Prev: slower memory in a faster motherboard
Next: Strangeness.. ASUS M4A78T-E "unlocks" AMD Phenom II X2 to X4 ??
From: Man-wai Chang to The Door (33600bps) on 4 Mar 2010 03:09
> What scares me:
> How good is this "unlock"?
> How reliable is the AUSU BIOS at the detection of an unlockable CPU?
> Could this be part of my "sleep?" problem?

You would know when to stop playing with ACC when the system became
unstable. Just remember that this "unlock" thing and its consequences
are NOT the responsibilities of AMD.

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.33
^ ^ 16:09:01 up 1 day 4 min 1 user load average: 1.02 1.08 1.09
不借貸! 不詐騙! 不援交! 不打交! 不打劫! 不自殺! 請考慮綜援 (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa
From: Paul on 4 Mar 2010 04:07
Man-wai Chang to The Door (33600bps) wrote:
>> What scares me:
>> How good is this "unlock"?
>> How reliable is the AUSU BIOS at the detection of an unlockable CPU?
>> Could this be part of my "sleep?" problem?
>
> You would know when to stop playing with ACC when the system became
> unstable. Just remember that this "unlock" thing and its consequences
> are NOT the responsibilities of AMD.
>

The four cores could be tested with Prime95 stress test, to determine if all
the cores are good. That is not as good as the testing they do at the
factory, but it is better than just blindly accepting that the cores work.
Prime95 multi-threaded version can be downloaded from mersenne.org/freesoft .

I saw some comments about unlock functions, in this article yesterday.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3755&p=4

The whole thing is a lottery, but it's fun to win.

*******

With regard to finding a solution for a "sleep" function, you could either
check vip.asus.com for other customer comments, or look in the Newegg
reviews and see if the problem is a common one.

M4A78T-E

http://www.newegg.com/Product/ProductReview.aspx?Item=N82E16813131366

http://vip.asus.com/forum/topic.aspx?board_id=1&model=M4A78T-E&SLanguage=en-us

It would appear at least one version of the BIOS for the board, is half
baked and should be avoided. And I've only read a few threads on vip so
far :-)

Paul
From: Man-wai Chang to The Door (33600bps) on 4 Mar 2010 07:51
>
> The circuitry is all there, but they disable it? I know that some
> circuit boards have uninstalled component locations, but all of
> the stuff is there and disabled?

AMD must have a good reason to not sell them as 4-core CPUs. Any one
knew why? Is it a secret? :)

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.33
^ ^ 20:51:01 up 1 day 4:46 1 user load average: 1.03 1.06 1.07
不借貸! 不詐騙! 不援交! 不打交! 不打劫! 不自殺! 請考慮綜援 (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa
From: Paul on 4 Mar 2010 20:28
Man-wai Chang to The Door (33600bps) wrote:
>>
>> The circuitry is all there, but they disable it? I know that some
>> circuit boards have uninstalled component locations, but all of
>> the stuff is there and disabled?
>
> AMD must have a good reason to not sell them as 4-core CPUs. Any one
> knew why? Is it a secret? :)
>

It allows them to get some value from defective quad core chips.

The quad core chip would be a relatively large silicon die.
If one or two of the cores had some small problem, the wafer
sort process could identify them as candidates for packaging
as dual core processors.

The AMD test will test all sorts of stuff that never
arises in normal usage. Compilers don't use or emit code
for all possible instructions on the processor. Only a
small subset of all possible instructions is generated
by the average compiler. (An evil person writing assembler
code can do that, if they want.)

The thing that is defective, might be something not normally
used. In other words, it is highly unlikely that the end user
will be able to figure out what is broken in the unlocked cores.
My suggestion to use Prime95 is only a bit of a test - by
no stretch of the imagination is it a complete test.

Or, the thing that is defective, might only be defective at
elevated clock rate. If you're making processors, you likely
test them at a few hundred MHz above the stated operating
frequency. That takes some aging and parameter shifting into
account over time.

Paul
From: Paul on 5 Mar 2010 02:48
John Doe wrote:
> Right... Even if the unlocked CPU will probably function normally,
> the idea that the other cores might be defective puts doubt in the
> idea of unlocking the thing. Unless it can be re-locked?

As far as I know, it is just a BIOS trick. I did manage to find
some web articles, which suggest there is an electrical signal
running from the Southbridge to the processor, and there is some
kind of communication. But so far, I haven't found any details
as to exactly what they send from the Southbridge to the processor.
In any case, the "unlock" would be a process that happens on
each BIOS POST. Turning ACC off in the BIOS, should stop the
unlock on the next reboot.

There is a picture of the BIOS screen here.

http://images.anandtech.com/reviews/cpu/amd/phenom2/555andmore/followup/ACC.jpg

Another thing I didn't know, is the ATI SB750 Southbridge, has its
own processor inside. There is an 8051 inside, and the ACC function
is somehow associated with that block. But the details aren't in the
documentation, as to what they do with it. Intel has a similar
scheme, but it is for IT control of computers in large corporations
(AMT). I don't get the impression the 8051 inside the SB750 is
quite that ambitious. It may be used to control fans or something,
or replace part of the functions of the SuperI/O chip. That is all I
could find. The "EC" in the BIOS screen, stands for "Embedded Controller".
The AMD documentation refers to IMC or Integrated Micro Controller.

Paul
 |  Next  |  Last
Pages: 1 2
Prev: slower memory in a faster motherboard
Next: Strangeness.. ASUS M4A78T-E "unlocks" AMD Phenom II X2 to X4 ??