Core 2 system...**EDIT** GOT Q9650

[badnewcastle]

Ok, I'm getting alot closer. I had to bump the GTL's up to .70x for (0/1) and (2/3). And .69x for GTL NB Ref.

And good god is this thing fast @ 4.03 Ghz! I hope I found stability, if it's not stable I think I'm getting closer. I could just tell the way that it booted into Windows all snappy and quick like.

Running 4.03Ghz|FSB @ 445|vCore @ 1.368|VTT @ 1.25|vNB @ 1.3 Ram @ 891 5-5-5-15 2T

@ Chizow, I did go back and play with the FSB I took it up to 460 w/multi @ 6 and stopped playing with it because that's higher then I want to be and I was running my old E6750 @ 3.6 = 450 FSB. Thank you for your sound advice at the very least you have reminded me to go back and check my details.

 

[EXman]

Originally posted by: Krakn3Dfx

Originally posted by: techmanc

Originally posted by: nyker96
from dual to quad? of course you will feel the differences. esp if you do any encoding.

I just love these wide open statements with and actually proof but guess 4 of anything is better 2. It does help sell computer upgrades tho

I use Adobe Premiere Elements, and going from 2 to 4 cores was HUGE.

Damn I just built a Core 2 with E8400....

Seriously considering dumping that for a Quad as i also use premiere.

 

[badnewcastle]

I think I'm over trying to hit 4Ghz or even 3.8 at this point. I'm very disappointed in either my motherboard or the q9650... I don't which is causing the problem but I expected to hit 4Ghz and possibly run 3.8 24/7... But nope I'm stuck at 3.6. It ran stable for a while under prime and occt at 4 and 3.8 but eventually errors out.

I guess I'm going to try and drop form 8GB of ram to 4GB and see what if maybe the extra 4GB's is just putting too much pressure on the NB.

 

[elconejito]

did you make sure you could run 460 (or even 450) FSB with the quad and a low multi? That way you can at least determine if it's your CPU or the mobo that's holding you back.

 

[badnewcastle]

Originally posted by: elconejito
did you make sure you could run 460 (or even 450) FSB with the quad and a low multi? That way you can at least determine if it's your CPU or the mobo that's holding you back.

Yea I did and for whatever reason we were good up to 460.

 

[n7]

Getting quads truely stable over 440 FSB on P5Qs can be lots of fun

It's doable though, you just need to have a lot of patience, & be willing to fight with GTLs a lot.

Getting your VTT + corresponding GTLs & NB + GTL bang on is absolutely imperative, or you will not achieve stability.
Getting these just right can be a ton of hassle though, so if you don't have the patience, you'll likely want to just be happy with a lower FSB.

Here are what i run 24/7 with my P5Q-D + Q9650 + 8 GB Mushkin 996580:
http://ense7en.com/pics/q9650/...%20passes%20STABLE.JPG

AI Overclock tuner: Manual
CPU Ratio Setting: 9
FSB Strap to North Bridge: AUTO
FSB Frequency: 445
PCI-E Frequency: 100
DRAM Frequency: 890
DRAM CLK Skew on Channel A1: AUTO
DRAM CLK Skew on Channel A2: AUTO
DRAM CLK Skew on Channel B1: AUTO
DRAM CLK Skew on Channel B2: AUTO
DRAM Timing Control: Manual

1st Information :

CAS# Latency: 5
DRAM RAS# to CAS# Delay: 5
DRAM RAS# Precharge: 5
DRAM RAS# Activate to Precharge: 15
RAS# to RAS# Delay : AUTO
Row Refresh Cycle Time: 55
Write Recovery Time: AUTO
Read to Precharge Time: AUTO

2nd Information :

READ to WRITE Delay (S/D): AUTO
Write to Read Delay (S): AUTO
WRITE to READ Delay (D): AUTO
READ to READ Delay (S): AUTO
READ to READ Delay (D): AUTO
WRITE to WRITE Delay (S): AUTO
WRITE to WRITE Delay (D): AUTO

