Core 2 Extreme QX9650 rig, my O/C swansong?

Phil recently posted here about how the 'mighty have fallen' w/r/t some of
the heroic tales of overclocks here in years past. I agreed, saying that
there is so much power in modern CPUs that the benefit of squeezing every
last MHz out of your system is hardly apparent other than when you run
(increasingly complex) benchmarks.

I remember my first 'epic' o/c, getting a Celeron 600 (Coppermine)
rock-solid stable at 900MHz, a 50% o/c, using nothing more complex to cool
it than a fairly small hunk of aluminium with fins with a fan screwed to it.
*That* was noticable, very much so. I was getting equal performance of a
high-end PIII out of a budget processor and my games flew.

That wasn't my first o/c, that honour belongs to a Pentium 90 that I coaxed
into running at 100MHz. A mere 10% o/c. Since then I've overclocked
everything I could to varying degrees, and if I couldn't o/c it (i.e.
laptops) I'd undervolt it. I left Intel for AMD when they went the way of
that abomination, the Pentium 4. My last AMD CPU, an unlocked Barton 2500+
(running at 3200+ speed) saw me through to my first Core 2 Duo (a direct
descendant of the PIII with very little of the despised P4 DNA).

It was about then, with my E4500, that I started to realise that I was
overclocking for the sake of it, going through the motions but really the
only noticable result of my efforts was that I was having to pay for more
electricity. I managed to get that 2.2GHz dual-core CPU running at 3.3GHz,
another 50%er, and justified it to myself by running SETI (which I'd been
doing from the first week of it 'going live' anyway). My electricity bill
and my SETI score were the only real indicators of my overclocking, I'm not
into bleeding-edge gaming so have always used merely 'mid-range' graphics
cards.

Then, in about 2006 I started building what, although I didn't realise it at
the time, was to (probably) be my last desktop machine. (I'm an invalid on
welfare with a compression-fracture of the lower back, a chronic-pain
patient in constant torment. I lost my home, my business and my life-savings
trying to work through my injury, refusing to accept that I couldn't
continue with my life-plan. I'd just started my business, borrowing against
my home and, pre-injury, after years of working for the man, I was set to
'make it'.)

Over the years I've had to abandon all of my cherished vices such as
tobacco, alcohol, playing pool, fishing, socialising etcetera etcetera ad
nauseum, due to lack of funds. The last bastion of my previously hedonistic
lifestyle that I clung to was building and overclocking computers. As I
couldn't afford to keep buying new hardware myself I often got my fix by
building and o/cing rigs for my more affluent 'friends'.

This last build of mine, that started on '06 and (almost) ended yesterday
started with an Asus P5KE-WiFi-AP, a mobo that supported the upcoming 45nm
CPUs, had Japanese-sourced solid electrolytic capacitors (as I'd re-capped
far more mobos than I care to think about) and a then staggering eight-phase
VRM. I'd just finished assembling a rig for a gamer friend which utilised
the P5K Deluxe and a 'G0' Q6600. It was a great board to use and o/c
(especially once Asus released a few BIOS updates) and the far cheaper E
version that I bought, other than some fancy SB / NB / VRM heatsinking /
piping and one or two 'features' that I'd never use, was essentially
indentical. They were built on the same PCB and used the same VRM.

It was always my aim to run a fast 45nm C2Q in it. I wasn't going to mess
with a 65nm quad for myself when, at the time my E4500 was doing everything
that I wanted with plenty in reserve. I would wait until the price of this
'new' CPU was within my reach. After a year or so I replaced the E4500 with
an E7400, a more efficient 45nm CPU that dissapointingly didn't o/c any
higher than the 'slower' 65nm CPU that it replaced. However it *was* more
efficient and ran a lot cooler.

The CPU cooler that I'd chosen to run in this system was a Thermaltake
Mini-Typhoon but with a Thermalright 'bolt-thru' kit instead of those horrid
plastic push-pins. It didn't have the numbers of something like a T.R.U.E.
or a Scythe but I've always liked a 'top-down' cooler for the way that they
keep circulation around the VRMs, essentially doing double-duty. I also
liked the all-copper construction and the fact that it wasn't massive. A
while before I'd used the copper-cored factory Q6600 HS/F that was discarded
from the above-mentioned friend's build to cool my E4500 to it's 50% o/c.

