Intel Core i5 2400S

Now that the Linux review of the Intel Core i3 2120 is published and there were initial benchmarks of the Core i5 2400S a few weeks back when looking at the state of Intel's "Sandy Bridge New Acceleration" architecture, the complete review of the Intel Core i5 2400S processor is here. The Core i5 2400S is meant to be an energy-efficient Sandy Bridge processor with a 65 Watt TDP compared to the i5-2400 non-S CPU that has a maximum TDP of 95 Watts like the other higher-end models, but this power reduction comes by scaling the CPU frequency back by 600MHz.

The Intel Core i5 2400S is a quad-core Sandy Bridge processor (four physical cores, but no Hyper Threading) that runs at 2.5GHz with a maximum Intel Turbo frequency of 3.3GHz. There is 6MB of Intel Smart Cache and there are Intel HD 2000 processor graphics. Of course, there are the normal features like the other Sandy Bridge CPUs: 32nm manufacturing process, SSE4.1/4.2, AVX, PCI Express 2.0, etc.

By dropping the TDP by 30 Watts, for a maximum TDP rating of 65 Watts, Intel hopes these energy-efficient Sandy Bridge CPUs with the "S" postfix will work their way into more home theater PCs and other low-power, low-noise compact systems. These energy-efficient CPUs come mostly due to down-clocking the non-S model. In the case of the Core i5 2400S, it's dropping the bus/core ratio from 31 to 25, which results in a 600MHz speed drop. While not being reviewed today, Intel also has Sandy Bridge CPUs with a "T" postfix that drop the maximum TDP down to 45 Watts.

Like the Core i3 2120, Intel Corp sent over the Core i5 2400S as a review sample to Phoronix. This was not an engineering sample but was a retail unit. The MSRP on the Core i5 2400S is $195 USD.

As this is a non-K Sandy Bridge CPU, the overclocking ability of the Core i5 2400S is rather limited, but that is not the focus of an energy-efficient CPU anyways where you are looking for low-power and low-heat output. For those interested in the /proc/cpuinfo output for this processor, it's available from this OpenBenchmarking.org log.