g6e.48xlarge

Analysis

The g6e.48xlarge is a gpu instance instance in the g6e series with 192 vCPUs and 1536 GiB of memory on AMD EPYC 7R13 (up to 2.6 GHz), which makes it memory-heavy at 8 GiB of memory per vCPU.

Within the same generation and size, the g6.48xlarge on AMD EPYC 7R13 costs $13.35/hour, 56% less ($12,250 a month).

Running it in N. Virginia instead of Tokyo saves $9,905 a month per instance: $30.13/hour against $43.70/hour, a 45% spread. It is not offered in Canada Central, São Paulo, Ireland, London, Singapore and Sydney, so plan around that if you need a presence there.

It carries 8 GPUs with 384 GB of GPU memory, so most of the price is the accelerator rather than the CPUs. It has local instance storage (4 x 1900 GB NVMe SSD), which is fast but lost when the instance stops. Network: 400 Gigabit.

It is the largest of the 8 g6e sizes. Dropping to the g6e.24xlarge saves $10,998 a month. Unit price is not flat across the series ($0.1184 to $0.4652 per vCPU-hour), so the size changes the rate you pay.

The newer g7e.48xlarge on Intel Xeon Scalable (Emerald Rapids) is 10% dearer at $33.14/hour.

On demand it costs $30.13/hour in US East (N. Virginia), $21,996 a month if left running, which puts it mid-table for gpu instance types at the 60th percentile per vCPU ($0.1569 per vCPU-hour) and the 40th percentile per GiB ($0.0196 per GiB-hour).