The Tt Mini-T didn't allow me to o/c it higher but did run it quite a bit
cooler. I wasn't overly-impressed with the three-wire fan on the Tt though
so, with a Dremel and some Araldite, I adapted the Intel Q6600 four-wire fan
to fit the Tt (and the four-wire fan header on the mobo) as they're both
90mm. Now the mobo could control the fan speed, keeping it at a silent 1200
rpm at idle and ramping up to 3000+ rpm as the temps went up. The Intel fan
moved a lot more air than the supplied Tt fan when called upon to do so
which suits me just fine as that's when the area around the CPU needs most
cooling.

Anyway, to the present-day. This week I *finally* took delivery of my fast
45nm C2Q, four years after the project started. It's a Core 2 Extreme QX9650
as mentioned in the subject. A 3GHz, 12MB L2 CPU with an unlocked
multiplier, a CPU that I've coveted since it was released in '07 (at
NZ$1,950 per unit). I got mine, a brand new (even though they were
discontinued in '09) OEM 'tray' CPU for NZ$400. I nearly didn't go through
with it as, for the last couple of years I've been using a ThinkPad T60
laptop (with a beautiful IPS 4:3 1400 x 1050 display, T7400 2.16GHz C2D, 3GB
RAM, 7.2K HDD...) as my day-to-day machine and it does all that I want it to
easilly.

However I wanted to finish the project that I started back in '06, even if I
only run the machine for maybe a couple days a month, so I put the QX9650
'on the plastic'. I probably don't need the processing-power that this rig
will provide but, more than anything I guess, I wanted to 'finish' my
project. The CPU arrived yesterday so, pre-fitting it I updated the mobo
BIOS (I was three releases behind) and cleaned the small amount of dust from
the fans and fins. Then I fitted the QX9650 with the last of my AS5 and
fired it up at stock speed.

There were no dramas and it idled at ~12ºC above room temp, ~6ºC above case
temp. There are four 7.2K HDDs in the case, in removable bays, set up as
hot-swappable. The case is an iCute and has a 24cm / 600rpm fan in the side.
I ran Prime95 for ~15 mins and the core temps got up to around ambient +
25ºC (A+25). Then, as I know that the mobo and the RAM both handle it with
aplomb and I like 1:1 FSB / RAM I upped the CPU FSB from it's stock 333
(1333 if you're into marketing terminology) to 400MHz* and left the
multilpier and vcore at stock. It fired up at 3.6GHz and I left it running
Prime while I Googled o/cing a QX9650.

(* I learned a long time ago that X CPU running at X GHz isn't the whole
story by far. The FSB speed is very important. For instance, as I discovered
with the unlocked Athlon XP T'breds and Bartons, 2GHz with a 200MHz FSB
needs a tad more vcore, runs slightly hotter and benchmarks higher than the
same CPU at 2GHz on a 166 FSB. Therefore, when overclocking an unlocked CPU
I decide on what FSB speed I want to run, then take it from there with the
multiplier. This board will run 500MHz FSB just fine but then I need to run
the RAM at 4:5 and experience has shown me that a 1:1 ratio is slightly
better. The board and the DDR2-800 RAM will run at 450MHz but I need to feed
it a little more VDIMM above stock and DDR2 RAM isn't getting cheaper. I
want the 6GB (2 x 2GB + 2 x 1GB) that I have to last the life of the
machine. Ergo my optimum FSB is 400MHz.)

This is turning into a long post but rest assured, it's likely my last ever
post about my overclocking.

Anyway, while the machine was running Prime at 3.6GHz I found many, many
pages dealing with o/cing the QX9650. I guess that's an advantage of buying
a CPU that's been around a few years. I opened a bunch of pages in tabs on
the laptop. Then, as Prime hadn't thrown any errors I re-booted the heavy
iron, setting the multipier to 10 (4GHz) while leaving the vcore at stock
(1.240V according to CPU-Z). However althoug it POSTed fine it wouldn't go
into Windows so I dropped the multi 0.5 (3.8GHz) and it booted fine at
default vcore. I set about some reading while I left Prime running.

