Phenom 9500 w/ MSI K9A2 Platinum

[KTE]

Nope, nothing at all. 121 fixed some PCIe bug IIRC, but no other difference and 113 is still my favored one. P0D gave abysmal performance.

[Emperor]

Nope, nothing at all. 121 fixed some PCIe bug IIRC, but no other difference and 113 is still my favored one. P0D gave abysmal performance.

official 1.13 bios ?

any of the bios allows for pci-e freq changing in bios?

[KTE]

BETA dated 16th Nov.

No PCIe changing yet that I've seen. But if you use the "DOT OC" options, they can oc the PCIe.

[Dstoff]

Hi all, Noob to the forum, and learning alot. thanx. Just started a new build, components in my sig. I have reloaded vista twice, I think AOD was killing it after first install. Now setting restore point between software installations. At work so won't get to play with it for hours. But, something driving me crazy. CPU fan running at 2800 is LOUD, and can't find anywhere in bios to change it. Solution? Thanx in advance.

[KTE]

Welcome to XS Dstoff.

In the BIOS there should be an option for "CPU Smart FAN Target" where H/W Monitor is. That'll help you be able to control fanspeed. As the manual says (attached).

Hope it helps.

.

[DaMulta]

The new .14 overdrive now shows the temps!!!!

[KTE]

Yep

[Brother Esau]

@KTE.....hey how many Bios updates have you guys seen on the Gigabyte and MSI please since release?

[Blacklash]

Vista x64 definitely carries a score hit in 3dmark06. It equals about 500 on the CPU and 750 total @ my max overclock.

There's my Q6600 G-0 @ 2.6GHz with Vista x64-

http://img147.imageshack.us/my.php?image=06263dud4.jpg

and my normal Vista x64 score-

http://service.futuremark.com/compare?3dm06=3553315

It's only equal to about 2FPS in the graphics tests in 3dMark06. So in actual impact it looks worse than it is. My OC score is with older drivers and I might score better with the latest. It was run in late October.

If you're all about squeezing those last few points out of a bench stay on XP, or dual boot.

[KTE]

Nice finding Blacklash, thanks.

Brother, Brother, as far as I'm aware, there have only been two public release BIOSes for MSI 790FX, the last older than a 5 weeks now and the others we have to FIND ourselves (BETAs) around were four which were older than 4 weeks again and hardly resolved anything. All major issues still remain.

BUT Gigabyte 790FX has had three offficial final releases which resolved all major issues in their second one, one latest came out not more than two weeks back to add ease, and at least four others that I knew of they could find around (BETAs).

Gigabyte support has been far far superior than MSI support. My experience with MSI this time has been exactly like what I have gotten by gorilla ASUS in the past. Afterall, I PAY, they have to appease my service for me to buy and support their products, not the other way round they seem to boast. I could easily go and buy ANYTHING else, plenty of boards better working around, and let alone Foxconn's upcoming board.

[Brother Esau]

Like I said thats why I wont patronize MSI ever again.....too bad its hind sight 20/20 though

[Emperor]

guys...do u use the bios or the pstates to change the vcore?

for me my settings are 1.537 something volts in bios and pstates is 040.



another thing that puzzled me is that after leaving my com powered down for the night, the bios automatically reduced the vcore in bios without changing the clock speeds and because of this i ended up with repeated reboots telling me my OC is unstable.

and the best part? it appears the max voltage is lowered from 1.6 to 1.5375 or so.

[KeZzZu]

15.7 sec wprime 32 run @ 2.74 ghz(4-4-4-3) unfortunately 2.750 ghz run screenie bananed(4-4-3-3)

Question: is there some good tweaks more? where i can tight up some more in memsettings?


http://img441.imageshack.us/img441/30/tight2yk0.jpg

[DaMulta]

Anyone run one on dry ice yet?

I guess I need to RMA my motherboard and do it then

[Daveburt714]

Hi Folks, and thanks for all the GREAT info on OC'ing Phenoms!!

I've been reading the forums for a couple weeks now and finally decided to join... Have you guys come to a consensus on what the BEST AM2+ Mobo is yet?

After reading alot of this stuff I'm leaning heavily toward the DFI board, even though my last few builds have all been MSI's... I'm trying to hold off until the 9600 BE's are released, but I think I'm going to go ahead and pick up a board now...

I already have the Memory for it (G.Skill (2x2) DDR2/1000)
http://www.newegg.com/Product/Produc...82E16820231145
so if you have any Mobo recommendations, or see any problems with this Mem on a particular board I'd appreciate a heads up!

Wish the BE's would hit the shelves already!!! This money is burning a hole in my pocket

[KTE]

Hi fellas, had to leave for some time. Hope your all enjoying your holidays, best wishes.

guys...do u use the bios or the pstates to change the vcore?

