Spectre 24.1 Release Now Available

The Spectre 24.1 release is now available for download at Cadence Downloads. For information on supported platforms and other release compatibility information, see the README.txt file in the installation hierarchy.

Spectre 24.1

Here is a listing of some of the important updates made in the SPECTRE 24.1 release.

Hardware Acceleration and Diversification

• Arm Platform Support: Starting this release, Spectre supports ARM CPUs with an ARM-specific installation available on downloads.cadence.com. 

• Multiple GPU Support: With this release, the GPU solution in Spectre supports the deployment of more than one GPU. While the performance gain from multiple GPUs over a single GPU is limited, the multiple GPU solution is important especially for designs that require more memory than that available on a single GPU.

• Additional Features Supported on GPU: The following features are fully supported
  • Assert statements
  • All dynamic design checks except dyn_highz, dyn_dcpath, dyn_floatdcpath, dyn_float_tran_stat, and dyn_nodecap
  • Direct EMIR analysis, which provides a significant EMIR performance gain while still providing SPICE accuracy
  • Aging and self-heating analyses
  • Transient noise simulations

Core Spectre

Spectre X

• Improved AX Performance: The performance of +preset=ax mode for Spectre X has been enhanced. For pre-layout designs, a performance gain of up to 20% over Spectre 23.1 is observed.
• Improved LX Accuracy: Spectre X +postlpreset=lx setting has been optimized to improve the accuracy of the post-layout optimization. With this enhancement, post-layout designs that needed +postlpreset=mx in Spectre 23.1 might get sufficient accuracy with +postlpreset=lx. There are selected post-layout designs that might enable designers to use +preset=lx instead of +preset=mx for achieving higher performance.
• Real-Time Self-Heating Performance: Real-time self-heating for N2 and N3 nodes is a challenging simulation. In this release, the related overhead compared to non-self-heating simulation has been reduced to an average of 30%.

Spectre RF

• S-Parameter Quality Checking: The S-Parameter checking and fitting tool as well as the Spectre nport element have been enhanced to support the Measurement Data Interface File (MDIF) format.
• XDP-PSS/QPSS Shooting: Distributed PSS/QPSS shooting has been introduced in this release. It enables you to analyze large designs, which require more memory than available on one machine. In the distributed solution, the memory demand is shared between different machines. Supported analyses are pss, pac, pstb, pnoise, pxf, psp, qpss, qpac, qpnoise, qpxf, and modulated and sampled pac.
• Lowmem=X for HB and HB Small Signal Analyses: The lowmem technology for HB and HB small signal analysis has been further enhanced. It lets you reduce the memory consumption of these analyses and keep the memory usage within a given memory budget.
• Multi-Tone Fast Envelope: The Fast Envelope technology has been enhanced to support multiple tones.
• Jammer Analysis: The modulated jammer analysis has been introduced in this release. This analysis helps you characterize RX desensitization because of large TX signal coupling to low level RX input. It currently supports only one modulated signal or one CW signal.
• New Signal Standard: Support wifi7 (802.11be) wsource in rfLib.

Spectre FX

• Spectre FX Interactive Environment: Spectre FX Interactive Environment is a new debugging environment for Spectre FX simulations running transient analysis. This environment provides features, such as force and release, which you can use to pause a simulation and modify the voltage of specific nodes in a circuit.
• Most Connection Nets: Spectre FX now reports the top 10 nets with most MOSFET connections in the biggest design partition. When the partition size is large, you can use this information to analyze whether there is a net with many connections.

Major Default Value Changes

• intrinsic_cap_merge Default Value Change: The default value of the intrinsic_cap_merge parameter has been changed from no to yes for the following features:
  • info analysis what=captab
  • dyn_nodecap dynamic check

Starting with this release, the internal nodes of a device model are not reported in the capacitance reports. Their capacitance value is merged into the capacitance value of device terminals.

• nport Interpolation Method Default Change: The default nport interpolation method for transient analysis has been changed from linear to bbspice when using the default setting nport_default_interp=auto_switch. This change improves the following:
  • The convergence when simulating designs with nport.
  • The quality of S-Parameter simulation results because the bbspice technology guarantees passivity.
• Power-Grid Views (UTI) Generation Default Changes: The following three options, which are essential for a proper power-grid view (PGV) creation, have been made available by default in this release:
  • eisopt save_gnd_caps=1 : enables the extraction of the net capacitance.
  • eisopt pgv_sum_cgnd_cc=true : enables the consideration of the coupling capacitance when extracting the net capacitance values.
  • eisopt uti_tap_cap=avg : calculates the average tap-device capacitance values over the UTI time window instead of getting the capacitance value from the first time point of the UTI time window.

Solutions and Flows

Spectre EMIR/Voltus-XFi

• EMIR Autostop: The autostop functionality has been added to the Spectre EMIR analysis. It lets you stop the EMIR analysis at the end of the (last) EMIR time window.

Legato Reliability Analysis: The following enhancements have been done in legato reliability analysis:

• Aging is now supported for Multi-Technology Simulation (MTS) TMI analysis.
• Legato aging or self-heating (SHE) enabling analysis can be run from outside of the reliability block.

Legato Analog Fault Analysis

• Spectre FX support: The direct fault analysis is now supported with Spectre FX.
• Circuit Activity Analysis in Fault Simulations: According to IEEE P2427, the defects that do not affect the circuit functionality are counted as undetected and excluded from the fault simulation. This activity-based fault reduction is realized by running an initial simulation to determine inactive circuit blocks, which are subsequently excluded from the fault simulation.

Legato Thermal Analysis

• Irregular Mesh Grid: A new format has been developed that lets non-uniform mesh grids to be stored and plotted by the Virtuoso thermal map. This is important for designs requiring an adaptive mesh, such as designs with trench structure.
• Subckt Block Rotation and Mirroring: The floor planning capability of the Legato Thermal Analysis flow has been enhanced by supporting subckt block rotation and mirroring.

Fast Monte Carlo

The fmcmethod=yieldest for yield estimation is now supported in core Spectre. By using this method, Spectre estimates the parametric circuit yield for each goal in the format of a sigma and failure rate along with uncertainties using confidence intervals. The additional parameters, such as fmcbudget and fmcsigmawidth, can be used to control the total number of simulations and the confidence interval width resolution to be achieved, respectively.

Please send questions and feedback to spectre_rm@cadence.com.

To receive Spectre release announcements like this one, and other Spectre-related information, directly in your mailbox, click SUBSCRIBE in the Subscriptions box.

Spectre Team