Metsähovi station report

The most significant development has been Jan Wagner's work in porting the 
Australian DiFX software correlator to the Cell processor (read: Sony 
Playstation 3).

The original DiFX program proved to be impossible to port to other major 
architectures efficiently. However, after using Daniel Hackenberg's matmul 
program as an initial example, we could demonstrate a six-station Cell 
correlator engine that processes raw data at 166 Mbps per each station on just 
one PS3. Aggregate processing speed is 1Gbps, the PS3 correlates faster than 
the gigabit Ethernet can receive data.

A small cluster of eight PS3s is capable of realtime correlation in a 
six-station 1 Gbps experiment. The correlator engine can be easily expanded to 
12 or 18 stations.

It is clear that all upgrade work on existing hardware correlators should be 
stopped, building a new software correlator is less expensive and adds new 
capabilities.

The Cell correlator engine is available in the SourceForge project 
"cellspe-tasklib".

Other new software from Metsähovi is Jan Wagner's "fuseMk5A". It allows all 
normal Linux programs to read scans from the diskpack. Several stations and 
software correlator users are now adopting it.

Our work in e-VLBI continues successfully. Lately we have performed a series of 
ultra-rapid UT1 experiments with Onsala, Kashima and Tsukuba at 128 and 256 
Mbps data rates, both in real-time and near-realtime. As far as we know, the 
correlation results were obtained at world-record turn-around time delay.

We have implemented several improvements to the Tsunami eVLBI software and 
provided increased support to its users. As Tsunami popularity increased, eVLBI 
experiments between countries become feasible.

It is also gratifying to see that other stations and institutes are starting to 
use the Tsunami protocol for eVLBI. Recently MPIfR and Wettzell have joined the 
group.

Tsunami is available on SourceForge under the name "tsunami-udp".

The old VLBA data acquisition system continues to deteriorate. Repairs to the 
baseband converters have resulted in only short term improvements while new 
components are now gradually failing.

On the positive side, we have received three of the new iBob-ADC systems from 
the SETI@home group. We will soon begin to develop and test 10 Gbps networked 
data acquisition both locally and over the Internet to Jodrell Bank.

We will explore the possibility of replacing the old decaying VLBA rack 
soon with a new system such as the iBob variants or a dBBC.