OC'ing The Thunder K7

Author : Luis Date : 13th August 2001

• Tyan Thunder K7 S2462UNG
• 2 x AMD Athlon MPs @ 1.2 GHz
• NMB 460w P.S.
• Morpheus GeForce 3 230/515
• 512 (2 x 256mb) of Crucial PC2100 ECC & Registered
• Hercules GameTheatre XP
• Aafreey 56x CD-Rom Drive
• Zoom 56k v. 92 Modem (No Broadband in my area yet :o( )

 

---

Concerns:

The first concern one would have with this project is "Will it work?" Well it was reported by some sites that the PK units will not work on the Athlon MP processors, but they were wrong. They not only work but work flawlessly. All of the new features in the Palomino core were unaffected by the addition of the PK's.

Another one that comes to mind is whether it will make the system unstable because of the extra trace length added to every pin by the socket. The answer is again no since my system is now overclocked and just as stable as before. Now whether you could reach a higher overclock if the trace length was shorter is unknown, but I doubt it would be more than 66 MHz, or half a multiplier point.

The last one I can think of is if it will damage the processor. Well, yes and no. Overclocking in general will shorten the life of your silicon, however, the only way I can see to damage the processor by using this is to chip it by handling the HSF without care or to burn it by setting the multiplier too high. Overclocking is not for newbies and should not be attempted unless you are sure you know what you are doing!!! Oh one more thing, Radeonic is not responsible for any damage or losses you occur if you attempt this.

This isn't really a concern but instead something to keep in mind. Both my Pally's are of the first stepping available, my board is also the first revision made. With this in mind and the fact that we now have two heat producing chips right next to each other, cooled only by quiet Taisol HSFs, one shouldn't expect these chips to hit a very high clock speed.

Background:

I Spent the first half of this year waiting for the release of the AMD 760MP chipset. At one point I checked every hardware news site out there about three times a day each. On June 4th this monster of a board was finally released! I was amongst the first people to receive the production version of the board. It took me another couple of weeks to round up the rest of my components, but even before they all arrived I already had my eyes set on overclocking this beast. That's when I faced my first challenge.

The Tyan Thunder K7 S2462UNG (Thunder K7) motherboard comes with everything included. Dual SCSI channels, dual onboard NICs, onboard video, 5 x 64bit PCI slots, 1 AGP Pro slot, 4 x 2.5v 184-pin Registered DDR DIMM sockets for Registered ECC PC2100 RAM, 40 + capacitors on the board, the almighty AMD 762 northbridge and the AMD 766 southbridge (an all-AMD board!!!). It has everything you could ask for and a bag of chips, but don't get me wrong, you'll pay for every bit of it when you see it's hefty price tag. Apparently the only thing left out is onboard multiplier settings and other FSB settings besides '266' and '200', which are actually '133' and '100' double-pumped. This does however make sense since the Thunder K7 is a server board and was never intended to be overclocked.

As I sat pondering how I would overclock my newly acquired twins I came upon very few choices. I either attempted the excruciatingly delicate, and hardly reversible, task of doing a hard OC on the chips by connecting the various 'L' bridges, or I tried to get my hands on two extremely rare pieces of equipment. That's when I heard that Wayne was running a deal on these babies. He managed to get us 100+ units and distribute them without pocketing any money.

As I waited for my new toys to arrive I frequented the Rad-Board to read up on the latest news. That's when I read a post that scared me stiff. According to Plycon, the PK-OCK7/EV6 (PK) units would not work on the new Athlon MP processors. I immediately went to AMD's website and downloaded the newest tech specs for both processors. After studying them for hours I came to the conclusion that Plycon must be wrong. The pins that control all ID settings, multipliers, FSB, and voltage, are exactly the same on both processors. Every other pin was confirmed as being continuous on the PK so there was no reason for them not to work. But as sure as I was I could not stop thinking of how devastated I would be if they were not to work.

So after such ordeal all I could do was wait. Until both of my PK's finally arrived by mail. That same day I took them up to my room, disassembled my case, and began to operate. So I'm done with my story, let's move on to the good stuff!!!

