LHC@home

At 1.00 a.m. + 17 seconds on Monday, 17 July, the LHC beams were dumped after only 9 minutes in collision due to a radiofrequency interlock caused by an electrical perturbation. Approximately 300 milliseconds after the beams were cleanly dumped, several superconducting magnets around the LHC quenched – i.e. they lost their superconducting state.
...
Unfortunately, in the case of the inner triplet magnet located to the left of Point 8, a small leak has appeared between the cryogenic circuit, which contains the liquid helium, and the insulation vacuum that separates the cold magnet from the warm outer vessel, known as the cryostat. This vacuum barrier is crucial for preventing heat transfer from the surrounding LHC tunnel to the interior of the cryostat (this is similar to the functioning of a thermos flask). As a result of the leak, this insulation was lost: the insulation vacuum filled with helium gas, cooling down the cryostat and causing condensation to form and freeze on the outside.
...
it is clear that an intervention with the inner triplet magnet at room temperature will be required. This incident will probably have a great impact on the LHC schedule, with machine operation unlikely to resume for at least several weeks.

I'm intrigued by the cryptic references to UFOs in that blog post - sounds suspiciously like someone's dropping their spanners on the machine. (Unless that's a dated baguette reference?)

Either confirm or exclude lepton flavour universality (LFU) anomalies, and further searches on leptoquark models
New experiements FASER and SND
Results from significantly upgraded LHCb and ALICE experiment, like new exotic hardron discoveries and studies on quark–gluon plasma (QGP)
Higher precision experiments and search of rare decay modes may find the first deviations from the SM

Meanwhile the LHC magnets remain at 2K, allowing more time to train the magnets.

At LHC beam energies, the electric currents are extremely high, up to 12,000 Amperes, and superconducting cables have to be used. Superconductivity is a low-temperature phenomenon, so the coils have to be kept very cold, just 1.9 degrees above absolute zero to be precise, or about -271°C. Even a tiny amount of energy released into the magnet for any reason can warm the coils up, stopping them from superconducting. When this happens, the current has to be safely extracted in a very short time. This is called a quench, and just one millijoule – the energy deposited by a 1-centime euro coin falling from 5 cm – is enough to provoke one. Magnet protection in case of quenches is a crucial part of the design of the LHC’s magnetic system.

When a new superconducting magnet is qualified for use, it needs to be trained. That involves steadily increasing the current until the magnet quenches, then starting again. At first, the quenches may occur at relatively low current, but over time, as the components of the magnet settle in, the current increases until the magnet can be operated routinely at its nominal current. If a new training cycle is started after an extended period during which the magnet is warm, the magnet usually restarts training at a value that is higher than first quench in the first training cycle but lower than the maximum previously reached. In other words, the magnet’s ‘memory’ is usually less than 100%.

RF will be warmed up to room temperature following the cooling tower fault in point 4.
No beam for the next ~4 weeks

Four days with success, Pythia8 - [boinc pp jets 8000 170,-,2960 - pythia8 8.301 dire-default 48000 158]

https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html
This might be a good starting point to understand why it is not a good idea to work with reduced precision.

I know reduced precision is bad. I don't have enough maths skills to understand any papers on it. But can't more calculations be done at FP32 to make it more accurate? Even if it requires several, it would still be faster than FP64 on most cards.