For your info:
The WUs of Rosetta usually work faster in the first 80% and then slow down for the last 20%. The science that is done changes in the alst 20% and needs more time.
The time also heavily depends on the type of WU. Right now, most WUs contain a string like "len10" which means lenght 10. But there are also WUs "len20" and and very few "len30". That means, they take twice or thrice as much time as the len10 ones. Thus, they'll appear stuck on 1% for a longer time. As far as the "topology sample" WUs are concerned, I guess they're all the same lenght, but they don't really tell.
On my X2-4400+ some of the len20 need as long as an 1:24h to pass to the 10% mark. Then again, they appear to be stuck at 80%, but they aren't. I haven't heard or read about any WUs stuck at any other mark than the 1% one. So, once you're past that, you should be safe.
I know, it's a difficult decision, but recently a lot of WUs get killed too early, altough they weren't stuck but just needed more time. Just check the name of the WU and the "lenXX" string. That can give you a rough estimation.
I just asked on the R@H forums if the gfx would actually tell you if a WU is stuck by not moving and whiggling any more. Stay tuned, I'll further investigate into that. Should anyone of you have a stuck WU, please open the visuals and see what it does.
🙂
for BMC
