Sign in with
Sign up | Sign in
Your question
Solved

P45 or E8500? One of them has become sleepy

Tags:
  • Intel
  • Overclocking
Last response: in Overclocking
Share
a b K Overclocking
October 5, 2009 8:02:12 PM

So in the weekend I left my computer on for a couple of days. Came back and found it crashed. It'd recently started crashing out of the odd game too. So my 'stable 4.37' was losing stability.

It got worse and worse until I was forced to lower my FSB, anything over 405Mhz will now not boot at all (it tries but if it makes windows it bluescreens). So now I'm at 3.8Ghz, and it's stable, and I've lowered all my voltages. So the machine is fine, still overclocking, I'm just down 500Mhz or so.

Now I'm putting this down to degredation on one of the chips, as nothing else has changed in the machine. I realise this is my fault for pushing voltages too high but I'm hoping to work out better which part has degraded and it might prove to be useful information for future E8500/P45 users.

So the settings, not current but what they were:

VCore: 1.575 (actual 1.52) Temps: 60C load (peak at 65 for split seconds)
MCH/NB: 1.48V
GTLRef: 0.67 x VTT
VTT: 1.36V
PLL: 1.52V
FSB: 465 (actual was 460, stupid Asrock)

Board: Asrock P45XE.

I tried lowering the multiplier on the CPU, keeping the same bus speed, I booted then got blue screens, I even went down to x6. So from that I'm gathering the problem is FSB related.

If its the CPU I'll save and get an i7 setup, otherwise I'll just get a new motherboard and try again :) 

The RAM is definately not the issue, it's rated 5-5-5-15 @ 1066Mhz, and in the tests it was running on these higher timings (not the ones in the sig) at 800-920Mhz on 1:1 mode. Voltage set to the rated 2.1V.

So the question, would the bus section of the E8500 be the most likely damaged part, or would the P45 Intel chip itself be the problem?
(PS I realise this is my fault, no finger pointing please, I'm not complaining, just edumicating myself)

More about : p45 e8500 sleepy

a b K Overclocking
October 5, 2009 8:22:41 PM

What's the HS/F you are using and is it securely fit, this happened to me on my E8400/P5Q Pro. Ended up re-applying thermal paste and re-installing my HS/F and this resolved the issue...

The reason i say this is because i had originally installed the HS/F and everything was good for over 9 months, turns out that a simple re-application of arctic silver did the trick,

I guess running at those temps 24/7 might shorten the length-per-application of thermal goop! :) 
m
0
l
a b K Overclocking
October 5, 2009 8:27:37 PM

Thanks for reply.

Using a Coolmaster Hyper NS520 with dual 92mm fans, one of which blows directly over the NB heatsink. HS bolts on with four nuts, and is definately on tightly. I used a thin layer of arctic silver 5.

After install on stock settings, the cores peaked at 45C load (core temp).
In the 60's should be fine, my P4 spent years at 80C (dont ask). I'm pretty sure its a voltage caused issue. This chipset/CPU is only around 3 months old.

Machine was running fine, nothing changed, then problems progressively got worse in the weekend.
m
0
l
October 5, 2009 8:35:53 PM

45nm chips will die if pumped too much voltage so be careful
m
0
l
October 5, 2009 8:39:22 PM

you have 1.5V which for any processor for current gen is a little too much, so try to not go beyond 1.36V
m
0
l
a b K Overclocking
October 5, 2009 8:44:17 PM

I know you guys are trying to be useful, but I already know the voltages are the most likely cause for the lower FSB stability. I knew, accepted, and took the risks, and I'm not complaining.

This post is about determining weither the settings would have damaged the P45 chipset or the Core2Duo CPU itself.

Hans - people run these chips at up to 1.8V, it may reduce lifespan but not quite this radically, I actually think its more likely the VTT and GTLRef which are FSB related.
m
0
l

Best solution

a b K Overclocking
October 5, 2009 8:45:36 PM

Yeah, 3 months since application of arctic silver 5 & HS/F shouldn't warrant any issues in regards to re-application. This is a fun one :/ 

If i had to guess, it's something on the P45 or perhaps the motherboard itself as CPU's are usually the last to go. those 8x00 series chips are solid! I guess what i mean is, on a CPU, if it's going to die, it'll just die, it won't get weaker, never have i seen a CPU slowly degrade, it either is or it isnt... Least in my experience.
Share
October 5, 2009 8:57:18 PM

Jonpaul, on the contrary. Chips can slowly lose the ability to overclock that they once had. I've lost some capability on my E8400 after running it at 4.05Ghz through the winter 24/7 and then into the spring (not noticing high temps for a bit.) I agree the E8xxx chips are very solid, but there is a point where they can't take it for a prolonged time. 1.575 volts would certainly do it. Btw, my chip is an original stepping hence the "lower" overclock and nearly 1.4V to get there.

