ASUS AMD Beta BIOS Releases

[Halk]

Dave - Why the Big Bang Conq? It's going to be FX990 chipset (at least I think it is... ). Just simply because it's the next chipset up, on a board designed for bulldozer rather than a board that happens to work... or are there other reasons to move? I was all keenly looking forward to choosing between FX990 at launch, but now that it seems my IV Formula is going to work, I'm thinking to myself why upgrade, FX990 isn't supposed to be that different.

[FlanK3r]

look at 790 and 890fx...Thubans at 890FX are better than at 790fx (specially working with RAM). And there will be new CPU...think, we will see more than one diference.

[Halk]

Yep, I guess. Although there's a bigger difference between 790 and 890 than there is (by what little I know) between 890 and 990. So this week I got news that Asus will update my BIOS to support Bulldozers, and out of the blue the guy who makes Speedfan emailed me with a beta of Speedfan that he says after I gave him some test results, is now fully compatible with Crosshair IV Formula. Events are conspiring for me to keep my Crosshair

At worst though I'll have the luxury of getting Bulldozer right away, and picking the right motherboard from the choices available.

[NEOAethyr]

I rebuilt my machine/board yesterday, got her s&m stable again.
Checked it twice last night and once this morning.
Now I'm happy, hopefully it stays that way.

The board lost half my mem again.
Loading defaults didn't fix it at first.
I went back into the bios, switched the setting's for bank interleaving and unganged mode to the opposite setting and back and then exit+saved the bios.
Then both gigs showed up and my everest scores was around 14k for copy instead of 11k.

Pretty sure this is the bios screwing up and not my parts lol...
I might switch back to the previous bios anyways to be honest.
I haven't tried to clock my cpu on this bios, because I totally forgot what I was using...

I would be nice if I could control the downspread ratio...
Seems like it might be diff, haven't paid that much attention to it until now.

[Daveburt714]

Dave - Why the Big Bang Conq? It's going to be FX990 chipset (at least I think it is... ). Just simply because it's the next chipset up, on a board designed for bulldozer rather than a board that happens to work... or are there other reasons to move? I was all keenly looking forward to choosing between FX990 at launch, but now that it seems my IV Formula is going to work, I'm thinking to myself why upgrade, FX990 isn't supposed to be that different.

I agree with you Halk ....
The fact that the C4F will accept BD is pretty sweet!
I'll probably hang on to it now and see where the new boards shake out as far as performance goes...

I've been testing the 3012 bios this weekend on my 1090T and I've made some interesting discoveries.

1) They took away alot of the granularity in voltage control across the board.
It's not really that big a deal, but I do kind of miss the detail we had with 1304.

2) CnQ seems to stay on, reguardless of what you set in Bios.

3) It looks like they put quite a bit of effort into LLC (this is a good thing!)...
I always used to run LLC @ 100% even though alot of the pros said it wasn't advisable...
I just couldn't get decent clocks with it off due to the V Droop.

LLC seems to be MUCH better on the new bios!
Neither Off, nor 100% seems to be the best option now...
Setting LLC @ 50% works the way I always thought it should (even though I still think there should be a 75% w/ no droop at all )...

Reguardless of CPU v's I've never seen the droop go more than .03v's @ 50% LLC.
Thats a huge improvement compared to what we had with 1304!

It doesn't look nearly as good in EVO where I could get away with 1.435 CPU core...
Now Evo shows 1.5 CPU, but in reality it's about the same (according to CPU-Z) as it was with the OV from 100% LLC.

Long story short, 50% LLC results in a managable droop and I don't have to worry about the crazy ov's I was with 100% LLC on 1304....

Now I just need to tune my Memory!

BTW: NEO, It could just be my kit, but I don't have an 1866 mem option, and I've never had any problems with the bios not detecting all my ram...

Oh, BTW... Here are the results of my weekend testing of 3012. My CPU craps out about the same place it always has, but the IMC and Mem clocking seem better.

[NEOAethyr]

1) They took away alot of the granularity in voltage control across the board.
It's not really that big a deal, but I do kind of miss the detail we had with 1304.

I haven't noticed this.

2) CnQ seems to stay on, reguardless of what you set in Bios.

It always did that on me...
It's not a huge deal though, it least it's on .

Long story short, 50% LLC results in a managable droop and I don't have to worry about the crazy ov's I was with 100% LLC on 1304....

This I'm not sure of.
I do think it was improved on actually, but I have no idea by the amount yet.

