This is an easy mod if you can fiddle with a PC.

It's especially helpful if you are running a locked down bios mother board such as most Intel MoBo prebuilts utilize.

I bumped my Intel Q6600 from a 1066 FSB to a 1333 FSB by taping just one pin. I took some black electrical tape and applied it to a piece of glass. Then I made a small rectangle using an exacto knife. This is a SMALL piece of tape. It's about 100 atoms by 120 atoms in size, haha.

Using the tip of the blade I then removed the tiny piece of tape from the glass and placed it onto the appropriate pin. Some guys use electrically conductive paint to connect this pin to another pin but that seems to be overkill.

Now the PC boots up and runs faster. And can be further overclocked from this higher base point. Woohoo!



Here is a view of the pin pads on the bottom of an Intel Q6600 Socket LGA 775 chip:

isn't it easier to change the BIOS settings? less risky... and if this is a voltmod... it can be done under the BIOS too



I don't see the point... if (for example) you got a piece of hardware that is not stable at 1333 then you won't be stable at that freq no matter what yuo do (changing a BIOS setting, or making a mod to the hardware)

anyway if a motherboard doesn't allow overclocking it's because it's not ment to be overclocked, it happens a lot with cheap motherboards... it won't be the best overclocker anyway... I think that buying a new mobo is a better solution if you don't want to burn something (or deal with stability problems)

Umm..most Q6600s live on Intel boards and not enthusiast mother boards.

In my case this is on an Intel G33 board...which is really quite excellent. It's 100% ready for the 45 nm ECUs and DDR3, plug and play. But it does not have user adjustable BIOS, which of course is the easy, conventional, well-mapped way to overclock.

No problem, hard core is fine with me. I got a great deal on a prebuilt Gateway so I went that way and learned and tried overclocking stuff recently for the first time despite the extra challenges. It seems like there is just a different way of doing things as you can get to the same place roughly.

The Intel G33 (and P35) board overclocks quite well. On stock ECU voltage (1.2625) it runs stable to somewhere around 3.2 Ghz +. I can bench it to 3.35 - 3.4 Ghz on 1.26 volts, but stability goes down and so I am running 3.2 for now as that is bulletproof. With some extra voltage I should be able to settle in around 3.4 - 3.5 air cooled with forget-about-it reliability. I'll plan to do that in a few days after going through the Intel documentation. There are all sorts of interesting things in there!

nice to hear that, every day I learn something new...
good luck!

If you want to lock the BSEL2 signal HIGH, it'd be better to tie it through a terminating resistor then simply opening the circuit. A whole lot less likely to cause, oh, say, ringing in that output line and an unstable clock...

Thanks for your thoughts here.

It seems to be working just fine. Last night and all day today at 3.2 Ghz via SetFSB, my usual applications plus common benchmarks. I think it can go higher with more ECU voltage, the memory is up to it.

There seem to be 3 main ways that people are doing this type of pin mod. One is via tape as shown above. Another is by tying the BSEL2 pin to VCC (two pins up) using conductive paint. And the third is by taping BSEL2 and then connecting the top of that tape to VCC (from two pins up, again with conductive paint). Mods like this are at your own risk like any other mods or overclocking.

Any comments on those other approaches?

Wow. Your poor chip. You grinded down the top and taped the bottom at the same time.

With all the stuff you've done to your Gateway, your first build is going to be awesome. What's the ETA on that?

LOL, thanks. Don't you like "lap jobs", haha? Next comes voltage mods via more pin modding. From what I gather about Intels big changes coming later this year...I think I'll hang with this stuff for some time and keep an eye open. Penryn is a tick while Nehalem is a tock. My guess is a whole new socket and related big MoBo changes (no more FSB) along with cheaper fast memory. As early as 4Q08. So why invest a bunch now?

The largest lags are:

1) Hard Drive. Bog Slow compared to everything else in a PC. Raptor Raid 0 is not enough to fix that, solid state hard drives are needed.

2) Software. So rare are the ones that can fully use the hardware in a useful manner.

3) Graphics Card drivers kinda suck. What a PITA.

4) Internet / Web Site connections

5) Waiting for more sex

A particular challenge has been heat in this small, tight case. I have that all handled now, and quietly. That took some effort and was much harder than this simple overclocking. One thing that impresses me is that the whole PC motherboard layout is legacy from the days when chips used 15 watts hence heat was basically a nonissue when this stuff was "permanently" cast in concrete. I'd lay things out much differently were I to change PC layouts at the MoBo level.

Wow. You should start a thread (or even a whole board) for modding cookie-cutter systems! There are so many people out there that want to...

I want a solid state drive for the OS. Can't wait until they come down in price.

They have the right idea. I'd rather put a 100k-ohm where they're putting the paint, though I reckon if you lay the right sized paint line you might accomplish the same thing.

Trust me - it may not show now, but at these speeds I personally wouldn't want to risk it. I'd want that line tied off rather than open-ended...

Well the paint commonly suggested is the nearest thing to paintable copper wire, so the resistance is low. However that can be changed by using the penciling technique.

So you prefer the concept of tieing BSEL2 to Vcc. Well which of those two strategies? The direct paint BSEL2 = Vcc, or the paint over the BSEL2 tape and connect that to Vcc?

How could I test for the issue you bring up? So far, temperatures are very good, performance is great, and any frequency I've monitored is steady with no surprises, stability is fine.

Thanks.

Oh BTW check out some pin details. Some are not used. Some seem to have barely contacted a pin. Some seem to have two pins feeding them (to carry more current?). Most are bullseyes though. Some of this could be lighting related...I'd have to examine the chip again under a magnifying glass. I also see from reading Intel's notes that the insertion process is rated for only about 15 times.

Well, now you're getting into something I haven't physically done since CPUs hit the multi-GHz range - because none of my scopes and crap were fast enough, and the circuits I build aren't that fast either.

And considering the length of leads in 65nm tech, there's every bit the possibility that it may be a non-issue.

It could even be that it simply goes contrary to everything I've been told or experienced in circuitry and it's just a subconcious thing.

After Googling the technique and seeing how widespread it really is, reality is probably different.

As to the other question - I'd probably try laying the paint line, then covering it with tape for the same insular purposes you had from the socket contacts...