Update: P5B Vanilla (16Feb2007)

[ziddey]

This should be good news for anyone looking for a cheap(er) board that'll fair pretty well. In the past, as I understand it, p5b vanilla only supported vdimm up to 2.1v, limited vfsb, no vmch, etc.

However, I just got a new p5b vanilla from newegg. It allows vdimm up to 2.45v, vmch up to 1.77v, vfsb up to 1.45, has voltage adjustments for vcore southbridge (pcie/sata), vich. Great news! Means way less modding to have to do on the board.

Still, vdrop and vdroop need to be worked on with pencil. Setting 1.40v in bios results in 1.35v at idle and 1.30v at load. This seems to be the only issue. Also, it seems that leaving vmch at 1.25v results in occasional inability to recover from a cold boot (p5b does cold boot every time it seems). Also, pcie for g71gs was stuck at x1. I'm not sure, but after I made the switch up to 1.55v vmch, pcie went right up to x16.

This board is so much less of a headache to work with than the ds3 ever was. The only thing I'm a little bugged is that posting seems to take longer than the ds3, for the initial screen to come up.

Also, the board has support for ahci with just the ich8! However, I don't have a floppy drive and don't feel like slipstreaming drivers right now, so I haven't tested it.

I'm currently using kvr667 while I rma my d9gmh's, so I can't climb that high on fsb. I am currently at 8x400 with ram 1:1 5-5-5-15 @ 2.15v. Can't get cas4 at all with these. Hopefully, once I get better ram, I'll be able to do more testing and hopefully settle at either 8x450 or 7x500 (now that nbcc has been disproved, no?).

Good buy!

edit: Also, it should be noted that the cpu reading is way higher than what it actually is. Coretemp reports about 47'c under load for me, which is about what it's been with two other boards I've used. Meanwhile, asusprobe reports 62'c.

Also, anyone know where I could get a bracket with the lead for the serial port header on the motherboard? I would have rather had a serial port instead of esata.



Update: Looks like it was too good to be true. Vdimm set to anything above 2.1v doesn't seem to operate correctly. I'll be getting a new battery for my dmm soon so I can see exactly what's going on. Hopefully, I'll be able to at least verify that vmch and vfsb settings are true.

[ziddey]

Here are some bios shots:







First shot shows everything at max. Next ones show options.

vdimm was checked with dmm. 2.10v set results in 2.12v
2.15->2.12
2.25->2.37
2.30->2.37

Something is still fishy but it does show that higher vdimms are kind of available. I didn't test any other's yet since these powerchips don't respond well to voltage. It's just too bad there's nothing in between 2.12 and 2.37 since that's where I'd imagine most of us would like to end up being.

[Eldonko]

Whoah, must be a new revision or a new BIOS. Can u tell me what BIOS you are using and what rev the board is?

[ziddey]

I'll try to look up revision for you. But it came with whatever bios was second most recent. I updated to most recent and it didnt' change available settings

edit: where does it say revision? it only says pcb made in china

[Eldonko]

Should say just above the nearest ram slot to CPU afaik. I have a regular ol vanilla with the lower max voltages. Does yours have the ICH8 southbridge like the others or did they throw the deluxe ICH8R in it?

[ziddey]

It's still ich8 regular. Above ram by cpu it enumerates the dimm slots. Then it says ASUS P5B. Unfortunately nothing else. I'll take a picture in an hour or so

[Eldonko]

Ok cool, thanks for the info.

[Boogotop]

subscribed

[ziddey]

I updated the post so it's not that cumbersome.

[Okda]

subscribed

me too

[ziddey]

Update on the vdimm settings. I got a new battery finally.

2.00v: 2.08v

2.05v, 2.10v, 2.15v: 2.12v

2.25, 2.30, 2.35, 2.40: 2.37v

2.45v: 2.55v.

so really the only high voltages are 2.12v, 2.37v, and 2.55v. for me, that means that max is still just 2.12v, which would have been there had I gotten an older p5b vanilla.

I've got a theory to test on a ds3 board, and if it passes, I'll try it out on the p5b. However, it's currently in the case so I'm kind of reluctant to fiddle with it since it's such a pain in the ass.

