Top 10 Lists
This is a semi-official, semi-serious, fully-enjoyable activity at FCCM, which has become a periodic tradition. It combines prediction of where the FCCM field is headed with a good bit of fun. Participants break into groups, and assemble a set of predictions for future FCCMs. All participants then vote on their favorites.
FCCM 2016 Top 10 Predictions for FCCMs in 2021
Organized by Mike Butts
11: Homogeneous FPGAs as we know them will no longer exist. FPGAs as we know them today will only show up as embedded components within complex, heterogeneous SOC devices.
10: FPGA-based machines begin colonization of Mars.
9: All large FPGA designs use NOC overlays or hardened NOCs.
8: Xilinx will be bought by another company (IBM? Google?).
7: Very few designers will use RTL-level tools – most designers of FPGAs will use high-level tools (C, C++, OpenCL) and/or IP integration.
6: Most large FPGAs have in-package high bandwidth DRAM.
5: FPGAs components will be a standard element of most high-end microprocessor architectures.
4: FPGA design tools will still need improvements.
3: FPGAs will have hardware double-precision FPU DSPs.
2: A large fraction of datacenter nodes will include FPGA logic.
1: The Top500 supercomputer list will include an FPGA-accelerated machine
FCCM 2011 Top 10 Predictions for FCCMs in 2016
Organized by Mike Butts
The top 10 predictions in 2011 took a technical turn with an email based submission and an online voting poll. 56 different predictions were submitted by conference attendees and posted during the annual FCCM demo night. While enjoying drinks and demos, FCCM attendees pored over the list and heartily debated the merits of the various predictions. On the second day and after a deep pondering, FCCM attendees voted on the predictions to make a top ten list.
A new feature of the top 10 is an award for the top prognosticator. The person that submitted the most top 10 predictions will be awarded the FCCM prognosticator award. For 2011, we had a tie for the FCCM prognosticator. Lesley Shannon and Andre DeHon both submitted two of the top 10 predictions and share the FCCM prognosticator award. More details about the FCCM 2011 top 10 can be found in the following presentation.
10. Mike Butts will do another Top 10
9. FPGAs will move to 3D by having stacked dies within one package
8. FPGAs will contain embedded GPUs
7. Dynamic Partial Reconfiguration will be accessible to the masses
6. Intel acquires an FPGA company
5. VHDL and Verilog will remain ubiquitous for circuit expression
4. Heterogeneous architectures will be in common practice and will be in market.
3. You can buy FPGA IP from an FPGA app store
2. Reconfigurable fabrics will be integrated onto Intel CPUs
1. A high-level, object oriented solution for FPGA design will be popular (it won’t be C to gates)
FCCM 2007 Top 10 Predictions for FCCMs in 2012
Sponsored by DRC Computer.
After enjoying some of the provided beer, we broke into six groups, each of which came up with 10 (or 20) predictions. After enjoying more of the provided beer, each group presented their predictions, then each person voted for their favorite five.
Moderated and compiled by Mike Butts. Results certified by the accounting firm of Dewey, Cheatham and Howe. Thanks to DRC for providing beer, and mugs for drinking it!
12. It didn’t fit.
11. Still no standard platform or tools, open source FPGA still a dream. (7)
10. We hate the tools more. (7)
9. Bill Gates changes his name to Bill Luts. (7)
8. All chips will be defect-mapped and self-healing. (8)
7. Heterogeneous architectures dominate. (9)
6. All devices will require on-line patches like Windows. (9)
5. There will be viruses and malware for FPGAs. (10)
4. An FCCM will be in the TOP500 Supercomputer list. (11)
3. Partial reconfiguration has been realized in a real-life app. (11)
2. There will be enough women at FCCM 2012 that they complain about the bathroom lines. Note: this item suggested by the female Top 10 participants…
1. FPGAs will have floating point cores. (26 votes)
FCCM 2000 Top 10 Predictions for FCCMs in 2005
After many of us enjoyed some evening refreshments, participants broke into five groups, which each came up with ten predictions. Each group presented their predictions to all, and each participant then voted for their favorite five.
Moderated and compiled by Mike Butts. Results certified by the accounting firm of Dewey, Cheatham and Howe, Philip Frieden and Jan Gray presiding. Thanks to Virtual Computer Corp. for all the fine Hollywood Blonde and Hollywood Red beer!!! (…and their generosity in picking up the tab for the room, the treats and the splendid prizes). And especially to all the participants!
14. It didn’t fit
13. Optical interconnect in FCCMs (7 votes)
12. A C derivative will be the dominant HDL (7 votes)
11. “Clippy” (that stupid Office assistant) becomes new hardware design aide (7 votes)
10. Linux ported to XC4010 (7 votes)
9. 80% of the communications infrastructure is reconfigurable (7 votes)
8. FPGA Systems-on-Chip are commonplace (7 votes)
7. We will merely dislike the tools (9 votes) (see #1 in previous years)
6. FPGA Barbie™ says “Layout is hard” (9 votes)
5. Internet Appliances all have FPGAs (9 votes)
4. JBits grows up to become JBytes (10 votes)
3. CPUs + FPGAs are a reality (10 votes)
2. Still using VHDL (and hating it), still talking about C to Silicon (10 votes)
1. Each FPGA has its own IP address (15 votes)
FCCM 1998 Top 10 Predictions for FCCMs in 2003
After many of us enjoyed a bit of beer, 41 participants broke into five groups, which each came up with ten predictions. Each group presented their predictions to all, and each participant then voted for their favorite five.
Moderated and compiled by Mike Butts. Results certified by the accounting firm of Dewey, Cheatham and Howe, Philip Frieden presiding. Thanks to Virtual Computer Corp. for all the beer!!! (…and the room, the yummies and the splendid prizes). And especially to all the participants!
12. It didn’t fit
11. CAD will be the killer app for FCCM (8 votes)
10. We will see the first conscious FCCM (8 votes)
9. A “savior” killer app for FCCMs will appear (8 votes)
8. Splash II apps will still appear at FCCM ’03 (9 votes)
7. The predominant FCCM language will be Java++… but someone will still be doing C to FCCM (9 votes)
6. FCCMs will place and route themselves (9 votes)
5. At FCCM ’03, there will still be a high number of gratuitous architectures presented (10 votes)
4. At FCCM ’03, Mark Shand will give a paper on interfacing to the Hubble Telescope using FCCMs (11 votes)
3. FPGAs in FCCMs will have hardwired functional units (ALU, FPU, MPY, PCI…) (12 votes)
2. 90% of embedded systems on a chip will contain FPGA logic for FCCM-type usage (14 votes)
1. We will still hate the tools (27 votes: two thirds!)
FCCM 1996 Top 10 Predictions for FCCMs in 2001
13. It didn’t fit
12. FPGAs used in FCCMs will have 4-LUTs, FFs, and not enough routing
11. FCCM apps will be downloaded from the Internet a’la Java
10. FPGAs used in FCCMs will have embedded functional units
9. Microprocessors will have FPGA logic (for x86 emulation ;-), but not in the API
8. Dynamic FPGAs with on-chip DRAMs in FCCMs
7. Time-sliced FPGAs will be commercially available, and used in FCCMs
6. FCCM languages will be Visual Verilog++™, gcc, and Matlab
5. Top 3 application areas will be communications, military, image processing, and one we don’t know of today
4. $100M/year FCCM industry with two public companies
3. FCCM ’01 will have 1K attendees and a 100 vendor trade show
2. 50% of the 1996 DARPA goals will be met
1. We will hate the tools