Virtual FPGA (V-FPGA) architectures are useful as both early prototyping testbeds for custom FPGA architectures, as well as to enable advanced features which may not be available on a given host FPGA. V-FPGAs use standard FPGA synthesis and placement tools, and as a result the maximum application frequency is largely determined by the synthesis of the V-FPGA onto the host FPGA. Minimal net delays in the virtual layer are crucial for applications, but due to increased routing congestion, these delays are often significantly worse for larger than for smaller designs. To counter this effect, we investigate three different placement strategies with varying amounts of manual intervention. Taking the regularity of the V-FPGA architecture into account, a regular placement of tiles can lead to an 37 % improvement in the achievable clock frequency. In addition, uniformity of the measured net delays is increased by 39 %, which makes implementation of user applications more reproducible. As a trade-off, these manual placement strategies increase area usage of the virtual layer up to 16 %.

No abstract

No abstract

Virtual FPGAs (V-FPGAs) are used as vendor-independent virtualization layers, to retrofit features which are not available on the host FPGA and to prototype novel FPGA architectures. In these usecases, the achievable clock frequencies of V-FPGA user applications are a major concern. The abstraction layer inherently induces overhead, but this aspect is reinforced by nonuniformity effects: When V-FPGA cells perform worse locally, basic architecture modeling generalizes these worst-case path delays to the whole device, limiting applications to a lower frequency than theoretically achievable. We propose three approaches to attenuate these effects: First we introduce uniformity metrics and manual V-FPGA placement strategies for more uniform placement, improving achievable frequency by 16 %. Second, we propose a framework for automated timing extraction, enabling individual characterization of each V-FPGA design. Third, after evaluating Vivado synthesis strategies, we extend the timing model for non-uniform timings, achieving improvements of up to 28 %.

No abstract