Basically, we know that we need to manipulate the resistance between one of the legs of the voltage generator and ground to adjust voltage. So, I'm going to try to see if shading nearby components will result in such a change. If anyone else has a loose board to try this on, knock yourself out . I think I'm going to first pull out my old ds3 and see if I can test it out against the known good.

[Eldonko]

You should just do the vmods, I did em all to mine and its a great board (other than no clockgen avail).

[ziddey]

The hypothesis tested true. I shaded on my ds3 as per directions and then measured the voltage generator and sure enough, resistance fell.

I got a new psu today so I guess I'm going to get some downtime and completely take apart everything and rearrange my radiator and everything. Will post back in a few hours hopefully

[ziddey]

Update: vdimm pencil mod found for p5b vanilla
Looking at the board vertically, just under the dimm slots you'll see the voltage generator. Under that are 6 resistors aligned next to eachother vertically.

I've found that the last one (one on the right) when manipulated will change the resistance measured between pin6 and ground. I'm still working on it and won't be able to test results for a while longer.

But of my initial testing, pin6 to ground is 550 ohm stock for my board. Shading in the resistor as much as I can with a #2. I can get it down to 520. I'm not sure how the scaling works yet since I haven't tested results, but I did try connecting the entire thing with ink, and it netted me somewhere in the 100's.

[ziddey]

I'm pretty pissed off right now. I got in a new power supply and was taking the time dissembling everything in order to remove 2 of the fans from the radiator and turn two of the fans around. But then I misinterpretted somehow and put the two fans in the same way they were before. Now there's no way in hell I'm going to tear it down again. Oh well.

As for vdimm mod, looks like I need to do a little more research. I got the resistance between pin6 and ground to decrease. But when I tried to boot, it fed vdimm as 1.45v instead of going higher. I used a qtip and rubbing alcohol to remove it and now it's back at 1.8v stock. ?!

[ziddey]

Hmm, I decided to shade up the resistor again. I only shaded for around half a minute, then blew off all the graphite dust.

I didn't measure before and after resistances this time, but when I booted at a known 2.08v, I got 2.02v measured. Alright, a 0.06v drop. Don't really understand why it's working in the opposite direction, but whatever. I try out what was a known 2.12v and am now getting 2.06v. Great. Same 0.06v. Try for the known 2.37v. I'm getting 2.30v. Alright, now we have some potential again. I feel more comfortable at 2.30v than 2.37. Just hope that my d9's will be able to post at 1.74v (can't imagine why not if it's running via spd at 667mhz).

I did some testing again with my kvr667 sticks (powerchips). I've found that 2.3v will let me post at 4-4-4-10, but it won't boot windows (corrupted files). 2.3v won't let it even post at 900mhz, whereas 2.1v would, although it wasn't stable. I guess I've already established that these chips don't respond well to higher voltages, but it's still looking good for my d9's.

It's a shame this won't benefit anyone except for those with new revision p5b vanillas. Hopefully I'll be able to figure out a vdimm mod to actually raise it instead of lower. But I could really use some insight from someone more experienced though. Why would decreasing resistance between leg6 of vreg and ground result in decreased vdimm, where connecting leg6 to ground with VR will result in increased vdimm?


edit: I just did some more testing on my ds3 board that I'm not using right now. I found that the resistor (smd capacitor?) used in the pencil mod has one end that's connected to leg6 and the other to ground. The resistance across it is the thus the same as between leg6 and ground. I'll have to scope the p5b next time I have it out to see if I can verify anything on the board to be the same. The problem is that in the case, I can't really work on it at all. So I'm hoping someone with a p5b vanilla will see this and volunteer to probe around with a dmm and hopefully find something .

