Hi all,
it's 9 hours from now until the start of the event. If you haven't
registered yourself yet, now is a good time.
🙂
You can still register anytime even during the event, but the sooner the better.
Your credits will be counted automatically for yourself
and for the TeAm, provided that you joined the TeAm at
our page at the SETI@Home web site.
Regarding bunkering. From
https://www.seti-germany.de/Wow/64_en_FAQ.html —
SETI.Germany's Wow!-Event FAQ said:
Every Tuesday Seti@Home has a weekly planned downtime. During this time no WU can be downloaded or returned. To bridge this time you can adjust your bunker to a suitable time level in the Boinc Manager. A maximum working buffer (bunker) of 10 days is possible.
Actually the S@H server lets each client hold at most
300 tasks in progress
(edit: see post #53 for actual limits). This is far less than a day if you have graphics cards, or a server with lots of CPU cores.
I have a host with three fast GPUs. For now I run three client instances on it in order to have a queue of 900 instead of 300 tasks.
Initially I wanted to use client 0 to run 2 tasks on card 0, client 1 for 2 tasks on card 1, and client 2 for 2 tasks on card 2. I configured app_config.xml for 0.5 GPUs per task, and I added respective ignore_nvidia_dev entities to cc_config.xml. Unfortunately, the clients ignored GPUs but not the ones which I told them to. Therefore I changed configs so that each client is running 3 tasks now, 1 task per GPU per client. My cards are big enough that they don't bog down with 3 tasks per card. However, in the past I ran only 2 tasks per card which was already close to the throughput optimum.
