Phenom 9500 w/ MSI K9A2 Platinum

[KTE]

Well well... I switched system off, cleared CMOS before wanting to flash a new BIOS around 40 hours back and the system, like the previous ones - did not boot for 20 hours in a row
It booted ~6am this morning randomly. I asked for further info today, and news has it, that this is a BIOS problem but not just on this board, on a few others too. It obviously is a problem, since that's 4 new Phenoms now, 3 boards, 3 PSUs, 2 different RAM kits and 2 different GPUs all of which work without errors in other systems.

So, first note:-
Out of all the 9850BE's I've seen so far, mine is the weakest and poorest one.

Second note:-
I still have those dreaded idling freezing issues and I'll explain how they occur more. I tested 2860CPU 2220HT 2220NB today, stock 1.296v ID / 1.280v LD volts. Don't recall idle temps, but load temps were 36-38C after 30minutes EVEREST stability test loads. I don't remember which version it was, but one of the last 4.20 versions of EVEREST fixed many issues and now, on Phenoms, the core load is very high and susceptibility to catch errors very good.
Anyway, after I switched it off, ran PassMark but still had the PrntScrn screenshot in Clipboard not saved. Just trying to open the menu 70 seconds after those benchmarks and it froze.

Dropped down one at a time to 2800. Still froze.
Then started increasing voltage and kept rebooting every time it froze.
I am now at 1.36v ID and 1.328v load, at 2806CPU. This is now full load and idling stable for 1 minute short of 5 hours, more than others. It seems I will need much higher voltage to get near 2900CPU stable. From a quick try, I don't think it's possible without 1.45v ID and 1.42v load minimum.

Load stable is a totally different story. Can do that at 2950CPU 1.4v without problems.
Idle stability, major problems unless you increase volts pretty high.

Third note:-
3000CPU is now no way stable, meaning even 1.55v load will not get it load/bench stable like before.

Fourth note:-
Barcelona, Phenom and Toliman are now being offered by Gateway, Acer, HP, Sun, and Dell. Only remember the Gateway E-9232T, E-9722R, E-9522R, E-9422R and GT series, Dell OptiPlex 740, Acer Aspire M5100, HP Pavillion, HP Compaq, Sun blade/cluster/server lines, HP ProLiant DL585 G5, Colfax servers and so on. IBM is going to offer the B3 too. 2.3GHz is max I've seen the server line offering.

Fifth note:-
Intels 45nm CPUs are finally becoming available here, although supply is very short or back-ordered for most models, esp. Yorkfields. But the price hyped since October compared to retail price is very different, the prices are not competing with Phenom at all, Intel 45nm and 65nm prices are too high on the majority. Q6600 G0 is the outlier, and it competes with the 9850 BE, being $20 higher in price here. Since HP and Gateway are stocking more AMD based desktop solutions at major retailers such as Best Buy and WalMart than Intel based ATM, competition is sparking up. Especially since for many months, AMD offerings were only low-price, low-profit segment orientated but now you have mid to high-end AMD systems being listed by OEM and ODM's squaring against Intel ones. I hope Intel drops prices, they're too high here

Here's PassMark 6.1 CPU and Memory benches:


Very easy CPU setting EV. MEM/Cache:

You could have also used the 3.6GHz result for the qx. Will try your sandra results file here today, i laos prepared a few other sandra benches on the weekend.

Oh I didn't have that imported, only imported 5 results. Just imported more and here's the 1066 4-4-4-12 result at 3-3.6G Kentsfield =>

Thx, seems i had too low chipset voltages in the way here. Dropped the 2.8/2.2 stability test in favour of 2.9/2.2 with 1.35V for the CPU.

What Vcore do you need to get 2.9G stable Achim?

Hmm tried to install ubuntu on the GBT780G with an 9600BE but i cant get the installer working even in save mode, gdm restarts every few seconds.

Never had issues with the HD's under linux. Always builded packages for the os im on with the installer from AMD, will try it on the other setup later.

That's what I tried, but I actually tried everything there is out there. Main limitation is it needs CD I don't have.

Still have no clear picture about memory voltages here, that will require more testing here. seems higher NB speeds lead to higher voltage requirements for the ram too atm.

I won't be able to test MEM much until I can get on 113 BIOS or better. But 113 although is best perf/oc BIOS, it does not support 1066 mode and has many stability problems. I'll try moving to it and post a EVE/Sandra b/w comparison with high and low NB using 800 mode. Might try WinRAR and PhotoWorxx too, they're high MEM dependent benches.

I probably didn't need all 1.5v, but the board was running it at 1.375 and it wasn't enough. The other thing is that since its only a 590 board I don't know what the nb or the l3 are running at. I'll put it back under the vapo later and see if I can do anything different. At least that board doesn't have vdroop from 1.33 down to 1.26 with this chip.