I think if you get a chip to completely blow, it would be an extreme case in which you have severely overvoltaged or overclocked it...or both.

Sorry, but I won't be much help in figuring which chip is half dead. But your reasoning on the MB chip seems reasonable. Overall both chips (P45 and E8xxx) are both VERY solid and wonderful chips, but I think you've done it, as to which...I wouldn't be sure. But do check your temps and see if there's been a major change in ambient temps affecting it....(Ie...first time you've turned on the heat for the winter etc.)
m
0
l
a b K Overclocking
October 5, 2009 9:06:20 PM

Temperatures are all low, case has 8 fans, internal temp was 18C when I was testing (yeah it was cold inside). Never had any temp issues, have been watching NB, CPU, and CPU core temps. Nothing feels hot.

I tried the same 465 FSB x 6 on the CPU, instead of x9.5. No stability change, same crash screen. So I'm more along the line of thought that VTT/GTLRef has caused the rapid wear. I don't see many people using VTT past 1.3V, but I needed to for stability, possibly this is a board design issue? As it was a cheap Asrock board.

I've lowered all the voltages now, as they're not needed for a 3.8 overclock.

I've read a few accounts about quad 45nm chips dying from being kept over 1.4 VTT, and I'm wondering if 1.3x volts will cause rapid damage (but maybe not quite instant death for a review/benchmark of article writers).

Not sure how I can tell if its the P45 or E8500 without buying more gear though.
m
0
l
a b K Overclocking
October 5, 2009 9:24:01 PM

Keep the thoughts coming :) 
I realise I probably wont get a definate answer, but I'm gonna go with most common opinion.
m
0
l
October 5, 2009 9:43:02 PM

Obviously, your vcore is way to high, so its best to start with a new cpu.

It only takes a little common sense to figure that out, but obviously you have none for running it that high in the first place.
m
0
l
a b K Overclocking
October 5, 2009 9:48:22 PM

People run higher V-cores than 1.52.
It's still running fine at 3.8Ghz at almost stock voltages, so I'm not worried.

If you had some common sense you'd understand I was trying to see how far I could push it for daily use.
If it wasn't for people doing this, you would not know what was safe for overclocking.

In my posts I said I was aware of the risks, and I wasn't complaining.
So keep the insults to yourself. I'll put one vote in for VCore.
m
0
l
a b K Overclocking
October 5, 2009 10:05:20 PM

Another vote for Vcore though, quote from on here after anandtech review:

"This means, 1.8V for 4.0Ghz E8400 is similar to putting 2.1~2.2V into 65nm Core 2s. I can guarantee you, unless you have a golden chip, or you're extremely lucky, your chip will fail within weeks, if not days."

I'm not near 1.8V but it's been at 1.52 for about 2 months now.
Although its the bus section, lowering the CPU clock via the multi doesn't change anything.
m
0
l
October 5, 2009 10:43:01 PM

With anything higher than 1.4V vcore you risk cpu damage. Your vtt of 1.36V is a little high, but below the accepted limit of 1.4V The only voltage that could possibly damage your motherboard is the MCH. I think the safe maximum is 1.5V, but usually higher than 1.4V doesn't help stability. My EP45-UD3P has only 1.3V MCH for 448fsb on my Q9550.

So, most likely it's the high vcore degrading the cpu, and not a problem with the motherboard or ram. But the only way to know for sure is to test them separately.
m
0
l
a b K Overclocking
October 5, 2009 10:56:23 PM

Thanks orangegator.

Oddly I needed my MCH/NB voltage that high to stay stable. I put that down to MB design perhaps, although 460 bus speed is starting to get up there.

It's at 1.26V now.

I think I'll buy another CPU first. My cooler uses nuts to secure it under the board, so if I get a new board, I'd rather it went in once and stayed there.

You've sparked an idea in my head of how to test this though.
If its the NB, I should be able to crank the NB voltage right up (temporarily) I should be able to get in to windows. If it makes no difference then it's the CPU.

It'd only be a quick test and the NB has good cooling.
According to Intel, instability due to degradation can be seen to by increasing voltage.. Anyone confirm if this should work?

