mattlock

Early results are looking fantastic.

My Rig: E8400 @ 4ghz, 2gb XMS2 DDR2 800, Zalman 9500.

F@H setup: VMware/Ubuntu on Windows XP

Results: 6 hrs into my first WU 40.4% complete 115 points per hr and climbing. At this point the WU will be done in 15hrs.

CPU temp: 48-50° (should get better once I replace the broken fan on my 9500)

With these early results I'm expecting the WU to be complete in a total of 14hrs or so.

For comparison sake my old FX60 did about 46 points per hr and completed the same 1760pt WU in around 38hrs, and my 5600+ @ 3.0 does it in around 28hrs on a native install of Ubuntu 7.10 64bit.

I'll post the final results in the morning.

Glaswegian

Nice cpu Matt! Some quick reading suggests it outperforms the quad cores.

mattlock

Thanks Glas. I'm really impressed with this thing. At some point I'm going to remove the OC and see what kind of folding results it achieves at the stock speed of 3.0ghz.

What's more impressive is the load temps while OCed to 4ghz. It doesn't break 50°c while folding. I'm about to replace the CPU fan now so that should get better.

Final result.....just over 14 1/2 hrs, or more than 2 1/5 times faster per WU than the FX60.

Glaswegian

So, just under 15 hours per WU? Wow!! I'm now saving my Xmas money - I think I'd like one of those!

I'm now at 25 minutes for 1%, even when running various other programmes at the same time. I'm still adjusting the cpu usage. cpu temps have not broken 50C yet - nice and steady between 47 and 49, although my big Zalman fan probably helps.

JohnthePilot

Has anyone tried this yet. It looks as though it could make a big difference when pushing the CPU to its limits, particlularly if OCing.

DonaldG

I too am using a Zalman fan - I have noticed that the vains collect the dust quite readily, especially in towards the centre and need cleaning out regularly to maintain good air flow... (I use a tin of compressed air to blast it away. )

Interesting paper on cpu cooling:

http://news.softpedia.com/news/The-coolest-PCs-around-43379.shtml

mattlock

The Zalman 9500 that Donald and myself are using already has a mirror finish, so I'm not sure lapping would do much. I could see it being useful on heatsinks that are have a rougher finish. I just bought a second one so I may lap the one with the broken fan a see if it makes a difference. (As soon as my torn rotator cuff allows )

JohnthePilot

Sorry to hear about that.

DonaldG

That is an interesting paper John. The only thing that irks me about it is that they refer to 'sand paper' when they should refer to 'wet & dry' carborundum paper.

Until I read that paper, I had presumed that a good quality heat sink compound would have the 'gap filling' ability to give near perfect heat transfer. Especially if you consider the 'voids' to be filled are very very small. That paper has made me rethink all my training dealing with RF power transistors (as tech director of electronics repair shop for 10 years! )

My QX6800 is still under guarentee & @ £650 a pop, I am reluctant to do that. until it is expired.

However, I have a little experience in lapping and would add the following comments, not as a criticism but as improving the speed of getting a perfectly flat surface.

First suggestion is to wipe the whole of both surfaces to be lapped with the felt tip pen & not just a big 'X'. That will reveal any high spots that need removing.

I suggest that the ink from the felt tip pen is replaced every so often to see the progress of flattening. Once the surface is flat, it would not be needed as the rest of the lapping process is basically a process progressively removing the depth of the scratches until a polished surface is achieved.

From time to time, use a vacuum cleaner extension with the brush attachment to remove the dust made up from an amalgam of metal and carborundum powder.

Start the grinding at the top of the sheet of paper and as it wears down, progress down.

He says to rotate the CPU 90 degrees. I would suggest that this is wrong - it should be rotate a few degrees, either to CW or CCW (but always the same way) this will help to reduce the grinding at an angle. (In lens grinding & polishing, the irregular pattern of rotation creates positive and negative errors which cancel each other out and helps create a perfect surface.

When changing to the next, finer grade of paper, vacuum every thing down before putting the finer sheet down That will ensure that there is not contamination of particle of grit that are larger that that to be used - the idea is to progressively 'polish' the surfaces and a larger piece of grit from an earlier grinding can and will ruin the surface. Likewise do not keep the different grades all together- keep them in separate envelopes to stop contamination of the finer grades with grains from the courser stuff.

Roll up your sleeves. That way, clothing cannot harbour contaminant grains.

Once you are working on the heat sink, I suggest a bucket of water with liquid detergent is nearby so that the current sheet of paper can be dipped in an the amalgam rinsed out (clogged paper cuts slower). Change the water when you change to a finer grade.

In the UK, most care repair shops will stock 'wet & dry' paper as do Halfords etc - I purchased my small tube of ice arctic silver from (I think) Amazon.

... just a few thoughts gained from studying optical lens grinding & polishing.

DonaldG

I did not notice the 'finish' on mine but a mirror finish does not mean a perfectly flat surface - it could be slightly concave or convex to the point that it could create a void. I dont know. It is perfectly possible that it has been ground to be 'optically flat'

It is the flatness that is the critical detail.

JohnthePilot

Many thanks for that Donald. If you fancy making some changes I'll put it back in new articles and you can add your comments. Use the quote button and make the changes you think necessary.

DonaldG

Thanks for the suggestion. Quite honoured. I would be delighted to do so.

However, I feel that it may be seen as a criticism of the original work and detract from an excellent piece of work - take the gilt off his work, if you know what I mean...

My experience in lapping & polishing is in polishing soft surface materials like perspex, hard surface materials like glass. My objectives have been to create an optically polished surface & in retrospect, I am wondering if an optical finish is overdoing it - I just don't know.

DonaldG

Sorry Matt, I didn't intend to userp your thread

mattlock

No harm no foul Donald. More of a FYI thread than anything, so I don't mind.

DonaldG

very noble, young man

kodi

I'm getting very jealous, I still haven't got my QX6850 online yet.
Had to RMA my new GigaSHYTE GA-P35C-DS3R motherboard which took 8 weeks to be returned only to find that they did not fix it (confirmed by the techs at the computer shop where I bought it) so it has been returned again.
It had better not take 8 weeks this time

DonaldG

I would DEMAND a full refund and buy another one elsewhere.

8 weeks...geesh...
I guess OZ law is similar to UK... You will have the right to a refund, not a replacement.

Mind you, you can always spend the time waiting by grinding the living daylights out of that CPU case

Glaswegian

Interesting stuff...

I'm now down to a steady 22 minutes per 1% - temps are steady at 51C. All seems OK- so far...

JohnthePilot

Hijacked with Matt's permission.

Glaswegian

I know - but I've been wanting to use that smiley for ages...