Hello everyone and Good day. My name is Mohammed Qasim. I am the co-founder and CTO of Eve Havelist and Thank you very much for making the time to join the presentation and I hope We follow up with questions that are useful to everyone

So I'm gonna give a little bit of an introduction about myself and basically I come from the chip world from a long Career in the in the wireless industry my background is in analog circuit design and I joined Texas Instruments when TI was starting the era of the

chip development for smartphones in general and then the Smartphones later and if you know a processor like OMAP then you'll find that the analog on OMAP

Partially 3 and OMAP 4 and a bit of OMAP 5 were my responsibility So if you hated or liked it, you may blame it on me And here are some of the devices that I was involved in indirectly at some point Yeah, I also I'm a thinker and I'm a hardware

enthusiast and and always design and look for what the chips do and in smart devices Or any devices, so that's my habit from since I was a kid. I Open things down. Sometimes I when I was a kid I didn't actually I wasn't able to bring him back

but now that's my I tear down the device device doesn't understand how they're constructed and what chips are in and That's a passion of mine as well pretty much for a lot of the new devices that are out there So I'm going to dive into the talk right away and

and it's basically talking about the chip design and how we can approach approach the problem of the inextricability and the restrictions to innovations For this industry, so I have a friend of mine who just put together this

this little abstract view and On the left side it says and it shows that as the process technology evolves from down in the feature size The development cost of the process and the designs around it they increase actually exponentially the number of

Capable companies or people who have access to this technology and can design with it goes down And that means the number of generally number of designs goes down So it's a very simple view now when the number of designs goes down or designers rather go down that means the The number of or the freedom to actually tinker and have more ideas

Move into silicon becomes restricted which in general restricts innovation This is another view. It's a financial view and this would probably be the last financial slide on this It is actually in it shows us over the years just the number of companies

Companies not individuals even that have the access and to advanced nodes Technologies or process technologies. It's just going down over time and that didn't change really over beyond 2018 since we're now in 2022 and the

The That by itself looks like the car companies as you as it converges you end up with a few car companies that have the market and The people who have ideas in the cars and engines They're just can't just come up with an idea anywhere and just say it works

so This is just the dynamics of the industry now on the other side of the of the industry the need So there's so-called the long tail I'm representing here with Orange and I intended to drive this way because I usually call it the very very long tail

And what is that is that it's the customized applications for specific markets or? specific use and typically people today the hardware developers they use what's available in the FPGA world or standard devices from

Standard semiconductor either processors or chips, which typically leads to different like living with the form factor or extra capability on the chip or I need the five Regulators, but I or three regulators, but I'm getting six or the other are just I need one extra

So there is what I call here and also that's a term that I got from somebody Friend of mine, it's a rice-sized compute my size meaning that I can for every problem that I am facing I you'd like to size the chip or the

Compute capability and the power especially if it's a battery-powered device to the application now on the left-hand side This is a traditional approach where smartphones have tens of millions Maybe you know tens of millions of units or hundreds of millions of units, which is

Justifies the development cost around these applications The problem with the yellow or the orange part is that there are lower volume and generally speaking It is hard to justify a cost of an ASIC for these orange areas now By and then and then the typical answer for that is the standard product

Then now we believe that The number of these applications are going to be tens of thousands so if you want to achieve both the right size compute and the tens of thousands you really need designer or You know, you can't just they have one standard product and you need designers around the world, which is the

problem so we want to have a thousand X developers roughly to be able to Deliver to that Now when people look at the link tail, they keep asking me. What's the killer app? I'm like, well, there's no killer app I don't know what the killer app is

the only way to have this work is to have people try and Communicate with the market or the users the end users and come back with iterations So in order to do that it was happening in the software world It is happening software world in a way that is efficient so that you can compile something

make it an app and put it out and get feedback and Improve and you can see that in the software world the only answer for the demand to meet the demand for the customization of hardware And chips is a community-based approach meaning that instead of saying I need a company or 10 companies to deliver to the world

No, I will just say we need the world to deliver to the world. We need to community Around the world to learn and become capable of making the custom applications and it's actually grows you know with the With the demand and it also has variety of expertise there. You can tell from the open source community. For example

