ABit AX78 Overclocking/Stability

[LIKMARK]

so, that probably means I should purchase highest multiplier I can, if its not a BE right?
so that the NB stays at a lower clock? translating to higher core OC potential?

Yes in theory. It should not be a problem at all as you can give NB lots of juice with this board anyways. You probably won't be able to overclock the cores that much that it becomes an issue in the first place.

[GenTarkin]

ok cuz, Im thinkin on getting the 8650 X3...you think multi wont be an issue for its NB then? cores will get unstable first? =)
Thanks

[hokiealumnus]

I can change the NB multi on mine. From the manual:

CPU-NB FID
This item (appears only with AM2+ CPU) selects the processor to north-bridge frequency.

and

[LIKMARK]

Strange, It must be cpu limited for me then. I have tested changing it with BAR EDIT too, and it dosen't work either. @hoki: you're running beta bios 13a too? (That pic from the manual was one of the reasons I bought the board)

[KTE]

I'm in Ubuntu 8.04 right now using AX78 with 9850, for over 3 hours now, benchmarking mainly.
I'm using stock BIOS and it booted and runs perfectly. I'm using my MSI board settings, 2700MHz stock voltage, 1.288v, just placed a heatsink on it but its not attached and there's only old TIM The PWM are running quite cool, cooler than MSI board and far cooler than DFI M2R. Only got 1 stick in yet, busy online yet.

http://imageupload.com/out.php/i111441_Screenshot1.png
http://imageupload.com/out.php/i111440_Screenshot.png

The 43/44C is BIOS reported temp.

Board fails POST with C1 using 1066/1000 RAM modules but works perfect with 800 RAM modules. It also boots very weird timings but has CPU-NB voltage control although there is no NB-FID option in this BIOS and I do have it on other boards.

[KTE]

In XP now;



Next I'm going to shoot for higher, 3000MHz -- easily with the CPU's range but let's see if the "idling instability" exists on this board too.

Volts is reported accurately so far..

[KTE]

3 in a row for me...

2.8GHz 9850BE stock voltage = 1.292/1.288v ID / 1.280 LD



These aren't just ss, I'm benching them trying to invoke an error - previously MSI and two GBT boards have found the core is 3055MHz 1.355v stable, so this is well within the load range.
Only trying to check if it suffers what they suffer -> idle instability.
All other boards failed at the above setting, freezing idling at anything above 2700 at any voltage... so far Abit has not froze or given any hiccups at this setting and its also the first BIOS.

Issues

-1066 RAM just does not run with this board - C1.
-How do I get unganged mode, there's no option as such anywhere?
-Also I cannot get my 800 RAM to run 1066 mode with Phenom in there (it can do it easily on other boards).
-Abit EQ keeps starting yet is not in autostart and keeps harping on about running as admin first time even when I've done it ten times and it still won't come on

The BIOS voltages work perfect for me.
0.1v NB adds 0.1v to NB.
0.02v CPU adds 0.02v to CPU, all the way up, I tried it.

[davidletterboyz]

The PWM are running quite cool, cooler than MSI board and far cooler than DFI M2R. .

Thanks for sharing. I'm looking for an affordable 9850-able mobo (I don't need multi gc). Probably will go for this or gigabyte 780 :P
BTW, what's the temp of the MOSFET? Do u cool it with a fan?

[KTE]

There wasn't a fan in the above shots, no. But since then I added one around the back, I always do it to rule out factors when testing as a good assurance. Fan is more around the back of the heatsink than the PWM area though.

PWM idles this at stock+CnQ+full fanspeed->
http://imageupload.com/out.php/i111723_pwmfs.png

PWM idles this at stock full speed+lowest fanspeed->
http://imageupload.com/out.php/i111725_pwmfsls.png

PWM under 5min load->
http://imageupload.com/out.php/i111724_pwmfsfs.png

Ambient is quite high, around 25-6°C and the "CPU" temp is taken from BIOS/Abit EQ gathered through ACPI. The "core" temp is taken from K10 sensors, its the same as 780G/790FX report.

