Brisbane X2 4000+ Testing (air, DI to come!) + DFI 590R/G issues

[mAJORD]

ahh got ya, I belive it may be called CPU-C51 voltage but can't turn it on atm to check. This is probably the one I'm thinking of.

Perhaps it's my asus board that refers to it as NB.

-edit nm , I've gotten off the drugs (the ones in my head) and realise what you're saying. HTT has nothing to do with the C51, my bad. I'm thinking LDT.

back to what I was going to say, perhaps it was the drive strength settings then? They're the only thing that spring to mind on this board. Can't find manual either so I'll just have to have a fish around. Thanks

[K404]

I went googling coz I cant explain things worth-a **** an I came up with this from the DFI boards..



It was LDT voltage I was thinking of, but hey its the same as HyperTransport (ahem....more or less) so its all good

[m411b]

On this BIOS, there is a setting for NB core voltage (C51 NB Core Voltage) and HT voltage (CPU/C51 HT Voltage)..
Never messed with either one of them, but I think I will try it as well.
Thanx, K404

[Kenetixx]

SO whats the lowdown on this board?
Worth getting or go for something else?

[m411b]

Here are all the voltage settings for this board:


And as usual the owners manual says nothing and helps absolutely none at all.

[m411b]

SO whats the lowdown on this board?
Worth getting or go for something else?

I like the board personally, yeah it has some flaws, but that was to be expected with AM2. Things will get better.

[ozzimark]

Have you any HTT voltage options? Back in the NF4 days, bumping that up a bit helped me no end

yeah, there's actually a bunch of options for the ht link on the dfi




tweak the board right, and it CAN do high htt. it's just finicky as hell. i used to have a wall around 340mhz or so... then figured out how to get it up to 360mhz, but that was the limit of my cpu at 9x.. so then i figured out how to boot at high htt and a multi other than stock (yeah, boot at 300mhz with 5x multi and whatch what happens )

woulda gone higher, but my ram wasn't set up to go much higher, and i wasn't really interested in trying to break 400mhz, since i have a 9x multi to work with

[Lightman]

Nice read ozzimark
Good to know this kind of stuff ...


mAJORD I'm waiting patiently for proper DI action

Cheers guys!

[mAJORD]

hehe, that's a crazy sempy. Very nice.

I'd love to know how you got past the 340 wall. mine appears to be the same on all CPUs.

I'll tinker with that HT voltage then if I run into probs

Cheers.

[prostoman]

Yeah, I would like to know how you got past that wall as well, and I also like to know why the mem voltage is set to 2,81V on the screenshot?

[mAJORD]

HELPPPPP

Take 2 is underway, but the problem has come back, and It is rather odd.

I am booting with the board at above 0 temps and just sitting in the bios slowly bringing the temps down with small amounts of DI

now the MB core 1 temp reporting is about 10deg high

core 2 temp reading is BUGGED , and dances between negatives and +200degC all the time.

Problem is the moment temps go down below I think -7 on the bios readout, the thing instantly shuts down, and refuses to post at all.

I can't get it to restart again until temps are much higher.

Has anyone had coldbug that acts like this? i.e a shut down, not just instability or even a reset, but a hard shutdown.

About ot throw an F2 windsor in which I guess will be allmost definatly coldbugged and see if the behaviour is identical, or at adifferent temps.


any suggestions , pls tell.



ED: it's done it twice tonight, first time was in WIndows, working perfectly on second. shut down the next.


everything is fine as it gets down to -5 on the

[mAJORD]

OK

I'm about ready to declare brisbane severely cold bugged


the F2 did not exhibit the rather severe (never heard of coldbug causing a total shutdown like this tbh, but whatever) problem. It would lockup in windows below about -15 on speedfan, -20 on my probe at bottom of pot (sounds about right) It shut down once when OCd but not at sub zero temps, so I'm still a bit puzzed about that one. (0 deg under load)

Still after feedback on the failure mode though...


There is only one more thing left.. the F3 5600+ I'm typing on now. Has anyone actually confirmed their F3 to be coldbugged?

Cheers

[BadNizze]

First of I realy apreciate (spelling) your work testing all of this new hardware! Big to you! Second I don´t think anyone has tried Rev F3 subzero yet! Go for it! I think it´s time for som 4Ghz AMD Cpu´z Have a good benching weekend to

[mAJORD]

Thanks Baz. I don't feel too disapointed as at least it's done now so we know where CB is on them (well this one anyway)

Here's hoping 65nm barcelona isn't as bad..

I'll give this F3 a try in the morning.

