Tyan Thunder K7 System Review and OC'n

The Case/PSU - Cont.

The Front of the case is pretty uneventful, and it has a door. Most doors suck ass, as you cant get them out of the way unless they are closed.

This one is double hinged so it will swing flat back against the side of the case.  OMG its a door that someone REALLY engineered!! Its got some serious LED Deficiency though.. This machine could use about 10 LED's for all the goodies the mobo got going on ( LAN LED Headers on the mobo etc..).  Its got nice open front vents to let the 120mm fan behind it breath easy.

What's that I spy? None other then the long since believed dead - ProCooling case badge!  The company FINALLY got them done and to Case etc. So if you want go and ProCool your rig.

Yes the peripherals are not top end.   That is a 6xDVD/32x CDROM - SCSI2, and a Yamaha 6/4/16 writer.  They get the job done so I don't complain and they work.  Which is more then I can say for higher end IDE devices I have tried... but that's just me.

Not much else going on here.... :)

The back you can see the caged in 120mm Fan. I will be removing the stock grates with a dremel in the coming weeks to replace them with the nice chromed wire ones.

You can see how the IO panel and everything fits.  Not much to say about this either ;)

That's the inside. Not nearly as cluttered as it could be.  besides for the misc. fan header wires, and power runs its pretty clean.

The Blue rounded wires were bought from KD Computers, I have them for the U160SCSI and the Floppy, I am waiting on a 50Pin SCSI blue rounded one to go with it.  The SCSI ones are LVD cert but do not claim U160 Cert.  I have used this IBM Ultrastar 73LZX 36.4GB HD on it for over 3 weeks with out any issues, in fact I used this wire to overcome some sync issues I had with the cable offered in the box!

You will also notice the second ATX style header is not running down with the rest of the Mobo power, I am running it behind the PSU and then down an area between the PSU and side of the case.  so its not in the air flow of the CPU's at all.  The Big ATX style connection ( 24 pin) is run around the front of the motherboard in front of the RAM sockets.   This also keeps it out of the air paths.  You can see the 2 Low Power 3 pin 120mm Sunon BB fans.   I wanted to use LV fans due to the noise. There is also a 3 pin 80mm fan infront of the HD blowing in. I removed the stock box that goes around the front 120mm fan to let it breathe easier to the inside of the case.   With 2 100watt CPU's you need all the air you can to keep it cool.

I tucked the extra data cables below the HD to keep the inner area of the machine clean, and stuffed the miles of extra PSU wire ( makes Enermax power cables seem short), bundled under the DVD drive and above the writer. 

The HSFs -

I am generally against HSF cooling in favor of water cooling, but for my main machine that runs 24/7 and contains a lot of data I use daily, I couldn't risk it.  So I went with the next best thing - the best HSF's you can find currently.

The fans on top are the high speed Sunon dual bb 4500RPM fans. They move 25CFM of air and are very quiet.This machine is in my living room so its got to be quiet, or reasonably quiet. If I were to run dual Delta 7200 rpm fans, my neighbors would be pissed.   I also would prolly go insane and start kicking the crap out of my machine in the middle of the night when I cant fall asleep cause it sounds like a Rolls Royce Jet engine is at 100% throttle in my PC. ( hehe gives you an idea of how much I hate those POS delta fans...) So instead of relying on high CFM on an inefficient and lacking design ( thermoengine), I chose to use a lower CFM tolerable fan, with a very efficient and functional design.  Unlike the CAK or other Cu fans the fin structure is very light on this HSF allowing 2 to be used on a mother board without the worry of busting the board in half :)

How cool did the HSF's keep the CPU's? Impossible to tell.   The BIOS reads 80Deg C for them.  Which is SO wrong its not even funny. The reason its wrong is that the Tbird CPUs do not have thermal Diodes like the MP's do. So the board is reading something that's no where remotely associated with temp ;)  As far as I can tell there is no CPU temp reading on the motherboard if the CPU doesn't have a thermal diode.

The CPU and OC'n -

The CPU's is an area that I am not really happy about at all.  We have a board meant to run 2 Athlon MP's but are running 2 thunderbird 1.2Ghz AXIA CPU's.   I was planning on running some Durons in here to test with, but mysteriously I had a Duron stop posting on a test rig I am running,  ( dead chip.. cant RMA as I was testing etching the bridges off on it :) So the review is being done only with Thunderbird cores.   This is not what I was planning, but eh, life goes on.. or so the song says :)

First thing you need to do is put the CPU in a board with a heat sink a lil worse then the one you will be using in your Thunder K7 ( so the core runs hotter, this will compensate for the extra heat inside the case having 2 Tbirds cooking next to each other). The board you put it in should be able to change VCore and Multiplier.  Now with a stock 266 FSB change the multiplier to as high as you can and still keep the machine stable ( run Prime 95 or something for a few hours). This will tell you the max you can clock the chip stable on the Thunder.  Do this with both chips and and figure the highest common speed.

Pics below are a Duron I used to test all this before I went and touched the Tbirds

This was a pic after the fact, only one I had to bear with me :)

Next look VERY closely at the bridges for L3, L4, L6 and if you are making VCore changes - L7.  L1 is useless on the Thunder K7 so I just left them unlocked.

Find the bridges that are CLOSED and see which ones match the sheet of paper you have just printed out.   Cross off all the ones on the piece of paper that are already connected on your CPU. ( use a mag. glass if needed to see if its connected or not.)  The ones that are closed on the CPU and open on the paper are what we are concerned with.  Mark those on the paper separately.

Now, we get into surgery.

Now its time to bust you the Diamond tipped bit. The reason we need diamond is because the ceramic is harder them steel, aluminum, glass, and Tungsten.  Diamond is harder though.  The bit needs to have a very fine point to it.

Now, have a beer (if your old enough of course). This will hopefully help steady your hand, and calm you nerves :). ( how many How 2's have you drink a beer?... man if only)

Now with your newly calmed nerves, line up the bit a little above the bridge you want to break, EXACTLY IN THE CENTER of the bridge.and with a

good deal of pressure ( not enough to break the CPU), drag the diamond tip across the trace. go back and forth JUST UNTIL YOU BREAK THE BRIDGE, if you keep going back and forth you will dig into the ceramic layers and cut the traces that go to the pins = dead CPU.

with a flash light and a mag. glass check to make sure you have broke the connection.  Do that for each that NEED to be broke, but do not do it to the ones you need connected later.  Also, the reason for doing the break in the middle of the bridge is so that you have the pads left to reconnect at a later time if needed.

Now this will perm void your warranty, not that many people ever are concerned with that, but you do need to know, if you do this, you are going to possibly kill your core, void your warranty.

We will use the pic above for the next part also.

Take your .5mm HB lead pencil and now ... on your still calmed nerves, draw a line over the disconnected bridges that your piece of paper says that you need to connect.  go back and forth to make sure you get graphite to fill the small valley created between the bridges by the laser cut.

DO NOT connect 2 rows of bridges together... that's a no no... so just do like in the picture, each bridge connects to itself only, NEVER to another bridge.  So please, color in the lines :)

If all goes as planned you do that to both CPU's.  Put them in the board you tested the max speed on to see if they are detected at the proper VCore and multiplier. If they do, strap them into the Thunder, and power it up and see what it reads.

2 x 1.4Ghz CPU's from 2x 1.2Ghz CPU's at 1.75VCore.

Now lets look at how Windows 2000 likes this piece of hardware