Bizarre: dying CPU or Power Supply or other?

[johnman 7]

Looks like pretty reasonable Protection - Note Power input surges and spikes are absorbed by the Battery - and the output is steady - Surprised to see that the Model that I bought over 10 years ago is still available - Rest assured  - I didn't pay $120 for it - Johnman B)

[Noel 4,492]

More help please...

 

I was very careful to follow Arctic Silver's method to prep and apply their product to my SB-E chip.  I do not recall Core1 running considerably warmer, about 4-5C even at idle, over the other 5 cores.  If I recall P3D/FSX tax Core1 the most, right?  I may just not have noticed the right-most readout from CoreTemp always read higher--it might have.  In any case with Prime95 running at 4.3Ghz temp never got over 69F, but that was with pre-cooled air coming into the case.  So questions:

 

1.  Does Coretemp ever mis-ID which core is which?  It appears the left most of 6 temp readouts is Core0, the right-most Core1, which seems weird.

2.  Do some cores either report differently or run hotter than other cores, independent of the heat management scheme?

 

Hate to take it back off because I believe I did  a sterling job of doing it right.

[Richard Sennett 2,499]

And what method was that - many have their own way of going it and the pea method has been superb for me

[Noel 4,492]

And what method was that - many have their own way of going it and the pea method has been superb for me

You can find it on their website by processor type.  I followed the one for I7 which includes they tell me SB-E.  

 

http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

 

Now when I look at the QP4 picture of how that method spreads, I don't like the looks of it because if you look at where the 6 cores live on this proc it is apparent the outer corners aren't covered well, if the picture is any indication of what actually happens.  That I 'tinted' both surface maybe solves that.  Also, maybe the heat cap over the processor way exceeds the edges of the actual cores.

[martin-w 3,235]

Hello Martin,

Maybe You are  trying to read too much into what I Said - as You Pointed out Hardware Failures are very rare to begin with - when they do occur - they are costly and time consuming - If it is Possible for Me to help anyone prevent or deter such a catastrophe - I feel that I am obliged to point out what My experience has been - this is of course MY Experience - Not trying to make any Body DO anything - not trying to Prove anything - Just trying to help - You of course are Free to Disagree - With NO Offense taken - actually - I looked at the Sinclare - and the Comadore - but with its Graphics and Audio - I Had to Have the Atari - as a Matter of fact I still have it - Don't know if it still works - I guess Way back in the back of my "Little Pea Brain" - I have an unfulfilled desire to drag it out and see if it still works - fulfill the moment - so to speak - Johnman B)

 

 

Nope, not reading too much into it at all. You said quite clearly that 80 - 90% of failures are caused by voltage instability. Clearly this is an exaggeration, over the top. My only objection. I'm not sure why you invented that figure. 90% is a massive figure, implying almost all failures are due to voltage instability, which is clearly wrong, hence why it's necessary to draw attention to that fact in order that others aren't misled.

 

And nope, I didn't point out at all that hardware failures are very rare. I didn't say that at all, not once.

 

I have no problem with you giving your advice, based on your experience, I have a problem with exaggerated, invented percentages that aren't factual. 

 

By the way, a UPS, [uninteruptable Power Supply] is a battery backup. In case you didn't know. Different name for the same thing.

 

I should also point out that a simple and cheap surge protector does a great job of protecting our equipment from spikes and other voltage issues. No need for a UPS [battery backup] in that regard. They even come with a £60,000 equipment warranty. I never fire up my PC without one.

 

http://www.ebuyer.com/663063-belkin-8-way-surge-protection-strip-2m-with-2-x-2-4amp-bsv804af2m?gclid=CJnBo6ywlssCFQk8GwodifsMqQ#fo_c=951&fo_k=9df9e1695bf6272c5651eba2d7f0f6e7&fo_s=gplauk?mkwid=sLwyhOxBe_dc&pcrid=51482415179&pkw=&pmt=

 

Of course, a UPS [battery backup] is a great idea if power loss is something that must be avoided.

 

 

I looked at the Sinclare - and the Comadore - but with its Graphics and Audio - I Had to Have the Atari - as a Matter of fact I still have it - Don't know if it still works - I guess Way back in the back of my "Little Pea Brain" - I have an unfulfilled desire to drag it out and see if it still works - fulfill the moment - so to speak - Johnman  B)

 

 

 

Actually, I recall a ZX81 as my first experience with computers. I worked as a photographer for an advertising agency. We photographed the latest high tech gear for schools and colleges. The ZX81 we had in the studio even had a flight simulator program. Basically a collection of mega sized pixels that with tons of imagination just about resembled an aircraft. How times have changed.