You can a myriad of expertise that can actually deliver Value and then the last thing is the natural selection. This is beautiful Is that people choose what they like to work on and what they're passionate about and probably what they're good at

Versus working in a restricted environment where you just have to do what you have to do Now, I'm gonna jump to e-fab list. It's a quick overview and why you have this exists I'm just gonna say it's for meeting that goal So in order to get these thousand X Counts in

designer We need to simplify the process typically the first process of making ships is very complicated. So We want to simplify to make it easy and adoptable by a lot of people And again, what is the number 1000 X don't quote me on it But just the multiplier that it is actually hard to achieve with the current traditional

Approaches so yeah, so we need the thousand X and maybe more maybe less But we need a huge multiplier of the People who are able to design now with either we're going to have to educate people or Change the concept or simplify the process

We need to have a combination because actually it it's time to simplify the process because I don't need to know Everything about the chip to be able to create a chip. That's You know actually Achievable a little bit through the FPGA. So that's what we want to do The example that I'm going to use here is that in the software the app stores

the the app stores before the app stores in order to develop a piece of software and market it and sell it or Expose it to customers. It's a very hard process When the app stores showed up, whether it's iOS or Android What happened is the there is a development tool set that is very

robust and Has unreliable and and then there is a business process structured flow of Structured flow quality control or at least feedback and the connections customer so now if we can do that then

You can see you remember in the in the software or the app stores it it just exploded into millions and millions of developers anywhere from You know kids all the way to Anyone any age that wants to do something on their own without have to worry to actually

structure a company or so and the process also becomes has become simplified because of the the Tools that are available now in the chain in the chip world If you think some of the chips are actually this you can be designed by one to four engineers and

There are chips that require a hundred people but my experience says you can get a lot done within Ten people if they know what they're doing and that's hard to bring in one room if you're hiring maybe but if you're Working with the community. You'll probably find these for, you know, ten people to do something really great Now in order to do that we need to democratize the access to the needed what's needed to design

It's not as simple as a development kit and then again business process and the connection to customers. So Now how this is how we're gonna go now in order to get that there are a little bit It's much more complex actually than just a development kit because it's not one owner. It doesn't come from one place

like Like Apple Google or other open markets places. So in order to produce a useful chip and That is usable by The users and could be the you know, the youth end user of a smart lock, you know

Something that fits in an air purifier or air quality sensors System you need to ask to know the access to the market You need to have knowledge that you may not have IP or blocks that are currently

Gated by generally NDA's and on access cost EDA tools similarly PDK and affordable manufacturing and a reduction of this of the cost of manufacturing So the there are partial solutions that exist and In their admirable work pieces, but it still doesn't provide that you know

I wanted once for all so what we wanted to do is a complete holistic solution and In order to do this, it's basically boiling the ocean So the right thing is to do it one at one at a time by solving everyone

with a different tactic and Generally speaking I will leave this the comments me for the questions That's how we did so the key is that instead of having just a developing kit then it becomes like a stack-in-a-box basically that it can be available to everyone and designers can collaborate and define and develop and commercialize their products for

application for long-term applications To start we started with the EDA. That's one aspect of it and the One of the things is that we want to basically to give the EDA Flow to anybody anywhere without cost or without the permission or a violation of any licenses

So we assembled a great variety of tools for analog and digital design From open source engines and we worked to create to use them to develop real chips We also on the IP The this is kind of a recent view. So these are examples of the places you can find

I'm sure there are more and maybe I've missed but I just quickly I Think given the programs that we're working on right now. You can go to you have to see what people are doing in in actual Silicon IP that is Submitted into the Google projects for example, and then there are other examples

for accessing IP in a structured way like labor cores or QSOT Generally Google is good to find. I'm always surprised with what individuals or groups have been doing in Improving what's available for design now in order as I said in order to improve that this is actually work

we we took it to put on ourselves to Develop a six and we started with a chip called Raven this chip is a base a risk five base the people are based on a PQRP of 32 from Claire wall and

The IP and or the blocks are analog are coming from ex-fab foundry and this chip is in 180 nanometer and What one thing we intended to do is to make that chip top-level open source So if you go to get hub at this location, you're going to find the top-level very long is there including behavioral models for the analog so you can simulate the entire chip in a

