Disappointing Overclock on E7200 & P35 - Any Ideas? (No Newbie Questions Inside :))

[PlumpDJ]

I have the following kit :

PSU : Seasonic S12 500 Watt
MB : ASUS P5K-E WiFi (01/03/2008 BIOS)
CPU : E7200 @ 2.53 / 266 (not sure what batch)
Cooling : Thermalrigjt HR-01 Plus + Sycthe 1200 RPM
MEM : Corasir DDR2 8500 1066

Many people seem to be getting decent overclocks but for whatever reason I cannot seem to get this thing Prime 95 Stable @ 3.5 Gig or more. It will post at 3.8 / 400 but Prime 95 fails straight away.

I've tried :

1) Upping all the voltages one at a time to the "Safe Max". Taking notes as I go.

Vcc = 1.4
FSB Termination= = 1.4
NB = 1.7
SB = 1.55

2) Disabling PCIE / CPU Spread Spectrum / Transaction Booster

3) Manually setting PCIE to 100

4) Dropping the memory down completely so it shouldn't be an issue

5) Full Load Temps are OK @ 55/56 with 1.36 Volts (Voltage drops a little under load but that's normal apparently)

6) +5V line seems stable at 4.94 and drops to 4.92 under full load but doesn't budge from there

If anyone has any ideas or advice with overclocking on the ASUS P35 MB it would be greatly appreciated. Perhaps I just don't have a good overclocking CPU.

 

[myocardia]

You seem to be doing everything right. Are you using 2GB sticks of system RAM? If so, you'll need to raise your Row Refresh Cycle Time to 54 or higher, and make sure it's also set to run @ DDR 533 (1:1, you can try raising it later). You'll also want to give this guide and this guide a good read. Just make sure that you realize the max voltages mentioned in both of them are for 65nm chips, not 45nm's. You'll want to keep your Vcore below 1.40v, and your VTT as low as possible, but definitely no higher than 1.375v for 24/7 use. Good luck, and welcome to anandtech.

 

[error8]

The bios version you have there seems to be very old. There weren't any E7200 in march if I remember correctly.
And keep in mind, for 4 ghz or so, you may have to up the Vcore a lot. My E7200 is at 4 ghz with 1,45 V in bios and 1,42 V cpu-z full load. I was at 1,41 V for 3,9 ghz and 1,37 V for 3,8 ghz. Of course, there are some people that are getting 4 ghz with less then 1,3 V, but those cases are rare.
Your NB will certainly die at 1,7 V. That is way to high. 400-421 FSB should be attainable with 1,4 V or so.

PS: the 1,45 V I'm using, it's way over the safe limit and you should avoid it.

 

[PlumpDJ]

Thanks for your replies. I'll keep trying. It might just be i've got a crap CPU which happens quite regularly apparently.

I will try raising my Row Refresh Cycle Time to 54. My BIOS Version is 1013.

 

[Tempered81]

1. Reset your motherboard to defaults (clear-cmos). Reboot, make sure your bios/firmwares are the most current & up to date.

2. Use this bios: 2008/07/28 P5K-E/WiFi-AP BIOS ver. 1102 - 1007.82 (KBytes)
link: http://support.asus.com/downlo...d.aspx?SLanguage=en-us

3. Set FSB to 422 x 9.5 [4009 mhz] manual
Set Vcore to 1.40v via bios (will be 1.31v-1.36v after droop)
Set Vdimm to either 1.9, 2.0, or 2.1v (depending on your Corsair 8500 spec which i believe to be 2.1v.)
Set Mem timings to 5 5 5 15.
Disable c1e e1st speedstep, tm2, etc.

FYI:
c1e = idle state = less heat/power consumed = less noise
eist = lower multi/cpu clock when full speed isnt needed = less power consumption/less heat
tm2 = lower vcore when the cpu is clocked lower = less power consumed/less heat = less noise

4. Save settings, reboot. Check temps/stability.

If this doesn't work, we can go over other possible settings for NB, GTLREF, VTT, etc. I would leave everything at auto for Clock Over-Charging Mode, Transaction Booster, Load-Line Calibration, FSB Termination Voltage, etc.


Edit: in your first post, i don't see anywhere you've manually set your vdimm. Also, which stability tests are you failing, and could you please provide some 4-tab cpu-z & coretemp/realtemp screenshots

 

[PlumpDJ]

Hi Jared,

Thanks for your detailed response. I will try as you suggest and see If I get any where.

1) My memory is running @ 2.1v (manually set) and well under it's 1066 spec during CPU OC.

2) I'm failing Prime 95.

3) My Vcc hits 1.384v during Full Load and doesn't budge from there.

