*Official Retail G0 Q6600 Overclocking Thread*

[Brother Esau]

are you using same test set up as in your sig the eVGA 680i with P30 bios, cause i would really like to know how
well your G0 does with FSB scaling on
the eVGA 680i with the new P30 bios, please post pics of any test of high fsb if possible many thanks.....

regards dyn.

@ Dynasty High FSB with what Stock Multi X9 or just in general and not being Picky

[Dynasty]

@ Dynasty High FSB with what Stock Multi X9 or just in general and not being Picky

whatever gives you most fsb ......I seen some quad G0's go as high as
425 Fsb with new P30 bios................

[drunkenmaster]

just to introduce a teeny tiny bit of reality in the situation.

on release, teh Q6600 has a WORST SITUATION of 105W's. ok in intels case its a percentage of the actual theoretical maximum based on their "average" usage numbers. but thats a constant method/percentage anyway.

but 6 months ago, the worst chips about wouldn't break the 105W, they have to be a little conservative aswell so even the worst chips are unlikely to actually hit 105W on the dot. either way, the process becomes older, they find new ways to tweak yields up a little, quality of chips increases, the wattage decreases a little further with every little trick they find. over several months they get the wattage down , after 4 months a few chips but not many actually run with 95W, but still the worst ones are at 101W's. another few months later and the worst chips are doing 95W now. so what do they do, its time to remarket and rebrand, its their low cost quad core, dell and other people want to hit the sub 100W mark to go in cheaper computers so they call the new stepping G0 and its rated at 95W now.

not all use 95W, some are still better. but its not as if, they walk in one day and stay, stop production on those B3's all using 105W's, we'll use this brand new G0 at 95W. they simply change the numbers they put on the cpu after they hit the point where no more chips were over 95W. in all likelyhood the last month or two of b3's have mostly been sub 95W aswell.


also, the next time someone does a look over here G0 results thread. stop destroying every thread. start a threads saying "where did you get your G0". and leave the show us your G0 thread, for people with G0's.

i stopped reading xtreme properly a long time ago because you click on a thread with one name, and nothing in it is relevant anymore.

i'm not a thread of track = banning cyber nazi, but jsut once in a while it would be nice to click on a thread that should give me an easy 1 page look at how G0's are doing, and not read 7 pages to find 3 freaking results.

[Hey_Its_Cole]

just to introduce a teeny tiny bit of reality in the situation.

on release, teh Q6600 has a WORST SITUATION of 105W's. ok in intels case its a percentage of the actual theoretical maximum based on their "average" usage numbers. but thats a constant method/percentage anyway.

but 6 months ago, the worst chips about wouldn't break the 105W, they have to be a little conservative aswell so even the worst chips are unlikely to actually hit 105W on the dot. either way, the process becomes older, they find new ways to tweak yields up a little, quality of chips increases, the wattage decreases a little further with every little trick they find. over several months they get the wattage down , after 4 months a few chips but not many actually run with 95W, but still the worst ones are at 101W's. another few months later and the worst chips are doing 95W now. so what do they do, its time to remarket and rebrand, its their low cost quad core, dell and other people want to hit the sub 100W mark to go in cheaper computers so they call the new stepping G0 and its rated at 95W now.

not all use 95W, some are still better. but its not as if, they walk in one day and stay, stop production on those B3's all using 105W's, we'll use this brand new G0 at 95W. they simply change the numbers they put on the cpu after they hit the point where no more chips were over 95W. in all likelyhood the last month or two of b3's have mostly been sub 95W aswell.


also, the next time someone does a look over here G0 results thread. stop destroying every thread. start a threads saying "where did you get your G0". and leave the show us your G0 thread, for people with G0's.

i stopped reading xtreme properly a long time ago because you click on a thread with one name, and nothing in it is relevant anymore.

i'm not a thread of track = banning cyber nazi, but jsut once in a while it would be nice to click on a thread that should give me an easy 1 page look at how G0's are doing, and not read 7 pages to find 3 freaking results.

