Questions about VRM

[Halk]

I'm in the process of upgrading my PC. A new radiator is going in, and new blocks for a new motherboard. Memory and CPU stay the same.

I have an Opteron 185, and it's watercooled. As far as processors go this will draw the most current from the motherboard and put the most stress on the VRMs of any socket 939 CPU.

Currently I'm running an Abit AN8 Fatal1ty. I have no idea if it has bad caps or not, it doesn't matter since it's going to be replaced. I can't really tell what phase the board is. At a guess I'd say 4, since there's 4 circular components with copper wire around them.

The motherboard I'm replacing it with is an Asus A8N32. This board has 8 phase VRM.

The capacitors on the board are Nichicon ones, so that's good news.

However the 8 capacitors next to the VRM circuitry I cannot identify. They are smaller, do not have the head of them cut, and are completely enclosed. I believe these are polymer capacitors. They have red markings on them in the same place as the "Sanyo OS-CON" from the good capacitors sticky. However the writing on them is not arranged in the same manner.

I've taken a picture, unfortunately it's burred - not particularly good with the camera I'm afraid.



The VRM is a very important part of the board for me. The power mosfets will all be watercooled. The other components I was not intending to do anything with - will they benefit from heatsinks?

And the most important question, are all of these components high quality components, or are any going to cause problems?

The stated voltage for the processor is 1.35V, however it'll be run at 1.58V or there abouts. It'll also be heavily overclocked, so it will draw a large amount of power.

Any information is appreciated.

[Halk]

Re: Questions about VRM

I've managed to get a much better picture.

[gg1978]

Re: Questions about VRM

Ya sure you're not overdoing it on Vcore?? IME, getting it too high tends to reduce overclock stability.


Then again i just do aircooling though.. I've got an Opteron 165 at 2.7Ghz (300x9), with only 1.45V needed to hold it perfectly stable.. As in 72 hours plus of Prime95, one instance on each CPU core..


Anyway, IME, the Asus board didn't work as well for overclocking as my older Epox board.. My Asus was an A8R32-MVP, while the epox is a 9NPA+ Ultra.. It seemed the A8R32-MVP had trouble keeping Vcore stable compared to the Epox.. Both have 3 phase VRM's..

[Halk]

Re: Questions about VRM

Well, increasing vcore increases the amount of heat the cores will generate.

Basically you start at stock, and find out how high you can get the CPU to go, and then increment the vcore and try again. Eventually you'll hit a wall where additional vcore either doesn't give a great deal of improvement, or it generates too much heat for your cooling to deal with.

I expect the cpu to continue going faster with more vcore up to the limit of the board, which is 1.58V I've been told. I wouldn't want to go above 1.6V anyway.

If it turns out that I don't get much of an improvement past 1.45 then I won't go higher than that.

The board was bought specifically because it has 8 phase VRM. The "true" SLi 16x thing is of no interest to me at all.

Before I start plumbing stuff in though, if the capacitors, or other components are not high quality, then they will get replaced.

So I need to know if all the VRM components are up to scratch

[Halk]

Re: Questions about VRM

Any opinions on those red caps?

[gg1978]

Re: Questions about VRM

The little red one's are Fujitsu polymer caps.. I know some of their older yellow ones had issues. I've got some on the A8R32MVP, though i don't use that board, as my Epox overclocks better.