BIOS 12 doesn't give me NB FID, going to try 13 B2 now. LIke I said earlier, MB is very good and excellent for the price, but it can't be in wreckless hands.

[ChrisTucker2004]

tell me pls if Bios 13B2 include the NB Fid.

I saw on the Bios Screenshots, that there is an NB Fid, cant you change him or whats the problem?

[KTE]

Chris, I am using that BIOS, I have flashed them all one by one - there is no such option in the BIOS unfortunately
I've tried boards where there was and they changed it fine - this one doesn't have it. However, I'm just going to use Bar_Edit to change it temporarily for testing.

This board does not run 1066 mode or 1066 mode RAM (2 kits) at all! One big peeve of mine, it does give plenty of options though but no POST if you try 1066 with 1200-1380 capable RAM. The chip runs it fine though.

This board has no option for Unganged mode anywhere!

Also I ran EVEREST comparisons across both boards.. AX78 at high/loose timings and 800 clocks performs better in MEM than the MSI 790FX with tight timings in 1066 mode

Ganged that is.

AMD OverDrive runs the quickest and smoothest I've yet seen to date on any board, wow. MSI is 2nd after this board but it really is no where near this good
Also, in AOD, CPU VDDC is a dead value.

Next up to try is Auto Clock...

[KTE]

OK now for the bad news

This board is just not made for high volt/watt CPUs. I would say stick to stock volts, simple as that.

I am 2750 stock volts stable... but I fed the CPU 1.4v 2.5GHz and the system shut off instantly - nothing but motherboard.
Then rebooted and it would not boot for a while - the board itself won't start.
Then tried 2.7GHz stock volts boot, shut down at Windows loadup as soon as the power draw topped 100W on CPU.
Then tried stock boot, booted up.
Now at 2.7G/2.8G/2.7G/2.7G using AOD at stock volts but I booted up 2.5GHz on all.

There is no;

overheating
rail tripping/faulty-weak component

anywhere - this is totally the motherboard not able to withstand high power draw by CPU. The shut off is instant, as if you tripped a power rail or overheated too much. But the fact is, none of those are happening. The tems are 30-40C maximum and the rails have 300W still spare working fine.

I fed CPU 1.5v in AOD and the system shut down there and then! Again motherboard, not my CPU because on other boards I can bench <1.6v on Phenom perfectly fine.

So with a 1.3v chip, I'm stuck to 1.3v on this board -> hold your horses though, I'm still experimenting

[LIKMARK]

@KTE: 1066 memclk mode runs, but with horribly loose timings (6 7 7 20 etc..). I have made choosing unganged mode once, and it involved multiple auto settings, so I have not been able to reproduce it.
Were you able to change cpu-nb fid with BAR EDIT?

[Jethro]

This board would surely be more xtreme material if not for the very weak power handling. I had to ditch 330+ fsb, everything was wonderful till NWN2 and she just refused to be nice. Even at current she still flakes out now and again. All voltages reduced to minimum and everthing well within spec and she still act's up now and again. Might keep it for now but its not gonna be my phenom board I'm quite sure.

Current 98&#37; stable settings = 300 x 11 1200HTT 943ddr 4 4 4 12 2t 1.5v cpu... 1.31v nb 2.00v pcie 1.3v HTT

[KTE]

Ignore post#237 by me.
I've figured it out -> it's the friggin BIOS not the board, 13B2 is very buggy!

It gets stuck at bootup with 2F/2B code, then keeps recycling it really quickly.
It also takes around 25 seconds to reach C1, just before system starts.
It will not startup many times getting stuck at the above codes at stock.
It will not allow voltage adding to CPU, it'll reboot before startup if you do.
It will not boot anything but stock.
BUT, thanks to MSR Editor and Sam's MSR Tool, I've been running/benching 2700/2800/2700/2800 across all my cores stock voltage since my last post here without fail (no idle freezing, which is what I'm trying to avoid)

Look here, 26C ambients: http://www.xtremesystems.org/forums/...5&postcount=80

I have even ran 2800/2900/2800/2800 stock voltage running the old and latest two x264 benchmarks for about 20 minutes full load, no problemo including the PWM being 45C max, which is excellent (now have a low speed fan over it) for any PWM as they're rated for 110-120C.

