Computing Sciences

Data movement remains a major bottleneck in high-performance computing workflows, making GPU-accelerated compression an increasingly important optimization. While NVIDIA platforms benefit from mature compression libraries such as nvCOMP, the performance of practical lossless compression on AMD GPUs remains undercharacterized. This paper addresses that gap by porting LZ4, Snappy, and Cascaded from CUDA to HIP and evaluating them across four AMD GPUs spanning three CDNA generations (MI50, MI210, MI300X) and RDNA 3 (RX 7900 XT), using both synthetic datasets and seismic simulation data. Our study makes three contributions. First, we provide a cross-generational characterization of lossless GPU compression on recent AMD accelerators. Second, we present a functional CUDA-to-HIP port of three widely used algorithms, establishing an open baseline for future AMD-specific optimization. Third, we analyze how architectural differences shape compression and decompression behavior across algorithms and datasets. The results show that the MI300X achieves up to 11×$$11\times $$ higher decompression throughput than the MI50, while RDNA 3 is competitive for compression workloads characterized by irregular memory accesses. We also show that transfers dominate end-to-end execution time on all evaluated platforms, indicating that the main benefits of GPU compression are most likely to emerge in GPU-resident workflows. For realistic seismic floating-point data, lossless compression ratios remain modest (1.03-1.04×$$1.03-1.04\times $$), suggesting that error-bounded lossy compression is a promising direction for future work.

This paper presents the first differential fiducial measurements of γγ → ττ using 1.93 nb−1 of Pb+Pb data at sNN=5.02$$ \sqrt{s_{\mathrm{NN}}}=5.02 $$ TeV recorded by the ATLAS detector. Events in which one of the τ -leptons decays into a muon and two neutrinos τ→ντν¯μμ$$ \tau \to { u}_{\tau }{\overline{ u}}_{\mu}\mu $$ are selected and are categorized into three regions by the presence of an electron or either one or three charged-particle track(s) from the second τ-lepton decay. The measurement is performed in events where both Pb ions remain intact and no neutrons are emitted. Differential cross-sections are measured for seven variables in three fiducial regions at particle level. The measurements are compared to theory predictions with different photon flux models and spin correlation effects. For the fiducial region with one muon and one electron in the final state, comparisons to next-to-leading-order electroweak predictions are also made. The transverse momentum (pT) of the decay muon, the pT of the visible decay particles of the other τ-lepton, the total pT, invariant mass, and pseudorapidity of the visible particles from the di-τ system, and the rapidity and acoplanarity of the visible decay particles from either τ-lepton are measured. A maximum-likelihood fit to the muon transverse-momentum distributions in the three regions before unfolding is performed to extract the τ-lepton anomalous magnetic moment aτ and electric dipole moment dτ, the latter for the first time in heavy ion collisions. The observed 95% confidence level intervals are −0.057 < aτ< 0.035 and |dτ| < 2.7 × 10−16ecm.