new mb (Sempron/GigaByte)

  • Thread starter Thread starter Flasherly
  • Start date Start date
F

Flasherly

Have had an inexpensive Gigabyte MB and AMD Sempron sitting around a
year or so. Finally put the mem, cpu, and dvd to it -- and see in the
(rudimentary) BIOS a provision for something along "turn on virtual"
machines. In the MB book it says something basic about OS that
support virtual machines. Guessing offhandidly that "could be" for
dual cores. Then I go off and stumble upon a read, that all Sempron
productions are dual cores with a core locked. Oh, boy -- a
curveball! Now, if I get lucky I can test whether mine will unlock,
real lucky - for a semi-stable, theoretical, free core. An existing
XP single-core install can't be shifted over to a new machine, a
dually, (hopefully) slipped the new MB drivers and once working (FU-
kludged-BAR), to turn on that BIOS switch, hmmmm... Just asking (a
dream come true, when what "I think" I've heard mentioned, is XP
requires a fresh install for dual core support).
 
Flasherly said:
Have had an inexpensive Gigabyte MB and AMD Sempron sitting around a
year or so. Finally put the mem, cpu, and dvd to it -- and see in the
(rudimentary) BIOS a provision for something along "turn on virtual"
machines. In the MB book it says something basic about OS that
support virtual machines. Guessing offhandidly that "could be" for
dual cores. Then I go off and stumble upon a read, that all Sempron
productions are dual cores with a core locked. Oh, boy -- a
curveball! Now, if I get lucky I can test whether mine will unlock,
real lucky - for a semi-stable, theoretical, free core. An existing
XP single-core install can't be shifted over to a new machine, a
dually, (hopefully) slipped the new MB drivers and once working (FU-
kludged-BAR), to turn on that BIOS switch, hmmmm... Just asking (a
dream come true, when what "I think" I've heard mentioned, is XP
requires a fresh install for dual core support).

"I think" I've heard mentioned, is XP
requires a fresh install for dual core support).

I think it's as easy as running msconfig, hitting the 'boot ini' tab, click
on the 'advanced options' and choosing the '/procnum' that are available.

Getting that second core unlocked so that XP sees it's available is another
question.
 
Flasherly said:
Have had an inexpensive Gigabyte MB and AMD Sempron sitting around a
year or so. Finally put the mem, cpu, and dvd to it -- and see in the
(rudimentary) BIOS a provision for something along "turn on virtual"
machines. In the MB book it says something basic about OS that
support virtual machines. Guessing offhandidly that "could be" for
dual cores. Then I go off and stumble upon a read, that all Sempron
productions are dual cores with a core locked. Oh, boy -- a
curveball! Now, if I get lucky I can test whether mine will unlock,
real lucky - for a semi-stable, theoretical, free core. An existing
XP single-core install can't be shifted over to a new machine, a
dually, (hopefully) slipped the new MB drivers and once working (FU-
kludged-BAR), to turn on that BIOS switch, hmmmm... Just asking (a
dream come true, when what "I think" I've heard mentioned, is XP
requires a fresh install for dual core support).

You're mixing a couple concepts together here.

http://en.wikipedia.org/wiki/X86_virtualization

"In computing, x86 virtualization is the facility that allows multiple
operating systems to simultaneously share x86 processor resources in
a safe and efficient manner"

"AMD virtualization (AMD-V)" Pacifica
"Intel virtualization (VT-x)" Vanderpool

"I/O MMU virtualization (AMD-Vi and VT-d)"

Your BIOS setting, could have been turning on/off the Pacifica.

To give an example, I have a copy of Microsoft VirtualPC I run a lot.
I can run a copy of Linux or Windows, inside VirtualPC, while my
regular OS (WinXP) is running. In testing it, I discovered that
hardware virtualization support (can be turned on and off in VirtualPC)
makes no difference at all. What it does make a difference in, is some
Linux LiveCD distros, *crash* depending on the setting used, because
the Linux software tries to "guess" whether it is in a virtual environment
or not.

*******

The other concept, is core unlocking. Or multicore operation.

If you go to Device Manager, (devmgmt.msc), and select the Computer entry,
and do Properties on it, it'll say something like "ACPI Uniprocessor PC".
That uses one core.

If you enabled a second core, and left that setting as is, I don't think the
second core would be used.