[mAJORD]

OK guys

X2 5600+ F3

defiantly less severe than the other 2, but unfortunatly htt limited in the cold

Ran fine down to -50 and beyond I assume (didn't go lower) HTT maxes out around 240 though at this speed, regardless of voltages. (tried the HTT tips)

I let temps ramp up and at around -20 it starts to get better, by 0 330 is possible, and actually gets a tiny bit better into the positive's

Anyway thanks to everyone for the tips to work around the HTT prob on this board, I will tinker with it at higher temps at somes stage. Will try to push the max of the F3 on air. This CPU may be a candidate for some TEC action I think. thanks to its high multi, The point where the HTT instabilty becomes a problem VS the extra VCore afforded by low temps may add up to some good results at just below 0. Seriously considering giving it a go.

I'd try the same with small amounts of DI but im sure anyone who's used it would agree it's not fun to regulated the temps


I'm currently typing on the Brisbane in the Crosshair for the first time. Going to see if it's HTT limit differs much from the DFI board, to make a conclusion on whether the CPU or the board is the prob.


Dry Ice testing Summary:

X2 4000+ G1 Brisbane:

Severe Coldbug @ approx -20 -10 actual
Characteristics: Total shutoff below these temperatures.
No boot below these temperatures.

X2 5200+ F2 stepping:

Coldbug @ approx -20
Symptoms: unstable at even defualt HTT approaching -20.
No boot at default below -20.


X2 5600 F3 Stepping:

Characteristics:

Less severe Coldbug beginning @ approx -20
Unstable @ high HTT progressivly below -10.
Will boot at default down to -50 and beyond (undetermined)
HTT instability:

Vcore 1.5-1.7 , NB and HTT voltages raised varying amounts.

@ -30c: will not boot above 240HTT
@-10c: will not boot above 260-70HTT
@0 + will boot up to 340HTT (board limited)


Anyone with access to sub zeroc ooling, pls do give it a go if you have the chance, I'd especially like to know the results of the brisbane and if they are as severe as what I found.

Cheers

[BadNizze]

Good finds! Now what is the highest SS with the F3 Cpu, SuperPi 1Mb? I´d love to see some 3600mhz action!

[Lightman]

Cool guide mAJORD!

Are you going to try max clock for F3??

[mAJORD]

Cool guide mAJORD!

Are you going to try max clock for F3??

On AIR.. Yes

-edit . Removed windsor SS to avoid confusion

I've posted the rest of the AIR testing in the following thread as it's not relevent to here really.

What I will say though is how much COOLER the brisbane core is at high clocks.. As you'll see if you check the OC vs Vcore results of the F3 in that thread, the max Vcore I could run for any length of time on the Typhoon was 1.45 or so V. resulting in a tad under 3.2GHz max clock

compare this with the brishane which I was able to push to over 1.5v on air and still keep temps in order..

these observations are indepenent of temp reporting as It's physically obvious from touching the sink just how much cooler they run at high vcore.

I am personally looking fwd to the next Stepping for Brisbane (if there is one) Assuming it brings improvements in their process, as the only thing the F3 has going for it is higher clocks on GOOD COOLING, and the fact no brisbane has 1mb x 2 cache

Rest of 5600+ f3 Results here:
http://www.xtremesystems.org/forums/...=134035&page=2

[mAJORD]

And one last thing for the 4000+

A 3dmark06 run with OC'd 8800GTX.. X2 @ 3.05ghz

not bad CPU score




http://service.futuremark.com/compare?3dm06=1366828

For ther record. 5200+ @ 3.18 and same Card yeilded 11090, so surprisingly close really.

[K404]

Do you have access to a mild chiller? Could turn things right around for ya.

If you really feel like keeping at it, there have to be some steppings out there that perform a lot better when cold. Would be good to see a list of steppings which bug

K

[m411b]

Excellent results mAJORD! I hope you feel like keeping on this, as I do not have the resources you do to push any further. I think the brisbane will become better and better in the future.

[ozzimark]

@ -30c: will not boot above 240HTT
@-10c: will not boot above 260-70HTT
@0 + will boot up to 340HTT (board limited)

sounds somewhat similar to my sempron. btw, i got that 383mhz shot by keeping the core temp at idle in the mid 30's..

[informal]

Amazing results mAJORD.
Looking forward what would G2 Brisbanes bring .They should be out from the factory around now according to the info AMD supplied in its June Analyst Meeting(new improved transistor in 2 months from Dec the 5.)

[Kenetixx]

nice work mAJORD, i need a new board this giga board is limited to 270 FSB POS it is . With a decent board mine would hit 3.2 i reckon

[TEDY]

damn you usa bastards

Where're you getting these 4000+ brisbane babies?