Bad i7 bin?

[myounage]

My D0 has always run a bit .. hot, and when downvolting to 1.125V core it seemed to run pretty well, even up to 3.4GHZ, but I was getting some BSODs, so I got rid of my OC and set Vcore back to Auto. Now, some tell me that Auto follows the processors onboard VID. is this true? If so, my chip overheated itself on the stock cooler(at all stock settings), and forced me into going aftermarket on the cooling. My current idle voltage on auto is 1.20 V. I have heard of people getting 4.0 GHZ with this voltage. I was getting BSODs with 1.125V core, 1.86 PLL, & 1.225 QPI. Is this particular chip seriously only capable of 2.8GHZ @ 1.20V? My load voltage for stock seems to be 1.24 >:\. other people report auto taking them as low as 0.85V. Did Intel ship me a crappy chip? My idle with a Corsair H50 w/ stock fan is 42 C on a non-OCed part. That is with speed step, mind you. prime pushes cores to 65C on blend .. and this is with aftermarket, liquid cooling on stock settings because of .. well, a 1.20-1.24V VID.

Is my mobo (P6T) supplying this chip what it actually needs according to vid or just giving it an arbitrary value? Could my BSODs have a secondary cause in my OC settings? The RAM is DDR3-1600, and I was running it @ 1604MHZ @ 1.64V, and the uncore setting that was available was 3207MHZ. Trying the next setting up just made the BSODs hit within minutes of boot instead of hours later during CPU-heavy computing. This is seriously pissing me off. My chip's running right up against thermal spec with liquid f'ing cooling, and if I try to lower the voltage to something more sensible I get bluescreens. Is this particular i7 bad enough to mail back to Intel, or am I stuck with what may be a POS? I'm getting temps people that are running 4.0GHZ get on inferior cooling. I don't know my batch number, but I'm not quite willing to break the seal of my thermal paste to find out. I hate reinstalling paste >:\.

EDIT: Batch 3913B, part 166. Seems like its a generally shitty batch from my googling.

ED: Alright, people are getting 4 GHZ on 1.16 stable via Linx. Nice. Meanwhile I can't run below 1.20 @ stock without BSODs. They even get lower temps .. awesome ..

/lengthy rage post.

 

[cyclone3d]

What power supply do you have?

What case do you have?

How good is the airflow through your case?

Are you 100% sure you have the correct amount of thermal paste applied as well as the H50 block mounted correctly?

Do you have the latest BIOS for your motherboard?

The rig in my sig requires 1.325v for a stable 4Ghz with DDR3-2000. With IBT, I get 82-83c under full load.

 

[myounage]

Quoted & inserted

What power supply do you have?
----> OCZ StealthXtreme 700W.(maybe 750, one of the two)
What case do you have?
----> NZXT Hush
How good is the airflow through your case?
----> Good enough, my temps match up to other H50 users using stock fans *at the voltage I'm running.* (1.2 it seems :\. Stupid VID.)
Are you 100% sure you have the correct amount of thermal paste applied as well as the H50 block mounted correctly?
----> Yes and yes on those. Reseated plenty of times & followed IC's instructions for IC Diamond. it is a few degrees cooler than my old silver paste, so I think I did it right. The H50 block does not wiggle if I try to do so so, and is def. flat. One thing that irks me is that cores 1 & 3 have been 3-4 C hotter .. always. Throughout all my reseats and even when I used the stock cooler.
Do you have the latest BIOS for your motherboard?
---->Not quite, I have one of the middle versions. Fairly recent at least, Nov 2k9 I think. It came out months after Intel started minting D0s though.
The rig in my sig requires 1.325v for a stable 4Ghz with DDR3-2000. With IBT, I get 82-83c under full load.
----> True, they're not all the same. 1.2V for stock settings does seem very high though .

 

[ShuttleLuv]

Stealthextreme is your problem right there bud.

 

[myounage]

Idunno, according to my voltage sensors its providing some damn stable power. Much moreso than my last PSU (650W BFG). That PSU's coltages would fluctuate quite a bit when the system was in heavy use ... this one remains well, well within tolerance. Just because you had a problem with an OCZ PSU doesn't mean it won't do the job faithfully. Not every PSU needs to be a $180 Corsair.