BIOS mainly but I try AOD too if it works.

for me my settings are 1.537 something volts in bios and pstates is 040.

P-States won't be that for CPU, it won't boot at that if enabled. That looks like NB VID to me. CPU VID is locked and not the VCore.

another thing that puzzled me is that after leaving my com powered down for the night, the bios automatically reduced the vcore in bios without changing the clock speeds and because of this i ended up with repeated reboots telling me my OC is unstable.

How do you know if it was the BIOS which reduced it? Did you start AOD and ever hit Apply? If you did, then that is what reduced it, since you have to choose max volts in AOD to get a high actual VCore value.

and the best part? it appears the max voltage is lowered from 1.6 to 1.5375 or so.

Yeah same here. Mine used to be 1.64V-ish and then suddenly went down to 1.458V. What's your CPU VID? When that changes to 1.20VID from 1.25VID, the max volts you can have also falls according to my testing.

15.7 sec wprime 32 run @ 2.74 ghz(4-4-4-3) unfortunately 2.750 ghz run screenie bananed(4-4-3-3)

Question: is there some good tweaks more? where i can tight up some more in memsettings?

AOD.

Anyone run one on dry ice yet?

Don't have equipment here, since I'm on holiday for a while away from home. Tried an industrial phase change unit about 2 weeks ago from work, which can usually take the adjacent object to -30ºC temps. CPU IHS reached -28℃ but it did not make a different to max bootup MHz, max HT ref. and max CPU MHz, so I didn't bother again.

I guess I need to RMA my motherboard and do it then

Why do you want to RMA the MB?

Hi Folks, and thanks for all the GREAT info on OC'ing Phenoms!!

I've been reading the forums for a couple weeks now and finally decided to join... Have you guys come to a consensus on what the BEST AM2+ Mobo is yet?

After reading alot of this stuff I'm leaning heavily toward the DFI board, even though my last few builds have all been MSI's... I'm trying to hold off until the 9600 BE's are released, but I think I'm going to go ahead and pick up a board now...

I already have the Memory for it (G.Skill (2x2) DDR2/1000)
http://www.newegg.com/Product/Produc...82E16820231145
so if you have any Mobo recommendations, or see any problems with this Mem on a particular board I'd appreciate a heads up!

Wish the BE's would hit the shelves already!!! This money is burning a hole in my pocket

Welcome Daveburt714

The DFI board has problems more than MSI one with Phenom that I've seen. I like the board (especially colors) but it seems to have a HT wall with Phenom lower than the MSI does which needs to be fixed through BIOS revisions. I would've picked it up actually but it is no where to be found around these countries.

As for the memory, it won't be able to run at the rated speeds with the MSI board so far.

[Daveburt714]

Welcome Daveburt714

The DFI board has problems more than MSI one with Phenom that I've seen. I like the board (especially colors) but it seems to have a HT wall with Phenom lower than the MSI does which needs to be fixed through BIOS revisions. I would've picked it up actually but it is no where to be found around these countries.

As for the memory, it won't be able to run at the rated speeds with the MSI board so far.

Thanks for the Welcome!!

I'm glad to hear you say that about the MSI board, because I actually ordered one last night ($90 cheaper too)... Now just waiting for the BE's
I've seen some really killer scores on the DFI's using X2's, I'm sure it's a good board.. But it sounds like it needs a little work before it plays nice with Phenom, whether it be bios or a possible revision...

Plus I had to hunt a little to even find a K9A2 in stock, so when I did I snatched it up.... If you wouldn't mind and you have a better BIOS than what is posted on MSI's site, could you link it for me? The one on the site is over a 6 weeks old....

At least the DDR2/1000 will be one less thing to worry about, and it's said to OC quite well, so if they ever do get the 1066 spds running, I should be OK!

[KTE]

Daveburt714: I have all the released BIOSes, not linkable online but they're on a HD that I can't access anymore. I'll try and see if I can get it to work and attach them on one of the posts here later on.

IMPORTANT NOTE: MSI board dropped in price very quickly on my side of earth...? Some MB manfacturers (namely, MSI and ASUS) are telling official reviewers that their motherboards need a new revision to solve so many of these problems (which is coming out very shortly), which is exactly the opposite to what we customers are being told, but no wonder we've had no help for over a month. So if I was you, I'd wait till a better, newer revision of the MB's are out to purchase one to keep. To experiment like I did, it's a different story.

Here's one (pretty new and quite decent) Phenom 9900/Spider vs Q6600 review where these MB MFG's told the truth to a reviewer: http://www.overclockersclub.com/revi...mon_9900/4.htm

QUICK NOTE

I don't have a Phenom since around 34hours now. It was sold and delivered to a friend incl. insurance and recorded next day delivery with P&P for $160.
That means, I cannot possibly test it or refer to it anymore beyond memory. My C2D laptop and P4 do not for some reason pick up my SATA drives used with Phenom so I can't access the information I stored on them. I can only speak from memory now, sorry. Waiting for BE now.

