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

[mAJORD]

This little beast turned up the other day so I got to work with some OCing on AIR on this new DFI 590 M/2.

May I first just say, this Board is giving me headaches..

On both the 4000+ brisbane and 5200+ windsor it will not go an inch past HTT 328 on 1x HDT and DDR2-400, so even if by a strike of miracle brisbane has no coldbug their is no way possible to achive a 100% OC on this board. Infact the max I'll ever be able to screen is approx 3.7ghz, the fact it may very well not get there anyway means I'll still give it a shot, if it hit's it though I'll be board hunting agian I can assure you that..

Now with ram at 533(266) It won't budge past 278 on the 4000+, and 258(!) on the 5200+ - What a peice of crap.. I have run the 5200+ at up to 290 on the CROSSHAIR with 533/266 DDR and 3x HTT, letalone 1x.

Overall not impressed at all with this board for the $$. on top of the above it has a slightly flakey feeling to it IMO. I certainly am much happier with my Crosshair being in my main machine now.

Unless some light can be shed on this (maybe it's a ram issue, I just don't see how) then despite it's great power supply, I cannot recommend this board for serious OCing.. Essentially whilst this brisbane core will run at 3.1 happily, It won't do so with proper RAM and HDT clocks on this board as it just won't boot with the HTT required to do so. I will be throwing the cpu in my Crosshair when this machine is free'd up. possibly before I put it on DI. this week

Some notes on temperature readings:

As seen in the table, there are ZERO even semi accurate temp readings to go by. The mobo is just not reporting them correctly, It can be safely assumed the Speefan readings are about 10deg high, as a 10deg delta from ambient is about typical at idle. I'm merely using them as a reference. Something I meant to do was record the stock clock Temps. they should be similar to the first Big typhoon temps though.

So with the above in mind all testing has been for max CPU clock only, See results below.. For the record, room ambient temps were 28-30deg C. Another hot day in brissy!

Legend:

x Mins + = Orthos was still running after x time, test stopped due to finding stability limts only.

TT BT: Thermaltake Big typhoon VX

Coretemp readings Horribly inaccurate, estimated 20deg low - core 1/2
Speedfan readings Also inaccurate, estimated 10deg high Core 1 reading only





IN summary, 2.8ghz is semi achiveable on stock volts with stock HSF. Someone on another forum is running theirs as such 24/7 so it seems pretty typical. A small bumpin vcore would be in order for long term on this chip.

3.1 long term is almost in reach, but temps started to get out of control and it was turning it's nose up at Vcore icnrease as you can see.

Of special note, these things seem incredibly cool at high Vcore compared to windsor. I just threw the 5200+ under the Typhoon and at 3.1gig 1.50v , inaccurate core temps asside the heatsink itself is much much hotter to touch after only a couple of mins of load.. I'm 99% certain I had a good mount on the brisbane but in light of the differences I might try remounting it and testing again. esp with the delta of 20deg. IHS may be warped aswell.

Max screen:



Max Stable:


DI to come...

Cheers

[flippin_waffles]

wohoo! nice! AMD results!

[Kenetixx]

Looking good so far, mine is ok on the cheap Gigabyte board, was going to get the DFI but the more i read the more i might forget it and it looks lie the Abit or MSI diamond might be a better choice. i am pretty limited with my board i feel.

[flippin_waffles]

Nice results kenetixx. BTW, where did you get that gadget on your sidebar for the core activities?

[Lightman]

Great work mAJORD!!

You done huge work for us with this testing and table. BIG THANKS!
Of course we are waiting for DI !

[mAJORD]

Nice one Kinettix, May I ask if the Vcore was needed, or you've added it for stability assurance?

No Probs Lightman, Best to do it properly :P .. I'm anxious for DI myself.. Tuesday I'll give it a shot. Pot needs a tiny bit more work and tube attached, Got everything else ready to go!

Anyway

UPDATES:

I had a brainwave and installed ClockGen.. Seemed to solve the major HTT issues once in winblows.. allowing me to break the 278 HTT barrier (@DDR533) and go right up to 320 and beyond.. Unfortunatly the MAX HTT @ DDR400 was only able to be pushed to 340.

As a result though, A new max Screen just for fun, and 1M pi

So close to a 3.3 screen, but not without quite a bit more juice.


http://valid.x86-secret.com/show_oc?id=171889

super pi after getting the ram and hdt up to speed..



Under Volting:

min voltage @ stock clock: 1.15



OC @ 0.1v undervolted: 2690mhz

[Kenetixx]

Volts read high on this board, measured with DMM it shows 1.37 but under load drops to 1.3, i only have a crappy TT PSU atm , awaiting my good PSU back from RMA, when that happens i will get true measurements in vcore, i can boot into windows at 2950 mhz and be game stable but fails Othros in 15 minutes, this board struggles with any FSB over 270 really, im pretty sure its the board not the chip.