If you do a "driver update" on the computer entry, you may be shown which
HAL types are eligible candidates. For example, promotion to
"ACPI Multiprocessor PC" can use more than one core. To work best, it
helps if interrupts are steered with APIC, as even interrupts can be
steered by the OS, to a particular core, for best load sharing.

I think it's possible, that you could leave the Multiprocessor HAL
in place, even if switching back to a single core.

What you have to be careful of, is the "ACPI" part of the Computer entry
in Device Manager. If you disable ACPI in the BIOS, you can force the
OS to respond, by changing the HAL. And then, it will be difficult or
impossible to fix.

So the transition between "ACPI Uniprocessor PC" and "ACPI Multiprocessor PC"
is relatively easy. Whereas, if you find "Standard PC" is the HAL value in
the computer entry, you're in much more trouble. (The last time that
happened to me, I did a Repair Install.)

http://en.wikipedia.org/wiki/Hardware_abstraction_layer

"hide differences in hardware from most of the operating system kernel"

*******

To do core unlocking, you need BIOS support. Perhaps even chipset
support. This is called "Advanced Clock Calibration" or ACC.

This was introduced with a particular AMD Southbridge. I think
perhaps even Nvidia managed to include support on one of their
Southbridge chips.

Later, it was included on other motherboards, but whether it
still needs particular Southbridges, I'm not sure.

http://en.wikipedia.org/wiki/AMD_700_chipset_series#Advanced_Clock_Calibration

AMD also has the option of disabling that entirely. Most likely,
by just removing the core and making the silicon die smaller. It
all depends on whether they feel this feature, reduces demand for
their slightly more expensive processors.

When you enable a second core, it must be tested thoroughly for
stability. I would do that with a Linux LiveCD, rather than messing
up Windows immediately.

Paul
 
I think it's as easy as running msconfig, hitting the 'boot ini' tab, click
on the 'advanced options' and choosing the '/procnum' that are available.

Getting that second core unlocked so that XP sees it's available is another
question.

Thanks for the (copied) tip -- duplicated Ok for msconfig: [procnum(s)
avail] section/screen;- now I'll search around off the Gigabyte MB
mod# both for setting up a unlocked multicore, or at least try that
BIOS "virtual switch" for what if anything happens. The MB mentions
its utility CD disc overclocks through software, which is kind of
weird because I didn't see anything at all in the BIOS directly
related to clock multiples, mem, or the FSB. Yes, anything else would
be be potluck as an intended cheap backup to a very old AMD
NewCastle754/ASUS with suspect capictatertots.
 
Then I go off and stumble upon a read, that all Sempron
productions are dual cores with a core locked. ...
I've heard mentioned, is XP
requires a fresh install for dual core support).

You have an XP OS, I'd try the multi core then run a repair install on it.
load all of the new drivers, just to see if it works - Your going to delete
it anyways.
----

No answer, just some notes I took:

AMD Locked cores
No cite, Google claims it might of come from here:
http://forums.amd.com/gaming/textthread.cfm?catid=259&threadid=110481

--------------------------ACC------------------
This is what I know. I'm sure you have the same information, but in case
there are any of those who have not done the research for SB750 Southbridge
reviews....

ACC has something to do with overclocking the Phenom line-up. From one
review, they speculated that it has something to do with how the board and
processor "talk" to each other or something. With ACC enabled in BIOS' that
support it, and if you have a Phenom I chip installed on that board, it can
allow anywhere from an extra 200-400mhz on average. There are different
settings which the user can set, for ACC's aggressiveness.

Phenom II has ACC hard baked into the chip. Meaning, ACC is already enabled
and ready to go on-chip. You do not need a BIOS which supports it, just as
long as the motherboard and BIOS support a Phenom II. When review sites
have used Phenom II chips in boards with ACC, enabling ACC did not do
anything to help raise clocks higher.

The only instance where having an SB750 Southbridge chip would be
beneficial is when you have a tri-core Phenom II and enabling ACC can
sometimes unlock the 4th core (generally core 03).
----

How many cores does your system see?