I agree 100%, it always seems like I have to read 10 pages of crap to get any information.

[synergy]

I'm still not sure what EXACTLY is *really* different
about the Q6600 G0 vs the B3 other than LABELING.

From Intel's PCN Document:

Description of Change to the Customer:
The Intel® Core™2 Quad processor Q6600 and Intel® Xeon® processors X3220 and X3210 will undergo
the following changes for the B-3 to G-0 processor stepping conversion:
• CPUID will change from 06F7 to 06FB
• New S-specs for converting products

Ok so they changed a couple of labels on the package
and in the ID codes stored in the chip. No big deal.

• The converting products will change from the 2005 Performance FMB(105W) to the 2005
Mainstream FMB (95W)

Ok, so they found that 99.99% of their actual parts
never REALLY used 105W and that was too conservative
of a maximum specified "possible" temperature, so they
lowered the specified number on the label to guarantee
that they'll all use 95W or less under whatever test
conditions they use. For all I know 99.999% of the
B3 stepping chips ALSO use less than 95W under the same
test conditions. It's not clear that they actually CHANGED
anything to make this G0 take LESS power, they may
have just changed their guarantee / test criteria.

BUT if they DID actually reduce the true "real world"
power consumption of G0 vs. B3, HOW did they do that?
You don't get less power consumption for free; you
could slow something down so that there's
less performance SOMEWHERE so that less power on
average is consumer.

They certainly didn't do a die shrink to 45nm for G0,
since that's not specified as a change, the voltages of
G0 and B3 are the same, and so on.

So is this just a reLABELING of the guaranteed temp.,
or do the G0 chips REALLY use LESS power doing the same
things as the B3s, and if so, WHY?

• The Electrical, Mechanical and Thermal Specifications remain within the current specifications.
Intel anticipates no changes to customer platforms designed to Intel guidelines.
o Tcase for the Intel® Core™2 Quad processor Q6600 and Intel® Xeon® processors
X3220 and X3210 on G-0 stepping has been increased by 11 oC. Tcontrol offset will
remain the same relative to increase in Tcase which will help reduce acoustics

Ok so basically they are allowing the chip to run
HOTTER on G0s because they reLABELED the max Tcase
value. Translation: they got too many reports of people's
CPUs hitting the thermal limit and shutting down, and/or
they got too many complaints of PCs that were too noise
with the CPU fan running at 100% 'always' because the
Tcase under load was near the old 'limit'. So they said
"ahh let it run hotter and quieter, it won't change much
the number of CPUs that burn out under warranty, but
it'll let us sell a lot more because they don't sound like
rocket engines and trigger too many temperature warnings".

Customer Impact of Change and Recommended Action:
Minimal re-qualification and/or validation is expected for the G-0 stepping conversion due to no feature set
changes between the B-3 and G-0 steppings. Thermal qualification may be required due to increase in
Tcase.

Ok so if your poorly ventilated micro-ATX was about
to melt down before, it'll be worse now because they
let the CPU fan spin slower and the CPU get hotter.
This has nothing necessarily to do with any REAL
electrical change between G0 and B3.

The Intel® Core™2 Quad desktop processors Q6600 and Intel® Xeon® processors X3220 and
X3210 G-0 stepping will require a BIOS update.

Ok why the heck would it need a BIOS update, REALLY?
Presumably BIOSes ALL know how to read
Tcase MAX, Tcontrol FAN, CPU ID, etc. out of ANY
similar Intel Core 2 DUO/QUAD processor, and though
the numbers stored for those values changed slightly,
they say NOTHING else of a FUNCTIONAL electrical /
thermal nature changed with G0 stepping.
The voltage did not change, the frequency did not change,
it has no new 'features' or instructions.....