BTW: NEO, It could just be my kit, but I don't have an 1866 mem option, and I've never had any problems with the bios not detecting all my ram...

1866 should only show up on the coming up am3+ cpu's.
To be honest, what's the point of even trying to see if it's a working ratio on thuban?, 266 is doable, I think it's only there to bring up the bottom line for oems and get everyone using a min 1866 speed).
Still, I'de like to see it on thuban if it worked, I can only hope that asus has already tried it regardless of official specs, and it didn't work so they didn't include it for our cpu's.
Otherwise they didn't check it...

It would be great if the cpu-nb llc had been improved on for lower voltages on that...
I currently use +0.24375v for that thing.
Seems to take a good 2 hours to check for cpu-nb voltages using ycruncher.
That's the only thing I've found so far that verifies it 100%.
I've been using 8hrs (the setting got me past the 6 or 7hr "final?" marker and past 8hrs...).
Other progs just get me to the 97.5% stable level.
Which is 0.1v+ less then what it really needs :\.
Otherwise I'de go 3-4 step ups from stock voltage to get 3ghz.
I haven't checked to see if I could go lower in along time now...
That's alot of voltage for 3ghz isn't it lol ?...
I hope you guys are checking with ycruncher, 2hrs for your cpu-nb voltages...

I still haven't figuered out what I'm gonna do with the cpu-vdda setting.
A few steps higher and the mem bandwith usually goes up a tiny bit.
But it never fixes anything, like other voltages or the llc bandwith prob.
It just seems to increase bandwith a tiny bit.
Lowering it to 2.2v didn't even do much, just lowered bandwith a tiny bit and really there was nothing else to it :\.

Lame Example:
cpu 100% llc = 15k.
cpu 50% llc = 20k.
cpu 50% llc + extra vdda = 20.2k.
cpu 100% llc + extra vdda = 15.2k...

That's kind of the way it acts, in a generic example.
The vdda brinds up the lower end a bit, the results aren't as random anymore and etc.
Going higher doesn't seem to let you decrease any other voltage.
Going lower doesn't seem to lower temps as far as I know, could be wrong.
The diff is more between 20-80megs perhaps at best.

[charged3800z24]

2) CnQ seems to stay on, reguardless of what you set in Bios.

Dave,
under the CPU settings you have to disable "CPB". This should fix your issue. if not, then that is weird...

[Daveburt714]

Dave,
under the CPU settings you have to disable "CPB". This should fix your issue. if not, then that is weird...

Thanks Charged, next time I'm tweaking I'll turn that off...

[-Px-]

There is beta 1902 for M4A89GTD PRO and release 1902 for M4A89GTD PRO/USB3 on Asus site, changelog contains only one fix: Improve EPU function. Later I'll test it for other fixes

[NEOAethyr]

Read bandwith is about 400megs higher.
Copy bandwith is about 2000megs lower.
L1, L2, L3 ECC now works, but isn't adjustable, improves cpu stability slightly.
Mem and cpu-nb clocking is worse.
Fsb spread spectrum is diff or working, I'm unsure which it is lol.

Oddities...
Cpu voltage regulation is tighter, good.
Cpu voltage goes 0.1v lower, I can't explain it yet, I'm on win2k3 and my power states are locked on the highest pstate.
Yet it acts as if another pstate is in effect, one lower then the highest.

I think it's either:
CnQ was never working in the 1st place, it's vid wasn't, but now is.
Or, turbo mode uses vid from one lv pstate lower now, so each pstate is now 0.1v lower then before.

Note, turbo was disable, but cpu cpb mode was enabled.
Also note cnQ and c1e was enabled, but they don't disable on the previous bios anyways.

Saving over the last cmos profile with the new bios, then going back to the previous, the older bios'es cmos profile in that last location is still intact but non of the others are.
Yeah that's an odd one...

I was able to verify that the previous bios'es ecc support is 100% broken.
This new bios definitely works but it's not adjustable.

I'm on the 1304 bios right now.
I'm not sure if I wan't to go back or not, this old bios clocks better.
But the bugs in it sorta outweigh the clocking.
But then there's the speed difference, the new bios has some extremely low bandwith scores.


My overclocking went to shiz though.
I can't boot 3.25ghz nb, nor 2133mhz mem anymore into windows.
On the older bios and I can post those speeds but can't boot windows.
On the new bios I can't post those speeds at all.
I have no idea why.