Feel free to pm me if you're interested. Thanks

[ziddey]

I apologize for no further updates yet. It's pretty hard to work on the motherboard with it in the case and it's too much of a pain to take apart and put back together. I've attempted to measure vmch but chickened out for fear of shorting terminals. I seem to lose stability around 460fsb. I haven't determined if vmch voltages are valid yet, but I did notice a burning smell after attempting to use 1.70v for a quick test. Heatsink heated up rapidly as well. However, no settings seem to increase fsb limit. Not 1.25, 1.40, 1.55, 1.70. I know the cpu is good for at least 500 (determined on a rev2 ds3).

[ziddey]

I've been looking into the bios for p5b vanilla with amibcp tonight and am going to sleep now but was wondering if anyone had a p5b vanilla they wouldn't mind if it got bricked. I'm going to see if I can figure out a way to enable the hidden voltage controls for you guys

edit: I've found that aisuite will crash on loadup with vdimm over 2.10/2.15 (both 2.12).

[ziddey]

Update. I've found vdimm pencil mod for vanilla board.

I may not have the same revision as you guys based on the pictures I've seen compared to what I've seen but...

Under my RT9214 IC, I have 6 smd resistors in a row. If you notice the 5th one (with board vertical and pci to left and floppy to right),

you'll see continuity between pin6 and one end. You'll see continuity between the other end and ground.

I've found that with #2 pencil shading, the best I could do was about 2v (slightly less). No doubt with a higher pencil, I'd imagine it'd be possible to get more.

But I made a hybrid of copper ink and pencil and managed to bring vdimm up to 2.18v (well 2.34v also, but that was too much for a
I would have rathered 2.10v as a baseline, but seeing how the p5b only has +0.3v as a max, and I want 2.5v as a max, I guess we'll have to compromise.

I've found initial resistance to be 568, with vdimm 1.8v.

With 508, I saw 1.96
420, 2.34
and 450, 2.18.

Hope this helps.

I had only cpu in for testing, and once vdimm of 2.18v was achieved, placed in my dimms to ensure of it. However, I won't be able to do testing since I don't have a video card to use at the moment. When I dissembled my watercooling to get the motherboard out of the case, I managed to wet the video card, so I'm waiting till tomorrow.

Update: Bios options for vmch are auto, 1.25, 1.40, 1.55, and 1.70.

Actual values were found for 1.25, 1.40, and 1.55 to be 1.30, 1.50, 1.60.

It's a shame though. I was kind of hoping vmch adjustment wouldn't do anything. Would have meant hope to get to 500fsb. Looks like not though, oh well.

I doubt that fsb is what's limiting me at this point

Finally found the revision. The silkscreen was barely visible that I missed it the first few times. It's 1.04G of vanilla.

[ziddey]

I've found something interesting.

cpuz and wcpuid both show just about 7x500=3500 at all times.

rmclock shows core clock just about at 3500 with throttle at about 3500 or bouncing lower than it.

however, with aisuite, I see that the cpu frequency is absolutely all over the place now that spread spectrum is turned on.

I've seen fsb dip as low as 480ish (~3360mhz) to as high as ~520 (~3640)

pretty violent swings. it's the only program i've seen so far that's reported such wild swings.

although it looks to be about 4% if not a little less in either direction

I just wonder what this means, since 1.45v isn't stable for 3.6ghz, but seems to be for 3.5ghz. Going to give it testing tonight and see if it survives

edit: Hmm, I got to thinking about the g71 voltmods and how people got better results sometimes after alcohol swabbing the entire pcb. Perhaps if I did the same to my board sometime, it might reduce emi and not necessitate the use of spread spectrum. ehh

[serialk11r]

Sorry to sorta hijack, and sorry about this super noobish question but:
What's a P5B Vanilla?????????? Is it the P5B thingy on newegg?

[ziddey]

yeah, vanilla means plain

[serialk11r]

Hmmm on newegg the P5B Deluxe open box is only 108 bucks???!!!!!!! Whoa I could make myself an I/O plate for it, and I don't even need those extra cables.
Sorry bout the hijacking.

[ziddey]

ack, I was about to mention that the refurb ones are reaaaaal cheap. If I didn't have one already, and dfi wasn't coming, I'd bite.

[dnottis]

I picked up the deluxe from Newegg, working great except for the 1x/16x PCI-e issue, but there is a workaround and that is working.