FWIW - if I break something worse, it's a non-issue, I can order new gear if I NEED to. At the moment 3.8 is fine, but I want to know what caused the drop (advice to other people as well as my personal knowledge).
m
0
l
a b K Overclocking
October 6, 2009 1:03:55 AM

Quote:

Hans - people run these chips at up to 1.8V, it may reduce lifespan but not quite this radically, I actually think its more likely the VTT and GTLRef which are FSB related.

that depends on the CPU. Most of those that run at 1.8v,etc are doing it for record breaking under LN2,Water,DIce,etc

ABSOLUTE MAX voltage for 45nm is 1.45v.

Also, OP, try putting a fan on the north bridge and see if it's temp related. This happened with high FSBs on the old P35-DS3L boards quite a bit.
m
0
l
a b K Overclocking
October 10, 2009 3:23:05 AM

Okay so I have some results for you guys :) 

The CPU is in perfect condition. As I suspected it was chipset related.

I went and bought a Gigabyte EP45-UD3LR, a motherboard which was cheaper than the Asrock P45XE but had the RAID feature (which I want for later). Anybody considering the P45XE might want to look elsewhere IMO.

I set it at stock, fixed windows, rebooted, made some voltage adjustments and set the clock speed to 4Ghz (bus 422). I've just finished a 30 minute Prime95 test with CPU peaks at 55C and no errors. The Asrock (as of last weekend) couldn't get the CPU over 3.8 stable, I ended up leaving it at 3.6. To get results I needed much higher voltages than what the Gigabyte is asking for.

I'll post up full details once I've tried raising my bus speed and seeing how high I can get without exceeding 1.45Vcore. I'm also limiting myself to MCH of 1.3V and VTT to 1.3V as I suspect it was these two voltages, or one of them, has resulted in degrading the P45 chip on the Asrock. Although this is just a once off case and I can't prove it - so it's just a theory at this stage.

Anyway I'm seriously glad I bought a motherboard instead of a new CPU, the gigabyte does so many things better.

- memory multipliers actually work over 400 bus speed, so my RAM is now 100Mhz faster
- DC voltage control on fan outputs
- its cheaper (here anyway)
- has ICH10R RAID
- VDroop control option + heaps more voltage settings and skew rates etc

Only negative is it only has one PCI-X 16 slot, but with the 5850 and 5870 out now, this seems irrelevant because by the time I need more graphics power I'll be moving to a new motherboard/CPU anyway.

So yeah I'll see how high I can get safely and report back.

(Odd side point, I guess it's standard now, but both motherboards have UV reactive RAM slots, I'm suprised this is never advertised)
m
0
l
a b K Overclocking
October 10, 2009 3:57:06 AM

Stable at 4.2Ghz now, same voltages, so looks like I can go higher.

I'm stopping because the NB temp is getting in to the 50's, this might change when the PC is upright and there is another fan to be hooked up yet. Also my Sentry 2 is beeping at me because the RAM heatsinks are over 30C, before they never got above warm, but now it's running at 1066Mhz (I set the FSB so the DRAM was dead on 1066) they're actually making some good heat.

I'll get a proper NB cooler, this one just has a basic Gigabyte heatsink, it has much less surface area than the Asrock.

So yeah, word for the wise, Gigabyte rules, Asrock fails.
m
0
l
a b K Overclocking
October 10, 2009 9:06:55 PM

^Nice. Thanks for the results. And yes, most (but not all) ASRock/MSI/ECS/Foxconn boards fail at OCing well.

As for the NB fan, grab an Antec Spot cool and direct it near there.

Quote:

(Odd side point, I guess it's standard now, but both motherboards have UV reactive RAM slots, I'm suprised this is never advertised)

Really? Never knew that as I rarely run UV lights. lol. It should look nice in a UV WCed set up.
m
0
l
a b K Overclocking
October 11, 2009 4:28:23 AM

mmm I want to do watercooling really bad, I really really don't need it, but just the DIY aspect makes it appealing, plus my case is designed for it (can hang a 240x120 rad off the top, with two 140mm fans above it.

For the NB the spot fan is a great idea, but I want to avoid adding another fan, even with most the fans running at or below 50% speed, it's still kind of annoying how loud it is.

So I was considering the Xigmatech model, which is available here, if I fitted it I'd have the big sink directly in the path of one of the CPU fans (which blow outwards, both of them), which will give it good air flow.

http://www.legionenterprises.co.nz/chipset-ram-coolers/...

It's offset, which is ideal, but not too far offset, so will be half way between the fan and video card which should be perfect.