Regardless 3 major things with the new bios.
Overclocking is worse for mem and cpu-nb.
LLC @ 100% is really screwed up performance wise...
Mem bandwith is 2000m lower.

I suspect the low copy bandwith prob is due to the way bank interleaving is handled.
On the previous bios, address bits 12 was the fastest setting.
On the new bios they probably picked a setting out of there butt to use and hence the lower bandwith.


Edit:
Been trying to prime or ycrunch to check for lower nb voltages or what i need for 3.5ghz cpu speed and above.
Everytime it hardlocked...
Now I know why, cpu temp 73.3c and rising, nb temp 51c and rising...
Looks like it's time to replace my cpu's tim again, it's been about a month, I get sick of that.
I'm gonna try making my own asap...

[Daveburt714]

Dave,
under the CPU settings you have to disable "CPB". This should fix your issue. if not, then that is weird...

Hey Jay, thanks Bro...
That fixed the pseudo CnQ problem I was having! That option may be useful on BD, but not so much for Thuban...

NEO, granted, I haven't done alot of testing with this new bios yet, and it could be my mem kit,
but I really don't see the Mem performace problems your talking about...
If anything, 3012 seems better with IMC/memory than 1304.

Reguardless, I'm kind of liking this new bios...
It gives me something to play with, and I didn't even have to spend any money!
How can that be a bad thing....

[-Px-]

Tested 1902 for M4A89GTD PRO/USB3 on two systems, only fixed thing is Ez-Flash in BIOS, broken in 1703, all other bugs (dual-cards, bsod with disabled chipset sata) are still there...

[F@ruk]

Hi there. I have M4A89GTD PRO/USB 3 with bios version 1703 and Phenom II 970, so my problem is cpu speed and voltage are always fluctuating not stable. Can be there any problem on mobo?

http://vip.asus.com/forum/view.aspx?...Language=en-us

Thank you.

[Tonyb]

I personally found 3012 not as stable with overclocking Dave and I had the strange business with sata 1 and 5 swapped around and only the boot drive showing. Neo the address bits are for the channel interleave not the bank and basically affect SSD routines, in 3012 I found the channel interleave only had auto or disabled.

[NEOAethyr]

A min ago I checked into using 100% lcc on the previous bios, the 1304.
Just for the heck of it, to see if the performance drop was there.

Results:
Read bandwith, 100megs lower, could be within margin of error but I don't think it is, could be though.
Everything else was equal to 50% llc.

Meaning, the new bios reduces mem copy bandwith drastically when 100% llc for the cpu is enabled.

It also seems a tiny bit less stable for clocking cpu-nb and mem.
And it's 2k lower in copy bandwith overall (I get 14k instead of 16k at 3.25ghz, 11k with 100% cpu llc).

One interesting note however, the new bios goes 0.1v for the cpu lower then the 1304 bios.
It's not by standered either, the bios uses the same voltage stock.
It just drops down that much with one of it's power states somehow.
I don't know much about it, but I think it's interesting.

Example.
Hwinfo32, says my voltage is ruffly 1.2v on the cpu right now.
With the new bios, it says about the same, but drops on idle to 1.1v...
I like, but I have no idea what exactly enables it to do that.
Probably the new cpb mode thing or something.


I checked because I was messing around trying to see what I could do from in the bios to lower my cpu temps, from 70c+ to the norm of 50c+ on load...
I just redid the cpu heatsink a month or 2 ago...
I finally get rid of all my probs, and all the sudden my cpu temps sky rocket :\.
Granted I haven't primed in about a month either...
Still, as ceramic should last longer then a month or 2 .
I took good care with applying the tim too.
Tonight I'm gonna have to redo it again.
I'm gonna have to get my hands on some baby powder and try making my own, I already got dow silicone (expired but should be fine anyways).

I figured I'de try 100% llc and lower my voltage.
Well I did but my voltage didn't lower much at all (from around 0, to -0.06xv or whatever).
Then switched back to 50% and my voltage result lowed a bit more, I know that on 50% llc my cpu is stable at 3.25ghz at the lowest vcore offset.
Which helped a little bit but my load temps are still around 70c.
That was the reasoning behind checking the 100% llc bandwith and such...

Yeah 70c load temps at 3.25ghz with the lowest vcore availible ...
I hate stock coolers..., they always seem to be right on the edge of stability at stock voltages.