3rd Information :

WRITE to PRE Delay: AUTO
READ to PRE Delay: AUTO
PRE to PRE Delay: AUTO
ALL PRE to ACT Delay: AUTO
ALL PRE to REF Delay: AUTO
DRAM Static Read Control: Disabled
DRAM Read Training: Disabled
MEM. OC Charger: Enabled
AI Clock Twister: Lighter
AI Transaction Booster: Manual

Common Performance Level [10]

Pull-In of CHA PH1: DISABLED
Pull-In of CHA PH2: DISABLED
Pull-In of CHA PH3: DISABLED
Pull-In of CHA PH4: DISABLED
Pull-In of CHB PH1: DISABLED
Pull-In of CHB PH2: DISABLED
Pull-In of CHB PH3: DISABLED
Pull-In of CHB PH4: DISABLED

CPU Voltage: 1.2375
CPU GTL Voltage Reference(0/2): 0.645x
CPU GTL Voltage Reference(1/3): 0.645x
CPU PLL Voltage: 1.54
FSB Termination Voltage: 1.26
DRAM Voltage: 1.8
NB Voltage: 1.24
NB GTL Reference: 0.65x
SB Voltage: 1.1
PCIE SATA Voltage: 1.5

Load-Line Calibration: Enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew: auto
NB Clock Skew: auto
CPU Margin Enhancement: Compatible

Of course, what works for one person may not for the next...you have to find your own mix of settings most of the time.

Here's 9x467 stable...voltages/settings are listed in the screenshot's name.
http://ense7en.com/pics/q9650/...5%20Blend%20STABLE.JPG

I didn't have voltages perfected then, but really, to get from 9x445 to 9x467 takes a lot of extra motherboard voltage that i didn't consider worth it, especially as i can run 4 GHz at such nice low motherboard voltages.

 

[badnewcastle]

Thanks N7... I've seen some of your posts while searching for this. I will keep firing at it. I guess one thing I'm concerned with is that my core voltage needs to be so high... I think if I can work with the GTL's I'll probably be able to work it down.

 

[spinejam]

Originally posted by: badnewcastle
Thanks N7... I've seen some of your posts while searching for this. I will keep firing at it. I guess one thing I'm concerned with is that my core voltage needs to be so high... I think if I can work with the GTL's I'll probably be able to work it down.

try to utilize templates like n7's / others to narrow down your settings. getting the gtl's worked out will definitely help you w/ lower volts. keep on trying b/c that chip should never run below 4.0ghz!

 

[n7]

A couple things to consider:

8 GB is a bitch to get stable...i love my 8 GB, but it makes things a lot more challenging...
I've been using all slots in mobos for so long now i'm used to extra work though.

Basically, you want to make sure RAM-related issues are completely out of the equation initally.

Make sure you set Clock Twister to Lighter or Light...it can make or break things w/ 8 GB. You can worry about tightening it later.

Set tRD (Performance Level) looser than normal while you tweak things.
Set it to something really loose like 12 till you get the rest of stuff worked out.

Don't try running your RAM @ its rated 1066 or higher for now; 8 GB @ high speeds is very difficult to keep stable when OCing the board heavily.

IOW, run 1:1 for now, underclocked.

Loose primary timings (5-5-5-18), tRFC @ 60+.

If you really want to take the RAM out of the picture, you could indeed drop to 2x2 GB...of course, if you want 8 GB though, that's an unacceptable compromise.

Make notes.

I have notepad files of my overclocks that are probably hundreds of lines long worth of settings i've tried stably or unstably...it allows me to sometimes just stop, look at what worked & what didn't, & make conclusions.
Otherwise i might have made so many changes i forget what caused the instability in the first place.

Read thru this: http://www.xtremesystems.org/f...howthread.php?t=202292
It makes my head hurt, but a basic understanding of what you are doing by changing GTLs helps.

