e8400 overclocking problems - Vdroop?

[The_Ollienator]

Hey there,

Decided to have a go at Overclocking my shiny new e8400 today and started off by going 400fsb and 1.3vcore.

Boots into windows fine, havent tested yet (though i highly doubt its stable first attempt, im going to push for 4ghz)

But, in CPU-Z. It's telling me the Vcore is lower than what i actually set it too and im very worried. I have heard about something called vDroop and i presume this is it.

Its hovering around the 1.2v mark and has dropped as low as 1.1v seems to variate ALOT



Any advice? In BIOS it shows as 1.3 but not on cpu-z

[Webby]

Don't worry about it. It is the same for most boards use the BIOS value for what your happy to go to so if you do not want to exceed 1.4v do not exceed that in the BIOS.

The only way you will know what the actual voltage reaching the CPU is, is to use a DMM (Digital Multi Meter) find the appropriate read points on your motherboard and check but it really is not worth the hassle.

[watercooled]

Sorry for thread hijacking, but what exactly is the difference between the E8400, E8500, etc ie. why can they have different clock speeds? Are less-well-performing chips sold with a lower stock clock?

[Webby]

Basically yes, the chips come off the production line are tested and then graded. However Intel sell more cheaper CPUs than expensive ones so better performing chips can be graded lower to fill demand.

As to what is different between the actual CPUs they have different Multipliers so a E8400 has a multiplier of 9 so its clock speed is 9x333 = 3GHz (where 333 is the FSB) the E8500 has a multiplier of 9.5 so its clock speed is 9.5x333 = 3.16GHz. The multiplier of the chip is limited to a maximum value (except in the case of Extreme edition CPUs) however it can generally be lowered, the FSB is not fixed and that is what people increase to overclock their CPUs.

In addition to multiplier differences the are different series of CPUs concentrating on Intel's processors you have two main distinctions E and Q, E are duel cores Q are quad cores and sub categories of X and QX which are the extreme editions for duel and quad respectively. Then the first number determines the series and its manufacturing process 65nm or 45nm. The first number also to a degree tells you the level of cache the processor has the higher the number (in the particular series and manufacturing process). The second number tells you the relative speed of the processor higher is faster. The next number determines the level of cache again within the series higher means more again. The final number is always a 0.

Hopefully that shed some light on it if not tough I'm not writing that again!

[watercooled]

Yes, it was very clear thank you. I understand about multipliers but like you said they are set by the mfr. So the chips within a series (don't know if that is the correct terminology) are physically exactly the same apart from manufacturing inaccuracies?

[Webby]

Yeah pretty much but every CPU will perform slightly differently.

[watercooled]

Hence different OCing potentials. Phenom X3s have a failed core - to me that sounds like a core is broken when AMD test it so they disable that core and sell the chip. Doesn't that mean the chip is more likely to have a shorter life expectancy as on Phenoms, all 4 cores are on the same die?

[Webby]

Nah, the core that failed may not actually have failed at all it may just have been turned off because the was more demand for X3 than X4 chips at that time, equally the reasion it failed could be a minute error in the core which will have no effect on any other aspect of the operation of the rest of the CPU. If they didn't think they would last they wouldn't sell them nothing looses you custom faster than selling products which die all the time!

[watercooled]

Good point. They've had plenty of praise and since what you're getting is a Phenom with a core disabled for a good price it's good for applications where all 4 cores cannot be used simultaneously. Good idea in my opinion. I wonder why Intel hasn't done something similar?

[The_Ollienator]

problem solved i think. I did a CMOS and BIOS settings reset and all my overclocking options returned to normal (it stopped rating everything as red for unstable etc)

Either that, or it didnt like my ram timings being tightened to 4-4-4-12

AND CPU-Z is showing correct voltages now.

Haven't experimented properly, but im running 4.05 at 1.3 currently. Next up is tightening the RAM,seeing if that works, then dropping voltages.

Current settings:

CPU = 1.3v
RAM = 2.1v
NB = 1.4
SB I/O = 1.7 (i may have got sb and nb mixed up)
problem solved i think. I did a CMOS and BIOS settings reset and all my overclocking options returned to normal (it stopped rating everything as red for unstable etc) Either that, or it didnt like my ram timings being tightened to 4-4-4-12 AND CPU-Z is showing correct voltages now. Haven't experimented properly, but im running 4.05 at 1.3 currently. Next up is tightening the RAM,seeing if that works, then dropping voltages. Current settings: CPU = 1.3v RAM = 2.1v NB = 1.4 SB I/O = 1.7 (i may have got sb and nb mixed up)

[moogle]

What's your mobo? Usually they have a VDroop option which reduces it when under load keeping the voltage near the same.

[The_Ollienator]

What's your mobo? Usually they have a VDroop option which reduces it when under load keeping the voltage near the same.

MSI p35 NEO-F

[moogle]

MSI p35 NEO-F

Ahh. In the Neo2 FR I couldn't find or remember the VDroop option. The voltage shouldn't be so low from the bios setting until you stress it though. Also the red thing you're talking about, if the bios/boards are similiar which I think they are, then that means (the red) that it's over the limit. Like if I set my ram voltage at 2.2 it's red etc and all those other voltages. As long as you don't really push it over the limit then it'll be fine, as in if you set the voltage on a setting just gone into the red zone.

[The_Ollienator]

got it running at 4.0GHZ on 1.25v currently. Had to loosen ram timings to 5-5-5-15 else it wouldnt boot. Stable for 1h30mins so far - leaving it to run for a long time.

[Webby]

Well yeah as you increase the FSB you also increase the memory speed if your still on a 9x multi then your FSB is currently 445 (or maybe 444 if your slightly under 4GHz) which means your memory is running at a minimum of 890MHz depending on your memory and the voltage you give it will determine what its max speed and at what timings.

The is not a huge ammount of DDR2 memory out there that can handle 900MHZ at tight timings its no coincidence that most DDR2-1066 has timings of 5-5-5-15 of course some memory will do it for example my OCZ Flex DDR2-1150 will run at DDR2-1000 at 4-4-4-12 but it was expensive (well actually it wasn't but it should have been!)

[The_Ollienator]

Well yeah as you increase the FSB you also increase the memory speed if your still on a 9x multi then your FSB is currently 445 (or maybe 444 if your slightly under 4GHz) which means your memory is running at a minimum of 890MHz depending on your memory and the voltage you give it will determine what its max speed and at what timings.

The is not a huge ammount of DDR2 memory out there that can handle 900MHZ at tight timings its no coincidence that most DDR2-1066 has timings of 5-5-5-15 of course some memory will do it for example my OCZ Flex DDR2-1150 will run at DDR2-1000 at 4-4-4-12 but it was expensive (well actually it wasn't but it should have been!)

Very true

Yeah, my fsb is at 445 so ddr2 is at 890mhz. at 4-4-4-12 with exactly the same settings otherwise doesnt boot. Loosen those timings to 5-5-5-15 (which is pretty good for 4x1gb anyways) and it boots.

I'm very suprised too - Its been on orthos for 2 hours so far and i havent seen it go over 45* (at 42* currently) and yes, tjmax is set right at 100*.

Very happy, very good chip.