After just acquiring my Q9300, I've now had the chance to review all 3 45nm Wolfdale/Yorkfield CPU in my systems and here are some measurements and observations:

System:
Asus P5E-VM HDMI
Scythe Ninja Rev. A (fanless)
OCZ 4GB DDR2-800
Seasonic 600W PSU
3X WD 1TB GP
Lite On Blu Ray



All systems were configured to run a it's minimum voltage for 3.0GHz to make a fair clock for clock measurement:

Intel Wolfdale Core 2 Duo Xeon E3110 (E8400 equivalent) (3.0GHz stock)
3.0GHz (1.072V) -> 75W Idle @ 27C
3.0GHz (1.072V) -> 123W (full OCCT) @ 44C

Intel Yorkfield Core 2 Quad Q9300 45nm Core 2 Quad CPU (2.5GHz stock)
3.0GHz (1.088V) -> 82W Idle @ 42C
3.0GHz (1.088V) -> 130W (full OCCT) @ 55C

Intel Yorkfield Core 2 Quad Q9450 45nm Core 2 Quad CPU (2.66GHz stock)
3.0GHz (1.072V) -> 82W Idle @ 53C
3.0GHz (1.072V) -> 141W (full OCCT) @ 68C

It's incredible the difference betwen all 3 chips. E3110 aka E8400 runs amazingly cool. So much so that I had to triple check the temp calculations. I could probably run a completely fanless system if i kept the clocks low with this CPU! Q9450 is just a power hog, running really hot when stressed. Although its quiet powerful, that extra 6mb of cache sets this it on fire when fully loaded. And the Q9300 is just the perfect combination in my opinion. It's powerful with 4 cores and is able to clock just as much as the Q9450 (albeit on a more stressed out FSB), but it is much cooler in its operation. This is a keeper for me for sure and which i would recommend to any HTPC/Quiet-PC enthusiast who needs the power, without all the thermals (and associated noise with it).

Note: I realize there is chip to chip variance for temps, voltage, but i have been following other owners experience and they do correlate with mine.

Damn i didnt think that the 9450 would be THAT much hotter than the 9300 at the same clock speed, they are extremely similar designs. To me it just confirms that a 9300 with some overclocking is the way to go. Although i might still wait for the next generation to come outlater this year or early next. But i expect temptation will get the best of me first.

I assume that's total system power draw from the wall? The Q9300 and Q9450 both had the same power draw at idle. That should mean the same amount of heat is being dissipated by the CPU. I suppose maybe slightly more from the chipset on the Q9300 due to a higher FSB to get the same clock speed. So if the power usage is the same and the cooling is the same, the temperatures should be very close. Are the thermal sensors on these chips suppose to report actual temperatures that can be compared between different CPUs?

I always took thermal readings with a grain of salt. I had an old P4 that always reported 60C at idle even when there was barely any heat coming off of it. Heck, it was 60C immediately after booting before it even had a chance to warm up. I think it sat around 70C under load but never showed any signs of instability or throttling.

Great Comparison!

I've recently been debating this very subject. If you're running them at 1.072 / 1.088 (vCore?), i'm very impressed. I didn't realise you could undervolt that much!

How have you done your stability testing at such low voltages?

-Mikey

Xbit says E8500 is 3.4W idle (actual processor power consumption). Any idea what is taking the remaining 72W? Any interest in disconnecting some drives to check?

ST. Very interesting. Can anyone else confirm his numbers too?

If you still have all three processors, I would be interested in how they perform at their respective stock. I don't overclock, but undervolt occasionally. I really only need an E8400, but if the Q9300 power consumption is slightly higher, I could go with that. I'm very concerned with heat and power.

The biggest difference between the Q9300 and Q9450 is the cache, so I wonder if that has a lot to do with it. It couldn't be the .5 multiplier.