So it is not a PWM problem but a BIOS one - I'm reverting to BIOS 12 now unless someone has one after BIOS 12 but before 13B2?

This is what I'm testing for 19 hours now, going very strong.



Still sticking to stock voltage, I know I have at least two good cores which'll do 2900MHz stock voltage.

Just a note: Phenom cores are arranged like this by AMD and this is how they're picked up by AMD Power Monitor/CPUZ;

core 0 - core 1
|.................|
core 3 - core 2

So anything you use which thinks;

core 0 - core 1
|.................|
core 2 - core 3

will get it wrong (i.e. MSR Editor).

@KTE: 1066 memclk mode runs, but with horribly loose timings (6 7 7 20 etc..). I have made choosing unganged mode once, and it involved multiple auto settings, so I have not been able to reproduce it.

Thanks for the help
I'll try to reproduce it then.. which BIOS are you running?

Were you able to change cpu-nb fid with BAR EDIT?

Yes ->

[LIKMARK]

Ignore post#237 by me.
I've figured it out -> it's the friggin BIOS not the board, 13B2 is very buggy!

It gets stuck at bootup with 2F/2B code, then keeps recycling it really quickly.
It also takes around 25 seconds to reach C1, just before system starts.
It will not startup many times getting stuck at the above codes at stock.
It will not allow voltage adding to CPU, it'll reboot before startup if you do.
It will not boot anything but stock.

You are running XP 32bit I presume. Under Xp 32 bit this bios was VERY buggy for me too, under Vista x64 there is "NO" problems (exept for general unwillingness to give good tight ram timings, and no logic (at least for me) ganged/unganged mode settings). When playing with ram timings I've got som serious no boot issues though, that can only be solved by taking out the cmos battery for a long time.

I'll try to reproduce it then.. which BIOS are you running?

13B2, B for buggy

Yes ->

Then it definitley is cpu limited for me. Very strange that some cpu's are limited like this. (it is a week 44 though, pretty early batch, that might be the reason?)

Anyways.. just bought myself a L164WS/B and a barcy, looking for som fun Just need to wait for it to be sent from the US

[KTE]

Yes, I'm running XP 32b and Ubuntu 8.04 32b with that.
I also had all those weird RAM timing issues, no boots so many times.

Is there any BIOS after 12 but before 13B2 you have?

My week 44 9600 was also CPU limited, could not downclock the NB on any board or BIOS while the others could (non BE).
Week 43 and 45 9500/9600/9600BE were all fine.

Good luck with the Barcy.. which one BTW?

[LIKMARK]

Yes, I'm running XP 32b and Ubuntu 8.04 32b with that.
I also had all those weird RAM timing issues, no boots so many times.

It is the achilles heel of this board IMO, If they could fix some decent ram values I think it would shine as the best entry level Phenom board.

Is there any BIOS after 12 but before 13B2 you have?

No, sorry.

Good luck with the Barcy.. which one BTW?

It's a 2347, 1.9GHz with too low cpu fid for real fun, but they're the only ones that is possible to get hold of at a reasonable price. Seller stated it was a B3, but I think they only comes in B2 flavour. Overclocking probably will be limited by cores before cpu fid anyways. (always hoping for the opposite to happen, though)

[ChrisTucker2004]

Bar_Edit is a tool to change the NB-Fid?

Then you would say, that this Board is better to oc as the MSI? MH 4 Phases vs 5.

[KTE]

Yes we change NB FID using Bar_Edit or WPCREDIT for a long while now
We change CPU FID, CPU VID and NB VID using MSR Editor/Sams MSR Tool aswell.
You can do it on any AM2+ supporting board, the registers we change are within the CPU.

