Go Back   Hardware Canucks > PC BUILDERS & TWEAKERS CORNER > Overclocking, Tweaking and Benchmarking

LinkBack Thread Tools Display Modes
  #1 (permalink)  
Old September 5, 2008, 11:59 AM
Top Prospect
Join Date: Aug 2008
Location: Montreal, Canada
Posts: 59
Default OCZ Vendetta 2 is badass!


So I finally got around to installing the Vendetta 2 on my q9550. Installation was a breeze, I bought the bolt thru kit by thermalright to make installation all the easier. Used Arctic MX-2 thermal paste, and even opted for the Noctua NF-P12 fan to install on the side of the heatsink. I did find that, unlike in the review of the OCZ Vendetta 2 on this site, it was much easier to install the four anti-vibration mounts onto the fan first and then attach the fan to the heatsink, in contrast to installing the mounts on the heatsink first and then attaching the fan. I installed the heatsink/fan in such an orientation such that the fan would push air from the case into the heatsink, and this air would subsequently be pushed out the case by the exhaust fans, which is what this guy did My Vendetta 2 install, with pics (Q6600) - Overclock.net - Overclocking.net


Stock intel heatsink/fan:
Idle temps @ 2.83ghz ~ 47 47 55 55
Load temps @ 2.83ghz ~ 65 65 68 68

W/ OCZ Vendetta 2
Idle temps @ 2.83ghz ~ 41 41 47 47
Load temps @ 2.83ghz ~ 45 45 53 53

So as you can see, the combination of Vendetta 2, NF-P12, MX-2 and thermalright bolt thru kit decreased my overall temps by around 15C on load.

OC results:

Since I was never really able to OC on stock cooler, I decided to perform a a mild OC from 2.83ghz to 3.4ghz by increasing the FSB from 333mhz to 400mhz. Since my RAM is clocked at 800mhz, then this is the best I can do to obtain 1:1.

Idle temps: 45 45 51 51
prime95 for 3 hours (will be doing a 12hour test): 54 54 58 58

So, it is staying under 60C at this overclock. I was worried at one point, however. I used OCCT to do a quick test and see if the overclock was stable. After 20mins, the computer restarted. I don't know why it restarted, since the temps were all under 60C, and the voltage was within range of the proc (1.2V, max = 1.3625)

I also ran 3dmark06 and 3dVantage, and had no problems whatsoever.

What do you guys think? Also, is there a preferred torture test in prime95 to test CPUs? I only did the blend test for 3 hours, and the large FFT test which leveled off after 10mins to the same numbers i saw with the blend test (maybe 1-2C higher).

How much more room do you think I have to overclock if I ever decided to get 1066mhz DDR2 RAM?
Case: Coolermaster Cosmos 1000
Mobo: Gigabyte X48-DS4
CPU: Intel Quad q9550
PSU: Corsair 750w
GPU: Diamond HD4870
RAM: 4gb Corsair XMS2 PC2-6400
HDD: 2 x 640gb WD Caviar
Audio: Auzentech Forte 7.1
OD: Pioneer DVR-216
OS: Vista Ultimate SP1
Reply With Quote
  #2 (permalink)  
Old September 5, 2008, 12:30 PM
DK2 DK2 is offline
Join Date: Apr 2008
Location: GTA
Posts: 820

Hey pretty good. Big difference over stock.

SuperPI will get a little higher temps, a degree or two, if you make run lots of threads, like 16.
I think its SuperPI??, there's a few PI ones out there.....
Reply With Quote
  #3 (permalink)  
Old September 5, 2008, 08:07 PM
Dashock's Avatar
Hall Of Fame
Join Date: Nov 2007
Location: Toronto
Posts: 1,955

Originally Posted by ricknash61 View Post

So as you can see, the combination of Vendetta 2, NF-P12, MX-2 and thermalright bolt thru kit decreased my overall temps by around 15C on load.

Man those Nf-P12'S are awesome ever since i got them i have been promoting them on all rigs i got 6 of them at the moment gonna add 3 more eventually i hope they come out with faster ones as they are soo whisper quiet that way they can push even mooore air.
Intel Xeon E3110 @ 3.0ghz-Then
Core 2 Quad Q9650 @ 3.8ghz-Now
PNY GTX 1050Ti (Coming)
Asus P5Q-E
Mushkin Ascent Redline 8000 8gb DDR2 @ 1066 5-5-5-15 2.05v
Corsair Cx600M
Samsung 850 Evo 250gb SSD
Noctua NfP-12's on Swiftech 320 rad
Logitech Z-625 setup
Sony X800E 43" 4K HDTV

Coolermaster Cosmos S/
Swiftech Mcp655/ MCR-320-QP/ 6X Noctua NF-P12'S/ Swiftech Apogee GT/ Swiftech Mcw-60/ Swiftech Micro-Res.

