According to [1], the K80 GPUs have the following specs: Chips: 2× GK210 Thread processors: 4992 (total) Base clock: 560 MHz Max Boost: 875 MHz

According to [1], the K80 GPUs have the following specs: Chips: 2× GK210 Thread processors: 4992 (total) Base clock: 560 MHz Max Boost: 875 MHz Memory Size: 2× 12288 Clock: 5000 Bus type: GDDR5 Bus width: 2× 384 Bandwidth: 2× 240 GB/s Single precision: 5591–8736 GFLOPS (MAD or FMA) Double precision: 1864–2912 GFLOPS (FMA) CUDA compute ability: 3.7 Is that a good deal for $1/hour? (I'm not sure if a p2.large instance corresponds to use of one K80 or half of it) How much would it cost to "train" ImageNet using such instances? Or perhaps another standard DDN task for which the data is openly available? ______ [1] https://en.wikipedia.org/wiki/Nvidia_Tesla#cite_ref-19 https://en.wikipedia.org/wiki/Nvidia_Tesla#cite_ref-19