Product Brief
FAQs
1. What's the maximum capacity that NitroSIM can handle (gates)?
A: 300 million logic gates. Up to an additional 4 GBytes of on-chip memory is available.Capacity refers to NitroSIM capacity, not to the actual design size that can be compiled. The design can actually be much larger, through use of hierarchy.
As a example, a design which has large modules, e.g. 1x25 million gates, 4x30 million gates and 2x40 million gates in NitroSIM technology occupies only 25+30+40=95M gates, even though the design size itself is 25 + 120 + 80 = 225M gates. If this design is instanciated 4 times on a board, the NitroSIM required gate-count would stil be 95M gates, even though the 4-instance board gate-count amounts to 900M gates. In these gate-count numbers, memories are not included. They are counted by total occupied memory space.
2. 300 Mgates on a single PCI card?? How's that possible?
A: Our technology is very different than traditional hardware acceleration technology. We are a simulation technology - our compiler compiles your design into instructions and executes them in memory - just like a simulator.Our technology requires, on average, only 3 bytes of storage per gate. Therefore, 300M gates require less than 1GB of storage. NitroSIM has over 2.5 GBytes of storage available.
3. Hardware accelerators use many FPGAs or custom processors and are large, bulky, power hungry and expensive. Simply put, how can NitroSIM implement a 300 million gate design in a single card? Aren't HW solutions capacity-limited? Why doesn’t NitroSIM have those problems? How did NitroSIM solve the capacity problem?
A: NitroSIM compiles the design to memory and the compiled data is executed in the VLIW coprocessor. Hardware acceleration systems synthesize the design netlist into multiple FPGAs or ASICs. Simply, our patented approach allows us to use high-density commercial memory technology resulting in a low-cost Bill of Material for the PCI board.4. NitroSIM is a simulation technology but includes hardware as well. I'm used to software-only simulation solutions. Can you explain this difference?
A: Software simulators also use hardware. They simply use the general-purpose processor that's on the motherboard (CPU). NitroSIM uses a patented technique of a VLIW coprocessor with extremely wide word width to parallel process the data - therefore we have a plug-in PCI card with the coprocessor and memory.5. Can I use NitroSIM to replace my current simulator?
No, NitroSIM does not replace your current simulator – NitroSIM speeds it up. Your current simulator and methodology are preserved, which makes NitroSIM easier to integrate into your flow. In NitroSIM, only the DUT logic is compiled to the VLIW processr (in future releases this will be enhanced). Your testbench still runs inside your host simulator. You will need to use your current host simulator to drive the NitroSIM hardware.6. Why is NitroSIM referred to as a hybrid simulator? Isn't NitroSIM just another hardware accelerator?
A: NitroSIM is a simulation technology. Like a simulator, NitroSIM compiles your design into instructions and executes them in memory using a coprocessor optimized for simulation. It doesn't use hardware accelerator technology which synthesizes and place-and-routes your design into FPGAs or even processors.The difference between NitroSIM and common hardware accelerator technologies has a huge impact for you. NitroSIM can accommodate design sizes up to 300 Mgates on a single card. A hardware accelerator based PCI-card doesn't have that capacity at all! Hybrid simulation compile times are comparable to simulators. Hardware accelerators have long setup and compile times. In addition, hybrid simulators have very low cost compared to hardware accelerators. What's more, is that NitroSIM can support 4-state and assertions along with simulator-like debug capabilities.
We realized that the difference in NitroSIM technology is so substantial that we found it necessary to classify NitroSIM differently - a hybrid simulator.
7. What languages does NitroSIM support?
A: For our initial release NitroSIM v1.0, Verilog is supported. SystemVerilog and VHDL are on our roadmap for future product enhancements.However, other languages (VHDL, SystemVerilog, SystemC, etc) in your verification environment, remain on the host CPU and are simulated by your simulator.
8. Does NitroSIM support gate level (regression) simulation?
A: Yes, NitroSIM supports acceleration of gate level simulations, but not with back-annotated (SDF) timing information. All of NitroSIM gate-level and RTL simulations are functional only.We have received interest from various customers for timing support and we are considering this as a future roadmap item.
If you like to cooperate with us in this respect, please contact us.
9. How large are the designs you've been working with? How much acceleration did you get?
A: The largest design to date was over 30 million gates. We've validated with customer designs that it met our 10x product claim.10. How difficult is it for users to adopt NitroSIM? What skills or expertise are required for adoption? Does LIGA offer services to support adoption?
A: NitroSIM is no more difficult than using the Verilog simulator. If you are familiar with a compiled software simulator, you will be quick to understand the NitroSIM compiler.11. What commercial simulators are supported?
A: We support industry-leading, PLI-compliant simulators.12. What makes NitroSIM different than what's already available in the market?
A: Our differentiation centers on capacity, performance, price and scalability. We boost the speed of software-based simulations, delivering significant cost savings and increased capacity in a small form-factor. We also provide additional simulation features and ease-of-use differentiators to hardware-based solutions available today.One of the latest features that we're now working on is behavioral acceleration. This is an industry-first feature, enabled only by hybrid simulation since it’s a true simulator technology.