 

[E4g1e]

Idunno, according to my voltage sensors its providing some damn stable power. Much moreso than my last PSU (650W BFG). That PSU's coltages would fluctuate quite a bit when the system was in heavy use ... this one remains well, well within tolerance. Just because you had a problem with an OCZ PSU doesn't mean it won't do the job faithfully. Not every PSU needs to be a $180 Corsair.

The problem here is that you got ripped off - not because the StealthExtreme 700 is a bad PSU per se, but because you paid the 700W price for what's effectively a 500W PSU. That StealthExtreme 700W is only capable of 500W under realistic conditions. (As Danny Bui stated in another recently-created thread, that StealthExtreme is based on the troublesome FSP Epsilon platform.) Still, that's better than your old BFG, which could barely supply 350W under realistic operating conditions.

 

[myounage]

The problem here is that you got ripped off - not because the StealthExtreme 700 is a bad PSU per se, but because you paid the 700W price for what's effectively a 500W PSU. That StealthExtreme 700W is only capable of 500W under realistic conditions. (As Danny Bui stated in another recently-created thread, that StealthExtreme is based on the troublesome FSP Epsilon platform.) Still, that's better than your old BFG, which could barely supply 350W under realistic operating conditions.

It was on a quite deep sale .. but still, its tested out properly. It doesn't account for why my CPU would not run @ lower voltages. Vdroop is working right, and w/ LLC I could make the voltage constant.

 

[cyclone3d]

It was on a quite deep sale .. but still, its tested out properly. It doesn't account for why my CPU would not run @ lower voltages. Vdroop is working right, and w/ LLC I could make the voltage constant.

Just because the voltage is steady doesn't mean that the amperage is.

 

[BinarySynapse]

Just because the voltage is steady doesn't mean that the amperage is.

Um.. V = IR says otherwise.

 

[cyclone3d]

Um.. V = IR says otherwise.

I've worked and owned too many computers that can hold the voltage steady but will still cause issues due to a seemingly obvious issue with amperage.

Also, what happens when you have a regulated power supply that will only put out a certain amperage at a certain voltage?

Just because it holds the voltage steady when you overload it, doesn't mean it it outputting more amperage than it is physically able to.

So if V=IR, and I is not enough, does that mean R goes up so that V=IR?

Say V = 4, I=2, and R=2.

What happens when I is 1... for I*R to equal 4, then R would have to change to 4.

 

[EricFX1984]

The problem here is that you got ripped off - not because the StealthExtreme 700 is a bad PSU per se, but because you paid the 700W price for what's effectively a 500W PSU. That StealthExtreme 700W is only capable of 500W under realistic conditions. (As Danny Bui stated in another recently-created thread, that StealthExtreme is based on the troublesome FSP Epsilon platform.) Still, that's better than your old BFG, which could barely supply 350W under realistic operating conditions.

there are better Power Supplies available but not for the cost... esp when you take into account the awesome rebates that OCZ has

hell I just bought a StealthXtreme 700 watt for $60 at Microcenter and then it came with a $20 MIR... not to bad for a $60 PSU, and even better for $40

the 600 and 700 stealth and gamextreme are 80+ and have pretty clean power (but not a clean as some)

the 600 stealth/game xtreme will do more than 500watts easy...

http://www.hardwaresecrets.com/article/OCZ-StealthXStream-600-W-Power-Supply-Review/451/7

hell the 700 was better than that

however the modXtreme has issues... they tend to only do around 500 to 600 watt but that is in a 45c room

they also had an issue with the modular output board failing for whatever reason

http://www.hardwaresecrets.com/article/OCZ-ModXStream-Pro-600-W-Power-Supply-Review/936

 

[-Dragon-]

R is pretty constant depending on the given load, most of a CPU is pure resistive load, there's a bit of capacitive and inductive loading, but at the end of the day they won't do what you want them to do. If the PSU can't keep up with I, V has no choice but to drop. If the PSU can hold the V, it can hold the I to, they are intimately related.

 

[cyclone3d]

R is pretty constant depending on the given load, most of a CPU is pure resistive load, there's a bit of capacitive and inductive loading, but at the end of the day they won't do what you want them to do. If the PSU can't keep up with I, V has no choice but to drop. If the PSU can hold the V, it can hold the I to, they are intimately related.