Digital simulator actually would be analog presented in real numbers. We also developed another chip called Ravenna. This was a based on a request of our customer and it has more resources, but it's also open source and

the that it's available for the Tinkering and adding additional features both of them are available on our platform so you can access in black box mode the IP These chips are Here we have four major chips that we started with now

On the left is that Raven and Ravenna risk five microcontrollers and then the top right is actually an open source Chip but the core is arm based So it is not when I say on open source chip I mean, I mean that the the top level connectivity or netlist now it has a lot of

Components that are not open source, but it is a Representation or a step forward in that direction and then in our bottom, right? This is a chip called Hydra. That is purely analog To for a characterization testing all of these it's important to say all of these have been completely designed and verified with 100% open source tools

period the Raven ship Was actually demonstrated in different places we partner with other people to make it a part of the other integrations and Then this is the Hydra chip just for those who want to take a look at it. Later It's an array of analog chip. One of the interesting things is that Raven as a top-level

Open source chip and the core being open source is the NEC one of the NEC research labs looked at it as this is an ideal for Trust through transparency So they wanted to look at as use it as a kind of an authenticator for a co-processor for a little bit of a rooted

Trust and they requested Raven we send them the boards and they developed several applications as you can see here but they also specked with us the Urbana chip and We delivered it to them Now I will this is what general overview of if I have list

but now as we published a raven chip all over the world and we had the attention of different places Especially Google and then Google's philosophy in this involvement was if you can have a continuous integration in software Why don't we have an continuous integration concept in the hardware or special the chip world design something?

Build it and measure it learn from it and then do it again and keep iterating in order to do that You really need to have a cost structure and availability and access to make it Worth their what so the Google program was it started as Google would fund six manufacturing runs

And that means in and you'll see why it's 240 chips And now when I say at least it's because it's actually getting more now There were four last year and there's another at least another for this year 22 and The first of which is in March 21st this year

This the submitted designs and by the way just to get into this you don't need to Go through big so you can start designing go to the list of calm just show the intent that you have a project and just Write the abstract of the project and then some if you finish Before the deadline meaning that you pass all the pre checkers

Then which are available automated and open source to you Then you're in so it's a first come first serve and Google's purpose from that is to just mobilize that the Wheel in terms of having people to learn and design and learn and and reiterate so

If your design is perfect, then you're not taking too much risk or taking risk So it is okay to get in and learn something. So Don't be intimidated by that chip concept because it is We've proven that, you know from a variety of ages and a variety of software and hardware experiences

things worked out here are some of the links you can go and start and we've proven as I said many of the users and designers have been Not almost half from the Google program Haven't designed the chip before and they have chips and they actually demonstrated a success

For more information go get it and one of the program I go to a fabulous of comments the the interesting things about the Google program is that The tools are open source and then you get a chip That is as a reference or a platform that you can modify

You can add to it your site and then you get development boards Not just the parts and abundance of parts. Obviously if you want to build your own boards The key component of that program is a partnership with Google Skywater and tablets the Started with the open sourcing the PDK which is a representation of the process technology

It's the models is the the libraries the the I always The different views and documentation so it is available without an MDA Which is a never happened in this in the history of this industry as foundry We actually open source the PDK of our reliable process as it's just to give our background

If you're familiar with the piece soft devices from Cyprus semiconductors Skywater, it was a spin-off is a spin-off from Cyprus back in 17 I think and The process that is being used the 130 nanometer is the process that's being used for piece of devices. So it is it is

truly professionally done for industrial grade manufacturing Now why people say what so what is a 139 and I work personally on 130 nanometer 17 actually almost 20 years ago. That was my part of the development of the analog back then well

Until today you can do a lot if you're gonna see go see what people did today since the last three years On this link. However, just an example I want to bring up is you know, if you look at the Intel chips themselves back in, you know You know back in this, you know the age of 2001 Which is roughly about 20 years ago, you'll see that even 180 nanometer you're able to get

the speed of it, you know 1.7 gigahertz, of course Intel has a lot of this customization and so to get there But I'm saying that transistor pitch doesn't mean you're gonna you know, you won't get performance and also some of the data that are

Bored from our friends in your practice. They have a very good annual report It shows here that in the microcosm of use in semiconductors from a Europe practice standpoint There's about 50% of the process technology or their project on the process or their customers. They're below

something that's bigger than 90 nanometer which is 130 110 130 and more and can see it now a part of the project for Google is that we develop Almost a new compiler, right for digital