[Richard Sennett 2,499]

You can find it on their website by processor type.  I followed the one for I7 which includes they tell me SB-E.  

 

http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf

 

Now when I look at the QP4 picture of how that method spreads, I don't like the looks of it because if you look at where the 6 cores live on this proc it is apparent the outer corners aren't covered well, if the picture is any indication of what actually happens.  That I 'tinted' both surface maybe solves that.  Also, maybe the heat cap over the processor way exceeds the edges of the actual cores.

 

Nope thats not what to do - others will say hog wash - well I tested - I did it that way and my way on my Titan X that I just did on my new computer build - when I pulled the waterblock off  after using said version I had the cpu showing a lot of area that now was metal to metal no coverage - your call but I'm ll set with that

[Noel 4,492]

Looks like pretty reasonable Protection - Note Power input surges and spikes are absorbed by the Battery - and the output is steady - Surprised to see that the Model that I bought over 10 years ago is still available - Rest assured  - I didn't pay $120 for it - Johnman B)

So's you know, this particular unit offers no conditioned power I learned today while at Office Depot.  I called APC and they gave me models that do.  In APC UPS'  Power Condition = Automatic Voltage Regulation.  So if you want a UPS with power conditioning, look for AVR.  

 

The next thing I learned from APC tech support is that it is not necessary to use a UPS to get voltage regulation.  They sell 600W and 1200W Automatic Voltage Regulators and they cost retail $41/$54 respectively.  Here's a summary of what this gizmos do, and it's exactly what is in their UPS' with AVR:

 

The Line-R automatically corrects brownouts (by boosting low voltage) and overvoltages (by stepping down high voltage) from the power utility service to levels that are safe for computers, as well as other sensitive equipment. APC by Schneider Electric’s Line-R provides the highest degree of protection from line voltage sags and swells, and has been designed for many years of reliable, maintenance-free service. 

 

I asked their tech people if it would still function to deliver conditioned power to a surge protector and they said absolutely and that that would translate to conditioned power thru the surge protector--it is important to plug the surge protection strip BETWEEN the Line-R and your electronics they tell me.   In any case, it's around $140 for their cheapest UPS w/ AVR, so will go for this device, and in the process pick up a few more outlets from Line-R to feed a couple of less expensive parts that currently I don't have room for on the Monster strip.

 

Here's the model info for Line-R:

 

http://www.apc.com/shop/us/en/categories/power/surge-protection-and-power-conditioning/voltage-regulators/line-r/_/N-e6eqkp#

[martin-w 3,235]

More help please...

 

I was very careful to follow Arctic Silver's method to prep and apply their product to my SB-E chip.  I do not recall Core1 running considerably warmer, about 4-5C even at idle, over the other 5 cores.  If I recall P3D/FSX tax Core1 the most, right?  I may just not have noticed the right-most readout from CoreTemp always read higher--it might have.  In any case with Prime95 running at 4.3Ghz temp never got over 69F, but that was with pre-cooled air coming into the case.  So questions:

 

1.  Does Coretemp ever mis-ID which core is which?  It appears the left most of 6 temp readouts is Core0, the right-most Core1, which seems weird.

2.  Do some cores either report differently or run hotter than other cores, independent of the heat management scheme?

 

Hate to take it back off because I believe I did  a sterling job of doing it right.

 

 

One core will always run hotter in my experience. Normal behaviour. Try RealTemp. Although I recall that delidded CPU's tend to have temps somewhat closer.

 

Absolutely agree with Rich! The pea sized blob in the middle is the best method. The guys at Innovation Cooling did some experiments a few years ago, and concluded it was the best method as it avoids air bubbles. The "spread it with a credit card" method is years out of date and not applicable to we enthusiasts.

 

 

it is apparent the outer corners aren't covered well,

 

 

 

Not the point. The die is small in relation to the Integrated Heat Spreader. The majority of the heat is in the centre. Don't worry about the far corners.

 

 

Noel... unless the power you get in the US is crap and vastly inferior to ours in the UK, just by a basic surge protector like the one I linked to and don't worry. That's all you need. Of course manufactures love to convince you otherwise to get your dollars.

[Noel 4,492]

Nope thats not what to do - others will say hog wash - well I tested - I did it that way and my way on my Titan X that I just did on my new computer build - when I pulled the waterblock off  after using said version I had the cpu showing a lot of area that now was metal to metal no coverage - your call but I'm ll set with that

I assume 'your way' means a pea sized (how about number of mm's) blob, yes?  Pretty bizarre Arctic Silver would publish this since it is always in their best interest to prove their product works well, but they also will err against too much interface medium since if it squishes in the wrong places it is electrically conductive. 