From my reading it seems that most all QX9650s will run at 4GHz on air, even
using the stock HS/F (which I didn't get with my OEM CPU) with just a small
vcore increase as long as the case is well-ventilated. They're all C1
stepping as far as I can tell so, other than the luck of the draw with the
silicon, identical. Most sources agreed that 4.25GHz was easilly achieveable
with water cooling and taking the vcore up to around 1.45V but that any
higher than that needed a big vcore boost and resulted in a *lot* of heat.

The best source that I found on o/cing these CPUs, which was referenced in a
lot of forums is http://www.anandtech.com/show/2404 "Overclocking Intel's
New 45nm QX9650: The Rules Have Changed" from December '07. It's well worth
a read IMO, even if you don't have a QX9650 or even a 45nm Socket 775 CPU.
It's a well-researched and well-writen article.

Anyway, a passage in that article made me change my mind about finding the
limits of my 'new' CPU. It basically re-stated what I'd said in response to
Phil's post a month or two back. I quote:
........................................
"We honestly believe a new direction in CPU overclocking may soon be upon
us. While there will always be those that continue to push processors to
their absolute limits, the majority of us will find our new "performance"
benchmark in efficiency. This makes sense though - the market has been
heading this way for years now and overclockers may have simply chosen to
ignore the obvious. The multi-core era we now live in places a heavy
emphasis on performance-per-watt figures and other measurable efficiencies."
........................................
Heh! I thought I was being original when I said similar here yet here it is
from a three-year old article.

So, my motivation in finishing what is likely to be my last 'new' build was
to get the best possible (or close to it) Socket 775 system and, by adding a
bit more money to what I'd already spent on it, have a machine that would
hopefully last me many years without being gutless.

So, for now I left my FSB at 400MHz but dropped the multi to 7.5 so that the
CPU is running at stock speed, 3.0GHz. However I *dropped* the vcore to the
lowest that my mobo will allow, 1.10V in BIOS, 1.04V in windows under full
load according to CPU-Z. The machine ran Prime95 all night with those
settings and the CPU fan didn't have to increase above it's 1200rpm minimum
to keep the cores below ~45ºC, A+25. There is plenty of power in this
machine to last me a long time at that speed as far as I can tell.

So, perhaps this isn't my last overclocking story. Maybe, in a year or three
I might have to squeeze some more grunt out of this CPU. Watch this
space.....

Thanks for reading this far (if you did). May all your boxen be cool and
stable.
--
Shaun.

"He who fights with monsters might take care lest he thereby become a
monster. And if you gaze for long into an abyss, the abyss gazes also
into you." Friedrich Wilhelm Nietzsche

I read all of it thinking ah!! the good ole days........
I feel that with every new chipset/CPU coming out nowadays
there really is not much of an advantage to overclocking unless
you want bragging rights...
If you check the "reviews" of the last few mobo you will note a vast
increase in power consumption...but then that never stopped anyone in the
past
and I feel it wont stop anyone except us older users now.
I run a Core2 duo that is oc'd by 40% on air and that's good enough for me.
I am in the market to buy another system..maybe the new Intel coming out
next month,Maybe a X58
chipset mobo and CPU to match...but I dont think I will be playing with
settings too much.

sorry to hear about your poor health

peter



If you find a posting or message from me offensive,inappropriate
or disruptive,please ignore it.
If you dont know how to ignore a posting complain
to me and I will be only too happy to demonstrate :-)
"~misfit~" wrote in message ...

Phil recently posted here about how the 'mighty have fallen' w/r/t some of
the heroic tales of overclocks here in years past. I agreed, saying that
there is so much power in modern CPUs that the benefit of squeezing every
last MHz out of your system is hardly apparent other than when you run
(increasingly complex) benchmarks.

I remember my first 'epic' o/c, getting a Celeron 600 (Coppermine)
rock-solid stable at 900MHz, a 50% o/c, using nothing more complex to cool
it than a fairly small hunk of aluminium with fins with a fan screwed to it.
*That* was noticable, very much so. I was getting equal performance of a
high-end PIII out of a budget processor and my games flew.

That wasn't my first o/c, that honour belongs to a Pentium 90 that I coaxed
into running at 100MHz. A mere 10% o/c. Since then I've overclocked
everything I could to varying degrees, and if I couldn't o/c it (i.e.
laptops) I'd undervolt it. I left Intel for AMD when they went the way of
that abomination, the Pentium 4. My last AMD CPU, an unlocked Barton 2500+
(running at 3200+ speed) saw me through to my first Core 2 Duo (a direct
descendant of the PIII with very little of the despised P4 DNA).

