The recently released Pwn Plug R3 is a departure from its predecessors in a number of ways, but perhaps none as pronounced as the move away from the low-power ARM architecture. Rather than stick to ARM, which is primarily used in smartphones and tablets, Pwnie Express has embraced the latest Intel Next Unit of Computing (NUC) architecture to bring desktop-like performance to the Pwn Plug line for the first time.
But what does that mean, in practical terms? Just how much faster is the new R3 compared to its most recent predecessor, the R2?
File I/O Performance
All of the Pwn Plugs use flash storage of some type, in the case of the R2 it’s a micro SD card, and on the R3 an Intel 525 Series SSD. The R3 obviously will have the advantage here, given that micro SD cards are designed to be nothing more than cheap mass storage, but by how much?
To get a very rough idea of sequential performance we can simply write a large file to the drive with the common Unix tool “dd”, which will report its write speed upon completion. The following command will write a 512 MB file filled with zeros, and report how fast it performed the operation:
time dd if=/dev/zero of=test bs=8k count=62500
The results of even this simple test are staggering:
The R2’s SD card can only manage a write speed of just over 12 megabytes per second, while the Intel SSD clocks in at an incredible 519 megabytes per second. There’s simply no contest, the R3’s storage technology is a generational leap above the R2.
The R2 is powered by an Marvell Armada-370 processor clocked at 1.2 GHz, while the R3 has an Intel Celeron at 1.1 GHz. The casual observer may look at these numbers and think both processors would be around the same in terms of performance, since they are operating at the same clock speed. But with vastly different CPU architectures like this, clock speed is completely meaningless when it comes to performance.
To get a better idea of how these processors actually stack up, we can use a simple tool called “sysbench”, which is available in both the R2 and R3’s online package repositories. This tool will calculate prime numbers up to a user-defined value, putting continuous computational stress on the processor. The command to calculate 5000 prime numbers looks like this:
sysbench –test=cpu –cpu-max-prime=5000 run
The results, once again, are completely one-sided:
The R3 is able to calculate 5000 prime numbers over 10 times faster than the R2, showing the stark contrast between the computational capabilities of the ARM chip versus its similarly-clocked Intel counterpart.
It’s surely clear that the R3 is a vastly more powerful machine than the R2, but some may make the case that the day-to-day usage of both devices are comparable enough that the raw power of each respective Plug is irrelevant.
To address that, let’s take a look at a valid real-world example. The infamous password cracker “John the Ripper” represents a practical demonstration of both raw computational power, and rapid file operations. It just so happens that John the Ripper even includes a built-in benchmarking facility for various encryption types which can be accessed like so:
The benchmark shows how many hashes per second John could perform against given encryption schemes on a particular piece of hardware. A higher number is better, as that means you can try more passwords in less time.
In truth, there’s hardly even a competition here. The R2 is completely outclassed by its successor. Both the simplistic benchmarks and the practical John test prove the same fact: the Pwn Plug R3 is a quantum leap forward in raw power, which directly equates to better performance in the field.