Functions, so that means if you know very long then you develop something in very long and recently we made that in Python using something like litex Did you compile your code and you get a functioning design? Maybe they're not the best performance. Maybe not the best area, but it works so you can iterate and

The intent here was to make it available the Google partner on With me is immensely always says I'm a software guy. I don't know what a DRC is. I don't know how to fix it I just want a complete design that works and manufacturable and I can learn from it So if you do that and you make it look like a compiler

I use the new compilers all over the place But I don't know how they work exactly and I get software that I actually use so if you apply this concept on the Digital design well, that means Software developers should be able to do that without a bling and that opens starts open the thousand X more at least

People that are capable of making a digital design Now it's also in the analog design same if you if you learn the prim the the basics or you want to go further also That actually moved there with the great help from the community

It's It's become possible now both of them This is actually a code again in the community. So we're working closely. This is a collaboration between the open road team and The flow it's called open lane that we have this developed and it is on several process technologies available now

Analog is the same and it's a collaboration with me. For example in this year Design steam and just by steam and the X team developers different shippers Now part of the project also to simplify so all of the things I'm going to notice

I'm going simplifying layer by layer. We use the concept of caravelle because that's the it's a When you look at it or look it up or it's a ship that has specific properties that the Portuguese use In back in the 15th 16th century well caravelle is a chip that is almost like a

Ship that has a cargo space or project space that you can put your own design in it. It has resources It's an open source chip and it's using Components from like open RAM and the big rb32 where which are open source

Generators and and IP and this chip will also open source and you because of the process is open. You'll see it all over the It's open on github and here's the some of the features of the on chip Features that you can utilize for your design. The way it's being used is this you the designer

designs the block and then it gets dropped into the chip and We then that produces a user specific caravelle the designer only focuses on the left side of the slide They don't have to worry about the caravelle itself except from the models and the behavioral

verification to make sure that things work and we plug it in into the Caravelle master now the The designer also gets actually 300 packaged parts plus five ports and I mentioned that early So it is straight up you can push it into the

the Is USB connector and Get working and the board is like here. So this is one example of the caravelle boards with a specific design on it now we Way, there are multiple directions multiple tracks that we're improving on right now again with a collaboration with the community

There's also one piece of news here is that because of what sky water Foundry did it got the attention of other foundries and then there will be you know There is a plan for a second boundary and you may hear about announcement of that suit Just the engagement here. We can see when the

sky water PDK was open it gave us like basically it's like a barrier of behavior against projects that are accessible for years and Also, it gets downloaded about 700 times a week. We also have the slack community that is rich with almost 2,500 people you can join and get your self-invite from there

Now the first shuttle went with Google was an interesting one. We called it in pw1 you can see the designs up there too, but it had a variety of You know Contributors from individuals to companies. It was quickly booked again. It's free for the designer and

funded by Google for the time being and then Bartlett w1 was actually to actually get this picture done. It's the first time The the open source hardware at least to my knowledge They get engraved on actual chip that is totally open source all the way to GDS No, there is no compromise here and in true just to mention the community contribution

the John McMaster here who's a pictured in the bottom left is actually took that chip decapped it and he has an electron Scanning microscope at home to actually do that. So that's kind of a community interest and then MPW to again

I'm not going to go all over the whole design, but you can see the links if you go to the link below you'll see the full overview and as well as the GitHub repositories, this is just a Candy for the I to see the two different 80 chips plus the test chips for the foundry up there

Now now the Google when we found that it's very important that people say I want to open source my design And I want to guarantee my spot and I want to control my schedule. So we created an offering called chip ignite Chip ignite is basically the same thing for Google just Shifted into a fabulous completely and to you actually to the designer. He says, okay, I'm gonna

I'm gonna get my own design. I don't have to open source it or not. That's up to me and then I will book my space for 200 bucks starting designing Start the line and then you will get actually 10 millimeter squares inside the chip, which is I mentioned earlier

But it is actually it's not a real state I always say it's a framed house with plumbing and electricity and then you bring your own appliance and these are some of the Pricing here, but the pricing is actually intended to not to show the dollar numbers. That's it through that comparison You can either get a hundred parts with a QFN

300 with the WSP which has never happened or if you're doing a low volume production you can pay 20,000 and you get a thousand units, which is Almost basically $20 per chip in all includes They these are options of chip ignite. Some of them are the same the same as the chip ignite as

The area as it is and then we got requests to say okay I don't want to use the management area or the logic here. I own an area all myself So we started offering that as well. Here is the schedule for chip ignite and And MPW's from Google, so these are two chip ignites

Data points one of them is April 8 the Google shuttle that's upcoming is March 21st And the next should ignite or second chip ignite in 2022 will be in June Go, you know can go to efellows.com to see the details the users of chip ignites not the Google program basically started with the universities there

Here Stanford University used it for a course there was a basically the whole course is design and end up with a Go to fabrication and then you come back and test it also die triple-e. They created a competition So the they actually got can chip ignites and then it was a great competition

You can you can view it and see how different we had like 50 for six proposals in it great designs Startups also said okay. I can use chip ignite as it is Basically, I can put my own designer like the blue in the middle here design and then I have a product basically for low volume

I don't have to optimize the area because it makes sense economically and I bet I can get the market very quickly to my custom This is back to the IEEE also shows that these were ten projects from all over the world The last six project you can see how people utilize one slot to multiple designs

Instead just having one slot for one project This is the part where I showed the university there's multiple aspects of the universities have started talking to us to actually adopt that either in the Courses or in parallel sessions like the capstone projects and graduate research and that's it

So I close here and then when you get the slides You'll find I'm gonna flash through this very quickly fun from the community that I borrowed and Just share different people have posted different pictures and you will be able to flash through them And these are both contributors from different places and they know the names

But you can find them also on Twitter and it shows how passionate and Competent the community and to actually get something done. This is a RAM test chip that you know Andrew Zonenberg Created a board for testing it

This is a met then obviously people know met then they used one slot to create a course that has 16 slots this is the Paraville, this is us. Actually. This is me on my desk and then other people like TNT or Sylvain Money this you know show you see the quality of the sharing how people share things and

incredible never happened before and Get to chance just to take a look at this and that's my last slide Thank you for questions. So I look actually

Okay, now I can hear

Thank You Mohammed for a great talk And if you want to read any questions from the chat You'll be welcome. For instance. Let's see

How much will that normally cost if not using chip ignite? Say another shuttle well, the most important thing I would like to say here is that

it the word shuttle is 10 about The word shuttle is actually kind of tricky as I mentioned a shuttle typically is it gives you an area on silicon so if you want to compare it would be Wouldn't compare that's the area because you're getting an area with as I mentioned with

The older resources around it to make work like logic analyzer an actual chip around your design. So You definitely if you compare it, so this would be that 10 millimeters for 10 a thousand and If you compare at this level Even if the cost is the same, but you wouldn't get the same

Resources around it and packaged devices on ports. Okay. We'll have a question about When this lack will be bridged through IRC maybe it is actually kind of the bridge 100%

so if you go to The IRC servers you're going to find the space Sky water dash PDK and I shouldn't actually I should have maybe I'll send the link later wonderful and What else is either a reference design?

It is a reference design The design but in the it's open as well, but the the blocks in it are not however, you can replace them now with a lot of the MPW's coming from The blocks are coming from Google and chip ignite open source, all right, I think I'll ask a question myself

Okay, what kind of interfaces the the key to has It is high speed Can you put OSB maybe?

right now on the Board has USB you can connect it to the the computer and then it comes with an open source just Flashing software and and then you can go from there. Now. The chip doesn't have USB yet. At least it is Right now it's either a UART or a spy interface

Out there and on top of that GP the general purpose iOS Yeah, the general purpose iOS maybe they're high-speed at what speed For this process as they are they they're not high-speed They're actually 60 megahertz maximum as they specified by the original design. That's

Cyprus however, we have a flavor where you can Designers on on the shuttles actually created their own like LVDS designs So you there will be a variety of highest higher speed. I those are coming from the community

Shown you'll see it because people everybody is actually feels limited by the 60 megahertz Great So I think we'll wrap up And then anybody which wants to continue the link will be available soon To these two to their own where you could you can continually the shed. Thank you

Great. Thank you so much Have a good day, everyone