Monday, August 5, 2019
Quality of Time
Atomic Rules TimeServo product, introduced in 2016, first commercialized the idea of an FPGA system timer. The ability to coherently deliver time to any component within an FPGA, in the clock domain that time is desired in, is a common need which TimeServo fills.
We purposefully divorced the business of keeping time from the process of protocol-based time transference such as 1588v2/PTP and White Rabbit. TimeServo keeps time... period. There are both closed- and open- loop modes for the phase-accumulator we term BigPhase to operate in. For the closed-loop Digital Phase Locked Loop (DPLL) mode, we turned to mathematician Steve Gabriel to get the inherent structure and corresponding coefficients just as our customer's desired. Most TimeServo users stick with Steve's tradeoffs for settling-time and stability. But we left them programmable from the control-plane just in case.
It's really important to understand your requirements when it comes to Quality of Time. And as you press down below microsecond accuracy towards nanosecond accuracy, things can get difficult quickly. There's a lot of good spec work in the time-centric IEEE standards, and in how organizations such as the UNH IOL can verify and validate performance. We believe there are way too many devices on the market touting "sub-nanosecond accuracy" - when that's far from the case.
Wednesday, January 10, 2018
The Role of an Advisor
For the past nine years, Atomic Rules has been evolving. Evolving from a single engineer with a singular vision, to an amalgam of a dozen engineers with a team vision. Along the way we've had a lot of help. Help from technology firms like ARM and Xilinx producing the canvas that we draw on. Help from businesses like BittWare who represent our IP products with a worldwide reach. And of course Bluespec Inc, for Bluespec SystemVerilog (BSV), so that our codes would be correct and concise, in a way other Domain Specific Languages can only someday hope to match.
I started Atomic Rules back in 2008 solely offering FPGA design services. We picked up some great clients along the way. A few years in, the idea of producing IP products took hold. Well chosen, customer-driven IP cores would be the future of the company. Not for the squeamish, we knew going in that production IP would require an investment in verification and support significantly above that of just FPGA design services. We got that right.
As of January 2018, Atomic Rules now has three full-production IP cores on the market. Available through Xilinx, available through BittWare, and available directly from Atomic Rules. Hooray for us! With such deep and solid investment in products, tens of engineer-years in the making; it’s time for Atomic Rules to pivot and focus even more on sales, customers and product support; and less intensely on speculative R&D.
We will, of course, keep on our existing services clients, continue our relationships with the dozen engineers who make up Atomic Rules, and continue to support and develop new products and technologies. As such, we have a new Chief Operating Officer to do much of the day-to-day triage; and will soon announce a new Engineering Manager. It's all good!
This week I started a full-time position as VP of Research and Development at an exciting startup, Skreens Entertainment, and thus have relinquished my position as CTO of Atomic Rules. Skreens understands that Atomic Rules shall remain a going and growing concern, and I will stay on in an advisory role. And while my daily (and nightly) passion is now making Skreens all that it can be; I'm fortunate that I get to watch the Atomic Rules rocket climb up to orbit as well.
Saturday, June 3, 2017
FPGA with AXI plus GPP with DPDK equals Arkville
Are we done? No way! Work is already underway for the 17.08 release. Check back here in August.
Tuesday, May 2, 2017
Hoplite Comes of Age
FPGA overlays are a peculiar thing. The proliferation of spatial heterogeneity in reconfigurable computing devices (Trimberger FPL2007, et al) has effectively killed off the promise of a relocatable circuit module. Overlays fight back, in part, by delivering an abstraction that helps hide the spatial speed bumps.
Kudos to Jan, Nachiket and others who have advanced this work. They are delivering value to the community that may not be fully appreciated until we have our second 25 year retrospective.
Sunday, March 19, 2017
The Road to Arkville
We dont expect this to be an easy journey. Why? Let's take DPDK and AXI one by one.
DPDK is the de facto Linux Kernel Bypass mechanism evolving now for over a decade for use when you need to do useful work on multiple cores without the kernel stack getting in the way. Most DPDK users work with merchant ASIC-based NICs or virtualized NICs. Despite the "Data Plane" in DPDK, not all users see it as just an I/O mechanism. There's more to it.
Also over the past decade, AXI, a proper sub-set of AMBA, has emerged in the FPGA world as a standardized hardware interface and API. If you are building a Green-box fleet of RTL accelerators to go inside your FPGA; you probably want to use AXI for the plumbing. If not, you are probably writing gaskets with a throughput and latency tax.
Atomic Rules understands that both the DPDK GPP software world and the AXI RTL gateware world are two different things. And experts in one are seldom experts in the other. The guiding light for Arkville design has been to take the established DPDK APIs and ABIs and implement a DPDK- and AXI- aware packet conduit between GPP and FPGA. Conceptually abstract, but physically (at product launch in May) to be PCIe. We're excited about all of this, and if you are too, please get in touch. Since you made it this far, here's a hidden link on our site about Arkville in alpha-testing.
Monday, January 16, 2017
Paced Packet Player
Wednesday, December 2, 2015
Exciting Times
- 20 nm FPGAs are in production
- 20 nm FPGAs have 25 Gbps+ capable SERDES
- 25/50/100 GbE technology-enablers
- Able to serve 10/40 GbE by running SERDES at lower rates
- The P4 language is rapidly evolving
- Game-Changing Broadcom Tomahawk
- Game-Changing Intel Red Rock Canyon (RRC)
- Timing closure at 400 MHz through architecture
- A UDP/UOE product, first mover for 25 GbE
- 10/25/100 GbE L2/L3 stacks using production IP