Overclockers - beware of voltages in excess of Intel recommended max on Penryn (45nm) processors !!

  • Thread starter Thread starter John Lewis
  • Start date Start date
J

John Lewis

See the following sad story:-

http://www.anandtech.com/weblog/showpost.aspx?i=428

Even if you use voltages nowhere as extreme as the Anandtech
reviewers, you may still be contributing to a longer-term
early-failure. These hafnium-gate devices may be a lot less
accommodating of overstress than previous gate technology. It would be
very wise to respect Intel's recommended voltage maximums and let
other 'guinea-pigs' ( or those with lots of spare cash) fill in the
MTBF vs voltage profiles of 45nm processors..

John Lewis
 
See the following sad story:-

http://www.anandtech.com/weblog/showpost.aspx?i=428

Even if you use voltages nowhere as extreme as the Anandtech
reviewers, you may still be contributing to a longer-term
early-failure. These hafnium-gate devices may be a lot less
accommodating of overstress than previous gate technology. It would be
very wise to respect Intel's recommended voltage maximums and let
other 'guinea-pigs' ( or those with lots of spare cash) fill in the
MTBF vs voltage profiles of 45nm processors..

John Lewis
Click to expand...

Early failure is pretty much always a risk with overclocking, even
when specific cases are not published. That's the primary reason I
don't do it.
 
John Lewis said:
See the following sad story:-

http://www.anandtech.com/weblog/showpost.aspx?i=428

Even if you use voltages nowhere as extreme as the Anandtech
reviewers, you may still be contributing to a longer-term
early-failure. These hafnium-gate devices may be a lot less
accommodating of overstress than previous gate technology. It would be
very wise to respect Intel's recommended voltage maximums and let
other 'guinea-pigs' ( or those with lots of spare cash) fill in the
MTBF vs voltage profiles of 45nm processors..

John Lewis
Click to expand...

And this was X-posted to the Nvidia and ATI videocard group for what reason?
Because PC's have vieocards in them?
 
See the following sad story:-

http://www.anandtech.com/weblog/showpost.aspx?i=428

Even if you use voltages nowhere as extreme as the Anandtech
reviewers, you may still be contributing to a longer-term
early-failure. These hafnium-gate devices may be a lot less
accommodating of overstress than previous gate technology. It would be
very wise to respect Intel's recommended voltage maximums and let
other 'guinea-pigs' ( or those with lots of spare cash) fill in the
MTBF vs voltage profiles of 45nm processors..

John Lewis
Click to expand...

As the geometry of silicon gets smaller, core or case temperature
becomes a far less reliable guide to the Mean Time before Failure.
Other mechanisms such as applied voltage begin to dominate.

John Lewis
 
_____

Because the original poster didn't dare post in overclocking newsgroups
where the contents would be evaluated. For example why boost VTT in the
first place? After all, the stated CPU Clock rate is only 453 MHz, just a
53 MHz / 400 MHz = 13% boost above the stock clock for the Intel X48 1600
MHz specified FSB (plenty of Intel chipset motherboards NOT specified for
1600 MHz FSB operation will operate at that speed without VTT boosts.) And
the sample is quantity one for the original post. And the original post is
both unsigned AND in a forum. And the added comments by the original poster
for this thread have even less provenance.

Phil Weldon
Click to expand...

The misfortune happened at Anandtech in extreme stress testing..
Probably the most respected PC hardware review site in the world for
its technical authority. Er, Gary Key, the author of the article is
one of their top reviewers. And he posted the warning as a practical
instance of what can happen if you play with voltages in excess of
manufacturer rating, as many other have found out recently when
overvolting memory. Why not read the full blog and Gary's implied
intentions before jumping in with both feet.

Oh, btw. I do have quite a bit of stress/QC experience over quite a
few years with standard digital silicon processes. It takes quite some
time to fully establish the failure corners. Regardless of Intel's
best efforts, I suspect there is quite a bit of learning still be to
done even by Intel in that direction on hafnium-gate silicon.


Nope.

Because those that read those specific newsgroups are likely be more
than casual PC users and highly likely to be in the
build-it-and-tweak-it-yourself PC community. I thought it easier to
reach more PC tweakers that way than try posting to a myriad of
motherboard newsgroups besides Asus.

You do leap to the wrong conclusions rather fast, don't you? Been
very long in these newsgroups or are you just a passing newbie?

John Lewis
 
hmm...

