IIT Delhi Researchers Develop India’s First Indigenous Micro-GPU: How the New Graphics Processor Works and Why It Matters

Researchers at the Indian Institute of Technology Delhi (IIT Delhi) have developed what they describe as the country’s first working, demonstrable, indigenously designed micro-Graphics Processing Unit (micro-GPU) from an Indian university, marking an important academic development in indigenous graphics-processing hardware.

The micro-GPU was developed by researchers in IIT Delhi’s Department of Electrical Engineering. The project was led by M.Tech. students Nammi Akash and M. Ravi Teja under the guidance of Prof. Jayadeva and Prof. Kaushik Saha.

The development is particularly significant because the researchers have demonstrated programmable graphics rendering on physical programmable hardware rather than presenting only a software simulation or conceptual architecture. IIT Delhi says the custom floating-point GPU engine was implemented entirely in Register Transfer Language (RTL) and mapped onto a Spartan-7 Field Programmable Gate Array (FPGA) platform.

The architecture is intended primarily for affordable embedded graphics and display-processing applications. Potential uses identified by IIT Delhi include industrial control displays, low-cost human-machine interfaces, e-rickshaw dashboard navigators, inland-water navigation terminals for small fishing boats, educational e-book readers and other embedded visualization systems.

The research team is now exploring a substantially larger 8–16-core vector-style graphics processor architecture, along with an optimized compiler and graphics software toolchain. IIT Delhi says the longer-term roadmap includes migration toward a proof-of-concept chip using a 65-nanometre ASIC process node.

Official IIT Delhi announcement:
https://home.iitd.ac.in/show.php?id=485&in_sections=Press

IIT Delhi Micro-GPU: Key Highlights

What Has IIT Delhi Developed?

The IIT Delhi team has developed a compact programmable graphics-processing architecture designed for implementation on an FPGA and potentially, at a later stage, as dedicated silicon.

A GPU is a specialized processor capable of performing large numbers of calculations required for graphics, image processing and increasingly other parallel-computing workloads. Modern high-end GPUs can contain enormous numbers of processing elements and are used for gaming, professional visualization, artificial intelligence and scientific computing.

The IIT Delhi micro-GPU should not be viewed as a miniature replacement for a high-end commercial GPU. Its present objective is different: to establish an indigenous, programmable and scalable graphics-processing architecture suitable for affordable embedded systems.

Is IIT Delhi’s Micro-GPU Really India’s First GPU?

This claim requires an important qualification.

IIT Delhi’s researchers describe their development as, to the best of their knowledge, the first working, demonstrable, indigenously designed micro-GPU from a university in India.

This is more precise than simply saying that IIT Delhi has created the first GPU ever associated with India.

The distinction matters because India has a broader semiconductor, processor-design and high-performance-computing ecosystem involving government organizations, academic institutions, startups and private companies.

The specific milestone highlighted by IIT Delhi concerns a working and demonstrable indigenous micro-GPU developed in an Indian university environment.

Who Developed the IIT Delhi Micro-GPU?

The project was led by two M.Tech. students:

• Nammi Akash
• M. Ravi Teja

The work was carried out under the guidance of:

• Prof. Jayadeva
• Prof. Kaushik Saha

IIT Delhi identifies the project with its Department of Electrical Engineering.

What Role Did IIT Delhi M.Tech. Students Play?

The involvement of M.Tech. students is one of the noteworthy academic aspects of the development.

Rather than being solely a faculty-led semiconductor project, the micro-GPU was led by postgraduate students working under faculty supervision.

The project required the integration of multiple areas of engineering, including arithmetic hardware, digital architecture, programmable computing, embedded systems and software/compiler considerations.

What Is a Micro-GPU?

A micro-GPU is a compact graphics-processing architecture designed for workloads that do not require the enormous computing resources of mainstream desktop, gaming or data-centre GPUs.

Such processors can be useful when a device needs graphics acceleration or programmable visualization but must operate under tighter constraints involving cost, power consumption, hardware resources or physical size.

Examples can include dashboards, control terminals, industrial displays and specialized embedded computing devices.

Micro-GPU vs High-End GPU: What Is the Difference?

What Is an FPGA?

FPGA stands for Field Programmable Gate Array.

