AMD FX OC'ers Club

[Super XP]

Darn ASUS, I want an official bios from them that will help enable the Crosshair V to let us disable one core per module. Any luck with this from other mobo companies?

My 8-Core @ 4.40GHz is still puring like a cat.

 

[AphexDreamer]

Ok so tried some new RAM same issues.

Lowered the CPU clocks to default with new RAM same issues.

Issue being, windows just cuts out no BSOD no nothing just shuts off all fans keep running.

Immediate restart and it does the same thing right when Windows loading logo is done. Wait a few minutes and restart and it seems to be able to boot into windows where it just fails again a few minutes later.

A few signs of relief. Loaded Optimized defaults in BIOS and system has been up and running for ten minutes now.

 

[nemesis.ie]

Ahh I get that message every so often!

I think it's a problem with fx only takes a restart to sort out for me, so there isn't even a problem I'm guessing, just false positive.

More likely motherboard (BIOS) I think - I get that too from time to time and I'm running a 960T at the moment.

Crosshair V motherboard.

It's well worth running some extensive RAM tests when you first get a new rig IMO.

 

[nemesis.ie]

Ok so tried some new RAM same issues.

A few signs of relief. Loaded Optimized defaults in BIOS and system has been up and running for ten minutes now.

What board/BIOS are you running?

 

[eidairaman1]

Asus Crosshair V.

check the signature

 

[nemesis.ie]

Asus Crosshair V.

check the signature

Likewise. I think I've seen more issues (stock setting at the moment) with the new 1402 BIOS which ironically claims to "improve system stability".

It may be worth checking out the EPU and other power settings, I plan to have a look there myself and maybe increase them a little.

 

[eidairaman1]

Likewise. I think I've seen more issues with the new 1402 BIOS which ironically claims to "improve system stability".

put it back to the previous bios version or get another motherboard...

 

[AphexDreamer]

What board/BIOS are you running?

I was having windows shut offs with BIOS 1301 usually when coming back from Sleep and tried everything there and it was bad...Note I never ever got a BSOD and never got any beeps only the CPU LED once and No boot device.


Then out of an act of desperation I updated to the latest BIOS 1402. Still had pretty much the same issues. Then once it booted into windows and when I checked CPU'z it was because it loaded the default settings.

So I turned AMP off and other power saving features left everything on auto and have been running and stress testing the CPU flawlessly at 3.3Ghz.

It seems like its my CPU. It is so odd to randomly just stop working? Perhaps it was damaged from 1.47 volts at 4.5Ghz?

Anyways I'm just happy to be using my PC and when I get back from Greece (visit family) I will look into slowly trying to overclock it again.

Is 1.47 to high? It would just barely hit 60C on full load. I'll try to aim for the highest speed I can get with the lowest volts next time around I suppose.

 

[nemesis.ie]

put it back to the previous bios version or get another motherboard...

 

[Mathragh]

I also occasionally experience random failures after sleep, or when its supposed to go into sleep state. After such failures i need to unplug the power, after which it reports that "the overclock failed". I guess it has something to do with this platform.

Only occurs once or twice every month, so it doesnt really bug me, but still.

I'm currently using the pre-last bios version, since there was a new one released a few weeks ago. Perhaps the newest bios solves these problems(it does say that it improves stability), but I doubt it.

 

[pantherx12]

I think I've damaged my FX don't seem to be able to overclock over 4ghz anymore ....


I got 4.5ghz @ 1.475 v working kinda ( passed intelburntestv2) but it fails at prime ( HW failure on several cores)

This sucks considering what my chip used to be able to do!


Maybe it was that 5ghz run I done in the winter .......

 

[AphexDreamer]

I think I've damaged my FX don't seem to be able to overclock over 4ghz anymore ....


I got 4.5ghz @ 1.475 v working kinda ( passed intelburntestv2) but it fails at prime ( HW failure on several cores)

This sucks considering what my chip used to be able to do!


Maybe it was that 5ghz run I done in the winter .......

I've slowly been trying to see if mine will want to go back up to 4.5 as well. I'm currently at 3.8Ghz with 1.29 volts so the lower volts is kinda cool in its own way. But I used to run 4.5Ghz with 1.47 volts all the time till I had random shut offs.

 

[cadaveca]

I also occasionally experience random failures after sleep, or when its supposed to go into sleep state. After such failures i need to unplug the power, after which it reports that "the overclock failed". I guess it has something to do with this platform.

Only occurs once or twice every month, so it doesnt really bug me, but still.

I'm currently using the pre-last bios version, since there was a new one released a few weeks ago. Perhaps the newest bios solves these problems(it does say that it improves stability), but I doubt it.

This is probably your VGA, actually. If you look at the release notes for the most recent AMD driver, you'll find reference to this problem. FOr me, it also occurs with 6-series cards.


As mentioned with AMD, it's not jsut the driver, it's also, in some isntances, an out-dated BIOS, but AMD definitely plays a role in it as well. Basically the driver has issues restarting ZERO-CORE cards. In my instance, it has issues restarting the secondary cards, according to BSOD logs.

if you get a BSOD, 0x000009f, then this is what's going on.