 

Quite frankly I don't see how either gets the entire surfaces completely covered--it's a round blob going onto square mating surfaces.  The tinting will help some for certain.

[Richard Sennett 2,499]

I assume 'your way' means a pea sized (how about number of mm's) blob, yes?  Pretty bizarre Arctic Silver would publish this since it is always in their best interest to prove their product works well, but they also will err against too much interface medium since if it squishes in the wrong places it is electrically conductive. 

 

Quite frankly I don't see how either gets the entire surfaces completely covered--it's a round blob going onto square mating surfaces.  The tinting will help some for certain.

 

Theres videos showing how well it works on YT using glass to show it - take my word for it do that way - as for size a little bit smaller than a dime - I dont have scale in front of me - I'll look for a video

 

Notice the spread out version near the end and all of the air bubbles - shows what I am talking about

 

[Noel 4,492]

Looks to me that none of these methods solve the problem of getting full contact.   The X method gave the biggest coverage which makes some sense.  The spread method creates bubbles.  The thicker line is similar to the blob method.  After looking at this video, then looking at this layout of 6 cores in a 3960X CPU, it strikes me the best approach is essentially a square blob--after all, these are square mating surfaces, not round ones.  All of these methods potentially skimp a little on the corners & memory controller of the die:

 

 

I think I will for my own satisfaction try a bubble-free square shaped blob.  I will test this out w/ a piece of glass on a hard surface to see how big a square will cover this surface area.  I won't outline a square, I will try a filled in small square.

[Richard Sennett 2,499]

I think I will for my own satisfaction try a bubble-free square shaped blob.  I will test this out w/ a piece of glass on a hard surface to see how big a square will cover this surface area.  I won't outline a square, I will try a filled in small square.

 

I think thats a great idea Noel - good luck - I will have to try that someday also but now now  :wink:

[Gary A 8]

 

 

APC Back-UPS® NS 700VA 8-Outlet Power-Saving UPS
Item: 171431 / Model : BN700MC

 

Your question made me inquisitive so I have been doing some looking at power protection.  Pure sine wave is still preferred top of the line recommendation.

 

The UPS in your selection is a Stepped approximation to a sinewave.   Compatibility of any non pure sine wave UPS with a PSU that has a Power Factor Correction circuit (PFC)  is undetermined and case by case bases if it will work.  Rule of thumb is spec UPS 50-100% higher than PSU wattage (this to ensure that a minimum of 80% of power is supplied to PSU).  Source    Also see Q11 Seagate -

 

A more prudent option may be surge protection only, while this device will not protect data during unexpected shutdown, it may provide protection from serious overload/surge/spike.  Trip Lite

 

Truth be told I own one UPS, I have many PC's, many flat screens and a host of other electrical/electronic gizmos; a serious surge will wipe out existence as far as modern age is concerned.  The one pc that is protected may prove useful in filing the insurance claim or otherwise typing my suicide note. :smile:

 

Where to begin and where to end, do we want to be Doomsday preppers cause ain't no surge protector gonna stop a direct lighting strike anyway.  That job requires a lightning pole.

[Gary A 8]

Here's the model info for Line-R:

 

I looked at that during my research and liked it but my reading indicated surge protection <3000 joules was a farce. I get the warranty and guarantees that most power protection folks offer against surge damage but many have exclusions in fine print. Besides, if enough of you want to hedge a $50 bet with me, I will insure your PC against a lightening strike.

[Noel 4,492]

I looked at that during my research and liked it but my reading indicated surge protection <3000 joules was a farce. I get the warranty and guarantees that most power protection folks offer against surge damage but many have exclusions in fine print. Besides, if enough of you want to hedge a $50 bet with me, I will insure your PC against a lightening strike.

As mentioned I have my primary surge protection between the AVR device and my various electronic components so if you're saying it's either surge protection OR an AVR device w/ suboptimal surge protection then that doesn't seem to apply.

 

In the end Gary, the whole thing is verging on insignificant, that is to say the risk for damage seems to be minimal.  I don't have mission critical anything on any of my devices, but if it helps reduce a rare but apparently possible injury to some microcircuitry component and helps avoid the annoying troubleshooting routine maybe it's worth the $54.   We don't have perceptible brownouts very often but once a year, blackouts a few times and I don't leave my PC running unattended very often anyway, which is to say if I noticed a brown out I'd just turn the PC off immediately.  Outside of the theoretical realm, as I say I see 80 machines on for 10h a day and truly, haven't hardly heard of a failure at our business.  My 13 y/o XP box still works, no UPS, just a cheap strip.  Many others here and elsewhere don't use UPS or power conditioners, so really it's hard to get accurate information to base a purchase decision on--for $54, not a big outlay.