Installation:

What you see on the right is what came inside the two little

boxes from Japan. Doesn't seem too complicated, except that the manufacturer forgot to add in the Dip Settings sheet, so I had to hop on here and write them down before continuing.As I had anticipated the hardest part of this project is the removal and placement of the HSF. I opted for Taisol's with stock fans in this occasion since the thought of hearing two Delta fans running at full speed almost made my ears bleed, maybe in the future I'll opt for better performing HSF's but I'm happy for now. After sitting in front of my case holding a Swiss Army knife and a screwdriver probing at the socket tabs for 10 minutes I finally got the HSF off my first processor. The second one was even harder to remove, eventually it came out really fast. Too fast for me though, I got a cut on my thumb from the blade i was using to probe the HSF clip open.

 

 

 

With heatsinks back in place I plugged everything in and while crossing my fingers I pressed the power button. POST!!! Yes! I wasn't even expecting the thing to POST, but my excitement was soon drowned by a solid black screen as the BIOS passed the torch to Windows 2000.

Oh well, have to lower my multiplier. I pulled out the heatsinks again and changed my settings back to 11x, which with a little math gives us a 1466MHz processor speed. I put everything else back in place and laughed at the mess I made with the Silver stuff spread all over my processor. Again as I boot up I was greeted by a very pleasant POST, but then the system hung while loading Win2k. I tried booting it one more time and got an even less reassuring BSOD.

Back to step one, I took my heatsinks off again and lowered the multiplier to 10.4x, and again my math genius deduced the processor speed to be 1400 MHz. Not as high as you see in most uniprocessor systems but considering my HSF's aren't exactly the best performing out there and the added strain and heat an extra processor adds to a system it is still a pretty hefty number.

As with my previous tries I received the almighty POST. I tried not to keep my hopes up but it was hard to do when I saw the Win2k progress bar fill all the way. As the Windows login screen prompted for my password I felt a shiver through my body, I had done something no man had done before (at least none that reported it ;o) ), I had overclocked a dual Athlon MP system.

EDIT: I was informed that someone had in fact OC'ed a dual Pally rig but I don't know any other details except that they took a diamond tipped drill to their $220 chips... That takes some balls if you ask me ;o)

Everything worked fine except when I tried to write this review on my newly clocked system. An overclocking review is not a review unless it has some before and after shots of some benchies. So I went to run Sandra's CPU benchmark and Multimedia benchmark. The system crashed in both of them. The interesting thing is that I could play Max Payne for more than an hour but it crashed using Sandra. I tried 3DMark2001 just for good measure and half way through the Matrix-like testing scene it also crashed. I was faced with a difficult choice, should I keep my system as is since it never crashes unless running benchies? Or should I stay on the safe side and move my multiplier down so it'll run everything thrown at it? Well I figured a half stable system isn't stable at all and went back to the drawing board.

I set the multiplier back down to 10x. Booted up the system at 1333 MHz and Win2k loaded up as expected. Now for the stability trial I ran three loops of 3DMark2001 on it. I couldn't even sit and watch them because I'd be pretty pi$$ed if I had to take the multiplier down again, so I went downstairs for a bite to eat. When I came back I was greeted by a success screen. Finally!!! Now not only do I have an overclocked dual Athlon MP, but I also have a stable one at 1333 MHz.

Before and After shots:

Check out these comparison shots of the Pally's in action.

Conclusion:

Yes!!! Overclocking an Athlon MP with the PK's does work. It was miss-reported by Plycon and other websites that these beauties would be useless when paired up with a Palomino core, but luckily they were mistaken. This overclocking project may not have reached the highest speeds out there, but it was an entrepreneurial endeavor and a great challenge. Maybe in the future someone will pair these up with a watercooling system or even some LN2 and take them to the next level, but I have accomplished what I set out to do. I'm now running a stable 1333 MHz on both my processors (could do 1400 but it has it's costs) and enjoying the feeling it gives, even though the speed difference is not that noticeable... It also gives me a lot of bragging rights at LAN parties ;o)

 

BCH

 

Related Links:

AMD

Tyan

Crucial Memmory

 

Home