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session
Manager\Environment

Mine:
Key Name: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session
Manager\Environment
Class Name: <NO CLASS>
Last Write Time: 6/28/2011 - 5:45 AM
Value 0
Name: ComSpec
Type: REG_EXPAND_SZ
Data: %SystemRoot%\system32\cmd.exe

Value 1
Name: NUMBER_OF_PROCESSORS
Type: REG_SZ
Data: 8 (my addition: I have 4 cores +4 threads)

Value 2
Name: OS
Type: REG_SZ
Data: Windows_NT

Value 3
Name: Path
Type: REG_EXPAND_SZ
Data: C:\Program Files (x86)\NVIDIA
Corporation\PhysX\Common;%SystemRoot%\system32;%SystemRoot%;%SystemRoot%\System32\Wbem

Value 4
Name: PATHEXT
Type: REG_SZ
Data: .COM;.EXE;.BAT;.CMD;.VBS;.VBE;.JS;.JSE;.WSF;.WSH;.SH

Value 5
Name: PROCESSOR_ARCHITECTURE
Type: REG_SZ
Data: AMD64 (my addition: Intel uses the AMD 64 standard)

Value 6
Name: PROCESSOR_IDENTIFIER
Type: REG_SZ
Data: EM64T Family 6 Model 26 Stepping 5, GenuineIntel

Value 7
Name: PROCESSOR_LEVEL (my addition: refers to "CPU family")
Type: REG_SZ
Data: 6

Value 8
Name: PROCESSOR_REVISION
Type: REG_SZ
Data: 1a05

Value 9
Name: SAN_DIR
Type: REG_SZ
Data: C:\Program Files\SiSoftware2010

Value 10
Name: TEMP
Type: REG_EXPAND_SZ
Data: %SystemRoot%\TEMP

Value 11
Name: TMP
Type: REG_EXPAND_SZ
Data: %SystemRoot%\TEMP

Value 12
Name: windir
Type: REG_EXPAND_SZ
Data: %SystemRoot%
 
Flasherly said:
I think it's as easy as running msconfig, hitting the 'boot ini' tab, click
on the 'advanced options' and choosing the '/procnum' that are available.

Getting that second core unlocked so that XP sees it's available is another
question.

Thanks for the (copied) tip -- duplicated Ok for msconfig: [procnum(s)
avail] section/screen;- now I'll search around off the Gigabyte MB
mod# both for setting up a unlocked multicore, or at least try that
BIOS "virtual switch" for what if anything happens. The MB mentions
its utility CD disc overclocks through software, which is kind of
weird because I didn't see anything at all in the BIOS directly
related to clock multiples, mem, or the FSB. Yes, anything else would
be be potluck as an intended cheap backup to a very old AMD
NewCastle754/ASUS with suspect capictatertots.

Did S754 have any dual core processors, at all ?

I don't see any here, so it would be hard to unlock them, if they
aren't present.

http://products.amd.com/en-us/Deskt...f7=&f8=&f9=1600&f10=False&f11=False&f12=False

Paul
 
http://en.wikipedia.org/wiki/AMD_700_chipset_series#Advanced_Clock_Calibration


AMD also has the option of disabling that entirely. Most likely,
by just removing the core and making the silicon die smaller. It
all depends on whether they feel this feature, reduces demand for
their slightly more expensive processors.

When you enable a second core, it must be tested thoroughly for
stability. I would do that with a Linux LiveCD, rather than messing
up Windows immediately.

Due to impurities in the silicon die onto which the CPU structure is
etched, a certian % of the CPUs produced unusable. At that point it is
cheaper to test what works and disable parts with problems and sell them
as cheaper models, than it is to recycle the dies and start from scratch.

This is the rason why CPUs are sold as Semprons -- there is not a
separate factory line specifically to make them, where AMD would pick
and choose how many cores to include; instead Semprons are imperfect
Athlons. Imperfect dual-core Athlons make Semprons with the damaged core
disabled; imperfect single-core Athlons make stripped single-core
Semprons with no extra core to enable.

The smaller the nm technology the greater the risk that an impurity will
cause a serious problem (smaller wires make it more likely an impurity
will be large enough to cut it off). The ratio of good CPUs / bad ones
depends on how the technology of producing the raw material has caught
up with the nm production technology and it's never quite good either way.

As with all chip technology though, the impurities cause a probability
of a problem rather than a certainty, by exposing the CPU to the correct
conditions it may still be operable -- for example in Intel CPUs,
Celerons were imperfect Pentiums, underclocked for stability or with
cut-off damaged parts of the L2 cache which represented a large part of
the die surface. In modern AMDs, it seems the distinction is purely
software, enabling you to use software to enable the damaged
components... of course at your own responsibility -- no factory
guarantee that they will actually function.

