Application of GONT in the fabless industry

Today we are talking about the application of GONT technology in the fabless industry and the service GONT.Fabless.

Welcome, friends! Today we are talking about the application of GONT technology in the so-called fabless industry. We will tell you about the service GONT.Fabless and answer the key questions. Read and share articles with colleagues!

Motivation:

The development of chips is an extremely expensive process. The cost of tens of millions of dollars (and even hundreds in IBM) to develop one chip (NRE costs in fabless inures). And here there is always a place for optimization. Due to the methodology of chains of added value in GONT.

GONT improves not only the model of selling computing software for microchips, but also a model for selling expertise for professionals in chip design.

To do this, “expert chains” will also be created, along with “computing chains”.

Market

The volume of the current world fabless industry is extremely large.
https://cloud.mail.ru/public/MfJP/rqQpkB2ip

Fabless Company and Worldwide IС Sales ($B)

The market is measured in tens of billions of dollars (even taking into consideration the old statistics).

KEY QUESTIONS FOR ANY GONT SERVICE

To transfer the existing Design Flow business to GONT, you need to answer a few questions. These questions apply to any new service in GONT!

1. What is the Value Chain?
Are there any chains for creating services in GONT? Does the already available service offer something to display on the GONT economy?

2. What is the Consensus?
What are the requirements for the adaptability of consensus?

3. What is AlGas?
Which microservices are integrated into the Value Chain chain?

4. What are the FUPE blocks?
What computing accelerators can I use?

Value Chain – Chains for GONT Tree

Consider the basic Value Chain (in the Fabless industry – Design Flow) using the example of Global Unichip Corporation (Taiwan).

FLOW consists of a very large number of potential service routes:

Silicon design/ development flow 

 

We can see a large chain of sequential tasks, where each task can also consist of a multiplicity of computational subtasks (micro-services).

At the same time, any subtask can be launched on different miners, which also forms a model for verification of each task (in GONT it turns out automatically, but this is what the customer needs: excessive verification of the chip!).

More detailed description of the chains you can find here – https://cloud.mail.ru/public/Cwf5/uCTpa2enS

From the UniChip workflows, chains will be built on GONT Tree (in the UniChip dedicated plane).

CONSENSUS

We will consider consensus in the expanded paradigm of “adaptive consensus” from GONT. What makes consensus more utilitarian and applicable to services.

Miners

We will consider 3 major software manufacturers for chip calculations:

  • Synopsys
  • Cadence
  • Mentor Graphics

    Holders of nodes with at least one of these software (at least one product from Synopsys, etc.) will be considered as GONT.Fabless service miners.

Those Fabless companies that purchase massively expensive software Synopsys, Cadence and Mentor can potentially become miners for other projects in times of idle software. GONT will provide them with such a showcase.

Consensus requirements

The requirements for consensus (and, consequently, for the miners) largely determine the GONT.Fabless transaction economy.

GONT-diagram of consensus

The Importance of Speed ​​TR – The Importance of Consensus.

For the empirical criterion of verification sufficiency, we may require the coincidence of the results of calculations from 2 miners (conditionally in the Cadence miner and the Sysnopsys miner).

For extra-expensive and expensive chips, you can connect 3 miners (Cadence, Synopsys, Magma) and expect the results of the chip calculations to coincide at the specified stage in the Design Flow chain.

The coincidence of the results in a certain computational comparison space is, of course, conditional.

AlGas

Design Flow of the chip consists of a very large number of micro-services and meanings. And all of them can be decorated like AlGas.

The semantic content of Fabless space in GONT can yield up to 1 million “molecules” of AlGas. Some of these meanings can become calculations (smart contracts).

PHYSICAL SYNTHESIS chains – AlGas

Prects flow

Detailing the chains leads to new senses and calculations.
Service GONT.Fabless is filled with AlGas and microservices.

Postcts flow

Fabless industry gives a wide variety of chains to create value and it is well suited to the GONT.

FUPE

FUPE blocks will be computational modules that give real acceleration to calculations. For example, it can be hardware emulation units or FPGAs.

Example:
Synopsys HASP –
https://www.synopsys.com/verification/prototyping/haps.html

HAPS can be considered as a FUPE block in GONT from third-party vendors and under it will generate a special FUPE AlGas.

Thus, FUPE space for GONT.Fabless opens up opportunities for business partnerships.

Complex AlGas and the mechanics of launching third-party programs in consensus in a GVM container

Complicated AlGas for Genesis gas.
Based on the service processor.

An uncontrolled process can not be done in consensus.

A regular AlGas chain can only wait for the end-of-work signal of an unmanaged chain (for example, PowerShell). This approach greatly expands the capacity of AlGas, but generates a greater number of potential errors.

Nevertheless, according to this architecture, the fundamental gas GONT – assembly of gVM cores.

One of the unmanaged processes from GVM when building a gVM core is the launch of the C ++ compiler (a commercial product from Microsoft – AlGas for them!).

Thank you for attention! Good luck!
GONT

Leave a Reply