It was about then, with my E4500, that I started to realise that I was
overclocking for the sake of it, going through the motions but really the
only noticable result of my efforts was that I was having to pay for more
electricity. I managed to get that 2.2GHz dual-core CPU running at 3.3GHz,
another 50%er, and justified it to myself by running SETI (which I'd been
doing from the first week of it 'going live' anyway). My electricity bill
and my SETI score were the only real indicators of my overclocking, I'm not
into bleeding-edge gaming so have always used merely 'mid-range' graphics
cards.

Then, in about 2006 I started building what, although I didn't realise it at
the time, was to (probably) be my last desktop machine. (I'm an invalid on
welfare with a compression-fracture of the lower back, a chronic-pain
patient in constant torment. I lost my home, my business and my life-savings
trying to work through my injury, refusing to accept that I couldn't
continue with my life-plan. I'd just started my business, borrowing against
my home and, pre-injury, after years of working for the man, I was set to
'make it'.)

Over the years I've had to abandon all of my cherished vices such as
tobacco, alcohol, playing pool, fishing, socialising etcetera etcetera ad
nauseum, due to lack of funds. The last bastion of my previously hedonistic
lifestyle that I clung to was building and overclocking computers. As I
couldn't afford to keep buying new hardware myself I often got my fix by
building and o/cing rigs for my more affluent 'friends'.

This last build of mine, that started on '06 and (almost) ended yesterday
started with an Asus P5KE-WiFi-AP, a mobo that supported the upcoming 45nm
CPUs, had Japanese-sourced solid electrolytic capacitors (as I'd re-capped
far more mobos than I care to think about) and a then staggering eight-phase
VRM. I'd just finished assembling a rig for a gamer friend which utilised
the P5K Deluxe and a 'G0' Q6600. It was a great board to use and o/c
(especially once Asus released a few BIOS updates) and the far cheaper E
version that I bought, other than some fancy SB / NB / VRM heatsinking /
piping and one or two 'features' that I'd never use, was essentially
indentical. They were built on the same PCB and used the same VRM.

It was always my aim to run a fast 45nm C2Q in it. I wasn't going to mess
with a 65nm quad for myself when, at the time my E4500 was doing everything
that I wanted with plenty in reserve. I would wait until the price of this
'new' CPU was within my reach. After a year or so I replaced the E4500 with
an E7400, a more efficient 45nm CPU that dissapointingly didn't o/c any
higher than the 'slower' 65nm CPU that it replaced. However it *was* more
efficient and ran a lot cooler.

The CPU cooler that I'd chosen to run in this system was a Thermaltake
Mini-Typhoon but with a Thermalright 'bolt-thru' kit instead of those horrid
plastic push-pins. It didn't have the numbers of something like a T.R.U.E.
or a Scythe but I've always liked a 'top-down' cooler for the way that they
keep circulation around the VRMs, essentially doing double-duty. I also
liked the all-copper construction and the fact that it wasn't massive. A
while before I'd used the copper-cored factory Q6600 HS/F that was discarded
from the above-mentioned friend's build to cool my E4500 to it's 50% o/c.

The Tt Mini-T didn't allow me to o/c it higher but did run it quite a bit
cooler. I wasn't overly-impressed with the three-wire fan on the Tt though
so, with a Dremel and some Araldite, I adapted the Intel Q6600 four-wire fan
to fit the Tt (and the four-wire fan header on the mobo) as they're both
90mm. Now the mobo could control the fan speed, keeping it at a silent 1200
rpm at idle and ramping up to 3000+ rpm as the temps went up. The Intel fan
moved a lot more air than the supplied Tt fan when called upon to do so
which suits me just fine as that's when the area around the CPU needs most
cooling.

Anyway, to the present-day. This week I *finally* took delivery of my fast
45nm C2Q, four years after the project started. It's a Core 2 Extreme QX9650
as mentioned in the subject. A 3GHz, 12MB L2 CPU with an unlocked
multiplier, a CPU that I've coveted since it was released in '07 (at
NZ$1,950 per unit). I got mine, a brand new (even though they were
discontinued in '09) OEM 'tray' CPU for NZ$400. I nearly didn't go through
with it as, for the last couple of years I've been using a ThinkPad T60
laptop (with a beautiful IPS 4:3 1400 x 1050 display, T7400 2.16GHz C2D, 3GB
RAM, 7.2K HDD...) as my day-to-day machine and it does all that I want it to
easilly.