[THE JEW (RaVeN)]

Everybody's moving on to Bulldozer and I'm still waiting for ASUS to release BIOS support for the M3N WS. Seriously, a workstation board doesn't receive the update to get 6core support, but older consumer grade M3N series boards do?!? Poorly played..........very poorly.

[trans am]

Everybody's moving on to Bulldozer and I'm still waiting for ASUS to release BIOS support for the M3N WS. Seriously, a workstation board doesn't receive the update to get 6core support, but older M2N series boards do?!? Poorly played..........very poorly.

you are complaining about bios support for a EOL amd ddr2 server board with an nvidia chipset? poorly played or call it whatever you want. no offense but maybe its time to move on. poorly played? more like I'm too cheap to play.
what do you want man? I guess a good analogy would be complaining to AMD to release a 9800 pro with pci-e support because you have a agp 8x board you can't part with.

[XRL8]

He bought expensive workstation oriented mainboard from a respectable manufacturer.He has a right to be angry about the lack of support.

@jew raven, have you tried implanting the new agesa to the bios ? Its basically what foxconn did for my board.It gets me 6 core support without cnq and proper turbo support but it works.

[THE JEW (RaVeN)]

you are complaining about bios support for a EOL amd ddr2 server board with an nvidia chipset? poorly played or call it whatever you want. no offense but maybe its time to move on. poorly played? more like I'm too cheap to play.
what do you want man? I guess a good analogy would be complaining to AMD to release a 9800 pro with pci-e support because you have a agp 8x board you can't part with.

Perhaps. But when consumer grade motherboards with the same chipset already have the support added, I would expect the same level of professionalism applied to their workstation boards. People paid a premium for those boards so the same (if not better) level of support should be directed towards those boards.

Seriously, if these boards are supported why wasn't the workstation board?

M3N-HT Deluxe/Mempipe
http://www.asus.com/Product.aspx?P_I...specifications

M3N-H/HDMI
http://www.asus.com/product.aspx?P_ID=hR9xA49ZHyU31AeA

Both of these consumer grade motherboards support the 1100t. I completely understand your sentiments were these boards also not supported.

I also would argue with the EOL statement. It's the only motherboard listed under AMD workstations for that socket. It's not listed in the legacy section. They delisted the M2N version to show it was EOL. They even sell it as being open to future upgrades themselves:

The M3N WS is the ideal foundation for a powerful PC. It delivers awesome power, dependable performance and unparallel multiple /O scalability for the most demanding tasks and future upgrades.

All I'm asking for is the same support given to a workstation that was given to a consumer grade motherboard. I don't think that's beyond expectations. I've already opened up a complaint ages ago, but nada. So, in my book, definitely poorly played. Especially when a DDR3 alternatively doesn't even exist in their workstation section. I can't even go out and buy a new one!

[THE JEW (RaVeN)]

@jew raven, have you tried implanting the new agesa to the bios ? Its basically what foxconn did for my board.It gets me 6 core support without cnq and proper turbo support but it works.

No, but I'm open to the methods. Quick link to a walk-through or should I just read through the thread here? I began poking around Rebel's Haven but, "real life" got in the way of me actually attempting the BIOS mod

[Tenkura]

It this one for BD? M4A89GTD-PRO-ASUS-2001.zip, ftp://ftp.asus.com.tw/pub/asus/mb/so...-ASUS-2001.zip

[BeepBeep2]

It this one for BD? M4A89GTD-PRO-ASUS-2001.zip, ftp://ftp.asus.com.tw/pub/asus/mb/so...-ASUS-2001.zip

No, only the USB3 version got a beta bios for Bulldozer so far.

[Tonyb]

Strange someone seems to have a 1902 BIOS ver. for the Crosshair IV Formula.
Screen shot here.

http://www.overclock.net/amd-motherb...treme-866.html

[Ccaution]

He was kind enough to upload it too

1902 -> http://www.mediafire.com/?gzkw847zeddfff6

[NEOAethyr]

Cool

I'll flash it in a few.



There seems to be a tiny bit of improvement.
Before with turbo with win7-sp1 my fsb and such would droop a bit.
It's much better now, I haven't checked on load though yet.

Bios cmos profiles "are compatible" with 1304's.
Cpu ecc still doesn't work, no special ageas ver, this is a pre-bd bios.

I expect it to act alot like 1304, but it might be a wee bit better in some ways.

I think the bios date for the 1304 is 12/09/10.
And for this 1902, 02/17/11 if I remember right.


Thank you guys for this bios.