Reply With Quote
  #4 (permalink)  
Old September 5, 2008, 08:26 PM
MpG's Avatar
MpG MpG is offline
Hall Of Fame
Join Date: Aug 2007
Location: Kitchener, ON
Posts: 3,143

3DMark/Vantage are rather weak indicators of overall system stability. If you're trying to find stability, I suggest using OCCT first, as it finds most errors much faster than Prime, which really saves time. Once you can run OCCT for two-ish hours, get Prime95 going for a final verification. Personally, I consider eight hours to be minimum, 12hrs good, with 24hrs being ideal (You'lll get different opinions on that part, though).

Also, it's possible that 1.2Vcore isn't enough to support that CPU speed. The 45nm processors need less voltage, but once you break the 3Ghz mark, they still need a little extra.

Getting 1066MHz ram might help some, although most quad cores don't like going past 450fsb, so using a 1:1 ratio will only take you up to 900MHz at the most. Before you make the investment, you might consider declocking your processor and experimenting to see how high you can take your current sticks of RAM. Most (not all) 800MHz sticks will go a fair bit past their rated speed with an extra tick or two of voltage.
i7 2600K | ASUS Maximus IV GENE-Z | 580GTX | Corsair DDR3-2133
Reply With Quote
  #5 (permalink)  
Old September 7, 2008, 04:43 PM
Dadi_oh's Avatar
Top Prospect
Join Date: Sep 2008
Location: Almonte, Ontario
Posts: 180

I have the OCZ Vendetta "1" on my second rig and I was blown away by it's performance for a $30 HSF. Applying the thermal paste takes a little thought to make sure it distrbutes across the heatpipes without most of it squeezing up between the pipes and the holder. I assume the second version is all that much better.

As far as overclocking goes... A few of my thoughts.

1) The 45nm parts are not as voltage tolerant as the 65nm parts. The 65nm could take much more aggressive voltages as long as you kept temps in check. However, the 45nm can be damaged by voltage before they even run into temperature problems.

2) I try to determine the limits of each piece of HW independently so that I know when I hit the overall limits of the combined overclock. First I lock the PCIe frequency to 100MHz.

3) I usually start with my RAM, increasing the divider with the CPU FSB at stock and running memory tests to find the upper limit. I usually just set my RAM voltage to the max supported by warranty (usually 2.2V on OCZ DDR2 although I have some 1100MHz OCZ that is warranty rated to 2.4V).

4) Once max RAM frequency is found then I drop it back to stock and go for FSB on the motherboard/CPU. I drop the multiplier as low as it will go (say 6X) and then set FSB to achieve the rated frequency of the CPU. Say 400MHz for a 2.4GHz CPU. Then start increasing FSB until you find max FSB. You will need to adjust Northbridge and CPU voltages accordingly as you go. Watch temps and keep them in a comfortable range (for you). I personally shoot for less than 60C but everyone has their own limits.

5) Now you know max memory frequency and max FSB for your motherboard/CPU you can start seeing what the limit of the overall combo is. Higher FSB has benefits in overall system performance, particularly for memory bandwidth benchmarks. So 500X8 will give better performance than 444X9 if you have a choice. That's why I lower my E8400 9X multi to 8X so that I can raise FSB to 485. I watercool my northbridge so temps are not an issue for me at this high FSB.

Have fun.
Rig1: Q9550 @3.8GHz Vendetta 2, P5Q-PRO, 6GB OCZ Reaper PC2-8500. 5870 1GB, GT240 Physx, Ultra LSP 750W, 2TB internal + 2TB external, 3X23" 5760X1080 eyefinity + GT240 Physx, HAF22 case 3X200mm : Win7 Pro
Rig2: E8200 @3.6GHz Vendetta HSF, DFI P45 LP-JR, 4GB 900MHz, 4870 512MB, X203W 20", Ultra Gladiator Case
Rig3: Q6600@ 3.5GHz H50, P5Q-PRO, 8GB, 8800GTS512, OCZ GameXStream 850W, Samsung 245BW 24" + 19" dual: Snow Leopard + Parallels Win7
Rig4: MAC mini Snow Leopard
Laptop: Macbook Pro 15"
Reply With Quote

Thread Tools
Display Modes