However I wanted to finish the project that I started back in '06, even if I
only run the machine for maybe a couple days a month, so I put the QX9650
'on the plastic'. I probably don't need the processing-power that this rig
will provide but, more than anything I guess, I wanted to 'finish' my
project. The CPU arrived yesterday so, pre-fitting it I updated the mobo
BIOS (I was three releases behind) and cleaned the small amount of dust from
the fans and fins. Then I fitted the QX9650 with the last of my AS5 and
fired it up at stock speed.

There were no dramas and it idled at ~12ºC above room temp, ~6ºC above case
temp. There are four 7.2K HDDs in the case, in removable bays, set up as
hot-swappable. The case is an iCute and has a 24cm / 600rpm fan in the side.
I ran Prime95 for ~15 mins and the core temps got up to around ambient +
25ºC (A+25). Then, as I know that the mobo and the RAM both handle it with
aplomb and I like 1:1 FSB / RAM I upped the CPU FSB from it's stock 333
(1333 if you're into marketing terminology) to 400MHz* and left the
multilpier and vcore at stock. It fired up at 3.6GHz and I left it running
Prime while I Googled o/cing a QX9650.

(* I learned a long time ago that X CPU running at X GHz isn't the whole
story by far. The FSB speed is very important. For instance, as I discovered
with the unlocked Athlon XP T'breds and Bartons, 2GHz with a 200MHz FSB
needs a tad more vcore, runs slightly hotter and benchmarks higher than the
same CPU at 2GHz on a 166 FSB. Therefore, when overclocking an unlocked CPU
I decide on what FSB speed I want to run, then take it from there with the
multiplier. This board will run 500MHz FSB just fine but then I need to run
the RAM at 4:5 and experience has shown me that a 1:1 ratio is slightly
better. The board and the DDR2-800 RAM will run at 450MHz but I need to feed
it a little more VDIMM above stock and DDR2 RAM isn't getting cheaper. I
want the 6GB (2 x 2GB + 2 x 1GB) that I have to last the life of the
machine. Ergo my optimum FSB is 400MHz.)

This is turning into a long post but rest assured, it's likely my last ever
post about my overclocking.

Anyway, while the machine was running Prime at 3.6GHz I found many, many
pages dealing with o/cing the QX9650. I guess that's an advantage of buying
a CPU that's been around a few years. I opened a bunch of pages in tabs on
the laptop. Then, as Prime hadn't thrown any errors I re-booted the heavy
iron, setting the multipier to 10 (4GHz) while leaving the vcore at stock
(1.240V according to CPU-Z). However althoug it POSTed fine it wouldn't go
into Windows so I dropped the multi 0.5 (3.8GHz) and it booted fine at
default vcore. I set about some reading while I left Prime running.

From my reading it seems that most all QX9650s will run at 4GHz on air, even
using the stock HS/F (which I didn't get with my OEM CPU) with just a small
vcore increase as long as the case is well-ventilated. They're all C1
stepping as far as I can tell so, other than the luck of the draw with the
silicon, identical. Most sources agreed that 4.25GHz was easilly achieveable
with water cooling and taking the vcore up to around 1.45V but that any
higher than that needed a big vcore boost and resulted in a *lot* of heat.

The best source that I found on o/cing these CPUs, which was referenced in a
lot of forums is http://www.anandtech.com/show/2404 "Overclocking Intel's
New 45nm QX9650: The Rules Have Changed" from December '07. It's well worth
a read IMO, even if you don't have a QX9650 or even a 45nm Socket 775 CPU.
It's a well-researched and well-writen article.

Anyway, a passage in that article made me change my mind about finding the
limits of my 'new' CPU. It basically re-stated what I'd said in response to
Phil's post a month or two back. I quote:
........................................
"We honestly believe a new direction in CPU overclocking may soon be upon
us. While there will always be those that continue to push processors to
their absolute limits, the majority of us will find our new "performance"
benchmark in efficiency. This makes sense though - the market has been
heading this way for years now and overclockers may have simply chosen to
ignore the obvious. The multi-core era we now live in places a heavy
emphasis on performance-per-watt figures and other measurable efficiencies."
........................................
Heh! I thought I was being original when I said similar here yet here it is
from a three-year old article.

