OC Micro and RAM

[elsaq]

Hello!

Here are the details of my equipment:

• CPU: Intel Dual Core 2 E6600 2.4 ghz @ 3.06 ghz (334x9)
• Motherboard: Asus P5B
• Graphics: Sapphire Radeon X1950XT - 256 mb GDDR3 / 256 bit
• RAM: 6 gb (2 x 1 gb / 2 x 2 gb) identical and in dual-channel, PC2-5300 / 667 (5-5-5-15)
• PSU: Seasonic 550wt

Well, as I mentioned in another post and with your help, I have already decided to change the graphics card to an HD 5850, but looking at my CPU, it might be a bit of a "bottleneck" at 3 Ghz.

I have thought about overclocking the CPU to 3.2 Ghz without touching the voltage, the problem is that to put the CPU at that, I would have to set 400x8 and then I lose the 1:1 ratio that I currently have with the OC at 334x9.

The solution I have thought of is to try to set the memory from 667 to 800 mhz, this way they would be 1:1 if I set the multiplier to 400x8.

I have never overclocked memory before, so I have no idea how to do it, although I imagine it's all from the BIOS, right?

The default voltage of these memories is 1.8v, and they are Kingston Value RAM DDR2 PC2-5300 667 Mhz, that is, they are very simple memories and I am afraid they will break because I just bought them recently.

Has anyone overclocked these memories to 800 mhz? Can you help me?

Thanks! Best regards!

[CNCBCN]

if you edit the signature you save yourself the trouble of putting the equipment in each question
I think Kingston doesn't allow OC, they are what they are, but maybe someone with more knowledge than me (which is easy in this OC thing ) will tell you another option

[elsaq]

Thanks for the signing advice, done

We will wait to see what they tell us

Thanks! Regards!

[elsaq]

And what do you think would be better, leave the micro as it is at 3 Ghz with the FSB:RAM at 1:1, or raise the micro to 3.2-3.4 Ghz and let it stop being synchronized with the RAM?

Or is there a way to leave it 1:1 at 3.2-3.4 without touching the RAM?

Thanks! Regards!

[amd125]

I have a Q6600 that is basically like two of your micros, with some Kingstons and I have it at 2.4@3.2 memos at 400 and multiplied to 8. I put some juice into the micro by increasing the Voltage in Bios and I leave the memories as they are, at 1.8 and that's it. I don't touch anything else, I'm too lazy to adjust other things, and that works well for me. That said, my memos are DDR2 800

Look what the VID of your micro is, with the Coretemp program and it will give you an idea of the OC capacity of your micro. Mine has a VID of 1.325, the worst for a Q6600 :nono: but it doesn't prevent me from doing an OC to 3.2. That said, putting more voltage into it than a micro with a lower VID. :mudo:

Try to lower the voltage to the memories a little and go increasing the FSB little by little, leaving the multiplier at 8.

Kingston is not a good brand for Overclocking but that doesn't prevent a moderate OC from being done to these memories.

[elsaq]

I have the VID at 1.3500v

If I increase frequencies but leave the stock voltages, is there any risk of something burning out? No, right? If the frequency I set is not stable, what could happen to the micro or the ram?

Thanks! Regards!

[Sylver]

If it is not stable for that there are stress tests, if it restarts it means it is not stable, so reduce the frequencies a little :sisi:
And of course to control temperatures from when you do the OC, it can be stable and the micro is frying due to poor cooling :nono:

Greetings

[elsaq]

Cooling is not a problem, I thought about it when I bought the micro, since day one it has been with a Tuniq Tower 120 and Artic Silver 5 thermal paste

A link to the heatsink: Tuniq Tower 120, the best CPU cooler with great cooling performance

Any Bench just for CPU?

Thanks! Regards!

[Obione]

There are no more eggs than OCear RAM to improve that, what you can do is give a little more voltage to memories and if that relaxes latencies.

[elsaq]

Well, I'll tell you:

I've been tinkering a bit with the BIOS and I've found some problems. The first one is that on the box of my motherboard it says that the maximum Rated FSB is 1333 Mhz (currently it's at 1334 Mhz, which is the same for all intents and purposes) but if I go further than that value the computer won't start up. That is to say, I have the FSB "locked" at 334x9 for the 3.06 Ghz of the micro, because if I increase the FSB more than 334, I also increase the Rated FSB, so there's no way to touch that value. Because if I lower the multiplier, I have to increase the value.

I don't know if I've explained myself well because I'm a bit confused to be honest, so, have I hit the "ceiling" with 334x9 at 3.06 Ghz for 667 Mhz RAM?? Or am I doing something wrong (which is more than likely)

Thanks! Regards!

[Obione]

You should be able to overclock it manually, another thing is that you might reach 400 per limit on the board or in micro, for memory I think that with what I have told you you should not have a problem, I suppose they will hold 800 with latencies 6-6-6-18 and 1.9V.

