LHC@home

We will upgrade the LHC@home web servers to the new BOINC server code with the "Bootstrap" theme on Thursday 7 December. The new style and layout can already be seen on the LHC@home development project..

During the intervention, from 09 UTC on Thursday, there may be intermittent availability of the LHC@home servers, so BOINC clients may back off and try to upload data later.

The new appearance of the site is nice ,
but maybe in order to improve the reading and a better usability :
1°) can you try to right-align the numbers on the status server page (or medium-align for time execution) ?
2°) can you put a direct link to vlhc-dev site and symmetrically a link to this site from the vlhc-dev site (easier to browse between them) ?

I agree to what Philippe is saying. From an optical point of view, some fine-tuning still needs to be done.

However, what makes me worry much more: in the personal settings on the WebPage, the maximum number of tasks that can be set for a given project now is 8 - contrast to 24 as before. Why so?
There are definitely cases where this turns out being very impractical. I would guess that many crunchers will be most unhappy with this change.

Thanks for your feedback.

Regarding the limit to number of jobs/cpus, we have applied the standard BOINC web code, with a checkbox from 1 to 8 and "Unlimited". In my preferences, I have the latter, but again it boils down to how much your host can absorb.

Nils,

we have to thank for the web-update.
The big Computers with more than 8 Cores can run 8 CMS, 8 ATLAS, 8 Theory AND 8 LHCb ;-))

Nils,

we have to thank for the web-update.
The big Computers with more than 8 Cores can run 8 CMS, 8 ATLAS, 8 Theory AND 8 LHCb ;-))

The maximum of 8 is for all sub-projects together. When you have 8 selected in your project preferences and 8 SixTrack In Progress,
no new tasks (SixTrack and VM-based projects) will load until you return a task.
So you have to select 'No Limit' and the number of tasks will be based on your local buffer preferences.

Which compounds the problem. The way these numbers work (downloading the "exact" number, not "up to" the specified number), I absolutely need more than 8 - preferably more than 24 - WUs.
4 core machine, 2 WU x 2 cores each -- not a big computer by any measure.

The underlying issue of course is the way LHC distributes work: the BOINC settings how much work should be buffered are ignored, and there is no reputation system removing "bad" volunteers. Instead a lot of manual fine-tuning is required.

On Fridays or with forseeable maintenance ahead I prime my system with new WUs. Since "max #" is interpreted as "download this exact number" and "do it asap" this works effortlessly without remoting in.

Nevertheless, I might still run out of work, because 24 ATLAS tasks might not be long enough. So, to make sure I get any additional work available on short notice, I set "max #" back to unlimited again when the queue is full. This way, once Sixtrack has spillover, I can get additional work for a total of ~48 queued tasks.

In contrast, if I were to set unlimited and no Sixtrack is available, my system chews 2 WUs ATLAS and won't buffer or get more until those have been reported. Those next 2 WUs still have to be downloaded of course. Several minutes are wasted. -- 2 active and 2 buffered should be feasible in unlimited. As a workaround limits should be increased, perhaps in larger increments 1, 2, 4, 8, 16 ... so the drop-down is manageable.

If I were to set 24 permanently and no Sixtrack is available, my system restocks 24 ATLAS and only checks for work once every hour. Hence when Sixtrack shows up, it is immediately distributed to other volunteers. Many of those systems have never (at least not in recent weeks) returned a valid result. Once it runs out of work my system is again wasting time waiting for any aborted or failed Sixtrack to be redistributed. Additionally the overall turn-over time for Sixtrack increases (by at LEAST one week) as well. -- The number of WUs a "bad" or new(!) volunteer is able to receive should be limited by reputation.

These are my observations over the last few weeks. Although ATLAS provides WUs very consistently, there have been times when I would have appreciated Sixtrack as a backup. I have looked into other applications besides ATLAS and Sixtrack, but am detered by the likelihood of failing a task that runs 24+ hours, depends on an active internet connection and likely can't be stopped/continued. I'm disconnected by my provider every 24 hours.

On topic: readability issues with the new design as well, especially on mobile. I would appreciate the return of the static menu which put the most frequently used pages within immediate reach.

Further feedback:

-- Message board: font for quotes too large.

-- Menu drop-down "Project" awkwardly left empty when not logged in. To recreate log out, click on Project in menu bar, drop-down is empty. Chrome 62.0.

A further detail :
The word "LHC@home" is not correctly translated in the french , catalan and japanese versions, in the first item of the home page which describes the project .
Only appears " %1 " in the item title.

Curiously , the catalan and japanese versions are mixed with french version...

Further feedback:

-- Message board: font for quotes too large....

+1

Further feedback:

-- Message board: font for quotes too large....

+1

Yes, I've noticed that on -dev as well, but it seems not always -- unless my memory's faulty...

---

Please repeat the title of the page in a headline ("All tasks for computer ..."). These were dropped with the update. Other projects combine headlines and the new design.

I suggest a different color be used for the text for the following links so they are easier to see:

1) "Show names" for tasks on the user tasks pages
2) "Switch View" on the user preferences pages