Robert

Yeah, interesting study. I too would like to see the effect from disconnecting the four harddrives. Also, the reason for the relatively high wattage for such an incredibly energy efficient setup might be the 600 watts PSU. The PSU might not be very effective at ~100 W. I will see if I can find a review.

EDIT: The PSU claims 70% efficiency at 75 W and roughly 80% at full load on your computer (130'ish watts).
http://www.pcper.com/article.php?aid=10 ... pert&pid=6



*Assuming that your PSU really is Seasonic S12-600.

Thanks. I check stability via OCCT utility running for a period of ~12 hours.

If I have time, I'll try to disconnect the HDDs (I have 4 WD GP 1TB HDDs now).....

Is that 75W that comes with with 70% from the wall or from the psu? Because if the topicstarter measured a pull from the wall of 77W, then I'd think his system would draw a typical 50ish Watts, because of the lower efficiëncy at that low power need.

I don't know if it's 75 W from the wall or from the PSU. It's impossible to tell from the article. But I agree with you. His system would probably draw 50-60 Watt from the PSU. I myself is contemplating to ditch my current GPU and motherboard and get a G35/630i chipset motherboard and run the lot off a passive PSU since I don't ever play games anymore.

May I ask what the purpose of this PC is, HTPC or workstation?

Cheers

The purpose of my setup is a pure HTPC build:



FYI - My measurement for the above system was done at the wall via a KILL-A-WATT meter.

What case are you using for your system?

I am looking at the same core for an htpc. Specifically, the Asus P5E-VM, an E8400 CPU, 4Gb of OCZ Reaper PC-6400 DDR2 RAM, but with a Zalman PS and a Zalman 9700 CPU Cooling HS/Fan.

Any reason that you chose the Scythe Ninja rather than a Zalman HS/Fan solution? Noise is my most important concern since this HTPC unit will reside in my bedroom (WAC is crucial), but do not want the CPU to overheat.

Thanks for your response

greyskies - I have always a big proponent of Sycthe Ninjas, especially their ability to run passively (no fans). With Intel's new generation of 45nm CPU - it is actually plausible now.

Below is synopsis of the case I use:

I'm running a somewhat similar system:

Gigabyte P35-DS3R rev 2.1
Intel E8200
Corsair TwinX 2GB DDR2-800
Scythe Ninja Mini fanless
Hitachi T7K500 320 GB
Pioneer BDC-202BK
Radeon HD 3450
Chill CP-450A4 450W PSU.

I've undervolted the CPU to 1.1v in BIOS, which is the lowest it can run stable. CPU-Z reports 1.04v

My power draw measured with Kill-A-Watt is 70W idle and 90W Orthos Small FFT. Loading the GPU with rthdribl HDR rendering adds an extra 10W.

The only fan is in the power supply, and the system is stable and temperatures are fine.

Naively, let me ask, why would you undervolt your CPU?

I have an E8400 on order from NE.com with an Asus P5E-VM motherboard, and 4Gb of OCZ DDR2-800 RAM. Should I consider undervolting my CPU, and what can I expect if I do undervolt it?

There is no reason why you wouldn't want to really.
there's a difference between under/over volting and under/over clocking.
Undervolting does not effect performance. You can safely run your CPU at stock or a small over clock whilst undervolting it. Improves thermals and power efficiency. There is a limit and it depends on the mobo and CPU, but generally as long as you do it in small steps you cant damage your CPU by undervolting. You do get instability if you push it too far, but it shouldnt cause long term damage.

I have used the Enermax VOSTOK matx case for almost a year now - purchased it on a whim. Although physically flimsy, it's a cool, quiet case after swapping sides to hide the vents which leak noise and replacing the toy psu and fans. Especially like the provision for two 120mm fans - the front one blows across the hd's. The tool-free capability of the dvd bays is both innovative and annoying. I would pay a little more for a case with this configuration if there was more metal in it. I'm currently looking for another one without the psu.