[Kenetixx]

flippin_waffles its on the WIndows Official Gadget site

[prostoman]

Wow, nice results mAJORD. Yeah, those mobos cant reach very high HTT, even with the help of ClockGen. My tops out at 334 or something. Still, very nice.

[Console]

Are these chips clower clock for clock vs a toledo ? because @ 3157mhz I would think you would be better than 29sec , I score 28.4 with my 939 system below 3ghz

[MAS]

Originally Posted by Console

because @ 3157mhz I would think you would be better than 29sec

s-pi depends upon memory clock as well
the less memclock and the higher timings - th e lower s-pi score

[Lightman]

Originally Posted by Console

Are these chips clower clock for clock vs a toledo ? because @ 3157mhz I would think you would be better than 29sec , I score 28.4 with my 939 system below 3ghz

1. Memory speed - important
2. Toledo 2x1MB cache L2 - Brisbane 2x512KB cache L2 - also important
3. L2 cache latency increased in Brisbane by 2 cycles compared to all older A64 - very important in SPi.


edit: changed 2ns to 2 cycles

[flips]

Is it possible for you guys to post the Stepping of your 4000+ Brisbane?!

I got the 0703WPAW with the 0400 Batch...just testing

[mAJORD]

Console , lightman. indeeed. Ram was underclocked slightly for the pi due to 533 divider, timing fairly loose iirc. Yes it is slower pi/clock than my F2/F3 1mb windsors.

FLIPS:



0703 SPCW hmm..

[flips]

Ok, yeah your seem to be clocking well, I'm at 2,8Ghz @ stock so far, but my Mainboard reads out the wrong default voltage so far...1,392...1,3 is default or not?

[m411b]

Nice clocks mAJORD! I wish my 4800 would do 3.2. It hits a wall at 3.1 even with tons of vcore dumped on it. Maybe its something in the BIOS, duno. This board really ain't thaat great though, you are right about that.
Keep the clocks comin man, very interested to see the outcome under DI.

[mAJORD]

Flips.. what are your temps? as per the table, It's safe to assume a truely stable 2.8 would be the limit at stock volts on this chip, but only if temps are kept under control..

Note 80mhz extra headroom the massive drop in load temps from Big typhoon (what a great cooler)

m411b, is 3.1 the absolute limit?, or just it's unstable?.

First DI run tommorow if all goes well.

[m411b]

Originally Posted by mAJORD

m411b, is 3.1 the absolute limit?, or just it's unstable?.

I think it could go further(much further) but can't get it stable no matter what I do.
Don't know if this helps anyone but here is my steppings:

[Aerou]

I thought it was 13 vs 20 cycles, so at 2100MHz the difference should be 3,3ns

[Lightman]

Originally Posted by Aerosupra

I thought it was 13 vs 20 cycles, so at 2100MHz the difference should be 3,3ns

AMD stated officially that it is 2 cycles increase....

I wrote by mistake ns in previous post .

[mAJORD]

OK Dry Ice update.. bit of a nightmare!

few teething problems. despite being pretty carefull ( I thought) Possibly found coldbug but can't be certain.

I managed to boot into windows at 3330 and 1.65v Temps just below 0 and falling. Everyhting was Ok, but became a bit unstable in windows while setitng up cpuz and other progs. It eventually froze.

when in bios it froze again and I couldn't get it to restart. The POST LEDs were also frozen on all digits, leading me to belive something other than the CPU was amiss (it normally displays POST Errors!)

At about this time I realised I'd missed a small section of insulation around the pot and condensation was starting to form everywhere.around a mass ice blob.

On top of this, I found out after disassembly somehow It seems my press join between al and copper expanded/contracted and leaked half the alcohol down around the cpu socket and got all under my insulation.. disolving the glue.
The board was soaked when I pulled it all off.

managed to get it to boot after a dry out thank god.

I'm going to fix up the pot problem and and coat a larger area in liquid electrical tape before applying foam over the top.

Next attempt: Friday

Sorry guys!

[m411b]

Hey atleast you tried! I'm subscribed, so keep us updated.

3.3ghz is awesome stuff. If the condensation doesn't penitrate agin, what do you think will be limit?

[K404]

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

[mAJORD]

M411. I have no idea, it all revolves around the coldbug / severity of, and is the main aim of the exersize.

K404: Yep, NB Voltage increase up to 1.3 from 1.2v IIRC. I'd want to cool it down somewhat aswell though.

It's so damn hot and humid over here atm too, It's made the condensation issue 10x worse. high 20s (c) even at night in the workshop dew pt is about 22deg C, humidity 73% when I was attempting it.

[K404]

I`m not sure if the HT voltage is the same as the NB voltage :S I`m months out of practice on AMD stuff, but I *think* The DFI NF4 boards had voltage options for both, implying they`re not the same?

You know your BIOS tho, so its your call