UV looks awesome but you only really see it at night, which is a bit of a bummer. I got some UV reactive cable sleeving and UV fans coming tomorrow, which should keep me amused for a while.

Next thing will be RAID, drives are so cheap these days, going to pair up some 500Gb 7200.12 Seagates. See if I notice the difference.
m
0
l
a b K Overclocking
October 11, 2009 8:56:58 AM

Got some good news, that should interest a few people.

The 500 FSB 'wall' on the E8x00 chips is actually just a hole. I read about it last week in a single article, although most sources called it a wall. I can confirm it is definately a hole.

Just booted and prime tested 533 FSB, running (stable) 533 x 8 = 4.25Ghz without raising any of my voltages above my pre-defined limits. Nothing is set to 'auto' either. Advantage here is my DDR1066 is running native 1:1.

NB temps jump quickly in to the 50's and the heatsink is hot alright, so I wont be stress testing or putting it under big loads anytime soon. Will order a proper NB cooling solution tomorrow.

533Mhz FSB!!! woot!
m
0
l
a b K Overclocking
October 27, 2009 1:54:29 PM

This is great news! i'm glad you went with my advice and got the motherboard!!! (Ps. Gigabyte has great OC options and boards) In all honesty, i have had 3 motherboards fail over the last 5 years and i have never seen a CPU fail yet. ***, i still have some of the same CPU's i've had for a while now and i ALWAYS OC the crap out of em... Now vote me best answer! :) 
m
0
l
a b K Overclocking
October 27, 2009 5:49:26 PM

haha all good, I put you down as best answer.

I am having an issue with the Gigabyte board though, seems on mine (and one other toms user now) that some video cards combined with turning on Host Clock Control will cause the system to fail booting. I bought another video card from a different manufacturer to test with (turns up today, I don't really want to swap in another motherboard) but I am certain the issue is a board fault.

Despite this it's performance is epic. I set the FSB to 533 for a laugh to try it, booted in to windows straight away without problem, my face dropped. I hear on a dual core the EP45 (even the UD3s) will go well in to the 600's too, and it needs so much less voltage to do what it does than the Asrock did.
m
0
l
a b K Overclocking
October 28, 2009 12:07:55 AM

well, another GREAT board is the ASUS P5Q Pro which i have in my sig, i was able to get my E8400 to 533 FSB with no sweat. of course i didnt keep it blasting at that FSB though, i kept a 400 FSB stable (with absolutely no voltage increase, mind you, yes, i had a golden chip in that E8400) 24/7 and it was VERY stable, no such thing as an issue with that OC... Thanks for the vote!

Ps. Sold the E8400 for $200.00 to an enthusiest friend of mine who put much better use to the golden chip.
m
0
l
November 2, 2009 10:48:17 AM

jonpaul37 said:
well, another GREAT board is the ASUS P5Q Pro which i have in my sig, i was able to get my E8400 to 533 FSB with no sweat. of course i didnt keep it blasting at that FSB though, i kept a 400 FSB stable (with absolutely no voltage increase, mind you, yes, i had a golden chip in that E8400) 24/7 and it was VERY stable, no such thing as an issue with that OC... Thanks for the vote!

Ps. Sold the E8400 for $200.00 to an enthusiest friend of mine who put much better use to the golden chip.



Sounds a lot like my chip! Runs 24/7 stable (12+hour Prime95 no error) on 400FSB with no voltage changes. I attribute most of this to the chip itself being a dirty little monster, but on top of that, my eVGA 790i board is such a great board. I love the features on it, the simplicity in OC'ing and the stability. 100% solid state capacitors, great NB cooling, and vdroop control (on top of many other features).

E8400's certainly are an impressive CPU, no doubt about it!!!
m
0
l
a b K Overclocking
November 2, 2009 5:59:14 PM

I wanted an E8600 for it's higher binning (don't care about multi).
But prices are still too steep for a dual core.

E8500 does me well enough. Mine isn't a golden chip, but still gets the job done :)  E8400 is epic value for overclockers though.

When I started this build my goal was 4Ghz, didn't take long until 5Ghz was the goal, but I realise 5Ghz daily isn't really achievable on anything at the moment. The 'how much voltage I need' curve is very exponential above 4Ghz.

jonpaul - yeah I was looking at the P5Q Pro Turbo, the Turbo-Key is something I badly want, when I'm not using the computer, I don't want it running the O/C settings. With the turbo key I could switch easily.

With Intel SpeedStep it jumps up and down all the time, and the voltage never decreases in C1. So being able to force a second set of settings is brilliant.
m
0
l
!