Generally i find there are certain values that work for certain FSBs...for example, my current VTT is 1.26 x 0.645 = 0.8127 & NB is 0.806...around those numbers is good for me for 445ish

For my 465+, i was needing 0.92-0.93 for both VTT & NB.

 

[badnewcastle]

Originally posted by: n7

Read thru this: http://www.xtremesystems.org/f...howthread.php?t=202292
It makes my head hurt, but a basic understanding of what you are doing by changing GTLs helps.

Thank you for that link! I found it from a post you did in another thread (I think) but I couldn't find it again. I downloaded the spreadsheet but without the thread I don't know how to use it. Thanks.

Originally posted by: spinejam
... that chip should never run below 4.0ghz!

This is precisely what I was thinking I'll keep trying.

 

[badnewcastle]

Still working at this... I don't expect to get it until next week sometime at this point.

I can't even get it to post at 445x9 unless I'm pushing 1.3 on vCore.

 

[n7]

Originally posted by: badnewcastle
Still working at this... I don't expect to get it until next week sometime at this point.

I can't even get it to post at 445x9 unless I'm pushing 1.3 on vCore.

What's your VID?

I needed around 1.4v with my previous Q9550 to do 3.95 GHz.

I got really lucky with my Q9650. VID is 1.15, & it does 4 GHz @ 1.24v, 4.25 @ < 1.3v.

You might be less lucky it seems.

 

[badnewcastle]

Originally posted by: n7

Originally posted by: badnewcastle
Still working at this... I don't expect to get it until next week sometime at this point.

I can't even get it to post at 445x9 unless I'm pushing 1.3 on vCore.

What's your VID?

I needed around 1.4v with my previous Q9550 to do 3.95 GHz.

I got really lucky with my Q9650. VID is 1.15, & it does 4 GHz @ 1.24v, 4.25 @ < 1.3v.

You might be less lucky it seems.

VID = 1.1000v w/out Prime95 (400x6)

Currently running 400x9 everything on Auto including vCore (I know Auto over-volts) and here is what I get from Core Temp and CPU-Z w/Prime95 running on all 4 cores:

vCore = 1.312v
VID 1.2500v

This is the fastest I've been able to stabilize.

 

[n7]

Okay, so your VID is 1.25v...not so great unfortunately, which is why you're needing 1.3v or more to do 4 GHz.

But on the plus side, you get nice low temps compared to most of us with low VIDs.

As long as LinPack is staying below 80-85C i wouldn't be too worried...IOW, up your vcore & see what happens

 

[badnewcastle]

I still need to stay 1.4v on the core or less right?

 

[soccerballtux]

1.325 is what is specified on intel site.

 

[aigomorla]

Originally posted by: badnewcastle

Originally posted by: LongTimePCUser
Does the Q9650 system feel faster than your core 2 system?
Was it worth the extra money?

It does feel faster and it's nicer in some games. Worth the $? Only because I had nothing here to spend it on at the time.

I just need to get it stable at 4ghz and it would be definate yes worth the $$.

I think I'm hitting a fsb wall. It's still not stable at 1.38v or 1.4v and I'm not going any higher. Seems like others are getting better results with less vcore.

E0 like low voltage.. there light footed balarina's while the kentsfield are opera fatties.

Try 1.3875 as max... Vtt needs to be around 1.4 which is FSB termination voltage on older boards.

Keep your ram divider low, there isnt a Q9650 class that CANT do 4ghz in its sleep.

 

[badnewcastle]

Originally posted by: aigomorla

Originally posted by: badnewcastle

Originally posted by: LongTimePCUser
Does the Q9650 system feel faster than your core 2 system?
Was it worth the extra money?

It does feel faster and it's nicer in some games. Worth the $? Only because I had nothing here to spend it on at the time.

I just need to get it stable at 4ghz and it would be definate yes worth the $$.

