Summary
| set | device | problems | Ipopt | current augmented_cudss | current condensed_cudss_bk | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| solved | SGM (s) | SSGM | solved | SGM (s) | SSGM | solved | SGM (s) | SSGM | |||
| ACOPF | CUDA | 66/132 | 63 | 18.19 | 1 | 64 | 2.74 | 0.452 | 64 | 3.08 | 0.464 |
| COPS | CUDA | 65/65 | 61 | 21.16 | 1 | 59 | 8.88 | 0.606 | 64 | 4.02 | 0.45 |
| CUTEst | CUDA | 1338/1338 | 1205 | 7.56 | 1 | 1169 | 7.78 | 1.01 | 1178 | 7.90 | 1.02 |
| Mittelmann | CUDA | 47/47 | 39 | 89.67 | 1 | 41 | 18.78 | 0.289 | 36 | 48.18 | 0.584 |
Solvers
| solver | device | revision | benchmark |
|---|---|---|---|
| Ipopt | CPU | 3.14.19 | – |
| current augmented_cudss | CUDA | 47cc85e9 | – |
| current condensed_cudss_bk | CUDA | 47cc85e9 | – |
CUDA: node5106: AMD EPYC 9135 16-Core Processor (64 threads), 755.1 GB of memory; NVIDIA RTX PRO 6000 Blackwell Server Edition, 95.0 GB, driver 13.3.0, CUDA runtime 13.4.0; node5603: Intel(R) Xeon(R) 6737P (128 threads), 1007.1 GB of memory; NVIDIA RTX PRO 6000 Blackwell Server Edition, 95.0 GB, driver 13.3.0, CUDA runtime 13.4.0; node5202: Intel(R) Xeon(R) 6767P (256 threads), 2015.1 GB of memory; NVIDIA RTX PRO 6000 Blackwell Server Edition, 95.0 GB, driver 13.3.0, CUDA runtime 13.4.0
Benchmark : the version fixes the instances, the options, Ipopt and HSL.
SGM: shifted geometric mean of the solve time, with a shift of 10 s and a failure counting as the time limit of a solve (CPU 7200 s, GPU 600 s, for every solver on those pages), over the problems that all solvers shown have run. SSGM: the SGM scaled so that Ipopt, or the best solver where Ipopt has not run, is 1.