From FPGA to Support Cloud to Cloud of FPGA: State of the Art

Table 4

Summary of the FPGA platform as a service.

Works	Application^1	Scalability of the cloud^2	Heterogeneity^3	Scalability of FPGA designs^4	Multitenancy^5	Sharing time^6	Solved challenges^7
Soares et al. [59]	Education	No^1	No (Intel Cyclone II)	No	No	No	C3; FPGA device (D7)
Morgan et al. [60–62]	Education	Yes^2	No (limited to one Xilinx Zynq SoCs and one Intel CycloneV SoC)	Yes	No	Yes	C3
Machidon et al. [66]	Education	Unknown limit	Yes	Yes	No	Yes	C3; FPGA platform selection (D7)
Machidon et al. [67]	Education	No^1	No (just one Xilinx Spartan 3E)	Yes	No	Yes
Zebu platform [68]	Industrial (emulation application)	Yes (limited to 64 maximum)	No (limited to Xilinx Virtex series)	Yes	No	Yes	C3; FPGA platform selection (D7)
HAPS platform [69]	Industrial (emulation application)	Yes (limited to 64 maximum)	No (limited to Xilinx Ultrascale series)	Yes	No	Yes	C3; FPGA platform selection (D8)
Microsoft [72]	Industrial	Yes	Yes (Intel Arria and Stratix Xilinx)	Yes	Yes		C3; FPGA platform selection (D8)
Amazon [71]	Industrial	Yes (one to eight)	No (limited to Xilinx Virtex UltraScale + VU9P or VU13P)	Yes	Yes	Yes	C3; FPGA platform selection (D8)
Alibaba [73]	Industrial	Yes	Yes (Intel Arria and Xilinx Ultrascale familties)	Yes (F1,2,3 instances FPGA)	Yes	Yes	C3; FPGA platform selection (D8)

¹Application: referring to the target type of application, either academic or industry. ²Scalability of the cloud: referring to adding or removing FPGAs in the cloud. ³Heterogeneity: ability to use different types of FPGA in a same cloud. ⁴Scalability of FPGA designs: ability of a hardware designer to target the designs to a multi-FPGA platform. ⁵Multitenancy: ability of an FPGA to be used by multiple different hardware designers. ⁶Sharing time: ability of application designers to execute their designs in different FPGAs in the same time. ⁷Solved challenge; indicates that the challenge is not fully solved and solved difficulties are listed.