:

Vcore: 1.25v
PLL: 1.53v
VTT 1.13v (and calculated GTL_REF's)
Mem Voltage: 1.86v
NB Voltage: 1.32v
SB Voltage: 1.50v

Am I correct in assuming, that no matter what will happen further on in while testing, the memory WILL NOT be the reason for failed overclock if left at the current settings which seem to run fine under memtest 86+?

If this is so, I only have to test various FSB speeds, linked/unlinked/synced mem setups, Vcore, PLL v, VTT v (and GTL's), NB v, SB v?

I'll rephrase that. If memory checks out like it's looking to, is it safe to assume, that any further overclocks with the memory running at these specs (1600 7-7-7-20 @ 1.86v) will NOT fail, because of the bad RAM settings?