Help with old Core2Quad Q9650

[Engr62]

I recently picked up a used Dell OptiPlex 755 small form factor computer that I'm planning to pass along to a family friend in need of a basic computer. The computer had the following specs:

CPU=Core2Quad Q6600
RAM=8 Gb DDR2
HDD=None
OS=None

I replaced the Q6600 2.4GHz with a Q9650 3.0GHz (I know, not a big upgrade--but I wanted to max this out as much as possible). To test out the system, I put an old 320Gb HDD in it and installed Linux Mint 17 on it. It seemed to run pretty good for such an old system (I did run a benchmark on it--more on that later).

I then removed the 320Gb HDD and replaced it with a 1Tb HDD that I used as a data drive in another computer and installed Windows 7 Home Premium on it. After installing two days worth of updates, it's finally up and running.

When I get a system up and running, I usually run a video conversion benchmark using the version of Handbrake and the video discussed in this thread. Over the past 3 years, I've run this benchmark on several systems I've had in my position. The video clip is converted to an Android-friendly mp4 in a single pass using Handbrake 0.9.9.5530 64-bit. The results from this benchmark for the systems I've tested it on are shown in the table below. If you scroll over to the right, you'll see a value for "FPS/Core/GHz." To get this number, I divide the frames per second (FPS) by the number of CPU cores and again by the CPU frequency in GHz.

+----------------------+-------------+-------------+-------+---------+----------------+--------+--------------+
|       CPU            | Base Freq.  | L2/L3 Cache | Cores | Threads |      RAM       |  FPS   | FPS/Core/GHz |
+======================+=============+=============+=======+=========+================+========+==============+
| Phenom II x4 965BE   |   3.4 GHz   |   2Mb/3Mb   |   4   |    4    |  8Gb DDR3-1600 | 109.44 |     8.05     |
+----------------------+-------------+-------------+-------+---------+----------------+--------+--------------+
| Core2Duo E6550       |   2.33 GHz  |   4Mb/0Mb   |   2   |    2    |  4Gb DDR2-667  |  36.54 |     7.84     |
| Core2Quad Q6600      |   2.40 GHz  |   8Mb/0Mb   |   4   |    4    |  4Gb DDR2-533  |  77.51 |     8.07     |
| Core2Quad Q8200      |   2.33 GHz  |   4Mb/0Mb   |   4   |    4    |  4Gb DDR2-667  |  74.06 |     7.95     |
| Core2Quad Q9550      |   2.83 GHz  |  12Mb/0Mb   |   4   |    4    |  4Gb DDR2-667  |  97.10 |     8.58     |
+----------------------+-------------+-------------+-------+---------+----------------+--------+--------------+
| Core i5-2500K        |   3.30 GHz  |   6Mb/0Mb   |   4   |    4    |  8Gb DDR3-1600 | 159.63 |    12.09     |
+----------------------+-------------+-------------+-------+---------+----------------+--------+--------------+
| Pentium G3258        |   3.20 GHz  |   3Mb/0Mb   |   2   |    2    |  4Gb DDR3-1333 |  88.78 |    13.87     |
| Core i5-4440         |   3.10 GHz  |   6Mb/0Mb   |   4   |    4    |  8Gb DDR3-1600 | 175.45 |    14.15     |
| Core i5-4670K        |   3.40 GHz  |   6Mb/0Mb   |   4   |    4    |  8Gb DDR3-1600 | 203.62 |    14.97     |
| Core i7-4770K        |   3.50 GHz  |   8Mb/0Mb   |   4   |    4*   | 16Gb DDR3-1600 | 210.11 |    15.01     |
| Core i7-4770K        |   3.50 GHz  |   8Mb/0Mb   |   4   |    8    | 16Gb DDR3-1600 | 244.44 |    17.46     |
+----------------------+-------------+-------------+-------+---------+----------------+--------+--------------+
 *Hyper-threading disabled

The "FPS/Core/GHz" gives a good indication of the performance of a CPU family. From the above table, the Core2Duo/Core2Quad CPUs get approximately 8 to 8.5 FPS/Core/GHz depending on the cache while the Haswell CPUs get approximately 14 to 15 FPS/Core/GHz for non hyper-threaded CPUs.

Now, back to this Q9650 system. Running the Handbrake benchmark on this system yields only about 80 FPS (well below the 97 FPS achieved by the Q9550 I previously had--granted, it was a different motherboard and RAM). Based on my previous results (8.5 FPS/Core/GHz), I expected the Q9650 to get about 102 FPS.

When I had Linux Mint 17 installed on this system, I ran the Handbrake benchmark with the only version of Handbrake I could install (0.10.5 x86_64), and I got 114.25 FPS for the conversion. The only other time I've had Linux install on a system was with the Q8200--in that instance I got 82.34 FPS (versus 74.06 in Windows 7). So the Linux version seems to be about 11% faster than the encode in Windows 7. If that same 11% speed boost hold true for the Q9650 system, the expected results in Windows 7 would be ~103 FPS (114.25/1.11) which matches the expectation based on the 8.5 FPS/Core/GHz results I got for the Q9550 a while back.

The motherboard in the Dell OptiPlex 755 is very basic and has almost no control over the CPU/memory settings in the BIOS. But there are settings for virtualization and such that could possibly be affecting the performance in Windows 7.

CPUid reports that the CPU is running at 3 GHz, so I don't think Windows 7 is somehow underclocking it.

Is there anything I can check in some obscure Windows 7 settings that might indicate why this system is only performing at about 80% of what it should? The 1Tb hard drive is a 7200 RPM drive, so I don't think that's causing the problem. Any help is greatly appreciated.

