Overclocking question

vbimport

#1

At the moment I use an intel Pentium Dual-Core E2180 (I wait for an inexpensive Yokfiled Quad). That CPU is an Allendale Rev. M0, and comes with FSB800/2000MHz. Since the Allendale Rev. L2 exists as an FSB1066 CPU as well, and since the M0 is even a higher and (hopefully) better Rev., would it be actually overclocking at all, if I set it to FSB1066/2660MHz?

Wouldn’t heat generation, voltage used, current used and power consumption (I mean current and voltage together at the same time) stay the same? I once read that 33% OC means 100% more power consumption. Is this correct at all? It would be zero % more at least in this case, because the same chip is sold as FSB1066/2660MHz as well, wouldn’t it?

I know that these CPUs are very good overclockers etc. But my question is, if it’s overclocking at all, when the CPU actually does not “over” clock at all.


#2

I’m also using an e2180. I have it currently at 3.0GHz, using a 3:2 FSB: DRAM ratio. My DDR2-800 memory runs at 900MHz when using that particular divider. If your using cheap memory that cannot accept a good oc on the ram, then using a 2:1 divider will underclock the memory and hence give it a large headroom to oc the cpu.

Power consumption does increase when oc’ing but not by the amount that you have read, its much less. Also most people that are oc’ed don’t run their computer at 100 percent all the time. Your can test the difference in power consumption with a ‘kill-a-watt’ power meter.

Your MB is very OC’able and so is your CPU, but it depends on a lot of things, like how cool your case runs, whether or not your using the stock CPU cooler, the quality of your RAM.

Increasing the cpu voltage is done gradually to make the cpu stable when oc’ing. More heat is generally created when doing so becuase there is a certain ‘leakage’ current that gets greater with increasing voltage. With the stock intel cooler you should probably be able to reach 2.8GHz.

Overclocking is a balancing act, and there are some tests that can be done prior to starting. The first is to find the maximum FSB the cpu is capable of. This is done by decreasing the cpu multipliler to the minimum, 6x for the e2180, and underclocking the memory with the 2:1 divider. Then increasing the cpu speed and testing its stability in windows with a program like Prime 95, or Orthos. Those two programs are capable of fully stressing the cpu cores to see if they are stable. After finding the max FSB, you then have an idea of how fast the cpu can go, but this is not factoring in the heat generation that will occur when rasing the cpu multiplier back up to 10x.

The next step is to find a safe speed and voltage for the RAM without stressing it to hard to insure long life. For example I decided to run my ram at ddr-900 with just 2.05V applied, the ram is capable of much higher 2.2V and much higher speeds.

There are some good basic guides to oc’ing a core 2 duo processor, that should give you more insight on the web


#3

Your MB has that Ai oc’ing, I would turn that off, since that service will probably make your attempts at oc’ing a lot harder and unstable. Realize that if you try to oc too fast without testing for memory stability and cpu stability that your can corupt the data on your HDD and will require reinstalling everything on your PC.

There is a bootable DOS program that tests memory, its called memtest86.

Also for your first question, your not able to change the fsb of the cpu from 800 to 1066, that is something that is set within the cpu. You were just changeing the ‘strap’ speed of the MB’s bios. There is a way to ‘pin mod’ the cpu but that is not necessary for this cpu because its already very easy to oc.


#4

First of all, you should use memtest86+ instead of memtest86
http://www.memtest.org/
It’s also runs on Linux not DOS.
//Danne


#5

First, many thanks for answering!

Well, I have the memory for that. It’s Micron chips :iagree:. My CPU cooler is not the stock cooler as well. It’s the ASUS V70 (although not expensive, it’s for up to Core 2 Extreme, 4 copper pipes, huge amount of aluminium and 80CFM). Chassis: A Scythe Slip-Stream Super High blows out and one Slip-Stream Medium blows in. The chassis has a complete mesh front and a CAG side door. Although the Medium blows air in to cool the HDD, the Super High and the PSU together (about 200m³/h!) still pull enough air through the mesh front and the CAG(!) worth of another Slip-Stream Medium! No heat from the GPU as well since cooled by the ZALMAN VF700-AlCu + clocked down to nVIDIAs stock speeds (core/shader/RAM). That’s concept, isn’t it. Just kidding. :wink:

So that all wouldn’t be a problem I guess. The problem is that I don’t want to overclock. But I wondered if setting the CPU to FSB1066 (which would result in a 2660MHz CPU) would be overclocking at all for the reasons I described in my original post.

That’s the only case I would do it. You know, if it’s an act of overclocking, but not in physically, since my very chip is sold as an FSB1066 and 2500MHz+ CPU as well.

And: Can’t change FSB? :eek: Why? I thought that’s the only way to overcklock a Conroe/Allendale, because the multiplier is hard-wired in the CPU??? Now I can see, that I don’t know much about ocing. Maybe I should just wait for the Yorkfield. But that one I won’t overclock, that’s for sure anyway.


#6

Okay so you don’t want to oc, :rolleyes: hehe

Yeah I mis read your first post. You can OC your chip to a 1066 fsb, but you would still have to go through the steps to make sure its stable. you would increase you bus speed from 200 to 266 or 267, because its quad pumped.

The fact that some cpus are rated at 65W and run at different speeds is because that is a theoretical value that is calculated. the version running at 2.0GHz is not drawing 65W and the 2.6GHz version is also probably not drawing 65W. They are probably the same chip just of different internal quality so that it runs within certain parameters, they then set the FSB strap or the multiplier and sell them accordingly. the 1066fsb version of this chip also has more cache, so it is a different chip altogether, but a part of the same platform.


#7

Many thanks [B]eric93se[/B] for explaining! :slight_smile: :slight_smile:

OK, let’s see if I understand this correctly:

My E2180 is NOT a 5GHz super-CPU. To the contrary, it’s a bad Allendale. That’s why it’s only an E2180. That Conroes and Allendales are good overclockers only means that most of them can indeed overclock a lot, because they are very high quality chips in relation to the clock speed range they are sold in. I would still kill my E2180 with electomigration and other things when operating at FSB1066/2667MHz. It’s just, that I kill it very very little. Although almost any E2180 would run let’s say 20 years at FSB1066/2667MHz, only another part that performed even cooler at FSB1066/2667MHz at intel’s testing is a true FSB1066/2667MHz CPU, running let’s say 22 years at that clock speeds. So theoretically it would be overclocking! Very little risk, but it would be overclocking.

Is this right? Well, no ocing for me then. :slight_smile:


#8

Yup you’ve got it.

Most people don’t need the fastest computer since they surf/type/email, so OC’ing is out of the question for them. Others who are running folding/games/number crunchers/video editing want the most out of their systems, so they OC. :smiley: