S5R3 Nearing Completion
Anonymous
16 Jul 2008 15:15:32 UTC
Topic 13679
(moderation:
This might be subject to change, but some preliminary info to end speculations here:
Most likely S5R4 will start with Applications very similar to the last Beta Apps, i.e. SSE2 App for MacOS Intel, switching x87/SSE App for Windows, switching G3/AltiVec App for MacOS PPC (which will remain the slowest one in comparison), and an App that switches SSE2/SSE/x87 Apps for Linux.
The next improvements targeted for the Apps are getting the Windows App compiled with MinGW, so gcc-specific improvements could be ported to Windows (SSE2 optimizations), and, more general, get the "weighted" Hough code up to the speed of the currently used "unweighted". As the main improvement there is the prefetching, I'd suspect systems with a faster memory interface (AMD) to be a little faster in that part at the beginning.
However before that we have to get S5R4 running at all, and an almost equally high priority is a major server code upgrade. I doubt that any major work on the App code will be done before that.
BM
S5R3 Nearing Completion
)
It currently looks like S5R4 will be in place next week.
We'll probably have a short downtime for the server upgrade, we don't know yet if we'll do this both in one rush at the same time.
BM
BM
The 32Bit system the old
)
The 32Bit system the old server is running on is currently at its limits [Edit]Various components struggle to get enough memory[/Edit]. One reason to upgrade software and hardware [Edit]This will also include an update of the DB structure for the new server-side software[/Edit].
It currently looks like the server will be down for the whole weekend to perform these, and we'll hopefully start the new week and month with an all-new server and a new run.
BM
BM
We've started to 'drain' the
)
We've started to 'drain' the S5R3b Workunit generator (WUG), i.e. told it to generate all remaining S5R3 Workunits and put them into the database. We hope that this will shift a bit of load from the main server to the database server. In a few hours you'll notice that the WUG doesn't run anymore, which is nothing to worry about.
Also we nudged some S5R2 and S5R3a workunits that got stuck, so we can hope to have these runs finished before Friday.
BM
BM
RE: Don't get me wrong, I
)
Actually in in the last weeks Einstein@home has been awarding 25% more credit on average than SETI@home. And the only reason for no credit adjustment being done within the run was just that it turned out to be too late for that (i.e. too close to the end of the run) anyway. We are currently making the final adjustments, but you should expect to get in the order of 20-25% less credit per CPU hour than in S5R3, especially if you ran PowerApps.
BM
BM
RE: Any outlook on the
)
We discussed many alternatives until the very last minute, but it seems that all other kind of splitting up the sky than we do now bears at least a high risk of reducing the sensitivity of the search.
So for now we'll stick to the S5R3 scheme and I will make another attempt on adjusting credits in the workunit generation. Once we have an SSE2 App for Windows then probably built with the same compiler we use for Linux and MacOS Intel, most Apps used out there should behave roughly the same from the timing aspect, so this should be easier to handle.
It would make things even easier if we would switch from equatorial to ecliptical coordinates in the skygrid, but that would require more changes than we have time left for; and before this is fixed in the WUG, the effect would be even worse.
BM
BM