Unlike a conventional processor whose internal hardware architecture is fixed during manufacturing, an FPGA contains configurable digital logic resources that engineers can program to implement custom hardware circuits.

This makes FPGAs particularly useful for developing and validating new processor architectures.

Researchers can implement a design, test it on physical hardware, identify problems, modify the architecture and repeat the process without manufacturing a new silicon chip after every change.

Which FPGA Does the IIT Delhi Micro-GPU Use?

IIT Delhi says the custom GPU engine was mapped to a Spartan-7 FPGA platform.

This means the current demonstration exists as a programmable hardware implementation rather than a dedicated mass-produced GPU chip.

Why Is an FPGA Important for GPU Research?

Designing a semiconductor chip and manufacturing it are separate challenges.

Fabricating a dedicated ASIC can require significant time, engineering resources and financial investment. Researchers therefore commonly validate digital architectures on FPGA platforms before committing to custom silicon.

The FPGA implementation allows the IIT Delhi team to demonstrate that its architecture works on real programmable hardware while retaining the flexibility required for experimentation.

Is the IIT Delhi Micro-GPU Already a Fabricated Silicon Chip?

No.

The currently announced implementation has been mapped onto an FPGA platform.

IIT Delhi’s longer-term roadmap includes moving toward a proof-of-concept ASIC using a mature semiconductor process.

This distinction is important when interpreting the development. A working FPGA prototype represents a substantial hardware milestone, but it is not the same as a fabricated commercial GPU chip.

What Does RTL Mean in the IIT Delhi GPU Project?

RTL stands for Register Transfer Language or Register Transfer Level, depending on context.

At this level, engineers describe how digital data moves between registers and how operations are performed by hardware logic.

IIT Delhi states that its custom floating-point GPU engine was implemented entirely in RTL.

This is significant because it indicates that the researchers designed the digital hardware architecture itself rather than simply programming graphics operations in conventional software running on an existing GPU.

What Is a Floating-Point GPU Engine?

Floating-point arithmetic allows computers to represent and manipulate numbers across a broad numerical range, including fractional values.

Graphics calculations frequently require operations involving coordinates, transformations and other mathematical values that benefit from floating-point representation.

The IIT Delhi prototype uses a custom floating-point GPU engine to perform programmable graphics rendering.

What Does Programmable Graphics Rendering Mean?

Graphics rendering is the process of generating visual output from data and instructions.

A programmable graphics architecture provides greater flexibility than a completely fixed-function display circuit because developers can define or modify processing behavior for different rendering requirements.

IIT Delhi’s researchers specifically emphasize that their objective was to create a compact but genuinely programmable graphics-processing architecture.

What Is Processor IP?

In semiconductor engineering, IP or intellectual property can refer to reusable processor designs, logic blocks or hardware architectures that can be integrated into larger systems.

IIT Delhi describes its micro-GPU architecture as a scalable programmable graphics processor IP.

This means the design concept could potentially be adapted and integrated into different hardware platforms rather than being tied permanently to one FPGA demonstration board.

Can the IIT Delhi Micro-GPU Be Converted Into a Real Chip?

That is part of the project’s future direction.

IIT Delhi says the architecture may be mapped either to programmable hardware such as an FPGA or to a silicon ASIC.

The team is exploring eventual migration to a 65 nm ASIC process node for a proof-of-concept implementation.

What Is an ASIC?

ASIC stands for Application-Specific Integrated Circuit.

Unlike an FPGA, which can be reconfigured after manufacturing, an ASIC is fabricated with circuitry designed for a particular architecture or set of functions.

Moving a processor design from an FPGA to an ASIC can potentially improve characteristics such as performance, energy efficiency, physical size and manufacturing economics at sufficient scale.

However, ASIC development also involves substantially greater engineering and fabrication complexity.

Why Is IIT Delhi Considering a 65 nm Process?

The team’s roadmap mentions eventual migration to a proof of concept using a 65-nanometre process node.

Sixty-five nanometres is a mature semiconductor manufacturing technology rather than a cutting-edge node used for the newest high-performance processors.

That can be an advantage for certain embedded applications.

Mature nodes can be attractive where the objective is practical, affordable and reliable silicon rather than maximum transistor density.