LP,
Jure
 
Jure said:
Due to impurities in the silicon die onto which the CPU structure is
etched, a certian % of the CPUs produced unusable. At that point it is
cheaper to test what works and disable parts with problems and sell them
as cheaper models, than it is to recycle the dies and start from scratch.

This is the rason why CPUs are sold as Semprons -- there is not a
separate factory line specifically to make them, where AMD would pick
and choose how many cores to include; instead Semprons are imperfect
Athlons. Imperfect dual-core Athlons make Semprons with the damaged core
disabled; imperfect single-core Athlons make stripped single-core
Semprons with no extra core to enable.

The smaller the nm technology the greater the risk that an impurity will
cause a serious problem (smaller wires make it more likely an impurity
will be large enough to cut it off). The ratio of good CPUs / bad ones
depends on how the technology of producing the raw material has caught
up with the nm production technology and it's never quite good either way.

As with all chip technology though, the impurities cause a probability
of a problem rather than a certainty, by exposing the CPU to the correct
conditions it may still be operable -- for example in Intel CPUs,
Celerons were imperfect Pentiums, underclocked for stability or with
cut-off damaged parts of the L2 cache which represented a large part of
the die surface. In modern AMDs, it seems the distinction is purely
software, enabling you to use software to enable the damaged
components... of course at your own responsibility -- no factory
guarantee that they will actually function.

LP,
Jure

Flasherly mentioned in his second post, the processor is S754, and
I don't think any of those have a second core. As near as I can determine,
all the processors offered for S754 were single core. Which means there
is not a second core to enable, unlock etc.

Paul
 
Due to impurities in the silicon die onto which the CPU structure is
etched, a certian % of the CPUs produced unusable. At that point it is
cheaper to test what works and disable parts with problems and sell them
as cheaper models, than it is to recycle the dies and start from scratch.

It's not at all easy to recycle--the main cost of the raw materials is
purifying them to the degree required for chip-making. In making the
chip a bunch of impurities were dumped in. If you recycle them you
must then purify the material again.>
the die surface. In modern AMDs, it seems the distinction is purely
software, enabling you to use software to enable the damaged
components... of course at your own responsibility -- no factory
guarantee that they will actually function.

But it actually might be a good chip. If they don't have enough
failures to meet demand for the lower-end chips they'll take perfectly
good chips and degrade them.
 
Did S754 have any dual core processors, at all ?

I don't see any here, so it would be hard to unlock them, if they
aren't present.

(These Google DOS attacks or whatever are going to force me into a
dedicated reader USENET feed).

Not sure, possibly not.

I bit off more than I should have first chewed with this Gigabyte. First
off it's not the Sempron I misquoted - AMD/Orleans/AM2.

1: I said "new build" - checked the purchase dates after some
(forthcoming) problems. Bought that thing 3 years ago (thinking it's been
sitting a parts shelf a year, not much more)! MB has some questionable
facets (reviewed) to it for its production timeframe (old chipset BIOS
issues potentially foisted off at the time of mass marketing). In
hindsight I could have reviewed that MB for better purchase options (where
have I heard that before?).

AMD Athlon 64 LE-1640 Orleans 2.6GHz Socket AM2 45W Single-Core
Processor ADH1640DHBOX
Item #: N82E16819103239
--Processors (CPUs) Return Policy $35.99

G.SKILL 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Dual
Channel Kit Desktop Memory Model F2-6400CL5D-2GBNQ
Item #: N82E16820231098
Limited Non-Refundable 30-Day Return Policy $31.99 2

GIGABYTE GA-M61PME-S2 AM2 NVIDIA GeForce 6100 / nForce 430 Micro ATX
AMD Motherboard
Item #: N82E16813128333
30 Day Return Policy
Gigabyte (626)-854-9338 option 4 www.giga-byte.com $48.99


2: Weird PS issues. Fired up OK w/ a Sparkle/Fortron 400W active. Next
time, no change straight into the wall, nada (CPU fan wants to spin but
won't). Tried an OCZ PS swap - nothing (as in dead). Both used PS units
(Fortron long-time heavily, OCZ a couple years). Both with prior
problems/issues evinced on the S756 (semi-addressed stably with with an
ANTEC BRONZE -- 650W? whatever and spanking new). Got the
Fortron-SPARKLE/Gigabyte back up eventually after coming off the outlet
shared with a 115V AC, to another empty house powerleg. Took a few times,
too. Power supplies, in my experience, a plagued PS-user, can offer a lot
of 1st-time experiences with unbelievable problems.