In my experience, if a case is wide enough to accommodate a 120mm rear fan above the mb i/o ports, there will be enough room for any cpu hsf currently available.

There are a couple of other small cases that have front and rear 120mm fans but the front fan doesn't flow across the hd's. Don't know why other manufacturers haven't figured this out yet. This probably isn't important for just one drive but all the available video content is requiring multiple drives...

enermax vostok marketing - wonder how many have been sold

The OP has confirmed these are AC power readings. Our SPCR review of the S12-600 is here: http://www.silentpcreview.com/article247-page4.html -- extrapolate the numbers a bit from the bottom table, and we get around 55W/96w, 62W/104W & 62W/113W in DC power for the 3 systems.

Nice system. Speakers will be very unhappy though in their current positions, as I'm sure you know.

Do you do any postprocessing, since you prefer the quad core over the dual core?

And if so how do the processor usage/percentage compare (I´m asking because I´m looking for an upgrade for my E6600, which, with my postprocessing settings (ffdshow, Avisynth etc. for optimized dvd-playback) is on around 70% cpu usage (both cores) - making it necessary to run my 2 120mm Yate Loons/Ninja at 80%= no silentpc...)?

Regards, Finn

ST-

Thanks for posting these numbers. I'm in the process of choicing either the Q9300 or the Q9450. I'm doing a lot of modeling in Excel 2007 now and it appears the cache does make a difference as the Q6600 is faster than the Q9300 even with the higher clock and FSB.

http://www.legitreviews.com/article/695/6/

Do you know what the default voltage on the Q9450 is? Did you try to undervolt at all?

It's scary - I'm thinking about just the same system except with only 2 drives and my S12-430 so the numbers should be within 5 watts.

Thanks!
Chris

ST,

By coinsidence I'm looking for the same ASUS P5E-VM HDMI + Q9300(or Q9450).
I'm not experienced in overclocking.
My current system is ECS G33T-M2 + Q6600 (oc 2.9GHz).

Is overclocking this ASUS really tricky ?
It would be really great if you can post parameters you played with.

Thank you in advance,

VD

Your Q9300 result is much higher than what I've experienced and read in other forums.

Mine never gets above 30C, ever- it is running 2-4 Virtual Machines at any one time, even after 4 hour compiles. (VM's for Domain Server, Web Server, Compilation Farm, DB Server). The case has rotten airflow, and it is literally stuffed with disks as well.

Lian Li A5 Case
Q9300
4 x 2GB Kingston PC6400 @ 1.8V
Apack BTF90, no fan
1 x 120 Nexus Front, In
1 x 80 Nexus Front, In
1 x 120 Nexus Back, Out
1 x WD Raptor 150GB 10K (boot)
4 x WD GP 500GB (RAID 5)
1 Areca 1210 Controller
Zalman Replacement Northbridge Heatsink
Gigabyte G33 series uATX board, CPU undervolted to 1.0V, FSB -0.1V
Seasonic S12II-430 PSU, ADDA fan swapped w/Scythe Slipstream

Under normal use, CPU runs between 18C-23C (lmsensors)
HD's run at 32-36C (smartd)

I'm not sure how we can have such differences with a Q9300. It must be mobo sensors.

Run Real Temp and tell us your Distance to TJ Max at idle and load.

what is your ambient temp? seems unlikely if ambient is also 18-23C.

I agree, unless your room is around 10c ambient i doubt that temp reading is accurate. whats your GPU/HDD temps under load?

Ambient: 17-23C
HD RAID: 45C peak under load.

Boot HD monitored, but not recorded as long as it stays under 50C.

Host operating system is Debian Linux, headless, haven't bothered to install X, much less Nagios- the only 16x PCIe slot is occupied by the RAID controller, graphics are through the G33/GMA3100- GPU Temps not monitored.

If you don't mind, how is your VM set up? I bought VMware 6 workstation, and just curious how well it performs with 4 VMs running.