The IIT Delhi researchers believe mature nodes such as 65 nm could potentially support economically viable indigenous graphics silicon for embedded-system applications.

Is 65 nm Outdated?

Not necessarily.

Process-node requirements depend on the intended application.

Leading-edge GPUs and processors require extremely advanced manufacturing technologies because they contain huge numbers of transistors and target very high computational performance.

Embedded display processors can have very different requirements.

A mature semiconductor node may remain useful when cost, availability, reliability and sufficient performance matter more than maximum transistor density.

What Is the Next Version of IIT Delhi’s GPU?

The research team is exploring an 8–16-core vector-style graphics processor architecture.

This would represent a substantial expansion from the current micro-GPU demonstration.

The roadmap also includes development of an optimized compiler and graphics software toolchain.

IIT Delhi Micro-GPU Development Roadmap

Why Does the IIT Delhi GPU Need a Compiler?

Hardware alone does not create a practical programmable computing ecosystem.

Software tools must translate higher-level instructions into operations that the processor can execute.

A compiler and graphics software toolchain could make the architecture easier for developers to program and integrate into real products.

This is why the IIT Delhi roadmap extends beyond simply increasing the number of processing cores.

What Is a Vector-Style Graphics Processor?

Vector processing allows a processor to perform similar operations across multiple pieces of data efficiently.

Graphics workloads often contain mathematical operations that can benefit from this type of parallelism.

The proposed 8–16-core vector-style architecture indicates that IIT Delhi is exploring greater parallel-processing capability in future versions of the design.

What Applications Could Use IIT Delhi’s Micro-GPU?

IIT Delhi has identified several possible applications for the architecture:

• Industrial control displays
• Low-cost human-machine interfaces
• E-rickshaw dashboard navigators
• Inland-water navigation terminals
• Navigation systems for small fishing boats
• Educational e-book readers
• Affordable embedded visualization systems

These examples show that the research is aimed at practical embedded graphics rather than premium gaming hardware.

Could the Micro-GPU Be Used in E-Rickshaw Dashboards?

Yes, this is one of the applications specifically identified by IIT Delhi.

A programmable graphics processor could potentially support dashboard visualization and navigation interfaces in low-cost electric mobility platforms.

Any actual commercial deployment, however, would require further engineering, system integration, testing and product development.

Could the IIT Delhi GPU Be Used in Fishing Boats?

IIT Delhi identifies inland-water navigation terminals for small fishing boats as another potential application.

This reflects the project’s focus on affordable digital-access systems where a compact display processor may be more appropriate than an expensive high-performance graphics platform.

Could the Micro-GPU Power E-Book Readers?

Educational e-book readers are also among the use cases identified by the research team.

Such devices require visual interfaces but generally do not need the enormous graphics capabilities of gaming or AI accelerators.

A compact indigenous graphics architecture could therefore be relevant to affordable educational hardware if it progresses through the additional stages needed for product deployment.

Can IIT Delhi’s Micro-GPU Run AI Models?

The announced research focuses primarily on programmable graphics and display processing for embedded applications.

GPUs as a broader technology category are widely used in artificial intelligence because their parallel architectures can accelerate many mathematical workloads.

However, IIT Delhi has not presented the announced micro-GPU as a replacement for modern high-end AI accelerators.

It would therefore be misleading to assume that the current FPGA prototype can perform the same AI workloads as contemporary data-centre GPUs.

Can the IIT Delhi Micro-GPU Be Used for Gaming?

The current project is not positioned as a gaming GPU.

Its intended applications involve embedded visualization, dashboards, control displays and affordable digital systems.

Gaming GPUs require extensive graphics pipelines, memory bandwidth, drivers, APIs and software ecosystems in addition to substantial computational resources.

IIT Delhi Micro-GPU vs AI GPU

Why Is Indigenous GPU Development Important for India?

Graphics processors have become important components of modern computing systems.

Beyond conventional graphics, GPU architectures now play major roles in artificial intelligence, machine learning, scientific computing and advanced visualization.

IIT Delhi’s announcement notes India’s dependence on imported GPUs and positions indigenous graphics-hardware research as one possible contribution toward reducing technological dependence over the longer term.

A single university prototype cannot eliminate import dependence, but it can help develop expertise in processor architecture, digital hardware, compilers and semiconductor-system design.

Does This Mean India Can Stop Importing GPUs?

No.

The IIT Delhi project should not be interpreted as an immediate substitute for the large commercial GPU ecosystem.

The current development is an academic FPGA-based micro-GPU prototype aimed at embedded applications.

Modern commercial GPUs involve sophisticated chip fabrication, advanced packaging, high-bandwidth memory, complex driver stacks, software libraries, large developer ecosystems and enormous engineering investment.

IIT Delhi’s development is better understood as a step in building indigenous graphics-processing knowledge and intellectual property.

Why Is Semiconductor Self-Reliance Relevant to the Project?

Processor architecture is one part of the broader semiconductor value chain.

A country seeking stronger domestic semiconductor capabilities needs expertise across chip design, electronic design automation, verification, fabrication, packaging, testing, embedded systems and software.

University projects can contribute by training engineers and developing indigenous processor intellectual property.

How Could Students Benefit From the IIT Delhi GPU Project?

Prof. Kaushik Saha highlighted the educational value of integrating arithmetic hardware, programmable architectures, compilers and embedded-system thinking within a unified framework.

This type of project can give engineering students experience across multiple layers of a computing system instead of restricting them to software or circuit design alone.

What Research Areas Does Prof. Jayadeva Work In?

IIT Delhi’s Department of Electrical Engineering lists Prof. Jayadeva’s research areas as including machine learning, neuromorphic engineering, VLSI design and optimization.

He is also listed as Head of the Department of Electrical Engineering and holder of the Microsoft Chair at IIT Delhi.

What Is VLSI and Why Is It Relevant?

VLSI stands for Very Large Scale Integration.

It concerns the design and integration of large numbers of electronic components into integrated circuits.

Knowledge of VLSI design is highly relevant to processor architecture and the eventual transition of programmable-hardware designs toward fabricated semiconductor chips.

What Is Embedded Computing?

An embedded computer is a computing system integrated into a larger device to perform a specific function.

Examples include automotive dashboards, industrial controllers, medical devices, navigation terminals, smart appliances and electronic instruments.

Embedded systems often have stricter requirements around cost, power, physical size and reliability than desktop computers.

These constraints explain why a compact graphics processor can be useful even when it is far less powerful than a desktop GPU.

Could IIT Delhi Commercialize the Micro-GPU?

Commercialization is part of the longer-term vision, but it has not happened yet.

IIT Delhi says the research team plans to seek funding support for:

• ASIC development
• System integration
• Further technology development
• Eventual commercialization

Moving from a university FPGA prototype to a commercial semiconductor product would require several additional development stages.

What Must Happen Before the Micro-GPU Becomes a Commercial Product?

1. Further architecture development and optimization
2. Expansion toward the proposed multicore architecture
3. Compiler and graphics software development
4. Hardware verification
5. ASIC design preparation
6. Semiconductor fabrication
7. Chip testing and validation
8. Board and system integration
9. Driver and application development
10. Reliability testing
11. Application-specific certification where required
12. Manufacturing planning
13. Funding and commercial partnerships

What Are the Biggest Challenges Ahead?

Building a working FPGA processor is an important milestone, but moving to a deployable graphics platform creates additional technical challenges.

These can include processor scaling, memory architecture, power efficiency, compiler maturity, graphics APIs, driver development, verification, fabrication, packaging and application-level integration.

IIT Delhi’s announcement does not claim that all of these stages have already been completed.

Why Is the Software Toolchain Important?

Processor hardware becomes substantially more useful when developers have software tools that allow them to program it efficiently.

A graphics architecture can require compilers, libraries, drivers and development tools before it can support a broad ecosystem of applications.

The inclusion of an optimized compiler and graphics toolchain in IIT Delhi’s roadmap therefore indicates that the team is considering the processor as part of a wider computing platform.

Does IIT Delhi Already Have High-End GPU Computing Infrastructure?

Yes, although this is separate from the indigenous micro-GPU research project.

In 2026, IIT Delhi announced an AI-driven high-performance computing facility equipped with 400 NVIDIA A100 80 GB GPUs. IIT Delhi reported approximately 250 AI petaflops of AI computing capacity and 10 petabytes of parallel file-system storage for that facility.

