 So we have the microchip and this is the FPGA solution right here Hi, hi, so who are you? I am Diptesh Nandi. I take care of the industrial vertical marketing for the FPGA BU so this is Previously micro semi which part of microchip now FPGA solution. What is this? Well, what we are demonstrating over here is a splash kit It is a limited feature set evaluation kit using which you can evaluate various functions of our next generation the current generation FPGA called Polar Fire The demo that we are showing over here is essentially a deep learning demo What we're trying to show is a tiny YOLO algorithm That is running on the Polar Fire FPGA and the hardware platform that we're using is the Polar Fire splash kit So you're doing a deep deep learning on the chip, but it's not a high power consumption or right? So Polar Fire FPGAs or micro semi FPGAs traditionally have been known for their low power We have flash based FPGAs and Essentially what that enables us to do is to consume up to 50% lower power than competitive FPGAs So that advantage is also derived in a machine learning solution Where we can run a machine learning algorithm up to two to four times Better in terms of giga operations per second per watt feature So that sort of differentiates us With other FPGAs that are available in the market and we are essentially trying to convey that Hey, we can do much better performance at much lower power footprint Does that mean you also two or three times less per performance? No, it does not mean that you have the best performance per watt I have we have the best performance per watt and that has also got to do with the math block architecture that we have So in our math blocks, which is the essential core feature for deep learning We can do up to four operations per cycle. What it essentially translates is we can do up to 25% higher efficient math blocks And that enable that along with the low power consumption allows us to do two to four times performance per watt in terms of FPGA acceleration of machine learning And So is the news here that now it's available or what's special right now at the embedded world? So there are two announcements that we are making of course polar fire Has been in production since last year end of last year So we have polar fire hundred two hundred and three hundred parts that are available for order entry We are also launching another kid a video development kid and that is what we're showcasing in a different book And is it around here? Yeah, can we go? Yeah, okay. So how far let's go just from the corner Okay, so this is a this is a big it looks like a bigger chip It is not a bigger chip. It's the package of this chip is slightly bigger That's the value that we bring across we are telling to our customers that they do not need a heat sink They do not need a fan because we are up to 50% more power consumption than any other FPGA What we also support is very small form-factor To 11 cross 11 mm packages So what we're saying is that you can put our FPGAs into very small form-factor cameras And since anyways in small form-factor cameras, you won't have space required for heat sinks or fans Our FPGAs will give you the best performance So what is this one doing? This one is a new kit that we are launching In embedded world It has two image sensors that goes into an FPGA through a VPC-SI to interface and We are using HDMI display output To showcase the display here in an external monitor now this kit is meant to be a Complete system solution enabler for our embedded vision customers We have additional features like FMC, FMC port using this customers can Evaluate SDI, Ethernet or USB connections. We have VPC-SI transceive functions We have display serial interfaces using which you can connect to Touchscreens and we also have HDMI 2 ports, which does not use any external files You can the Polar Fire can directly drive HDMI 2.0 solutions all in one chip So essentially for any vision customers if you provide This kind of a hardware platform, they will be able to Design and develop and evaluate Polar Fires performance in such kind of applications Where is the market? What is this market? This market is Spread across verticals in the consumer segment you have camera manufacturers frame grabber manufacturers in the industrial market You have industrial machine vision camera manufacturers and frame grabbers in aerospace and defense this can be Audio and AR or VR headset or a binocular Manufacturer so this has application across verticals. This has applications across Different kind of applications and there's an arm logo on this This arm logo on the on the on the chips that means there's what kind of arm is running here So we have a soft core arm That means the soft core Soft core means the Polar Fire does not have a hard CPU hard controller inside So we run a soft core ARM Cortex M1 along with RISC-5 soft core So we support both the different types of processors M3 you mean or it's M1 M1 Soft soft as a run like just on the FPGA. Yeah, it's runs on the FPGA right to do whatever Software you want you can do whatever on software you want with the M Cortex M But then over there you're talking about a RISC-5 implementation So, how's it gonna be different? So so so we want to serve both set of customers now The reason we have chosen to go along with RISC-5 is because we have definite set of customers Who want to evaluate the complete instruction set the RISC-5 provides now? If you partner with RISC-5 then they can do deep sensing of our instruction set And they can be absolutely assured that the instructions that they're getting is Applicable for safety critical critical applications now There are other set of customers who develop very small core RISC-5 They need very small core RISC-5 implementations They might not need extra features that ARM provides, but moving into a RISC-5 They can change the architecture without have to pay without having to pay a lot of architectural licensing fees So that's the reason we are moving towards this file So it's about a license, but you're talking about soft Implementation of the R. Yeah, so you're gonna do a soft implementation of the RISC-5 too. It's not a hard So this polar fire also supports soft respect implementations We have three ports available right now with 32 IMA and data and cache support With both the AXI bus and HV bus interfaces and all those three are part of our MI5 MI5 ecosystem and is offered free to our customers for evaluations