Edit: Put the chip back under the vapo. I decided to try the 7x ht link to see if it would help, and it got me to boot up at 262 htt. Problem is that I can't get windows to start at that speed so it doesn't really mater. It always freezes at the windows logo before the logon screen.

What if you increase voltage to 1.6v?
Usually either voltage is all you need for what you're seeing or its a core limit. But if its a core limit, usually you can never POST with it.

I'm really starting to think that its the memory controller on these B2's that hate the cold rather than the entire chip. I just booted into windows with 260x11.5, I have the memory set to ddr2-400 setting, nb and ht link at 9x. If I set it to 533 I'm willing to bet that windows won't boot because the imc can't handle the cold. I'll do some more testing later, need to get to school soon.

Hmm.. thanks for the tests. I am willing to try something similar soon too, just need to figure this damn one out first!

Could someone with the MSI plat please try beta 1.13 with the 9850 and post there findings as far as muilt's go? If they work or if they don't.

If I clear CMOS, I'll be stuck for 1-5 days without boot. So....

I'll give it a go in a while. Multis should work fine, worked fine on 9600BE.

not sure if anyone has notice but, there are 3 new drivers at MSI's website. Ethernet, Audio, and Chipset.

For version 1.0 of the board? Chipset are just the ones released in ATi Catalyst 8.3, Audio/Ethernet I did see and try, had no issues or difference to earlier ones for me

KTE: Which bios are you getting 1066 unganged?

P0J

I have a question as i'm trying to decide which board to buy. Can the PCI-E x16 slots be used for anything other than GFX cards? I want to pick up a higher end board only because of the better pwm's i dont need crossfire and i'd like to have some expansion options. I know that according to the specs you should be able to, i'm just wondering if it actually works.

What do you have in mind? I can try and test if you let me know, they should work.

Aight, I'm done tinkering with my 9600. I don't want to or plan to buy aftermarket cooling. I figured I'd try and maximize my setup. No increasing ANY voltage to keep temps as close to stock as possible. I think I've done that. Video card is at 830MHz core / 1205MHz memory. Here are my results. 100% stable wooooooot, 100MHz OC!

3DMark06
http://service.futuremark.com/compare?3dm06=6174397
Crysis 1.2 | Medium Quality | 1680x1050 | DX9 | 32Bit
48FPS

Everest - Photoworkx
http://www.jmbat.com/media/phenomoc/...t4_everest.jpg

Screenshot
http://jmbat.com/media/phenomoc/phen...int4_11004.jpg

Great, but I am sure you can get more, you have one the lowest bin models I've come across - 1.2VID/1.2v 9600BE. It should be a better clocker than most others around, try below 2600 stock volts. Can it not go any further with unlocked multis?
Which BIOS are you on? What's your MEM rated for?
Try 212HTx11.5, don't change any voltages and check - it's a much better performance setting than what you're on and it looks like it should be stable at stock voltage. You are fine as long as you have sub 70C load temps in those utilities. That's well within rated spec, on the stock cooler, with no fan, I was reaching 55C load at just stock.

What version of everest are you running? EDIT: Oooops, it says in the screen shot... sorry.

Jack, has yours arrived yet? I noticed you had the Plat. on there, which version is it? IDT v1.0 [ones we have] retail in US anymore and the v2.0 sucks with poor support, be warned.

Thanks, for the file, no logs here before.

I went to CPU DID,HTT,HT and NB default speeds, left VID's at my last configuration (CPU=1.468;CPU-NB=1.331;DRAM=2.21;SB-PLL=1.30;NB-CORE=1.290;NB-PCI=1.250;NB-HT=1.320) and AOD runs OK, changed HTT in 2 10MHz steps then froze.

Oh man, no wonder. 10MHz is too high, it will freeze for anyone. +2MHz is what you should be going up until you start reaching within 25-30MHz HT of your core limit, by then, start going up by 1MHz.
Also, do you need all those extra latter voltages that high? There usually is no need for them above stock since they have nothing to do with the CPU or its clocks and chipset is unaffected by AMD CPU oc's.

AOD boots fine until HTT=245 from bios, after that "Fire Works", maybe have to do with BOOT HTT setting in BIOS, cuz I have to set it to 250 if I wanna boot at HTT=250, or else (25) error.

Hmm.. not sure but that's a typical sign of instability usually. When you have unstable HT values and AOD tries to access RD790 PLL, AOD will either crash or freeze the system - happened with me on all my chips

Are you sure you're config is stable above 245HT? Can you boot over that fine with CnQ enabled for instance?

All this testing was done with and without your page settings file.

Last Log file.

Log file is full of errors, mainly those of accessing ports or PLLs. It's not specific enough to debug though.