So it seems confusing why you'd NEED a BIOS change;
even the "GENUINE INTEL Q6600 G0" or whatever
processor model description ID string is hard-coded
into the CPUID instruction text (as far as I recall),
so even just to display the text identification of what
your CPU is, the BIOS would need to do NOTHING
new compared to the way ANY other Intel Core2
DUO/QUAD processor has always worked.

So enlighten me, EMPIRICALLY is there any
PERFORMANCE difference or
ACTUAL RUNNING TEMPERATURE difference between
B3 and G0 Q6600 when the
Vcore / Frequency / CPU fan speed / program running
is forced to be the same in a fair test?

Sounds like they changed NOTHING useful, or if they did,
they MIGHT have lessened the ACTUAL power/temperature,
and IF they DID it, I have yet to see a good explanation
of HOW and IF there were any performance sacrifices to
do that.

Sure you could find some manufacturing process change
that might get you 10% better thermal/electrical efficiency
by using a lower resistivity metal or better quality silicon
or lower loss dielectric or whatever, but since the
HighK dielectric and 45nm and other major fab. process
related changes are only happening for PENRYN et. al.
what's the deal with THIS change, really?

[dinos22]

So is this just a reLABELING of the guaranteed temp.,
or do the G0 chips REALLY use LESS power doing the same
things as the B3s, and if so, WHY??

you will find that the binning voltage was reduced

VID value you see in people's coretemp screenshots

[synergy]

Thanks for the reply.

I guess I don't TOTALLY understand how VID works in
the Core2 lineup.

AFAIK there's a VID/FID table of certain divisors that say
that at each SpeedStep or ACPI performance level
the processor should run with X voltage and F frequency.

When you're running at MAXIMUM STOCK performance
the frequency will be whatever max stock frequency is
for your model of CPU, and the VID will be the maximum
it'll be for any defined state.

BUT didn't they start making the ACTUAL voltage for
EACH INDIVIDUAL CPU different within some allowed
range? I recall that AMD did that a while back.
I recall that INTEL made the MAX CASE TEMP. value
stored in the CPU *different* for *each* CPU based
on their production testing if I understand that right.

So didn't they also start making the VID change slightly
for each CPU they make depending on testing at that
time also?

Also what ELSE does VID determine except for VCORE?
If you go into your motherboard's BIOS and instead
of using AUTO you manually FORCE VCORE=1.40
and manually FORCE FSB=390MHz or
whatever, isn't it essentially ignoring VID and FID
totally at that point?

Or does VID *also* control other voltages besides VCORE
e.g. some kind of I/O power supply voltage or whatever
else that the VRM might still be supplying at values
depending on VID in the table corresponding to
"full power" speed step even when you've manually
overridden VCORE?

So if I'm guessing right (please correct me if I'm wrong!),
then VID listed in the software just doesn't matter
if you've set VCORE manually?

Also IF VID is actually potentially different for each
and every CPU out there, do we typically see different
values of VID even for peoples different chips of the
same stepping? e.g. do ALL the B3 chips out there have
the SAME VID/FID table or not?

It seems like a lower VID table entry for any given
FID could just mean "yes, we realize that 99.999% of
all CPUs we make can overclock at least 10% on stock
voltage, and conversely they can mostly all run at
stock speed if they're undervolted by 10%, so we'll
just lower the VID voltage by a few percent for
any given FID step and guarantee that they'll work
by quality control so that everyone can save some
noise/power". In that case, there may be no real technical
engineering change that MADE the chip run faster at
lower power, because in fact we all know that they could
be undervolted/overclocked at least a little even with B3.

you will find that the binning voltage was reduced

VID value you see in people's coretemp screenshots

[GripS]

Some valid points made. Seriously contemplating just giving up the 5 bucks to tankguy's if these 'G0' stepping aren't all they're cracked up to be.

[rogueagent6]

Before everyone starts freaking out because of high temps with a storm on a Q6600... Cathar, who designed the Storm, has stated numerous times that the cooling patch on the Storm is TOO SMALL FOR QUADS.

Sheesh.

Go visit the water cooling forums once in awhile.