I brought my complete PC recently Overclocked proccessor (quad 2.4 OC to
3.2) and OC Vid as advertised. Wonder how mine will fare over the 12 months
that is a legal requirement for warranty on goods sold in the UK. As a unit
I've been informed, this would take precedent over individual item warranty
requirements laid down by say, Intel or Nvidia. Also the selling company of
the PC has the liability over individual parts manufacture.

Interesting :)

Alan
 
You do leap to the wrong conclusions rather fast, don't you? Been
very long in these newsgroups or are you just a passing newbie?

John Lewis
Click to expand...

Passing newbie?....lol....Who's the clown who's never heard of
alt.comp.hardware.overclocking?
 
You are bonkers if that is what you think. More so if you are in any way
affiliated with that site or its advertisers.
Click to expand...

Not at all. However, I do happen to be a hardware engineer
specializing in computer-based professional technical instrumentation,
so maybe I do have some ability to discern technical competence in
this field
Anandtech is perfectly average in every way.
Click to expand...

Really ? And what makes you come to that conclusion ?

Perfectly average? Just like you ??

Your moniker betrays your pomposity. Try applying to
Anandtech as a reviewer some day. Or to Ars Technica as
a journalist.

John Lewis
 
Really ? And what makes you come to that conclusion ?
Perfectly average? Just like you ??

Your moniker betrays your pomposity. Try applying to
Anandtech as a reviewer some day. Or to Ars Technica as
a journalist.

John Lewis
Click to expand...

I'm sure Mr.E.Solved won't sleep tonight now. It just breaks out hearts
when some non-entity like you blows through the n/g cross-posting and
talking out of his ass. Perhaps you can post some irrelevant gem again in,
say, 2011 or 2012.
 
Somewhere on teh intarweb "John Lewis" typed:
Not at all. However, I do happen to be a hardware engineer
specializing in computer-based professional technical instrumentation,
so maybe I do have some ability to discern technical competence in
this field
Click to expand...

Or not. IMEE when people start claiming quals without backing the claims up
it's usually a defence mechanism and chances are that it's bullshit.

Anyone with half a brain (or more) would have just posted this to
alt.comp.harware.overclocking. After all, if there was one group where this
post would have been right on topic and open to knowledgable discussion
that's the one.

By their actions you will know them.
 
hmm...

I brought my complete PC recently Overclocked proccessor (quad 2.4 OC to
3.2) and OC Vid as advertised. Wonder how mine will fare over the 12 months
that is a legal requirement for warranty on goods sold in the UK. As a unit
I've been informed, this would take precedent over individual item warranty
requirements laid down by say, Intel or Nvidia. Also the selling company of
the PC has the liability over individual parts manufacture.

Interesting :)
Click to expand...

If you managed to overclock without raising any voltages higher than
Intel's/nVidia's/ memory-supplier recommended maximum, the core temp
of the CPU does not exceed a conservative 60 degrees C and the GPU
core does not exceed a conservative 80 degrees C while running the
stress benchmarks, you should be in great shape.

John Lewis
 
The OP
Click to expand...

I was just passing along the Anandtech blog as a friendly warning for
those far less knowledgeable about 45nm technology than yourself. You
do work for one of Intel's process-evaluation group no doubt?

John Lewis
 
I was just passing along the Anandtech blog as a friendly warning for
those far less knowledgeable about 45nm technology than yourself. You
do work for one of Intel's process-evaluation group no doubt?

John Lewis
Click to expand...

You delusions about your vast and far reaching knowledge regarding "45nm
technology" is of no consequence to me. You don't know me, or anyone else
who posts here on a regular basis. So don't presume to. I don't give a shit
if you invented the internet along with your unique ability to "discern
technical competence". I put you in the same league as Skybuck and DaveW.
Actually, I take that back. They do have some redeeming humour qualities
they are unaware of. You are a just a puffed up, self styled "expert" with
zero credibilty. A usenet search of your postings speaks volumes about
your expertise, sources and knowledge base. Like the one back in Feb where
you told First of One that the 8800GTX was an "archaic" technology. Of
course, since you're a long time regular, unlike me who's just a passing
newbie and and have just discovered usenet. You still haven't figured out
how to post to just one group either.
 
John Lewis said:
... in the context of the time to failure in the Anandtech article....

next question?
Click to expand...

Here's a tip. If you use zero volts the CPU will last a lot longer than the
estimated MTBF! <g>
 
Here's a tip. If you use zero volts the CPU will last a lot longer than the
estimated MTBF! <g>
Click to expand...

MTBF is not a lifetime measure. MTBF is a failure probability
during the lifetime period. For example HDDs have a lifetime
of 5 years, and during that time the MTBF holds.

Arno
 
Back
Top