I can acutally make this problem occur myself, by overclocking the card just past stability. The system will hang, and it will reboot saying "overclock failed". Why a VGA BSOD, screws with the mobo BIOS, I am not 100% sure, but it seems that AMD is at least partially aware of the problem, given those release notes.

 

[pantherx12]

I've slowly been trying to see if mine will want to go back up to 4.5 as well. I'm currently at 3.8Ghz with 1.29 volts so the lower volts is kinda cool in its own way. But I used to run 4.5Ghz with 1.47 volts all the time till I had random shut offs.

Same here, I was running 4.6 24/7 around the same voltage if not less but then some hot dats came round so I went down to 4ghz to keep temperatures under 60c since then I've not been able to go back.

Even 4.4, 4.3 etc it's not stable at sensible voltages, hell even with high voltages they can trip up.

 

[Mathragh]

I was already wondering why there was such a big gap in voltages between the people running at around 4,4GHz and those above. With my chip, it seemed to be the case that when the core temperatures went over a certain point(around 60 degrees in my case) the chip all of a sudden needed a lot more voltage to keep stable further increasing the temperatures.

in my experience, lowering the voltage actually can improve stability of bulldozer at times, since this wil lower the temperatures. This way i could lower the voltage from 1,41V to just 1,38 while overclocking it only 50MHz less. It seems like you guys(or atleast panther) choose the high temp/high voltage route, whether consciously or not, and might in this way applied more voltage then needed.

Ofc, the difference between chips and their respective needed voltages might also be the explanation.


@cadaveca: thanks! This may very well solve the problem also. Now you mention it, I did not seem to have this problem when I still had 2 8800GTX's, and I also didn't notice it since i upgraded drivers(although thas wasnt really long ago, so its perhaps too early to totally rule out).

 

[pantherx12]

I was already wondering why there was such a big gap in voltages between the people running at around 4,4GHz and those above. With my chip, it seemed to be the case that when the core temperatures went over a certain point(around 60 degrees in my case) the chip all of a sudden needed a lot more voltage to keep stable further increasing the temperatures.

in my experience, lowering the voltage actually can improve stability of bulldozer at times, since this wil lower the temperatures. This way i could lower the voltage from 1,41V to just 1,38 while overclocking it only 50MHz less. It seems like you guys(or atleast panther) choose the high temp/high voltage route, whether consciously or not, and might in this way applied more voltage then needed.

Ofc, the difference between chips and their respective needed voltages might also be the explanation.


@cadaveca: thanks! This may very well solve the problem also. Now you mention it, I did not seem to have this problem when I still had 2 8800GTX's, and I also didn't notice it since i upgraded drivers(although thas wasnt really long ago, so its perhaps too early to totally rule out).

I was the dude on this forum pestering everyone to use lower voltages my cpu has just degraded. (requiring more volts)

 

[AphexDreamer]

Yeah I recall reading that Bulldozer scaled better with lower temps but alas I'm on air and in Texas so I fell victim to the need more volts due to excess temps case.

 

[Super XP]

My OC is still running great with low temps. I plan on selling this one day to fund a Piledriver CPU providing its worth an upgrade.

 

[Irony]

@ SuperXP: I would buy it from you if you upgrade in the next 6 months lol.

 

[Mathragh]

My OC is still running great with low temps. I plan on selling this one day to fund a Piledriver CPU providing its worth an upgrade.

If Piledriver performs atleast 15% better then my current chip(whether from improved IPC, or just higher clocks with the same powerdraw/heat), I'm planning on doing the same thing^^

Edit: Well, lets not fool myself here, I'd probably be happy with 10% aswell lol.

 

[eidairaman1]

well for it to have clock meshing should say something about it

 

[Mathragh]

well for it to have clock meshing should say something about it

Yep, and the IPC improvement of the current trinity chips compared to bulldozer is also promising atleast "some" improvement. Also, I guess they didnt implement an improved stepping for the current gen of bulldozers so they can show a bigger difference between the upcoming vishera based FX processors and the current ones running fairly unoptimised silicon. Atleast, if the last gen of phenoms and their incremental speed increase every couple of months is anything to go by they should've atleast been able to improve on current BD chips a bit now in the same way.

 

[Inceptor]

Yep, and the IPC improvement of the current trinity chips compared to bulldozer is also promising atleast "some" improvement. Also, I guess they didnt implement an improved stepping for the current gen of bulldozers so they can show a bigger difference between the upcoming vishera based FX processors and the current ones running fairly unoptimised silicon. Atleast, if the last gen of phenoms and their incremental speed increase every couple of months is anything to go by they should've atleast been able to improve on current BD chips a bit now in the same way.

They might still do that, if the BD chips are not EOL when PD is released. Just to extend the brand lifetimes, especially if Piledriver is the end of their discreet cpu line. I think it all depends on fab resources.

 

[lindy]

if Piledriver is the end of their discreet cpu line.

I don't think it will be the end of the cpu line. It will however be the end of the AM3 socket.

 

[Super XP]

IPC improvement

It's been speculated many times before we heard about this Clock Mesh thingy, that Piledriver may perform in around 10% to 15%. But now with this clock mesh talk, new speculation now puts it at a possible 20% in most and 30% in some cases.

It will be quite interesting, but AMD needs to ensure they continue support for discrete CPU's unless they convert there entire line into discrete high end style Fusion chips.