Arcmanov's rig evolves...

[Arcmanov]

HA!!!  Did some digging, and found an old Socket A HSF.  Of course, I sting off the fan ONE TIME
and 'chook it een' as per that pic.  IOH idle temp is now down to 58C, which should ensure longevity.

Based on what I've read, MSI has since listened to all the complaints, and have actually improved the
IOH cooling on newer revisions of their X58 motherboards.  GJ MSI.

[Arcmanov]

So yuh boy started playin up with the new X3 today...



Point of note: couldn't use my Dominator RAM because of the HSF.  Unfortuantely, on all
AMD mobos, the HSF can only be oriented one way, and this particular HSF (Thermalright Ultima-90)
overhangs the 1st DIMM slot.  Good thing I had an extra set of Crucial DDR2 'lying around'.

...and the Top Deck Station makes this sort of testing so easy.


1st run at stock (2.8 GHz) and 2nd run at 3.4 GHz.
 

Specs...

MB: MSI K9A2-CF (790X chipset)
CPU: AMD Phenom II X3 720 Black Edition
RAM: 2 x 2 GB Crucial Ballistix Tracer DDR2-1066
GPU: [Sapphire] ATI Radeon HD4850

No change in idle/load temps so far.
Funyy how Everest does not see the increase in CPU clockspeed as an overclock,
as the bus speed is unchanged.  Unlocked multipliers are da shiznit. 

Testing continues...

[W1nTry]

BRING THE NOISE!!!

That aside Arc, the Sunbeam CCF would solve your RAM slot overlap seeing as YOU did recommend it to begin with.

[Arcmanov]

Update...

Was able to post very 'comfortably' and boot into Windows, and run a benchmark...



...and I THOUGHT everything was fine and dandy, until I decided to stress-test with LinX 0.58.
Within 5 seconds I was b!7ch-slapped with a BSOD.

So of course, madd tweaking commenced, and I was finally able to get stability at 3.6 GHz...



Notice the substantial increase in CPU voltage. From 1.342V to 1.460V.
Load temp topped out at 61C.  *ouch*  A bit too hot for my liking.
This 92 mm fan I have on my Ultima-90 seems to have reached its efficiency threshold, so its push-pull ihmc.

Testing continues...

[W1nTry]

61? wow, I never hit that even at 3.7GHz... and you're heatsink dare I say is better than mine (marginally but still). Interesting... what is your ambient temp arc?

[Arcmanov]

Well, the addition of a 2nd 120 mm fan did help, a bit.  Load temp is now down to 57C.
As for ambient temp, I cannot tell, but it was very hot in here today, and I don't have AC
unfortunately.  I goin by a pardner tonite for a LAN sweat, and his game room is AC-cooled,
so test rig will be making a trip there for some testing too. 

With that said, it seems that 3.6 GHz is the maximum I can squeeze out of this CPU on this board.
It seems you were right W1nTry, when you suggested that the SB600 soutbridge would limit the OC
of this particular CPU.  Still, I'm VERY impressed as to the overclockability of this CPU.

Sadly, my LAN rig still pwns it. Far less my main rig.

[W1nTry]

Sadly, my LAN rig still pwns it. Far less my main rig.

How much did those cost in comparison? do a price/performance for we nah. Some first hand experience with different platforms, cost and their relative performance, we'll use 3DMark for arguments sake.

[Arcmanov]

Well, lets see...

Intel Core i7 920 = 280 US
MSI X58 Pro = 197 US
3 x 2 GB Corsair XMS3 DDR3-1333 RAM = 101 US
HD4870X2 = 450 US

Net total = 1028 US

Max 3DMark06 score achieved = 25101



Intel Core 2 Quad Q9650 = 325 US
Gigabyte GA-EP45-UD3R = 133 US
2 x 2 GB Crucial Ballistix Tracer DDR2-1066 = 65 US
Sapphire ATI Radeon HD4870 512 MB = 171 US

Net total = 694 US

Max 3DMark06 score achieved = 17066



AMD Phenom II X3 720 BE = 150 US
MSI K9A2-CF = 90 US
2 x 2 GB Crucial Ballistix Tracer DDR2-1066 = 65 US
Sapphire ATI Radeon HD4850 = 120 US

Net total = 425 US

Max 3DMark06 score achieved = 14640



All prices quoted from Amazon...rounded off to the nearest dollar.
Of course, these costs will differ significantly when you add other things
like case, PSU, HDDs, HSF, etc, but you get the general trend here.

You get what you pay for.

Still, the AMD plaform is a killer deal.  I'll drop the 4870 into the test rig
to see what the result would be.  Might not be far off.

[W1nTry]

So basically using a VERY crude method of finding bang for buck (3DMark/Cost in USD) we get:

Core i7: 24.41
C2Q: 24.59
X3 720: 34.45

Still, the AMD plaform is a killer deal.  I'll drop the 4870 into the test rig
to see what the result would be.  Might not be far off.

Yuh think?

[Arcmanov]

As per W1nTry's request, here's a 3-loop run of 3DMark06...



A healthy difference of 354 points...less.  I guess this can be deemed 'more accurate'?

[Spazosaurus]

hmmmm less eh?

[TriniXaeno]

very interesting

[Arcmanov]

I tried that small bump in bClk, and thought it was stable, until I happened to spy the
actual CPU speed during the testing.  Observe...



Looks normal enough, but check this screenie I took during the test...



+1 to the first one to explain whats happening there.

I won't talk about the 3DMark06 results.  Captain Awesome betta reckonize.

[Spazosaurus]

LOL. Yes, I recognised. Hail to the king bla bla bla. Cocaine is a helluva drug.

Linx normally causes that. The bclk seems to drop to 20 under extreme pressure.

[Arcmanov]

Its a 'new' day...

New OS, new drivers, new 1600 MHz Dominators, and so, a new benchmark run was in order.



The proof is in the pudding...so now I can officially lay Vista to rest.  The ride wasn't always fun, but it was always eventful.
Thank you M$, for taking the Vista backlash in stride, and giving us this wonderful OS a lot quicker than you originally intended.

[kgfalby]

Nice score!!!!

[Arcmanov]

http://img193.imageshack.us/img193/7540/capture17102009104704am.jpg

[Crixx_Creww]

...*- _ -* ah hope you could see dis cut eye properly thru de internets...an hear de hard watery steups....
buh 4am doh? lolz

[Spazosaurus]

Ahhh boy, yuh finally drop it on we eh?

I think it will take a few days for the message to soak through to the GATT masses.

[Arcmanov]

Well, my subtle clues were only fully grasped by the most techiest of GATTers.

That pic should provide some more info.