However, now I can speak with details and more thoroughly. My brain is free at last.

I've done many tests since our dear friend Lightman posted recently about Northbridge overclocking and if you remember, I had posted about it in this thread 2nd December with 3.12GHz NB speed readings by CPU-Z and many benches HERE. Start reading on from Post#69.

I noticed many many discrepencies later on, as did Sammi/Tony/lukija and after much testing and investgation with help from lukija, we decided the high Northbirdge reading in CPU-Z are faulty. It will not validate those high Northbridge reading CPU-Z files, no way. In fact, I will post comparison benches later on (when I can access the HD somehow) of same CPU MHz different NB MHz, and you will see that actual ≤9x multi Northbridge frequency change has a very noticeable performance effect even at +/-200MHz difference compared to these "errorsome" readings having zero performance difference whilst showing +/-1000MHz difference in speeds.
Another very easy way to determine what your Northbridge speed is, is to set a "high HT link speed". If you have a high Northbirdge speed from the BIOS, the HT speed can also be that high without problems. Try it.

You will see, that overclocking through changing northbridge multiplier with Phenom will only get you a faulty NB speed reading in CPU-Z and faulty component readings in all other monitoring software. The HT speed will not change from what the NB multiplier 9x would have it running at in any case. Meaning the NB pseed has not changed past what it was at 9x multi. Go and try it.

Why would Sammi say to overclock Phenom, "drop the NB multi"?

You see, inspect this:
Messing around I could get 3.12GHz on Northbridge speed by raising the NB multi through BIOS at very low volts..what?
However, when I tried coventional 9x multiplier (stock) CPU overclocking with max volts, it failed bootup past NB 2GHz-2.2GHz repeatedly, even in AOD.. ?
OK.
So I drop the NB multi to 5x and then overclock CPU and can reach +500MHz CPU speed beyond what I can with a 2GHz NB.. !?

Why would this happen if the NB could reach 3.2GHz NB on stock volts stable?

Eggsactly. You get what I mean.

Remember I've been saying I cannot OC Phenom anymore past 2.5GHz? Well, here's the crack.
I gave it one last shot before delivering it. THis time once again, I managed 2.84GHz with 5x NB multi (screenshots/CPU-Z files not uploaded yet).
As I mentioned very early at the beginning: The problem and bottleneck with Phenom is NB speed. Drop the NB multi to OC the CPU. Above 2GHz NB will start erroring randomly. At the start when your Phenom is new you may get upto 2.3GHz on NB speed at 9x multi, but soon if you keep pushing to this, this is going to stop you're oc on CPU MHz and lower it. Soon you won't be able to oc high CPU/NB MHz again unless you drop the NB multi/MHz to keep it below 2GHz at all times and then try and oc the CPU.

The only way to overclock the Northbridge so far with locked edition Phenoms is to leave the NB at 9x multi and oc high on HT ref. Then it'll continue going up in-sync and you see some very good gains sometimes past 100% CPU MHz scaling, which is due to the IMC/Mem MHz rising alongside the CPU MHz.
Hope it helps clarify much confusion. Thanks.

[Daveburt714]

@KTE.. So if I read that correctly, MSI and Asus, at least, are admitting that the boards actually need a revision because there is no Bios fix that will correct the problem... And that article was dated just a couple days ago!!

I hope I misinterpreted that, but I don't think I did...

I wonder if Gigabyte is in the same boat....?

[KTE]

Yep, that's true. Many problems are board hardware related rather than BIOS is what they are saying. That article is dated 23rd December and that's official news. I know new iterations are in the works but I didn't think they were so soon.

I don't know about the new iteration release situation with Gigabyte DS4/DS5/DQ6 boards but their 790FX boards are better for you than both ASUS and MSI as of present, judging by support/compatibility issues. Far better, even the BIOS and overclocking.

But I wouldn't buy a board yet. Wait for January 1-7 for them and maybe some more by other MFGs.

[Daveburt714]

Hmmm... That kinda sucks since the board is already on it's way!

I did check the companies return policy and they offer a 1 year replacement/repair policy.... Not sure how ethical it is, but I may just use it for a few months and then return it. Honestly, I don't think I'll feel too bad since they sent me a faulty product in the 1st place....

Thanks for the heads up KTE!

[Emperor]

Hi fellas, had to leave for some time. Hope your all enjoying your holidays, best wishes.

BIOS mainly but I try AOD too if it works.

P-States won't be that for CPU, it won't boot at that if enabled. That looks like NB VID to me. CPU VID is locked and not the VCore.