So, my motivation in finishing what is likely to be my last 'new' build was
to get the best possible (or close to it) Socket 775 system and, by adding a
bit more money to what I'd already spent on it, have a machine that would
hopefully last me many years without being gutless.

So, for now I left my FSB at 400MHz but dropped the multi to 7.5 so that the
CPU is running at stock speed, 3.0GHz. However I *dropped* the vcore to the
lowest that my mobo will allow, 1.10V in BIOS, 1.04V in windows under full
load according to CPU-Z. The machine ran Prime95 all night with those
settings and the CPU fan didn't have to increase above it's 1200rpm minimum
to keep the cores below ~45ºC, A+25. There is plenty of power in this
machine to last me a long time at that speed as far as I can tell.

So, perhaps this isn't my last overclocking story. Maybe, in a year or three
I might have to squeeze some more grunt out of this CPU. Watch this
space.....

Thanks for reading this far (if you did). May all your boxen be cool and
stable.
--
Shaun.

"He who fights with monsters might take care lest he thereby become a
monster. And if you gaze for long into an abyss, the abyss gazes also
into you." Friedrich Wilhelm Nietzsche

Somewhere on teh intarwebs peter wrote:
I read all of it thinking ah!! the good ole days........

Yeah, I know what you mean. I used to really enjoy the days when squeezing
every last (stable and sustainable) MHz out of your CPU was not only a
challenge but you noticably reaped the rewards for your efforts.

I feel that with every new chipset/CPU coming out nowadays
there really is not much of an advantage to overclocking unless
you want bragging rights...

Exactly! With these C2D/Q CPUs I found that the days spent tweaking settings
and stress-testing for stability just weren't giving the rewards that they
used to as, most of the time the cores are running at well below 40% usage
anyway. (As I mentioned I'm not a bleeding-edge gamer nor do I do a lot of
encoding. Also I can't afford to run SETI 24/7 like I used to so the extra
capability of an o/ced, already very powerful CPU is almost never used.)

If you check the "reviews" of the last few mobo you will note a vast
increase in power consumption...but then that never stopped anyone in
the past
and I feel it wont stop anyone except us older users now.

Ahhh, OK. I've pretty much stopped reading about the latest hardware, either
on websites or in magazines. It was too much like I was teasing myself as it
makes me 'get the itch' to try my hand again, see what I could squeeze out
of one of these new beasties. There's no point making oneself yearn for the
unattainable IME, it only leads to discontent with current equipment.

Also, I don't know about elsewhere but power seems to go up way more often
here than it used to and in larger increments.

Anyway, don't these newer systems basically dynamically overclock themselves
if the demand for CPU cycles gets high? Where's the fun in that?

I run a Core2 duo that is oc'd by 40% on air and that's good enough
for me.

Yep, that's a bloody good result.

I am in the market to buy another system..maybe the new Intel
coming out next month,Maybe a X58
chipset mobo and CPU to match...but I dont think I will be playing
with settings too much.

Can't say I blame you, as I mentioned, not only are they way more powerful
than most users need but I think they have a feature (turbo-boost?) that
means they dynamically OC themselves! Boring. It seems that the art of the
O/C is dead. No more point in spending months researching, reading reviews
(and this group, which used to be *the* best source of info and informed
advice), then up to weeks tweaking settings, optimising airflow patterns and
stress-testing the results.

sorry to hear about your poor health

Thanks, I just re-read my post and didn't realise that I'd harped on about
it so much! Way OT, I think it's just that the gaping wound in my available
credit was still raw combined with the sadness at my final acceptance of the
fact that something I'd enjoyed for years, essentially my last indulgence
(as I could sort-of justify it to myself), was now redundant. :-(

Best of luck with your new system when you get it Peter. You can always load
up CPU-Z and watch as it O/Cs itself!

snip my lament on the death of the art of O/Cing
--
Cheers,
Shaun.

"He who fights with monsters might take care lest he thereby become a
monster. And if you gaze for long into an abyss, the abyss gazes also
into you." Friedrich Wilhelm Nietzsche