Catalyst 10.8 beta 5 (8.76.2) Discussion thread

IBT does fine for say an hour.

 

4 hours of Prime 95 is not to be considered stable, its nowhere near enough.

 

Stock CPU's should be deemed faulty if they error at stock settings. When I say stock settings I mean all correct voltages and timings for ram etc etc are set and verified to be correct by manufacturers specification.

 

So no chance for a mobile version?

 

CCC's SSAA and AF manual settings are broken for DX9 mode games. I have tested with WoW and other DX9 games. 10.7 and 10.7 opengl beta work fine. In-game AA always work fine.

 

An hour :3eyes: A 10 minute pass should be enough to determine stability.

 

10.8 b5 totally crashed when I opened youtube. Uninstalled and installed 10.7 preview instead.

 

10.8b Crashes when It uses DXVA.

 

Using MPC x64 with EVR Sync here, no crashes.

 

Just found out a issue with MSI Afterburner. If I play DIRT2 with it on in the background, I have these microstutters in game. But If I close down MSI Afterburner, and specifically the OSD, the game runs fine with no problems. So you guys who have problems in games, try disabling msi afterburner to see if it goes away. If it does, we can tell unwinder about it? I really need MSI afterburner so closing it down fully is not the perfect solution for me, but atleast I know where to start?

 

Just disable anything being logged in AB OSD for temperatures etc. I have to keep it open or CCC overrides my custom fan profiles and you don't want AB constantly bothering the GPU about temperatures and clocks anyway.. especially with ATI drivers lol.

 

As for the side issue everyone is talking about here, 2 hours of a Linpacker (Intel Burn Test, LinX, OCCT Linpack) is the same as running 10 hours of Prime95

But back on topic, I've been running games, so far everything checks out fine, I've yet to have any problems with anything so far

I've not tried the video enhancements with internet videos as I've heard this is causing more crashes then good, and I hate crashing my PC needlessly so I'll wait till I hear this is more stable

 

Ibt alone doesn't insure stability but a mix of tests. on my old E8500 i was able to pass 100 passes of ibt yet i would still fail blend orthos/prime 95. I really do think doing 24 hours of blend and 20 passes is good enough

back on topic. do these drivers come with the optimizations of 10.7, 10.7 beta?

 

Likewise, no crashes here.

I do wonder if some of the more convoluted ways people try to upgrade drivers causes most of the issues. I know it has been said before but like some other here I never have to use Drive Sweeper etc. I just uninstall the old drivers, quick reg clean after and reboot. Install new drivers. Generally, for me, no issues, install after install after install....

 

I saw someone mention somewhere that they closed firefox and that helped with an install problem, and i thought to myself, holy ****, do you guys install drivers with all kinds of crap running in the background.

I'm seeing people running ati tools this tools that, rivatuner, afterburner, drivercleaner, ram optimizers, bootup helper, notepad booster, solitaire enhancer, gpu-z, z-my-gpu, bios editors...
The other day under 10.6 i had 2 instances of GPU-Z running and loaded BC2, it froze on the loading screen, i then terminated the process, closed GPU-Z and re-ran and it booted fine.

I usually, close running programs, disable my second monitor, unplug RJ45, disable AV, remove any desktop backgrounds, essentially minimize any additional work the system has to do while installing drivers. Have had a good track record so far.

Yes, there are issues here and there, but i'm curious how much peoples habits like hallryu says comes into play. we should start a thread and post screenshots of how people install drivers.

 

8.76.2 here and no problems at all

 

All you guys on here arguing about what it takes to be stable should look up what the common consensus is.

In all actuality, no system can be deemed truly stable until it's passed a 24 hour run of Prime95, and any STOCK/DEFAULT SETTINGS processor that fails Prime95 at all, even in a 24 hour run - is defective.

I can't tell you how many times and how many computers I've run IBT for 10 passes and passed, only to fail Prime95 in the first 3-4 hours.
IBT is excellent at pushing the processor to new levels of heat - nothing heats up the processor like IBT, but heat isn't the end-all for an overclock, and most certainly isn't the only prerequisite for failing an overclock.

Prime95, by design, finds Mersenne Prime numbers, which puts the CPU through algorithms made to push the processor's abilities...and is the best stability benchmark on the planet for determining a processor and RAM's ability to provide a stable environment, and it's NOT like running any other bench/burn-in...not even close.

Don't think for a second that anything less than 12 hours of Prime95 is stable. In fact, I challenge any and all of you that haven't put the time in with P95, to do so now - and post your honest results in a thread in the correct forum... I bet more of you will fail P95 in a 24 hour period than you think.



Now, with that said, if a GPU can run Furmark for 30 minutes straight with maxed out settings, then your video card is ruled out as the problem, and so it *obviously* lies somewhere else.


I have to digress - I went way off topic in the wrong forum for this discussion.

These drivers work excellent! Sniper, PoP, AC2, AvP, ME2 and Singularity are all working excellent - no drawbacks to any of them with these drivers.

Psychlone

 

Easy for you to say. Hard to sleep with a computer that has a bunch of noisy fans going all the time.

CPU is not OC'd at the moment and all other clocks tested at stock when crashing ensued. Setting OC on the GPU made no difference at all.

Currently on 10.7 Preview and so far only once has BC2 dropped me out. No errors nothing. Just *bink* right to the desktop. I would think if it were hardware related there would be a message from windows saying that something went wrong.

 

has anyone tested these with operation arrowhead?

 

FYI, you do know that Linpackers have to be 100% accurate as well or they fail the test, if the results of the calculations are faulty, it will fail as a result

OCCT Linpack is a great easy way to show if it passes or fails, even if it doesn't BSoD or restart your computer

And yes, Linpackers are -JUST- as accurate as Prime95, plus work your system harder, Prime95 only really stresses your computer every 3 hours into the benchmark, and at that for only 35 minutes, Linpack from the start will max stress your CPU/Ram for calculations

Both are fine tests, but if you pass like 2 hours of linpack and fail 8 hours of prime, you were never truely stable

2-3 hours of Linpack is -FAR- more stress then your PC will ever be subjected too, if it passes that, its pretty much guaranteed to be called stable

As for Prime, I wouldn't call prime stable unless it passed 6 hours of tests

Also FYI, the reason Linpackers find errors sooner then prime is cause linpackers by nature use ALL available ram, which gives the computer a pounding, Prime95 will never use all available ram for a test

EDIT: FYI, 10 passes of LinX is only like 22minutes of testing, thought you should know

EDIT2: I just checked, 10 passes of Intel burn test is only 12 minutes of testing, I'd hardly call that a good stable CPU test