Evidence & sources

Reports, datasets & manuals.

Each study belongs to a named model, reference and measurement window. Use the complete source data alongside the figures; do not compare unlike error definitions as one score.

Download the evidence manifest and SHA-256 hashes

Runtime performance

Prepared-step measured median. Not a complete host runtime or deadline guarantee. Correctness checks verify device execution, not model fidelity.

Read report

PDFtdse-runtime-performance-report.pdf5,013 KiB
Source identity · SHA-256
5ca3e51b935d0c25b2397ec16bba7586668c5843be47e28c1aa551d3f279c7aa
CSVtdse-runtime-performance-matrix.csv3,617 KiB
Source identity · SHA-256
adb403b8e4b8043a965c6c9540f77c43d68aae64b6533ea444c09deeec588e08
CSVtdse-runtime-platform-qualification.csv7 KiB
Source identity · SHA-256
585762bcf593478c6a2fdd259c7c4cad6ae329ad1c89d223aaf2a591cb3cd1d3

Public case / ieee118-frequency

IEEE118 / frequency sensitivity

Waveform and spectrum relative RMS error (relative ratio). Published Gold reference, 1 microsecond step.

One-port, full-system Gaussian case. Not a host-partition qualification.

Read the study Download dataset (JSON)

Configuration, scoring window & source identity

Configuration: Bus 12; DUT 50 microseconds; nh 1024; six displayed frequencies, complete sweep retained
Metric & reference: Waveform and spectrum relative RMS error (relative ratio). Published Gold reference, 1 microsecond step.
Window: Per-view waveform window and published spectrum extent
Limits: One-port, full-system Gaussian case. Not a host-partition qualification.

Source identity: 5c41c32944f36f3cb41f9837547e2bb5926ee7f857a62732a853ccda3bb657b5
Preserved public snapshot; no simulation was rerun to restyle these figures.

Public case / ieee118-timestep

IEEE118 / timestep sensitivity

Mean burst-core relative error (relative ratio). Published Gold reference, 1 microsecond step.

Fine-step results can favor TDSE Off. Do not substitute burst-core scores for full-record NRMSE.

Read the study Download dataset (JSON)

Configuration, scoring window & source identity

Configuration: Bus 12; 2000 Hz Gaussian; 22 steps and 50 event centers per step
Metric & reference: Mean burst-core relative error (relative ratio). Published Gold reference, 1 microsecond step.
Window: Burst core; separate two-cycle metrics retained in source
Limits: Fine-step results can favor TDSE Off. Do not substitute burst-core scores for full-record NRMSE.

Source identity: 5f42085a3f93b919ef2bfc4bd7a6d3432d90d61a88d8f3ade0b281a38247978d
Preserved public snapshot; no simulation was rerun to restyle these figures.

Public case / ieee118-twoport

IEEE118 / two-port overlapping bursts

Per-port waveform and spectrum relative RMS error (relative ratio). Published Gold curves.

Full-deck two-port evidence, not a partitioned-host result. Legacy introductory single-drive wording is retained; the figure's drive model is the two overlapping bursts.

Read the study Download dataset (JSON)

Configuration, scoring window & source identity

Configuration: Full system; buses 12 and 16; 1600 / 2500 Hz bursts separated by 0.5 ms
Metric & reference: Per-port waveform and spectrum relative RMS error (relative ratio). Published Gold curves.
Window: Per-port displayed waveform and spectrum windows
Limits: Full-deck two-port evidence, not a partitioned-host result. Legacy introductory single-drive wording is retained; the figure's drive model is the two overlapping bursts.

Source identity: 35305ee254f12c8ab9acdcf8d3def652ee62ac8f34239f9d973a8daef40d8239
Preserved public snapshot; no simulation was rerun to restyle these figures.

Public case / ieee118-fault

IEEE118 / representative short fault

Voltage/current waveform and spectrum relative RMS error (relative ratio). Published Gold curves.

Representative supporting case, not a uniform claim over switching edges and coarse steps. Display filtering is not a new simulation input.

Read the study Download dataset (JSON)

Configuration, scoring window & source identity

Configuration: Bus 12; DUT 50 microseconds; trial 14; 0.005 ohm fault; 5 ms clearing
Metric & reference: Voltage/current waveform and spectrum relative RMS error (relative ratio). Published Gold curves.
Window: Representative fault windows; current display uses the declared two-point moving average
Limits: Representative supporting case, not a uniform claim over switching edges and coarse steps. Display filtering is not a new simulation input.

Source identity: 6b3d890006d414dc40cd7bb640c55d119d4f7162a0826303c2b48921c05c68a7
Preserved public snapshot; no simulation was rerun to restyle these figures.

Public case / activsg200-timestep

ACTIVSg200 / timestep sensitivity

Mean burst-core relative error (relative ratio). Published Gold reference, 1 microsecond step.

Crossover occurs later than IEEE118; TDSE Off is lower through 30 microseconds in this sweep.

Read the study Download dataset (JSON)

Configuration, scoring window & source identity

Configuration: Bus 113; 2000 Hz Gaussian; 22 steps, 20 event centers per step
Metric & reference: Mean burst-core relative error (relative ratio). Published Gold reference, 1 microsecond step.
Window: Burst core
Limits: Crossover occurs later than IEEE118; TDSE Off is lower through 30 microseconds in this sweep.

Source identity: 21b48e3e81a53f1c5453915ce5490b25ed7034fda4a0f456d5062b9d979dbc12
Preserved public snapshot; no simulation was rerun to restyle these figures.

Public case / nrel240-timestep

NREL240 / timestep sensitivity

Mean burst-core relative error (relative ratio). Published Gold comparison.

Compact source contains summary values, not a full reproducibility bundle. Do not infer omitted solver settings or trial count.

Read the study Download dataset (JSON)

Configuration, scoring window & source identity

Configuration: 22 published timestep points
Metric & reference: Mean burst-core relative error (relative ratio). Published Gold comparison.
Window: Burst core
Limits: Compact source contains summary values, not a full reproducibility bundle. Do not infer omitted solver settings or trial count.

Source identity: abda56440383269dc748c2673456f0fb11bacdcbc8e12039274d1e12444f3491
Preserved public snapshot; no simulation was rerun to restyle these figures.

Product manuals

The complete formal Runtime and Circuit books are synchronized from the SDK sources. Download identity and chapter coverage are recorded separately from benchmark evidence.

Read online Runtime PDF Circuit PDF Coverage & source hashes

Manual edition & source hashes

SDK revision: cdffb34da. Source content SHA-256: 8c398f175f54cca90f5ff6e5104a5c0bac4f0313b3b41b94788e8f967d93eda9The hashes identify the edition; they do not assert that a PDF was rebuilt after every source edit.