4) Raising Vcc definietly improves the amount of time that Prime 95 runs before reporting an error but it seems quite high for the Clock Rate I'm getting...

5) Temps do rise up to about 64 after about an hour and a half

 

[PlumpDJ]

Tried what you suggested but wouldn't even POST at that setting. (422 x 9)

I suspect I just don't have a good OCing CPU.

 

[PlumpDJ]

It will post at 400 x 9.5 but fails Prime 95 straight away. Don't really want to boost my Vcc anymore and my past experience with this CPU / MB Combo tells me It will only get to about 3.5 or 3.6 Gig.

 

[mancunian]

Originally posted by: PlumpDJ
Tried what you suggested but wouldn't even POST at that setting. (422 x 9)

I suspect I just don't have a good OCing CPU.

Unfortunately, I think you're right bud. 3.6 seems to be your limit. As pointed out earlier, you have done all the right things.

 

[clarkey01]

Originally posted by: PlumpDJ
It will post at 400 x 9.5 but fails Prime 95 straight away. Don't really want to boost my Vcc anymore and my past experience with this CPU / MB Combo tells me It will only get to about 3.5 or 3.6 Gig.

Mine BSODS after 5 minutes of prime95 but is stable for 5 hours of Orthus (Blend , CPU, Large FFT)

 

[Tempered81]

well looks like it might just need a lot of Vcore. I've read some reviews where sites are maxing out e7200's stable at only 3.6g-3.8g with an astounding 1.45V or greater. Looks like there is no guaranteed 4G with E7xxx series.

1.38v under load only gives you .02 volts to play with before you enter red zone. I would just set that to ~1.4 under load and call it a day with the highest mhz you are able to reach & maintain stability. You upgraded bios & ran 5 5 5 15 timings?

 

[Jyve]

It would seem that the latest batches of 7200 seem to require a major bump in voltage to hit the 3.8, of which I have one. I have to set my vcore to 1.44 or something like that in the bios and after boot/vdroop I'm idle @ 1.392. This is for 400x9.5. My temps @ idle using realtemp are 34/41 and under load something along the lines of 58/65. Seems to be well within an acceptable range for me. If I only get a year or 2 out of the proc then I'm good, but I expect it'll go longer than that. This is on a gigabyte p35-s3g.

 

[Tempered81]

i guess moral of the story is don't expect guaranteed 4ghz out of e7200's.

 

[mancunian]

Originally posted by: Jyve
It would seem that the latest batches of 7200 seem to require a major bump in voltage to hit the 3.8, of which I have one. I have to set my vcore to 1.44 or something like that in the bios and after boot/vdroop I'm idle @ 1.392. This is for 400x9.5. My temps @ idle using realtemp are 34/41 and under load something along the lines of 58/65. Seems to be well within an acceptable range for me. If I only get a year or 2 out of the proc then I'm good, but I expect it'll go longer than that. This is on a gigabyte p35-s3g.

I think you're right. This one is one of the first batches and does 3.8Ghz with just 1.3v. It seems later batches just don't clock as good.

 

[PlumpDJ]

It would seem that the latest batches of 7200 seem to require a major bump in voltage to hit the 3.8, of which I have one. I have to set my vcore to 1.44 or something like that in the bios and after boot/vdroop I'm idle @ 1.392. This is for 400x9.5. My temps @ idle using realtemp are 34/41 and under load something along the lines of 58/65. Seems to be well within an acceptable range for me. If I only get a year or 2 out of the proc then I'm good, but I expect it'll go longer than that. This is on a gigabyte p35-s3g.

That's pretty high isn't it Jyve. I don't think it's worth risking your CPU for a few extra Mghz, which probably doesn't translate into a huge amount in gaming FPS given the CPU is well overspec. I know Intel say ~1.37 on their website and even I'm currently running it at 1.41 volts nervously.

 

[PlumpDJ]

I've updated my BIOS and double checked memory timings so I suspect I've probably just got a bit of a dud OCer.

Oh well so much for the budget OCer... I mean 3.6 is 'OK' but I was expecting 3.8-4 without too much hassle.

 

[M1A]

I think for the few extra bucks the 8xxx series are better.....