How do you know if it was the BIOS which reduced it? Did you start AOD and ever hit Apply? If you did, then that is what reduced it, since you have to choose max volts in AOD to get a high actual VCore value.
Yeah same here. Mine used to be 1.64V-ish and then suddenly went down to 1.458V. What's your CPU VID? When that changes to 1.20VID from 1.25VID, the max volts you can have also falls according to my testing.
AOD.

i mean the cpu vid is 040. i noticed there is some issue with 235HTT. on and off the system hangs at different times despite being on the same load. errata ? 235x11. i've tried vid = 40 ,41 and 42. strangely it doesnt like odd numbers.

not sure waht u meant by wont boot when pstates enabled. for me, the bios always uses the Pstates to adjust the multipliers for nb and cpu, voltage for both cpu and NB.

i know it isnt AOD because for this system i never installed AOD because i felt that it was too problematic for me.

my settings for 235x11.

BIOS CPU volts =1.545

P states=
cpu fid = 6 (x11)
cpu did = 0 (1 )
cpu vid = 042

nbfid = 04 (x8)
nb did = 0 (1)
nb vid = 042

another thing is, i tried to set a NB multi of 7 by using NB fid = 10(14) and nb did = 1 (2) so the resulting multi is 7. but it doesnt work at all. but nevermind.


BTW do you know what are the voltages for the various VIDs??

[KTE]

Try it Dave and see if its better with the upcoming BIOS. I heard it will have most issues ironed out.

[KTE]

i mean the cpu vid is 040. i noticed there is some issue with 235HTT. on and off the system hangs at different times despite being on the same load. errata ?

Nope, oc instability.

i've tried vid = 40 ,41 and 42. strangely it doesnt like odd numbers.

I've no idea how you've manged to manipulate CPU VID. It should be locked and it was for mine. It controls core amps AFAIK, and high value is lower.

Stock 9500 is 28 = 1.25V/1.2V
Lower value is higher VID.

Don't have the exact mapping with me here since I can't access the HD they're on. 29 is something like 1.168.

i know it isnt AOD because for this system i never installed AOD because i felt that it was too problematic for me.

The new BETA is very good.

my settings for 235x11.

BIOS CPU volts =1.545

P states=
cpu fid = 6 (x11)
cpu did = 0 (1 )
cpu vid = 042

nbfid = 04 (x8)
nb did = 0 (1)
nb vid = 042

Firstly, I highly doubt you need the highest volts for 2.5GHz. You need to drop NB multi to 5x (00) and then start oc'ing. Bootup at 225HT ref. It should get to ~2.6GHz through AOD with 1.4V stable.

Try 240 HT ref. through the latest AOD and if that doesn't work, then through the BIOS whilst having low NB and HT.

Secondly, your NB/CPU DID options are different to mine. Mine were 1 and 2, rather than 0 and 1.

Thirdly, you're running lower VIDs than stock on both NB and CPU. That will give you bad stability even if it boots. Put them back to stock or lower if you can, i.e. CPU 28 and NB 36, or lower because that's higher VID=better possible OC.

another thing is, i tried to set a NB multi of 7 by using NB fid = 10(14) and nb did = 1 (2) so the resulting multi is 7. but it doesnt work at all. but nevermind.

You'll have to expereiment why. I don't have it to try at all. Maybe it's because of the Mem speed.

[Emperor]

Nope, oc instability.

I've no idea how you've manged to manipulate CPU VID. It should be locked and it was for mine. It controls core amps AFAIK, and high value is lower.

Stock 9500 is 28 = 1.25V/1.2V
Lower value is higher VID.

Don't have the exact mapping with me here since I can't access the HD they're on. 29 is something like 1.168.

The new BETA is very good.

Firstly, I highly doubt you need the highest volts for 2.5GHz. You need to drop NB multi to 5x (00) and then start oc'ing. Bootup at 225HT ref. It should get to ~2.6GHz through AOD with 1.4V stable.

Try 240 HT ref. through the latest AOD and if that doesn't work, then through the BIOS whilst having low NB and HT.

Secondly, your NB/CPU DID options are different to mine. Mine were 1 and 2, rather than 0 and 1.

Thirdly, you're running lower VIDs than stock on both NB and CPU. That will give you bad stability even if it boots. Put them back to stock or lower if you can, i.e. CPU 28 and NB 36, or lower because that's higher VID=better possible OC.

You'll have to expereiment why. I don't have it to try at all. Maybe it's because of the Mem speed.

I am not using the official msi bios. it;s closer to a custom 1.

so this what i will be trying.

cpud fid 4
cpu vid 28
cpu did 1

nb fid 01
nb vid 36
nb did 1

HT 5x
HTT = 225
cpu voltage is 1.537