The g3.4xlarge pairs 16 vCPU with 122 GiB of RAM, the 7.63 GiB-per-vCPU ratio that marks a GPU instance. In the name, "g" is graphics, "3" is the generation and "4xlarge" is the size within the g3 family. It runs on Intel Xeon E5-2686 v4 (Broadwell) at 2.3 GHz, and AWS introduced it in July 2017. AWS has since shipped g4ad, g5, g6 and g7; g4ad.4xlarge is the direct replacement at this size. Rackspace Spot runs one GPU class, gpu.vs2.xlargeplusplus — a single NVIDIA A30 with 24 vCPU and 64 GiB, in San Jose only. Against this instance's 1 × NVIDIA Tesla M60 that is a fair swap for inference and media work — the A30 carries more memory and more throughput per card — and it prices about 88% below the US West (N. California) on-demand rate.
Match qualityComparable GPU
g3.4xlarge pairs 1 GPU (8 GiB VRAM) with 16 vCPU. Spot's GPU node uses the NVIDIA A30 (Ampere, 24 GB VRAM) with 24 vCPU and 64 GiB of memory — more GPU memory, more CPU and newer hardware than g3.4xlarge. For inference, batch transcoding and single-card fine-tuning that is an upgrade rather than a compromise. The catch is location rather than hardware: US West, San Jose is the only Spot region that offers it, so latency to your users and egress to your data are the things to check.