 

[Ketchup]

Is handbrake using all the cores in the Windows install?

Is there anything in Device Manager that isn't identified? Or isn't identified correctly?

 

[LTC8K6]

Ram speed?

 

[JeffMD]

Get rid of windows 7, windows 8 supports much better task scheduling for better multi core support.

 

[CorsairDemon]

What are your temps like? The new chip is actually 10W less than the Q6600, but if it's being thermally throttled, it's possible you may not know.

It's obvious that your chipset supports the CPU, but it doesn't look like the Dell Optiplex 755 Officially supports the Q9650 chip. It could be an incompatibility that was unforeseen or unknown.

Another thing to look at is drivers. These optiplex machines do like to have specific drivers for IMEI and their chipsets. Try using only the drivers provided from Dell instead of drivers from Windows Update or the ones that come reinstalled with windows.

Also, disable IME with the hotkey before boot. Another feature to not have enabled in windows to waste cpu time and resources on, of course that is if you don't use it.

Power Settings should be switched to high performance. Make sure your BIOS is up to date.

 

[Engr62]

Thanks for all of the suggestions.

Is handbrake using all the cores in the Windows install?

Is there anything in Device Manager that isn't identified? Or isn't identified correctly?

Handbrake is using all of the cores (per Task Manager). I don't have TM running during the actual benchmark as it slows things down. But I've run the test with TM on, and all 4 cores are pretty much tapped out--overall CPU usage stays between 98% and 100%.

Ram speed?

It's DDR2-667. Since there are no overclocking settings in BIOS on this machine, it's running at a FSB of 333 MHz (the CPU is quad-pumped at 1333 MHz and the memory is 667 MHz). The timings are 5-5-5-15 2T.

Get rid of windows 7, windows 8 supports much better task scheduling for better multi core support.

I had an unused of Win 7 Home Premium laying around, so I installed that. I plan to upgrade (upgrade may not be the proper word ) to Windows 10. I didn't want to buy another copy of Windows for this setup. But, I wanted to finish trouble shooting it before upgrading.

What are your temps like? The new chip is actually 10W less than the Q6600, but if it's being thermally throttled, it's possible you may not know.

It's obvious that your chipset supports the CPU, but it doesn't look like the Dell Optiplex 755 Officially supports the Q9650 chip. It could be an incompatibility that was unforeseen or unknown.

Another thing to look at is drivers. These optiplex machines do like to have specific drivers for IMEI and their chipsets. Try using only the drivers provided from Dell instead of drivers from Windows Update or the ones that come reinstalled with windows.

Also, disable IME with the hotkey before boot. Another feature to not have enabled in windows to waste cpu time and resources on, of course that is if you don't use it.

Power Settings should be switched to high performance. Make sure your BIOS is up to date.

I'm not sure what my temperatures are. I haven't installed CoreTemp on this system yet--I was having trouble installing it due to some firewall setting blocking the InstallIQ used by CoreTemp. I'm not sure what setting in the Windows firewall is blocking the installation, but I was afraid to mess around with it too much.

Like you said, I figured since this is a lower TDP chip than the Q6600, temps wouldn't be an issue--I guess I better check these.

A Dell representative on their message boards confirmed that the Q9650 works with BIOS rev A20. I have BIOS rev A20 on this machine, but I've just used the default drivers. I didn't want to mess with using drivers other than what Windows installed as I plan to upgrade to Windows 10. Perhaps I should go back and look for Dell-specific drivers.

 

[CorsairDemon]

Definitely do not allow that installIQ to go through, it's malware. Sounds like you downloaded the coretemp with bundled software.

http://www.almico.com/speedfan451.exe

Here is speedfan, well known and reliable. It will show you sensors in your machine that it can read. Although, some of the settings are a bit beyond me. Use at your own risk with the settings other than the temp readings.

Some drivers from windows updates and built into windows are either non-functional in some cases or very unstable. I've seen a lot of random CPU usage from Intel Management Engine before, and disabling it seems to help.

 

[Engr62]

I booted the system up last night and ran the test again. This time, I got 103.08 FPS which is just about exactly what I expected for this system.

I made no changes to drivers or anything. I checked the Windows Update log as well, and it indicated that no updates had been installed since 6:45 PM the previous day. I rebooted several times after the updates were installed, but there must have been something (drivers maybe?) that finally finished installing after I rebooted last night.

Thanks for all of the help.

 

[Ketchup]

Sounds like Windows working its magic, lol. Glad it's back up to speed.

 

[CorsairDemon]

Yeah, glad it's normal again. It seems to be that updates were most likely doing something in the background, just for curiosity, was there a svchost using a good bit a ram at any point in the past few days or during the testings?

 

[Engr62]

Yeah, glad it's normal again. It seems to be that updates were most likely doing something in the background, just for curiosity, was there a svchost using a good bit a ram at any point in the past few days or during the testings?

I checked the CPU Usage and Memory on the "Performance" tab of Task Manager, and nothing was out of the ordinary. I didn't, however, look at the "Processes" tab and sort on Memory. I'll bet you're right, though. It was probably svchost.

I tried to avoid this by checking for updates, installing them, rebooting, and let the system settle down.

I've found it common to have a system run sluggish until Windows gets all of the updates installed. But this seemed to be way longer than I've encountered before.

The performance of this computer under Linux versus what it was doing under Windows 7 until last night indicates that something must have been going on with the Windows updates.

 

[John Connor]

Be careful with those updates. Soon enough Windows 10 will want to sneak on there.