 

[Jyve]

As it stands now, you are probably right...more headroom on an e8400 I'm sure. Probably a few months ago I would have disagreed as the e7200's were just beasts. As for the high voltage, yeah, it is a bit on the high side, but like I said...the temps all look good to me. I can't imagine that this thing would die out on me in 6 months but who knows. And as for the difference in 3.6 and 3.8, you're right...all but unnoticable. But I would know. So say I've got kind of a dud proc and can MAYBE get 3.6 @ 3.67 volts or so after vdroop. Now we're only talking less than .4 volts and not a huge difference in temps I would imagine. What does that do, tack on a few months of lifespan on the thing? The price of e7200's keeps dropping, though I imagine not much more. I'm thinking WORST case scenario is the cpu dies in a year (which I doubt will happen) and I have to get a new one. I'm coming from a 939 opty 165, so the main reason for this upgrade was to get into the 775 socket intel. Now that I'm here, its only a matter of dropping a new cpu in it.

 

[VirtualLarry]

Originally posted by: jaredpace
tm2 = lower vcore when the cpu is clocked lower = less power consumed/less heat = less noise

No, that's not what TM2 is. It stands for Thermal Monitor 2. It's the feature that keeps your CPU from cooking itself. It causes the CPU to multi- and voltage- down when Tjunction hits max.

You want to ALWAYS leave this ENABLED. It may save your CPU's life.

 

[Tempered81]

Thermal Monitor 2

Now let's examine the most interesting technology ? Thermal Monitor 2. In brief, the advantage of TM2 versus TM1 is that it can drop the actual processor clock (that holds true in case of Pentium M) together with a significant reduction of its voltage (as is already known and demonstrated by EIST that a processor can operate at a lower voltage with reduced clock; to a first approximation, the clock-voltage dependence is linear). In its turn, it results in a noticeable reduction of CPU heat dissipation (the total power-voltage dependence, including the clock component is cubic) and consequently of its operating temperature. Well, enough of the theory, let's proceed to practice. The procedure is the same ? stop the cooler and monitor the CPU temperature sensor.

http://www.digit-life.com/arti...ermal-features-pm.html

 

[VirtualLarry]

Originally posted by: jaredpace
Thermal Monitor 2

Now let's examine the most interesting technology ? Thermal Monitor 2. In brief, the advantage of TM2 versus TM1 is that it can drop the actual processor clock (that holds true in case of Pentium M) together with a significant reduction of its voltage (as is already known and demonstrated by EIST that a processor can operate at a lower voltage with reduced clock; to a first approximation, the clock-voltage dependence is linear). In its turn, it results in a noticeable reduction of CPU heat dissipation (the total power-voltage dependence, including the clock component is cubic) and consequently of its operating temperature. Well, enough of the theory, let's proceed to practice. The procedure is the same ? stop the cooler and monitor the CPU temperature sensor.

http://www.digit-life.com/arti...ermal-features-pm.html

Yes, that's all well and good, but that's not what you wrote. You suggested that it lowered voltages and temps when the clock speed was lowered. In reality, TM2 is the mechanism that lowers both clocks and voltages, lowering temps, when the cpu is overheating.

Why would you ever want that disabled?

 

[mancunian]

Originally posted by: Jyve
As it stands now, you are probably right...more headroom on an e8400 I'm sure. Probably a few months ago I would have disagreed as the e7200's were just beasts. As for the high voltage, yeah, it is a bit on the high side, but like I said...the temps all look good to me. I can't imagine that this thing would die out on me in 6 months but who knows. And as for the difference in 3.6 and 3.8, you're right...all but unnoticable. But I would know. So say I've got kind of a dud proc and can MAYBE get 3.6 @ 3.67 volts or so after vdroop. Now we're only talking less than .4 volts and not a huge difference in temps I would imagine. What does that do, tack on a few months of lifespan on the thing? The price of e7200's keeps dropping, though I imagine not much more. I'm thinking WORST case scenario is the cpu dies in a year (which I doubt will happen) and I have to get a new one. I'm coming from a 939 opty 165, so the main reason for this upgrade was to get into the 775 socket intel. Now that I'm here, its only a matter of dropping a new cpu in it.

That's pretty much how I see it.

Intel kinda spoiled overclockers with the Core 2 chips as well. An E7200 at say, 3.4 is still a sizeable 900Mhz overclock.

We can't really complain if it doesn't overlock by 1.5Ghz.

 

[Tempered81]

okay, you're right. I worded that wrong. Apologies. Don't mess with TM2.

 

[formulav8]

Well, I basically have one of them SLAVN 7200 dud cpu's and couldn't get 3.6ghz stable until I raised the vcore above 1.3v. I was really hoping to get 4ghz though.


Oh well, I will have to make do for now.



Jason

 

[Jyve]

Originally posted by: formulav8
Well, I basically have one of them SLAVN 7200 dud cpu's and couldn't get 3.6ghz stable until I raised the vcore above 1.3v. I was really hoping to get 4ghz though.


Oh well, I will have to make do for now.



Jason

Fairly sure mine is a slavn as well.