Whouoooo: -4200 machine today ?
 

What's happened today ?

[CODE]
Status Summary Report 15 Mar 2004 22:00
63088 Machines Applied on 45812 Accounts[/CODE]

[CODE]
Status Summary Report 14 Mar 2004 23:00
67356 Machines Applied on 45889 Accounts[/CODE]

=> -77 accounts
=> -4268 machines

It seems we loose a big participant :cry:

Don't worry. A good fraction of the machines listed in the statistics never return a result. The speed is still good:

The virtual machine's sustained throughput* is currently 14115 billion floating point operations per second (gigaflops), or 1172.5 CPU years (Pentium 90Mhz) computing time per day. For the testing of Mersenne numbers, this is equivalent to 504 Cray T916 supercomputers, or 252 of Cray's most powerful T932 supercomputers, at peak power. As such, PrimeNet ranks among the most powerful computers in the world. (*Measured in calibrated P5 90Mhz, 32.98 MFLOP units: 25658999 FPO / 0.778s using 256k FFT.)

Uh oh, now things really are getting bad:

The virtual machine's sustained throughput* is currently 13115 billion floating point operations per second (gigaflops), or 1089.5 CPU years (Pentium 90Mhz) computing time per day. For the testing of Mersenne numbers, this is equivalent to 468 Cray T916 supercomputers, or 234 of Cray's most powerful T932 supercomputers, at peak power. As such, PrimeNet ranks among the most powerful computers in the world. (*Measured in calibrated P5 90Mhz, 32.98 MFLOP units: 25658999 FPO / 0.778s using 256k FFT.)

We've lost an entire Teraflop!

A clue to why the throughput has dipped may be a recent post to the "Finally on line again, what happened?" thread:
[QUOTE=PrimeCruncher]Speaking of outages, what's up with mersenne.org? It seems to have been down for a while now, at least an hour.
[/QUOTE]But if it was down hours ago, you'd think the backlog of results would've been processed by now. Unless there are machines that can't retry now for some reason, (the weekend?).

The throughput is also recalculated hourly, right?
There are sizable variations during the day, and also daily ones on weekends - but not at the Teraflop scale.

[QUOTE=Maybeso]The throughput is also recalculated hourly, right?[/QUOTE]

I know that the # of Teraflops statistic changes hourly.

It seems to be getting worse:

The virtual machine's sustained throughput* is currently 13017 billion floating point operations per second (gigaflops), or 1081.3 CPU years (Pentium 90Mhz) computing time per day. For the testing of Mersenne numbers, this is equivalent to 464 Cray T916 supercomputers, or 232 of Cray's most powerful T932 supercomputers, at peak power. As such, PrimeNet ranks among the most powerful computers in the world. (*Measured in calibrated P5 90Mhz, 32.98 MFLOP units: 25658999 FPO / 0.778s using 256k FFT.)

[QUOTE=jinydu]It seems to be getting worse:[/QUOTE]

People joining the search after the discovery got rid of it...
:unsure:

[QUOTE=ET_]People joining the search after the discovery got rid of it...
:unsure:[/QUOTE]

What does that mean?

[QUOTE=jinydu]What does that mean?[/QUOTE]

It means that people that installed Prime95 without finding a new prime decided to unload it and choose a mre "interesting" project (or none at all).

Luigi

Maybe cumulative statistics would be effective at retaining people. For example, displaying the total number of FLOPs (not FLOPS) and CPU years since the start of the project.

[QUOTE=jinydu]Maybe cumulative statistics would be effective at retaining people. For example, displaying the total number of FLOPs (not FLOPS) and CPU years since the start of the project.[/QUOTE]

I'm sure George and Scott are aware of it... :rolleyes:

Luigi