What about the voltage regulators? Does R stay contant with them?

Also, I highly doubt the onboard voltage sensors operate fast enough to detect ripple.

 

[BinarySynapse]

What about the voltage regulators? Does R stay contant with them?

No, effectively resistance drops linearly as current increase allowing voltage to remain constant. They also keeps voltage constant by either shunting current to ground (linear VR's) or by switching the circuit off and on rapidly to create a lower average voltage (switching VR's).

Also, I highly doubt the onboard voltage sensors operate fast enough to detect ripple.

They're usually barely accurate to detect voltage.

 

[RooK]

get a UPS.

That will condition your power out of the wall so your PC doesn't go down when there is a power flicker for half a second, and it will provide cleaner power to your entire rig.

Best investment i've made in a while... maybe 100 bucks to run my desktop, laptop, 3 monitors for about an hour if I wanted, or just the laptop and 1 monitor for 3 or 4 hours.

 

[BinarySynapse]

get a UPS.

That will condition your power out of the wall so your PC doesn't go down when there is a power flicker for half a second, and it will provide cleaner power to your entire rig.

Best investment i've made in a while... maybe 100 bucks to run my desktop, laptop, 3 monitors for about an hour if I wanted, or just the laptop and 1 monitor for 3 or 4 hours.

While this is good advice (that I need to follow ), I haven't seen anyone mention power flickers.

 

[DaFox]

get a UPS.

and it will provide cleaner power to your entire rig.

That is generally false unless you buy an insanely high end 'online' UPS sadly.

 

[-Dragon-]

What about the voltage regulators? Does R stay contant with them?

Also, I highly doubt the onboard voltage sensors operate fast enough to detect ripple.

Voltage regulators use these brand new magical devices called transistors with a sensing feedback loop to very quickly change circuit resistance to maintain the desired voltage. In different configurations these same magic devices can be used to rapidly vary voltage in things called current regulators in situations where you want constant current over a varying resistance. Transistors are neato, you should keep your eye out for them, I hear they might someday become popular in the new field of "consumer electronics".

As magical as they may be they do have limits with how fast they can change their internal resistance so to help them out we team them up with another recent invention called the capacitor! Capacitors can actually store up charge and as voltage drops will release its stored electrons to buffer the voltage drop.

If it DOES have problems with ripple, the issue would be with the VR circuits on the motherboard, NOT with the powersupply, if it was a PSU problem the voltage would droop and be unable to recover, you COULD see this, if the CPU is getting undetectable voltage drops and ripple that's causing it to malfunction then the capacitors and voltage regulating units around the CPU are to blame.

 

[hdgamer]

I had the same problem. I had to reset the QPI clocks to default and increase my ram voltage to 1.66V. I think it was the ram voltage that did it though, even though your Bios will flash red, 1.66 is perfectly safe for your cpu/ ram.

 

[myounage]

What about QPI voltage? I've ehard figures as high as 1.3V needed to run DDR3-1600 @ decent timings. Also @ my RAM's advertised timings I sometimes get a little glitch where the cursor messes up when changing cursors, ex text select to normal. Hard to describe, but it gets bigger, stretches, and shows some green. Only visible for a small fraction of a second. Should I back off my timings or increase QPI core or what?

ED: to clarify this does NOT happen when ram timings are set to auto.

 

[-Dragon-]

I thought I read somewhere that all the voltages on the chip had to be within something like 0.3 or 0.35V (pretty sure it's 0.35V) of each other, so if you're running at the max 1.65V on RAM QPI voltage needs to be at least 1.3. Personally I just got some of that G.Skill Eco 1600, 7-8-7-24 @1.35V stuff and keep all my voltages as low as possible for the desired clocks.

 

[myounage]

Mhm, managed to down my vcore to 1.15 with stability @ 3.36GHZ(160x21). Just needed to up my QPI to keep up with my DDR3-1600. Stock QPI is fine w/ stock RAM timings, but the rated timings on the package (8-8-8-24) takes 1.237 QPI and 1.66 Vdimm from what I can tell.