Those imported high-performance accelerators serve a very different purpose from the institute’s experimental indigenous micro-GPU.

IIT Delhi GPU Research vs IIT Delhi HPC Infrastructure

Why Should the Two IIT Delhi GPU Developments Not Be Confused?

The presence of NVIDIA GPUs at IIT Delhi does not mean IIT Delhi designed those commercial processors.

Conversely, the indigenous micro-GPU project is a processor-design research effort developed by IIT Delhi researchers.

One provides advanced computational infrastructure; the other develops indigenous processor architecture expertise.

What Does This Development Mean for Engineering Education in India?

The micro-GPU project demonstrates how postgraduate engineering research can move beyond theoretical processor design into working hardware prototypes.

Students involved in such projects can gain practical experience in digital design, processor architecture, FPGA development, floating-point arithmetic, embedded computing and hardware-software co-design.

These skills are relevant to India’s expanding semiconductor-design and electronics ecosystem.

What Does the IIT Delhi Micro-GPU Not Yet Prove?

The announcement does not establish that the processor is already commercially competitive with mainstream GPUs.

It also does not establish mass-production readiness, commercial pricing, fabrication yield, benchmark performance against commercial GPUs or a finished developer ecosystem.

Those would require additional development and testing.

Key Confirmed Facts vs Future Plans

IIT Delhi Micro-GPU FAQs

1. What has IIT Delhi developed?

IIT Delhi researchers have developed a working, demonstrable, indigenously designed micro-GPU architecture implemented on an FPGA.

2. Is it India’s first GPU?

IIT Delhi’s precise claim is that, to the researchers’ knowledge, it is the first working, demonstrable, indigenously designed micro-GPU from a university in India.

3. Who developed the IIT Delhi micro-GPU?

The project was led by M.Tech. students Nammi Akash and M. Ravi Teja under the guidance of Prof. Jayadeva and Prof. Kaushik Saha.

4. Which IIT Delhi department developed the GPU?

The project is associated with IIT Delhi’s Department of Electrical Engineering.

5. Is the IIT Delhi GPU a real hardware prototype?

Yes. IIT Delhi reports programmable graphics rendering on a Spartan-7 FPGA platform.

6. Is it already a silicon GPU chip?

No. The currently announced implementation is FPGA-based.

7. What FPGA does the project use?

The architecture has been mapped onto a Spartan-7 FPGA platform.

8. What language was used to implement the GPU?

IIT Delhi says the custom floating-point GPU engine was implemented entirely in Register Transfer Language or RTL.

9. What is RTL?

RTL is a hardware-design abstraction used to describe how digital data moves between registers and how logical operations are performed.

10. What is FPGA?

FPGA stands for Field Programmable Gate Array, a configurable hardware platform commonly used to prototype and implement digital architectures.

11. Is the micro-GPU designed for gaming?

No. The announced focus is affordable embedded graphics and display-processing applications.

12. Can the IIT Delhi GPU replace NVIDIA GPUs?

No such claim has been made. The current micro-GPU serves a very different embedded-research objective.

13. Can it be used for AI?

The announced project focuses on graphics and embedded display processing rather than positioning the prototype as a high-end AI accelerator.

14. What applications could use the micro-GPU?

IIT Delhi identifies industrial displays, human-machine interfaces, e-rickshaw dashboards, inland-water navigation terminals, e-book readers and other embedded visualization systems.

15. Can it be used in e-rickshaws?

E-rickshaw dashboard navigation is one of the potential applications specifically identified by IIT Delhi.

16. Can it be used on fishing boats?

IIT Delhi lists inland-water navigation terminals for small fishing boats as a possible application.

17. Can it be used in educational devices?

Yes. Educational e-book readers are among the potential use cases identified by the researchers.

18. What is the next stage of the project?

The team is exploring an 8–16-core vector-style graphics processor architecture.

19. Will IIT Delhi develop a compiler for the GPU?

An optimized compiler and graphics software toolchain are included in the research roadmap.

20. Will IIT Delhi fabricate the GPU?

The roadmap includes eventual migration toward a proof-of-concept ASIC.

21. Which semiconductor process is being considered?

The researchers mention a 65 nm ASIC process node.

