Mohamed Shalan - OpenROAD on SkyWater 130nm
Join live on YouTube on Tuesday July 28 at 16:00GMT https://lnkd.in/gCyMuPp
Jul 27, 2020
[FOSSi] OpenLANE: Open Source 130nm PDK
Join Mohamed Shalan for the 2nd talk in the Free and Open Source Silicon (FOSSi) Foundation Dial-Up series is on Tuesday 28th July, he will talk about OpenLANE on the first-in-the-industry Open Source Manufacturable SkyWater 130nm PDK
Unlike the wider software world, Electronic Design Automation (EDA) open-source landscape has been fragmented for a long time, requiring significant effort and knowledge in a variety of disciplines to assemble a working ASIC flow. This has changed with projects such as Qflow and OpenROAD that aim at developing open-source toolchain for digital layout generation from RTL. OpenLane is an automated RTL to GDSII flow based on available opensource EDA tools configured/tuned for the SkyWater 130nm PDK. OpenLane main objective is to generate a clean layout from RTL designs in less than 24-hours with zero human interventions. OpenLane has been used, successfully, to tape-out a family of test chips (striVe).
Join live on YouTube on Tuesday July 28 at 16:00GMT https://lnkd.in/gCyMuPp
Join live on YouTube on Tuesday July 28 at 16:00GMT https://lnkd.in/gCyMuPp
Jul 24, 2020
[paper] Vectorizing Device Model Evaluation in Ngspice
Vectorizing Device Model Evaluation in Ngspice circuit simulator
Florian Ballenegger, Anamosic Ballenegger Design
Preprint July 2020
Abstract: A method improving the execution speed of electrical circuit simulation using vector processing is proposed. The BSIM3V32 semi-conductor device model for the open-source Ngspice simulator has been re-written for evaluating multiple device instances of the same model at once using Single Instruction Multiple Data (SIMD) processor instructions. While parallel evaluation of device model was already available using multiprocessing, the proposed method can achieve the same speed-up using less processor resources, thus allowing to do more parallel independent simulations for statistical analysis.
In Conclusion: Only the BSIM3V32 device model was modified to use vector processing. Other device models would of course also benefit from the proposed method. In particular interest would be the EKV model https://github.com/ekv26/model, as the calculations in this symmetric model are more linear with fewer conditional branches and could be vectorized more efficently. The source code of the modified BSIM3V3 model is available at https://www.anamosic.com/pages/ngspice.html
Softbank talks to #Apple and #Nvidia about #Arm sale https://t.co/pgShLwpAz1 #semi https://t.co/KOS9UIIhO1
Softbank talks to #Apple and #Nvidia about #Arm sale https://t.co/pgShLwpAz1 #semi pic.twitter.com/KOS9UIIhO1
— Wladek Grabinski (@wladek60) July 24, 2020
from Twitter https://twitter.com/wladek60
July 24, 2020 at 10:47AM
via IFTTT
U.S. Senators C.Schumer and K.Gillibrand pushed through a massive semiconductor manufacturing incentive package worth as much as $25 #billion that could benefit #GF and #IBM, both are in the Capital Region and the Hudson Valley.https://t.co/lah21UaJsZ #semi https://t.co/Zxeyxds702
U.S. Senators C.Schumer and K.Gillibrand pushed through a massive semiconductor manufacturing incentive package worth as much as $25 #billion that could benefit #GF and #IBM, both are in the Capital Region and the Hudson Valley.https://t.co/lah21UaJsZ#semi pic.twitter.com/Zxeyxds702
— Wladek Grabinski (@wladek60) July 24, 2020
from Twitter https://twitter.com/wladek60
July 24, 2020 at 08:11AM
via IFTTT
#Intel conceding the battle to #ARM and #AMD as 7nm processors delayed even further https://t.co/FHOPn7AA0O #paper
#Intel conceding the battle to #ARM and #AMD as 7nm processors delayed even further https://t.co/FHOPn7AA0O#paper
— Wladek Grabinski (@wladek60) July 24, 2020
from Twitter https://twitter.com/wladek60
July 24, 2020 at 06:40AM
via IFTTT
Subscribe to:
Posts (Atom)