Gigabyte and ACC issues

[NCspecV81]

anyone here have issues applying ACC on their gigabyte board with Phenom II x4's? I can't make it into windows regardless if I've adjusted them or not with ACC enabled.

[chew*]

Problem is according to AMD phenom II was not / is not designed to work with ACC and I quote there is no benefit to running ACC on phenom II......

Of course we do know it unlocks chips but it obviously does diff things to the chips now that it was not doing to phenom I. Other than for unlocking chips they probably should have done away with it altogether on ddr III boards due to the fact PH I is not supported..........My guess is ACC function is not even part of the spec for AM3.

[NCspecV81]

I've actually had good results from enabling it on other boards. Most recently the MSI GD-70.

[chew*]

I've actually had good results from enabling it on other boards. Most recently the MSI GD-70.

Regardless it's unsupported and not part of "spec" however i have seen it help in rare cases on mobo's that tend to have a fair amount of droop......Ive said this before but if it does help than chances are your missing something in your manual tuning that ACC is "fixing"

I've played with a few times in all types of conditions, its actually given me worse results with chips that are natively 4 core by model designation, if i had to guess i used at least 15 various boards testing it.

I think if anything it has a placebo effect on users. People are so convinced it actually does something that they truly think it helps when in actuallity it's doing nothing.

[BeepBeep2]

Problem is according to AMD phenom II was not / is not designed to work with ACC and I quote there is no benefit to running ACC on phenom II......

Of course we do know it unlocks chips but it obviously does diff things to the chips now that it was not doing to phenom I. Other than for unlocking chips they probably should have done away with it altogether on ddr III boards due to the fact PH I is not supported..........My guess is ACC function is not even part of the spec for AM3.

In a guide from AMD for "tweaking" Phenom II's said that you can use ACC to tweak weak cores to up total clock frequency. I need to look it up again...

[chew*]

In a guide from AMD for "tweaking" Phenom II's said that you can use ACC to tweak weak cores to up total clock frequency. I need to look it up again...

That was it's purpose in PH I when ocing the chips at all was a strain.

[BeepBeep2]

No there was a guide made after macci, sampsa, and SF3D gathered at the XS Las Vegas event and it was an AMD .pdf but it's nowhere to be found on their site. I was reading it just a few days ago and should have bookmarked it. I know it showed that you can use ACC "to reach higher frequencies on the weaker cores of your Phenom II processor"

[Lastviking]

No there was a guide made after macci, sampsa, and SF3D gathered at the XS Las Vegas event and it was an AMD .pdf but it's nowhere to be found on their site. I was reading it just a few days ago and should have bookmarked it. I know it showed that you can use ACC "to reach higher frequencies on the weaker cores of your Phenom II processor"

+1 i have seen that to... but i have not any luck with it...maybe on some boards it works well.

[BeepBeep2]

Here is is:

Completely inaccessible from a search engine...I looked for it about 6 times then gave up. It was in my history from Thursday on my netbook.

http://game.amd.com/us-en/content/pd...ning_Guide.pdf

In case of the AMD Phenom II CPU, the “Advanced Clock Calibration” (ACC) feature can be used to
improve the tuning margin of specific CPU cores. Typically the CPU core that has the lowest tuning
margin by default is the likely candidate to see benefit from ACC-feature.
ACC can be enabled thru the AMD OverDrive™ utility or from the BIOS menu. Select the “Per Core”
option and set the value of ACC to “+2%” for the CPU Core that has the least amount of headroom. Set
the rest of the CPU cores to “0%” (default value).
Note that the system needs to be restarted before the new settings will be applied. Restarting the
system is required each time the value(s) of ACC are adjusted.

[BeepBeep2]

Adding to the OP, I'm running the board in my sig right now with ACC on, Per Core: Core 0 +2, Core 1 0%, Core 2 +2, Core 3 +2.

Booted into windows fine, Phenom II 965 BE C3.

EDIT:

I'd also like to know why my idle temps at 4016 Mhz 1.50v are 33c now, 6c lower than the 39-41c from before. I'm more than slightly confused.

[crazydiamond]

i know a few other guys around here also said their phenom II oc's benefited from tuning ACC

[Mavhenz]

http://game.amd.com/us-en/content/pd...ning_Guide.pdf

This is absolutely GOLD!
Don't have time to read it now, but I will!
That thing is a Noob question shield!

Mavhenz --> happy!

[BeepBeep2]

forget it, ACC normal mode doesnt even set...it reverts to "disabled"

hybrid I lose core temps and it has no effect...just for core unlocking.

[NCspecV81]

yeah I can't get ACC to work at all. I'm bummed.

[FlanK3r]

lol, with ACC u got better temps?!

[NCspecV81]

put the 955 on my gigabyte...now lets try this!

[BeepBeep2]

ACC doesn't even turn on...*frets*

Hybrid does nothing but turn my temp sensor off...normal...does nothing and wont turn on.

[BeepBeep2]

Flashed F8 on my board and I'm experimenting...first I saw hope...no more BSOD's and individual cores failing in OCCT...

Now, after messing with "Per Core" values I'm getting random IRQL_NOT_LESS_OR_EQUAL BSOD's and DRIVER_IRQL_NOT_LES....with an occasional "An attempt to execute non-executable memory blah blah"

My theory is that "set in stone" optomized values are set for each core as they leave the factory however sometimes they can be a smidge off. Setting +2 on my first and last of the four cores definately helped stability. +4 just tanked into the ground...core one took it but setting +4 on core four ended in me not being able to boot windows.

