EPT![[image loading]](http://i.ebayimg.com/t/AMD-Phenom-II-X4-955-3-2-GHz-Quad-Core-HDZ955FBK4DGI-Processor-/00/s/MTIwMFgxNjAw/$(KGrHqZ,!ooFBsd(QfqwBQjPNhBq8w~~60_57.JPG)
It doesnt need to fit in perfectly. A perfect CPU would just drop right into the socket. If it doesnt fit quite right in, test it by just sticking a corner or edge in, see if it'll go in. Then lightly keep pressing around, to see how far it'll go in. Usually you can figure out exactly where that bent pin that's giving a problem is, and fix it up, even if it's in the middle (it'll sort of teeter). You should never have to force the CPU in or use any force, but you will have to wiggle it and push it in edge by edge, corner by corner if the pins are really bent out. You really can't 100% bend a pin back 100% straight, especially if you have a ton of them, but you can get close.
As for broken pins:
Here's a bunch of pics of my processor with broken pins and fucked up pins, and how i fixed it:
This is the processor. Many of the pins are still bent, even after I straightened many out. On the left side you can see they are still slightly bent, and in the right you can see a really fucked up pin, and 2 missing pins. The CPU won't smoothly go into socket like a normal chip, but with this CPU, I take this worst corner, and put it in first, and then VERY LIGHTLY push on each edge to sort of crimp it in. You never want to force it in, but with bent pins you might not be able to smoothly drop it in, but have to sort of do this edge by edge to get it in. It's more about wiggling it in, one side/corner at a time, than trying to force it in.
Note that this is after I had straightened out the pins as much as I could (as much as necessary). This didn't drop in smoothly, as I described above, but it will fit (although, this pic shows which corner I have to drop in first).
![[image loading]](http://i.imgur.com/gsgngl.jpg)
You actually don't need to worry about getting the perfect length of wire. What you do is, is you get a piece clearly 2x the length of a pin. Then, stick it into a pin hole, and slightly bend it, so you know how long it 'needs' to be. Then, simply cut so it's still like a good half a pin extra.
![[image loading]](http://i.imgur.com/W1M93l.jpg)
Then, very simply, smash smash smash it in with a phillips head screwdriver (has to be phillips, the pin holes have a slight indention to them, just as the cpu pins have a slight, circular, base at the bottom of each pin). This is actually a MUCH better method than trying to use finesse and cut the perfect length, literally .001mm too short, and you won't get contact, too long, and it'll possibly hit another pin and screw things up royally. You also won't be able to pull it out if it's just slightly too short (or too long, even). This also ensures very good contact with the CPU, and makes sure it won't touch any other pins, the best of both worlds, and more convenient, and easier to do.
![[image loading]](http://i.imgur.com/NwQsBl.jpg)
Close-up. You can see I got a little overzealous and mashed the actual socket, the plastic is quite soft. Not a big deal though, I actually probably used a little bit too much wire here.
![[image loading]](http://i.imgur.com/xrwd9l.jpg)
This is a picture of the replacement pins I made before using the smash smash smash method. I found what is easier than trying to cut to length, was cut to 2x the length + a bit, fold the wire, and then stick it in, with the bend up top. I still smashed it in a bit, but I suppose I used more finesse here.
![[image loading]](http://i.imgur.com/YLA6gl.jpg)
In total, 5 replacement pins.
![[image loading]](http://i.imgur.com/hyc0cl.jpg)
If you really screw up, you can always just turn the motherboard upside down and the pins will shake out. Use tweezers to place the pins in.
I've hit 3.7ghz@1.504vcore so far, and the chip doesn't look like it's going to stop anytime soon. I will be testing 3.8ghz and above later this week, but I know it can last at least 10 passes of IBT and 2+ hours of p95 on 3.8@1.536vcore. Note that most C2 revison 955's hit 3.6ghz 24 hour prime95 stable, so the chip has already proven to be well above average. I have not seen any reports of 3.7-3.8ghz prime95 24 hour stable on C2 revision, but I have heard general reports of 'stable at 3.7-3.8', although they are in the very, very small minority, and I'm not sure they were 24 hour prime95 tested.
C2 revisions are generally thought to hit 3.6-3.8ghz at most though. I think the record for 'stable' (dont think it was 24 hour prime95 stable) was around 3.8.
