OC old man C2D E6600 + Asus Commando

[Rober García]

Hello!
For the older folks here, this is sure to bring back memories.
I've been away from the OC world for a long time, but I have this team laughing at home for office work and little more, and I've been tempted to try again. I've overclocked without any problems from 266x9=2400 to 350x8=3150
Before I go into detail, I have the following problem. To test performance, I'm using the 3dMark downloaded from the official site and these are the results:

                Total  Graphics    Physics        Combined 266x9 DDR2 800  4311  5487/25fps   2048/40/21fps 4540/18fps 315x9 DDR2 630  4922  6105/28fps   2434/44/24fps 5355/22fps 330x9 DDR2 660  4812  5828/26.5fps 2465/45/25fps 5477/22.5fps 350x9 DDR2 700  4620  5495/25fps   2444/45/25fps 5387/22fps 375x8 DDR2 750  4313  5494/25fps   1976/36/20fps 5147/21fps 

Physics Points/8threads/24threads

I'm surprised by the results, especially the 375x8 which is worse than 266x9. It can't be because of the memory, because it's precisely the one with the highest memory frequency. Can someone explain the results?

Equipment:
C2D E6600
Asus Commando (P965)
ATI HD5770 1GB (No OC)
4x1GB (G.skill good + KVR bad)

[Sylver]

Hello very much @Rober-García

I'm not particularly into OC, but I thought that at full load at 375x8 it might reach very high temperatures causing performance to drop, unless it has very optimal cooling and that's not the reason. I've come across some equipment that had the same problem, with worse performance at higher frequencies, and I never really knew what it could be due to.
Let's see what the frequency gurus have to say.

By the way, our esteemed mayor @cobito had a Conroe like that as his main computer until recently and he had it well tuned, he can surely give you some advice.

It's very interesting to see threads about OC on the old 775s today, it was a glorious era.

Best regards!

[cobito]

Well, in fact, until three years ago, I had an E6600 that had been running at 3GHz since day one. I had it in a P5B Deluxe and the limitation in that case was the memory (DDR2 666MHz). I don't remember the values, but despite the OC of the micro, the voltage was considerably lower than the stock one, so it was a very cool machine. Then I changed the memory to 800MHz and I could overclock it, but from 3.2 or 3.4GHz (I can't remember), the temperatures forced me to increase the fan speed and it was no longer so quiet, so it stayed forever at those 3GHz. In my case, I don't remember specific test results, but I do remember that in GTA 4, I got a noticeable increase in performance when I overclocked it from 3GHz. As @Sylver says, it looks like a motherboard temperature problem. Another thing I would try (if you're not using Windows 10) is to use only the pair of G.skill modules, to see what happens. If 3Dmark goes wrong with that amount of memory, I invite you to use [our benchmark](https://bm.hardlimit.com/download.php) which has reduced RAM consumption.

[Rober García]

@Sylver said in OC viejuno C2D E6600 + Asus Commando:

Muy buenas @Rober-García

I'm not particularly into OC, but I thought that at full load at 375x8 it might reach very high temperatures, causing performance to drop, unless it has very optimal cooling and that's not the reason. I've come across some equipment where the same thing happened, with worse performance at higher frequencies, and I never really understood why.
Let's see what the frequency gurus have to say.

By the way, our esteemed mayor @cobito had a Conroe like that as his main machine until recently and he had it well tuned, so he can probably give you some advice.

It's really interesting to see threads about OC on the old 775s today, it was a glorious era.

Saludos!

Thanks for the reply! In principle, temperature isn't a problem, or the sensor is broken haha. I haven't changed the thermal paste in a long time. I have a Thermalright 120

[Rober García]

@cobito said in OC viejuno C2D E6600 + Asus Commando:

ta hace three years, I've been with an E6600 that from day one has been running at 3GHz. I had it in a P5B Deluxe and the limitation in that case, were the memories (DDR2 666MHz). I don't remember the values, but despite the OC of the micro, the voltage was considerably lower than the stock one, so it was a very cool machine. After that I changed the memories for some at 800MHz and I could overclock it, but from 3.2 or 3.4GHz (I don't remember), the temperatures forced me to increase the fan speed and it was no longer so quiet, so it stayed forever at those 3GHz.

Hello cobito! Unfortunately I'm with Windows10. One day I turned on the computer and I found it (it updated via Windows Update). In fact that's what forced me to rummage through my RAM modules and connect the KVR I found. And it's a pain because the motherboard's overclocking software doesn't work on windows 10. I use setFSB to test limits "quickly"
I'll check the temperature again, but by AIDA they don't go above 45º and the voltage is stock 1.325v...