I think I'm hitting a fsb wall. It's still not stable at 1.38v or 1.4v and I'm not going any higher. Seems like others are getting better results with less vcore.

E0 like low voltage.. there light footed balarina's while the kentsfield are opera fatties.

Try 1.3875 as max... Vtt needs to be around 1.4 which is FSB termination voltage on older boards.

Keep your ram divider low, there isnt a Q9650 class that CANT do 4ghz in its sleep.

Mine is an E0, I will go for 1.38 max and see if I can get it worked out.

 

[badnewcastle]

Is there anyway to tell what might be the problem based on which test in Prime95 fails? I fail during the linpak part 800000 FFT length 8k.

Also, put a new fan on my cooler and it dropped the temps pretty well though 1 core seems to be consistently higher then the others...

According to Real Temp I sensor test I get this:
core 0|core 1|core 2|core 3
15 11 11 11

 

[aigomorla]

Originally posted by: badnewcastle
Is there anyway to tell what might be the problem based on which test in Prime95 fails? I fail during the linpak part 800000 FFT length 8k.

Also, put a new fan on my cooler and it dropped the temps pretty well though 1 core seems to be consistently higher then the others...

According to Real Temp I sensor test I get this:
core 0|core 1|core 2|core 3
15 11 11 11

11C?

holy crap you outside in the snow?

 

[badnewcastle]

I think that is the Temperature variance in celcius over the 10 second test... but the curious thing is that the CORE 0 temperature varies greatly compared to the others and the it is usually 3-7 degrees higher then the other cores except during Linpak when all become equal.

 

[aigomorla]

Originally posted by: badnewcastle
I think that is the Temperature variance in celcius over the 10 second test... but the curious thing is that the CORE 0 temperature varies greatly compared to the others and the it is usually 3-7 degrees higher then the other cores except during Linpak when all become equal.

idle is variable load.

while full load is max envelope in which the CPU can dish heat.

You only look at idle to see if you have a mount going. After that it becomes useless.

Then you use load to see how well your mount is by how even your temps are.

Analogy is....

when you test a car for MPG, you wouldnt sit it on idle and count down how long it can go would you? the whole point in idle is to test to see if your engine is making funny noises, then you rip her on the highway to see what your MPG is.

 

[badnewcastle]

I'm still trying at this and I finally got it stable for more then 15 mins last night then Prime crashed (not just failed) at 1 hr. 12 mins with these settings:

1.3750 vCore ~ 1.58 vPLL ~ 1.4 vVTT ~ 2.1 vDRAM ~ .67x GTL (0/2) ~ .67x GTL ~ 1.24 vNB ~ .65x GTL ref ->most stable I've been able to achieve

I've been reading alot trying to figure this thing out... wow it's been tough so far. I've been able to gather these general rules, I believe they are accurate but not totally sure:

1. vPLL needs to be .2-.4v above vCore
2. vVTT needs to be .02-.03v higher then vCore BUT NO HIGHER THEN 1.4v per anand!
3. Higher VTT then lower GTL Ref... (not totally sure about this one)
4. vVTT and vNB should not be that close?
5. vNB I'm lost here...
6. vCore needs to be lower then 1.38 per vVTT max @ 1.4v
7. GTL (0/2) GTL (1/3) don't have to be the same...?

 

[badnewcastle]

I think my core 0 is bad... it constantly fails prime while the res of them fly on through the rest of the test. I left for 9 hours and core 0 failed on the 800000/8k portion of prime like it always does and the other 3 cores fired away without failing... it also is running 3-7c higher then the other cores... constantly.

 

[aigomorla]

Originally posted by: badnewcastle
I think my core 0 is bad... it constantly fails prime while the res of them fly on through the rest of the test. I left for 9 hours and core 0 failed on the 800000/8k portion of prime like it always does and the other 3 cores fired away without failing... it also is running 3-7c higher then the other cores... constantly.

try running it on stock and see if core 0 will fail.