Abstract: A data acquisition system using Nvidia Tesla K40 GPUs is being used to accelerate the rate of data acquisition for a new particle physics experiment at Fermilab. A new measurement of the anomalous magnetic moment of the muon is being performed.
The most recent measurement, performed at Brookhaven National Laboratory and completed in 2001, shows a 3.6 standard deviation discrepancy with the standard model value of g-2. The new measurement will accumulate 21 times those statistics, measuring g-2 to 140 ppb and improving the uncertainty by a factor of 4 over that of the previous measurement.
The data acquisition system for this experiment must have the ability to create deadtime-free records from 700 s muon spills at a raw data rate 20 GB per second. Data is processed online in a layered array of networked commodity processors with 28 GPUs working in parallel to perform a fast recording of the muon decays during the spill. The GPU processing code is written in Cuda and achieves a factor of 100 in data reduction by processing each digitized sample in a parallel thread. A Cuda-based Monte Carlo simulation was also used to prototype the system and set design parameters. Data taking for the experiment is underway and will run for two years.