22. Why use 65 nm instead of a newer process?

The researchers believe mature process nodes such as 65 nm could provide an economically viable route to useful indigenous graphics silicon for embedded applications.

23. Is 65 nm suitable for embedded systems?

Mature nodes can remain practical for applications where affordability, availability and sufficient performance matter more than leading-edge transistor density.

24. Is the IIT Delhi GPU commercially available?

No. It remains a research development.

25. Does IIT Delhi plan to commercialize it?

The team says it plans to seek support for ASIC development, system integration and eventual commercialization.

26. Has funding for commercial production already been announced?

IIT Delhi says the team plans to seek funding support; the announcement does not state that commercial production funding has already been secured.

27. Does the project reduce India’s GPU import dependence immediately?

No. The prototype represents an indigenous research milestone rather than an immediate replacement for imported commercial GPUs.

28. Why is the project important for India?

It contributes to domestic expertise in processor architecture, programmable graphics hardware, VLSI, FPGA development, compilers and embedded systems.

29. What is graphics processor IP?

It refers to a reusable graphics-processing hardware architecture that can potentially be implemented or integrated into different systems.

30. Can the architecture be mapped to both FPGA and ASIC hardware?

Yes. IIT Delhi states that the architecture may be mapped to programmable hardware such as FPGAs or to silicon ASIC implementations.

31. What is an ASIC?

An ASIC is an Application-Specific Integrated Circuit fabricated with hardware designed for a defined architecture or application.

32. Why prototype on FPGA first?

FPGA prototyping lets researchers validate and modify hardware architectures before undertaking the more expensive process of fabricating dedicated silicon.

33. Is this a student project?

The project was led by two IIT Delhi M.Tech. students working under faculty guidance, making student-led research an important component of the development.

34. What areas of engineering does the project combine?

The work combines arithmetic hardware, programmable processor architecture, graphics processing, embedded systems and compiler/software considerations.

35. Does IIT Delhi have other GPU infrastructure?

Yes. Separately, IIT Delhi has announced high-performance computing infrastructure containing 400 NVIDIA A100 80 GB GPUs.

36. Is that HPC system the same as the indigenous micro-GPU?

No. The NVIDIA-based HPC system provides computing infrastructure, while the micro-GPU project involves designing an indigenous processor architecture.

37. How powerful is IIT Delhi’s micro-GPU?

IIT Delhi’s announcement does not provide a standardized commercial-GPU benchmark that would support a reliable direct performance comparison.

38. How much does the micro-GPU cost?

No commercial price has been announced because the technology is currently a research prototype.

39. When will the IIT Delhi GPU become commercially available?

No commercial launch date has been announced.

40. Where can readers check the official IIT Delhi announcement?

The official IIT Delhi press release is available at:
https://home.iitd.ac.in/show.php?id=485&in_sections=Press

Why IIT Delhi’s Indigenous Micro-GPU Matters

IIT Delhi’s micro-GPU should be viewed as an indigenous processor-design milestone rather than as a finished challenger to the world’s largest commercial GPU companies.

The researchers have demonstrated a programmable graphics-processing architecture using a custom floating-point engine implemented in RTL and mapped onto a Spartan-7 FPGA. That provides a physical platform from which the design can be expanded, optimized and potentially migrated to silicon.

The project is particularly relevant because its ambitions extend beyond a single hardware demonstration. The research roadmap includes an 8–16-core vector-style processor, an optimized compiler, a graphics software toolchain and a possible 65 nm ASIC proof of concept.

Equally important is the intended market. Instead of beginning with expensive gaming or data-centre applications, the researchers are considering affordable embedded systems such as industrial displays, e-rickshaw dashboards, navigation terminals and educational devices.

There is still a substantial distance between an FPGA prototype and a commercially manufactured GPU. ASIC fabrication, verification, software development, system integration, funding, testing and commercialization all remain future challenges.

But the IIT Delhi development demonstrates something important at the university-research level: Indian students and researchers can design a programmable graphics architecture, implement it at RTL level and demonstrate it on physical hardware.

If the project successfully advances toward multicore architecture, compiler support and silicon implementation, it could contribute not only to a specific embedded graphics product but also to the broader development of indigenous semiconductor-design expertise in India.

Official source:
https://home.iitd.ac.in/show.php?id=485&in_sections=Press