Maybe. What I did was order another PS, 3-day delivery from Newegg, its
housebranded stuff, a Roswill marketed to compete with "the bronze" types
for $29 on a $10 discount code marked already from $65 to $39. $29
shipped at a 1/4 bad reviews over 80. We'll see. If I don't, what I may
be looking at my first time up with Gigabyte is a replacement.
 
Update - just "smoked" the FORTRON 400W. Big Blue flash out the back,
behind the fan, which I'll salvage off both the OCZ. Seems they'd be fans
built to last, & should be good hard-drive plane candidates. Good news,
found a destitute PSP 300W in a white foam bag off The Shelf of Forgotten
Parts. Looks brand new & never used. Fires right up, though believe that
MB may be wanting a 500W unit, especially w/ PCI express video. Beggars
and limp-alongs for the MB's video chipset, with any luck before a new PS
arrives to set up -a- HD. Killed more than a few PS units, but this is
the first time I've smoked one. If painted w/ temperature-release perfume
coatings, they wouldn't stink up a house so bad.
 
"I think" I've heard mentioned, is XP
requires a fresh install for dual core support).

No there's a program in the XP resource kit called UPTOMP.EXE
that you can use to change the HAL from single core to multi-core.

Looking for the XP version (I have the NT one) I found what Paul
was talking about (ACPI) this link is meant to complement his post.
http://support.microsoft.com/?id=309283
 
No there's a program in the XP resource kit called UPTOMP.EXE

Forgot to mention I just found this file yesterday and the reason for
two post and two suggestions. Actually forgot about the first one until the
two were together.

I've had the NT Resource Kit forever and I'm still finding programs I'd
forgotten about . I had absolutly no use for them at the time and figured I'd
never use a program called (ie:)"UniProcessorTOMultiProcessor upgrade utility"
yet kept it/them only because they were part of the resource kit - I like to
keep things intact or originial as possible.
 
No there's a program in the XP resource kit called UPTOMP.EXE
that you can use to change the HAL from single core to multi-core.

Looking for the XP version (I have the NT one) I found what Paul
was talking about (ACPI) this link is meant to complement his post.
http://support.microsoft.com/?id=309283


Squeezed by, still SP2 (not 3) here.** I'll keep in mind UPTOMP. It's
all backburner from here on, though. I'd mentioned -mis-id'ing- what
I do have, an ORLEANS 2.6G unicore, not an AMD rebadged/locked Sempron
core.

Couldn't say more as yet -- cleaned everything up in an aluminum
Lanboy I may prefer over a Rosewill Antec 300 steel clone, the
*modest* Gigabyte MB/Orleans went into (after smoking a Fortron server
grade 400W PS). Tested w/ a cheap-o 300W extra, and now waiting on
delivery of Newegg's Rosewill clone of higher-end "Bronze" (Antec's)
spec'd PS (500W-ish recommend in the MB, coincidentally).

Checked Gigabyte's MB specsheets on this, and with BIOS revisions, the
sky appears pretty much the limit for AMD's array up to 6-cores.
(Seems somewhat "shrunken" with around 30 platform offerings these
days). Nice to know, anyway, for thingies to keep an eye out for --
damn MBs are like crackcandy for the unwary. . .must have a hi-end
wick CPU cooler replacement, a PCI-express GPU, extra PCI HD
controllers. . . $200 6-core potentials.

Well, for Gigabyte's one below the low-end of a solid state MB
capicitors (*more* yumyumies) -- there's no "secret" BIOS, ACPI
hotkeys or whatnot. Best I can go figure is download AMD's Orleans
spec sheet for verification that the provided MB software shows true
on everything else this BIOS won't do.

It's dated popularity I gather was in Hewlet Packard mass production
line. Many people were then upgrading this MB for the Orleans in
server configs. Your basic workstation otherwise now.

** MS: Computers equipped with multiple processors that support
Advanced Configuration and Power Interface (ACPI) processor
performance states, require Microsoft Windows XP Service Pack 2 (SP2).
Without these updates, computers that are equipped with these power
management-capable, mobile, dual-core processors may experience
decreased performance or unexpected behavior.
 
Back
Top