[Rober García]

Hello!
I've been testing with prime95:
360x9 gave me an error after a short time. It doesn't seem to be as much of a temp problem as a cpu limit, right? It gives it too quickly:

350x9 has been 75 minutes without any problem, cpu temperatures: 60-62:

Oh, I haven't changed the thermal paste for over 10 years. If I have two hours someday I'll do it, because dismantling this heatsink I think involves dismantling the whole board.. I can't remember now

[cobito]

With those results, I would try to increase the processor voltage a bit while monitoring those temperatures.

[Rober García]

@cobito said in OC viejuno C2D E6600 + Asus Commando:

With those results, I would try to increase the processor voltage a bit while monitoring those temperatures.

UFF, I left it for 3 hours with the prime at 375x6 and I left and when I came back it was at 87º. I will change the paste, clean the dust and continue later :s

[Sylver]

@Rober-García x6? If that's less than stock, right?

Best regards!

[Rober García]

Well, I couldn't take it anymore last night and I opened everything, cleaned it and removed the dust... which I had a little bit:

@Sylver yes, 266x9 is the stock. I've lowered the multi to see if it was a problem with the FSB but nothing, the fsb is going like a shot. After cleaning I put it at 410x8 (3280mhz) and it's been an hour and a half in the prime without a problem. Maximum temperature 51º core, 37º MB, 47º CPU
By the way, my other computer is also a 5600+ Brisbane

[Sylver]

@Rober-García It was too strange that with underclock it raised the temperature so much...

Now it is at more normal values, of course it had dust to stop a ship

You can already try the 3DMark to see what results it gets.

The Brisbane is a great processor, it has given me great joys, although right now it is not with me. In the future I hope to recover it, when I have space

Greetings!

[Rober García]

Hello again.
With everything clean I have run the tests again... with the same results. I rule out the temperature issue because they are really stable.
I have monitored the voltages during the tests in case there was any peak, but nothing, stable.
I am starting to think it is the motherboard.
Sometimes when I start up I get an error in Overclock, and it loads the default configuration... but I haven't done any OC now ?...
Does any of you know of a situation where this happens, or any way to check it?

[Sylver]

Man, it reminds me a lot of an Asus K8V-X I had, which got worse over the years until it finally died, and gave that error from time to time... I have no idea about the specific reason, I'm convinced that some guru who occasionally stops by should know, but it's just speculation. The battery I imagine is fine, right? To rule out the most basic thing... Regards

[cobito]

I don't remember if I got a message or not, but my P5B Deluxe had a pin bent on the socket that was precisely used to define the bus frequency. This meant that sometimes, the processor would not be configured at its nominal frequency even when a clear cmos was done. Perhaps if dirt has entered the socket, a pin has been bent, thermal paste has entered, or anything else, something similar might be happening to it.

[Rober García]

Thanks for the replies. I'll check the battery and the dirt. I think I changed the battery recently, but it can't hurt to check.

More tests: I've lowered the HLBM and I'm doing tests and something weird is happening or I don't understand the results, because no matter what FSB I set on the e6600, I get almost the same results. It only cares about the multiplier

FSBxX CPU  Mem Timmings MonoThread MultiThread 375x6 2250 1:1 4-4-4-12 9.48k      15.43k 325x6 1900 1:1 4-4-4-12 9.42k      15.25k 182x6 1090 1:1 4-4-4-12 9.69k      14.97k  375x9 3375 1:1 4-4-4-12 13.52k     22.16k 182x9 1642 1:1 4-4-4-12 13.83k     22.55k 

[Rober García]

@cobito No dirt or paste, but the processor was full of a gelatinous substance that I don't remember seeing before. I didn't remove it because I thought "if it has been there for more than 10 years it can't be harmful"

[cobito]

@Rober-García Those results don't make sense. I've been reviewing the validations you've sent and the only one under 2Ghz gives a multi-core score of just over 12k (which matches the frequency), but I don't see validations at 1090 and 1642Mhz that give those 14.97 and 22.55k respectively. If you could repeat those two configurations and validate the results, I could look more closely at what might be happening in the performance charts. So far, in the 6 submissions you've made with the micro both over and underclocked, I don't see anything out of the ordinary.

[Rober García]

@cobito You are right. I was overclocking from Windows with clockgen and cpu-z recognized it, but it seems that the FSB was not being modified... as is evident from the results. I have done the OC from BIOS and it does vary the results accordingly. Also with superpi I had the same behavior.
When I have time I will check the pin you mention

[cobito]

@Rober-García If it was a failure in the CPU-Z frequency measurement and you can make the modifications via BIOS, I don't think you have any pin problems. In fact, I don't recommend removing that micro if it has been placed for a few years, because that's where a pin can bend no matter how careful you are.

[jordiqui]

@Rober-García what memories those coolers. Have air solutions changed much?