Abit board is 770, MSI board is 790FX.
Both are very different categories, different price, different offerings.
Abit 770 is tons better than the MSI 770/790X no doubt.
MSI 790FX v1 has far less problems with Phenoms, much better support/compatibility but also less BIOS options for oc unless you go to 1.13b, which is one of the best performing/overclocking BIOSes I've seen. It has everything you need.

They are different boards, I would wait for a good BIOS for Abit board if you're planning to get that with Phenom B3, but for less than the price of GBT 780G, it's one excellent price:perf board!

-MSI board is the best 790FX offering in terms of price:perf at the moment, without a doubt in my mind. Best RAM clocker aswell. But I would not use it for over 1.50v load 125W CPU clocking, go DFI/ASUS 790FX for that.
-Best high end clocking, over 1.5v is DFI/Sapphire and ASUS 790FX, but all three are too expensive for what they're worth, not worth their price and only better if you go over 1.5v because of the PWM and because of better oc BIOS options.
-ASUS 790FX is far more stable than the DFI, DFI just has too many issues and weirdness, like always.
-Abit, IMHO, is the best 770 I've come across so far including the ASUS M3A coming second - they just have to resolve some major BIOS issues: RAM/NB FID. I would also limit the voltage to sub-1.4v with 125W quads personally.
-ASRock 780G is better than GBT 780G aswell, just to add and GBT 740G is not good, to say the least.

I have hit 224HT with AX78 fine so far, all stock voltages: http://valid.x86-secret.com/show_oc.php?id=355810
Haven't tried more because of bad BIOS.

Where the heck is cutie nowadays by the way?
Did he go crazy for that girl in his av or something?

***I know you're reading mate***

[hokiealumnus]

Ok, I see several of you have memory timing / OC'ing issues. This was an answer to KTE's post in the abit support section. It may (hopefully) clear up some issues you are seeing:

Most likely the reason you can't OC your memory is the tRFC timing. That timing on this board is measured in ns and are selectable individually for all four DIMMs. By default, the empty slots are set to 75ns. Unfortunately, for whatever reason, it'll change one populated slot to the setting it should be at (either 105ns or 127.5ns, depending on your memory) but it treats all three other DIMMs as empty with respect to that timing. Anyway, long story short, change the DIMMs that you have memory in to 105ns and try again. You can probably clock your memory higher at that point. If not, try changing both of them to 127.5ns.

Secondly, regarding your multipliers, are you all disabling Cool n' Quiet? If not, the option will not be present. You have to disable CnQ to be able to change your multiplier.

[KTE]

Perfect timing. I just ran all my tests
My post to Abit is here: http://www.xtremesystems.org/forums/...d.php?t=186671

Ok, I see several of you have memory timing / OC'ing issues. This was an answer to KTE's post in the abit support section. It may (hopefully) clear up some issues you are seeing:

Secondly, regarding your multipliers, are you all disabling Cool n' Quiet? If not, the option will not be present. You have to disable CnQ to be able to change your multiplier.

First of all, yes you are spot on. I checked about 45 mins back and was intending to post on it; the issue of no POST and weridness is on all BIOSes and it is caused when you move tRFC from 127ns to 105ns or lower. Sometimes USB keyboard stops working and graphics screw up this way. If you change it back to 127ns, it all works fine.

Secondly, I don't have CnQ enabled, it just does not show up there.

The rest of the RAM issues are timing independent. I've tried the highest there is, no 1066, no unganged.

[ChrisTucker2004]

thx KTE:

Can you show me a link for Bar-Credit?

[LIKMARK]

thx KTE:

Can you show me a link for Bar-Credit?

You sure you want bar credit, it could get you into financial issues if you can't control yourself LINK: http://www.tweakers.fr/baredit.html

[ChrisTucker2004]

thx

I hope i can control myself, but i must try it. Could anybody say me which line i must change for the NB-Fid?

Or an Tutorial?