How I've fixed my CTD!

[StefX 0]

Hello,

I want to share my fix with you so I hope it will help some of you.

Firstly I did not have any issue before I've installed the new ATI drivers 9.7.

Than after installing the 9.7 I had the famous CTD Arma2.Exe error!

This happened when I was testing my mission in the editor with about 100 AIs. The mission is basically Russian forces doing an air raid on a US camp this involving a lot of heli and armour.

So when I had those CTD I first did the basic trouble shooting:

-uninstalling of 9.7 then did a driver cleaner

-installing previous drivers 9.6

- did a registry cleaner

This did not solve it! I was starting to be really pissed of about this and thinking "bis fix this now!" and looking desperately for the 1.3 patch! :mad:

Than I've tried something else, few weeks ago before this problem I've reduced the Vcore of my CPU to 1.2V (Intel Q9550) which is overcloked at 3.4.GHZ.

I did this to save some power and to reduce the temperature. When id did this I have tested with OCCT and played ARMA fine.

As I was adding more and more AI in my mission I tried to add some volt to the CPU "just in case off"

I’ve added 0.2v from the bios which provided a 1.21v minimum during load.

Then I have tested and I have now no more crash!

This brings me to some conclusion: If you have OC your CPU bear in mind that if you play a mission with a lot of AI your CPU will be well stressed. So much that even some tool that test the stability of you system might run fine BUT arma with lots of AI will failed!

my spec:

CPU: inter Q9550 @3.4GHZ

4 giga of RAM

Video: 4890 1GB

VISTA 64 bit SP2

Edited July 29, 2009 by StefX

[FraG_AU 10]

I think this relates to the fact that you did insufficient reliability testing of your OC! But yeah definately a good idea for some that are having issues. (Not having a go at you btw!)

Conclusion: Ensure 100% stability first! And always try stock everything before looking for solutions in software that could be explained by hardware running out of spec.

I'd hate to say that a high % of people who OC actually have very unstable systems they just dont stress adequately to see it fail.

LinX, HiperPi, P95, Furmark, OCCT, Memtest == Your best firend. I am yet to find something that will beat a 32M HiperPi (Ie one for all cores), 4hr Memtest, 25 pass LinX, 24hr P95 (usually in that order), and furmark to ensure stability.. OCCT is another great tool that gets better and better using some of the prementioned tools.

[StefX 0]

I am agree my system was not stable anought, but I did test it with OCCT and it was running fine... Everthing was fine even ARMA2 in multi player...

only ARMA2 single mission with lots of AI made it unstable.

Al least I discovered the ultimate OC tools!

[Eugefunk84 10]

If I remember from my VLSI/MMIC class in grad school, increasing the operating frequency of a chip requires you to increase the voltage to allow faster signal swing on the control logic lines. By overclocking and REDUCING voltage, you put yourself in a pretty precarious state.

When overclocking, you may or may not have to increase your voltage, but you certainly should never drop it.

[FraG_AU 10]

[quote name=Eugefunk84;1384662

When overclocking' date=' you may or may not have to increase your voltage, but you certainly should never drop it.[/quote]

Off topic, but this is definately not the case. Certainly most cases voltage increase is required, but for example the i7's, many can achieve decent overclocks (say 1000mhz over stock) while at lower then spec voltage.

Just remember processors are binned according to economics not their capability in most cases.. While you say a 2.6Ghz chip needs say 1.2v, there will be a variance (normal distribution) below that mark so most chips can probably get away with 1.15 or even 1.1 .. EG. I have a i7 d0 which can do 4.0ghz at "stock" volts and 3.6 at well below stock (and we are talking 100% stable)..

Anyway just being a smart arse but making a correction :)

[StefX 0]

If I remember from my VLSI/MMIC class in grad school, increasing the operating frequency of a chip requires you to increase the voltage to allow faster signal swing on the control logic lines. By overclocking and REDUCING voltage, you put yourself in a pretty precarious state.

When overclocking, you may or may not have to increase your voltage, but you certainly should never drop it.

Of course I did raise the default voltage :j:. What I've tried to do is after I have raised a good OC at 3.4ghz stable( 1.28v) I've tried to use the lowest (1.22v) voltage to save some power and make sure that my grandchild will see some white bears...

Everything was OK only ARMA2 single mission with lots of AI showed problem until I've put more volts

[apollo4war 10]

Ensure 100% stability first!

Yeaahh... that's the big problem. Since overclocking can be so easily done by almost everyone, there are many many system's out that can't run stable when fully utilized. i. e. when a new game is installed that can really use all the hardware power.

But 100% stability? When will that be reached? - When all is at normal clock's :)

or... the hard way: start P95 (or Everest System Stability Test on low priority) and keep it running, start OCCT Graphics Error Detection (with CPU stress still running), and (if you can) add some Disk I/O stress (with Everest :D).

Keep both running a day :eek: - if all is still running 24 hrs. later without any error, then you can say: coool.

keep it running for a week, and you can say: it's pretty stable. (but almost - not 100% stable)

Finally - after that, if any error appears, then you can say: i think there must be something wrong with my software, drivers or operating system.

...and never trust any 3DMark Score that only runs <3 times loop ;)

[StefX 0]

This is the hardcore test! but as long I can play arma I am fine now!

---------- Post added at 02:45 PM ---------- Previous post was at 02:31 PM ----------

Appolo, how much volt did you put when your reached 3.4ghz?

Edited July 29, 2009 by StefX

[apollo4war 10]

*gg* - yes, you can easily use arma2 for testing ;)

I never done 3,4 GHz. I started with FSB450, CPU at 1.3V and 1.325V FSB which give the Q9550 (factor 8.5) 3.825 GHz.

Higher clocks won't run stable with my CPU, and higher voltages results in >70°C while stressing (CPU is C1 stepping and water cooled), and voltages <~2.7 wont run stable. And this way i can run my DDR3 memory synced on 1800 MHz, which give up >1840. I've done stressing with ~2.8V without errors, and give it finally 1.3V to increase stability.

So i keep it running on FSB450, now for over a year without any failure.

Voltage settings are very different on any Mainboard, BIOS Version or CPU. Even if it is the same Q9550 C1, there can be other voltages needed. You should try the voltage settings on Auto in the Bios, and try it out. Anyway i think 1.25 is a good start for 3.4GHz (also keep FSB voltage low) - but Auto setting will be enough for that on most mainboards. When reducing CPU voltage you must do the hard way to be on safe side - but i think reducing it is the wrong way to safe energy ;) - better turn off your lights :D

PSU can also affect overclocking capabilities and voltages...

Which hardware do you own?

Edited August 1, 2009 by apollo4war

[StefX 0]

here are my spec:

Edited July 30, 2009 by StefX

[Razorman 10]

over 100mhz fsb overclock (see sig,stock/original fsb is/was 333mhz) & system runs superbly, been overclocking for years & love it, sure did need it too with Arma 2 ;)

[apollo4war 10]

here are my spec:

I've made a quick test with my system: setting CPU-Voltage to Auto and FSB to 400 like your's, my mobo set it to 1.15V. I keep Prime95 running in the background while working for about 3 hrs without failures.

This can't be compared directly to your Asus P5B Deluxe Board. You should keep your setting as long as it works stable ;) or search through inet for other forums for overclocking your mb.