[sinar]

.i think if anything it has a placebo effect on users.

qft

Imo, same thing with 955 C3 compared to 965 C3(which binned higher by manucftr).

[BeepBeep2]

Well, Macci is the one that stated a positive ACC value can push more Mhz out of weak cores on a Phenom II CPU...;-)

[sinar]

Same settings with ACC disabled

[NCspecV81]

which gigabyte board is that with acc enabled? I still can't get it to work.

[BeepBeep2]

which gigabyte board is that with acc enabled? I still can't get it to work.

Gigabyte GA-MA790X-UD4P with F8 bios for me...

[Kei]

Regardless it's unsupported and not part of "spec" however i have seen it help in rare cases on mobo's that tend to have a fair amount of droop......Ive said this before but if it does help than chances are your missing something in your manual tuning that ACC is "fixing"

I've played with a few times in all types of conditions, its actually given me worse results with chips that are natively 4 core by model designation, if i had to guess i used at least 15 various boards testing it.

I think if anything it has a placebo effect on users. People are so convinced it actually does something that they truly think it helps when in actuallity it's doing nothing.

I was in the same thought as you are Chew regarding ACC and PII's though I'd heard mumblings on the net that some people had found a little help from it. I decided to try it out myself after I also came across that guide specifically mentioning Phenom II and ACC. The guide states that you are supposed to use the 'Per Core' option not auto or anything else even given an mention.

I finally decided to try it out myself on a core by core basis. I loaded up the system with stock 1.35v (PII 955), and clocked each core independantly using K10stat. For stressing I decided to use LinX with a quick 5 minute run to start out. I settled on 3.8Ghz for the core speed that I was testing which was no problems for Cores 0 & 1. Core 3 made it to 3.7Ghz and a few seconds on 3.8, Core 2 didn't make it at all at 3.8Ghz so I started there. I ran the test a few times and never made it anywhere (as in less than 5 seconds).

I went into the bios and left everything the same, but enabled ACC Per Core with a +2 on Core 2 only. I went in and ran the test again and actually made it somewhere which was surprising, but still failed in 30 seconds. Ran the test again with +4, and made it about 1 minute or more. Went the other way with ACC (like I was used to wiht Phenom I) and tried out -2%......ran for the full 5 minutes without a problem. I ran it again without the ACC and it failed just as fast as normal, added the ACC -2% and it ran just fine.

I did the samething for Core 3 and ran the test for 5 minutes, no problems at all. I tried to run it for 10 minutes on Core 2 but it failed at 7 minutes so I bumped the volts 1 click and it passed everything I threw at it after that regardless the length of time.

Without using ACC on those specific cores (after finding out exactly what the core needed individually) I wasn't able to run at those speeds on the stock voltage at all. It seems to me that somehow it does indeed help in that situation. Of course everyone will not have the same results nor the same % needed, some may not get any help at all.

It was interesting though to see that it actually did help even on a Phenom II.

Kei


Board for test wasx ASUS M4A78T-E with Phenom II 955 processor (haven't tried it much on my Gigabyte MA770T-UD3P yet except the -2% on Core 2 when running early tests which still helped)

[BeepBeep2]

I was in the same thought as you are Chew regarding ACC and PII's though I'd heard mumblings on the net that some people had found a little help from it. I decided to try it out myself after I also came across that guide specifically mentioning Phenom II and ACC. The guide states that you are supposed to use the 'Per Core' option not auto or anything else even given an mention.

I finally decided to try it out myself on a core by core basis. I loaded up the system with stock 1.35v (PII 955), and clocked each core independantly using K10stat. For stressing I decided to use LinX with a quick 5 minute run to start out. I settled on 3.8Ghz for the core speed that I was testing which was no problems for Cores 0 & 1. Core 3 made it to 3.7Ghz and a few seconds on 3.8, Core 2 didn't make it at all at 3.8Ghz so I started there. I ran the test a few times and never made it anywhere (as in less than 5 seconds).

I went into the bios and left everything the same, but enabled ACC Per Core with a +2 on Core 2 only. I went in and ran the test again and actually made it somewhere which was surprising, but still failed in 30 seconds. Ran the test again with +4, and made it about 1 minute or more. Went the other way with ACC (like I was used to wiht Phenom I) and tried out -2%......ran for the full 5 minutes without a problem. I ran it again without the ACC and it failed just as fast as normal, added the ACC -2% and it ran just fine.

I did the samething for Core 3 and ran the test for 5 minutes, no problems at all. I tried to run it for 10 minutes on Core 2 but it failed at 7 minutes so I bumped the volts 1 click and it passed everything I threw at it after that regardless the length of time.

Without using ACC on those specific cores (after finding out exactly what the core needed individually) I wasn't able to run at those speeds on the stock voltage at all. It seems to me that somehow it does indeed help in that situation. Of course everyone will not have the same results nor the same % needed, some may not get any help at all.

It was interesting though to see that it actually did help even on a Phenom II.

Kei


Board for test wasx ASUS M4A78T-E with Phenom II 955 processor (haven't tried it much on my Gigabyte MA770T-UD3P yet except the -2% on Core 2 when running early tests which still helped)

I'm getting the same type of results here...auto does complete squat, but if I set per core I gain some headroom. Even my best core seems to prefer -2%...I've got core 0 @ +2, core 1 @ -2, and cores 2/3 @ 0%.

I've gotten two cores tweaked, and now I'm moving on to the other two...it's time consuming but in the end it's worth the extra 50-75 to 100 Mhz