[elsaq]

Well I have touched some voltages that were on AUTO and I have set them fixed (not to fluctuate, I have looked at information)

• FSB: 1.30 v
• NB: 1.25 v
• SB: 1.50 v
• ICH: AUTO

Changing the processor to I noticed a minimal improvement, unnoticeable but there it is, in the SuperPi, before touching that, it took 18,408 seconds to do 1M, after touching that it does it in 17,596 seconds.

The data is insignificant but it is already a step in the right direction, now I will get on with the frequency, in principle I am going to try this:

• FSB: 365x9
• VRAM: 1.90 v (will I not burn them? they are some kinsgton valueram, I don't know if they will hold up…)
• RAM: 730 Mhz (365x2)

We will see what happens!

Goodbye!

[elsaq]

I've been tinkering a bit and I've managed to get it to boot and enter Windows with the following settings:

• FSB: 356x9 – 3.202 Mhz

• VCore: 1.3500 v

• RAM: 712 Mhz 5-5-5-15

• VRAM: 1.90 v

The orthos doesn't last even 1 minute, and I get this error:

"FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 0 minutes 29 seconds - 1 errors, 0 warnings.
Execution halted."

In the 29 seconds that the Orthos has lasted, Core Temp has marked 46ºC as the maximum temperature, so the temperatures seem to be okay for now, right?

What should I do now to make it stable? Should I increase the VCore to more than 1.35v??

The stress test for the RAM also fails, but I don't know if it's because of the processor or the memory. Should I leave it at 1.9v or increase it to 1.95v? 2.0v seems like too much and I don't want to reach that number, so I'll set 1.95v as the maximum.

I'll continue tomorrow as it's getting late!

Best regards!

[Bm4n]

It would be better for you to first test if the micro's oc is stable by lowering the memory divider so that you don't have any problems and then oc the rest. I imagine that in order for the memory to go up you will have to increase the latencies to cas 6.

[elsaq]

I continue with my attempt to reach 3.4 Ghz, I tell you: I have noticed problems with the Vcore, that is, I set it to 1.31v, for example, and in the Hardware Monitor of the BIOS it marks that the voltage of the core is 1.28v. The same thing happens if I intend to go beyond 1.35v, so that in the Hardware Monitor of the BIOS the reading of the voltage of the core is 1.366v, I have to set the Vcore to 1.41v, which is the real voltage? Is it normal for the voltage to drop so much between what you set and what it gives? What is the real voltage then? The one marked by the BIOS sensor or the one I set manually? Thanks! Regards! I EDIT: Currently I have it at 1.31v and it marks 1.28v, but it drops even more in FULL, when passing the Orthos it drops to 1.24-1.25v, is the voltage in FULL the real one?

[Obione]

The real one will be whatever it is, the thing is that you have voltage drop under load (understand it as a dynamic value). If you don't have the LLC (Load Line Calibration) option you won't be able to fix this Vdroop. Then there's the Vdrop, which is the difference between what's set and the "real" voltage. Try that with CPU-Z to see what it shows you.

Bm4n the memories are already 1:1 but well, there will always be latencies as you say.

elsaq relaxes the latencies if they are not already at 6-6-6-18 (you can look it up in CPU-Z). Focus first on the micro that's what's going to give you the most performance, and then if you want you can adjust the latencies a bit more aggressively.

[Bm4n]

If you can't lower the memory timings any more, do the opposite, first OC the memory by raising the divider and lowering the CPU speed. It's a manual process to OC part by part, this is the only way you can know for sure what is stable and what isn't. The only thing you're doing now is taking a gamble by putting high values on the ram without knowing why it's not stable.

Once you have the CPU without OC, you can see what values the memory supports, but normally if it's a high OC, it's necessary to increase the latencies.

And as obione says, vdrop and vdroop is normal and you should take it into account.

[elsaq]

Thanks for the help! ;D

For now I'm going to leave it as it was before at 3 Ghz stable, the micro has been like this for 5 years and I don't want to risk it now by increasing it 200 or 400 mhz and crash it, and on top of that I just bought the RAM so that it crashes too xD

I wanted to increase it because in a week or so I will change my graphics card, from an X1950XT to an HD 6850, and I fear it will make a "bottleneck" and that's why I want to increase it to 3.2-3.4 Ghz, but well, when I buy it I will test it at 3 Ghz and if I play games well I won't worry hehe

Thanks! Regards!

[Bm4n]

Not everyone is suited to overclocking, you need to have knowledge and a lot of patience, it's not just about increasing values. Anyway, as a consolation, I'll tell you that 200MHz will do little to improve the performance of the graphics.

[incrediboy]

the only real voltage is the one that is obtained by measuring it on the motherboard with a polymeter / multimeter / tester. if you have one of these it would be much easier to know the real voltage.