I want to take a set of Raspberry Pi cluster, used to do the front end of the site, the website is written in PHP, no database.
I want to know how many raspberry pies are needed to achieve the same performance as a E3 server (or other PC server), and has anyone done similar testing?
Reply content:
First of all, we use a more idealized counting, someone on the web made a floating-point performance comparison test between Raspberry Pi and several desktop-level CPUs (Linpack Benchmark Results
), Paste the results:
Pc:
Double Precision 100x100 compiled at 32 bits Opt No optCPU MHz MFLOPS MFLOPSCeleron C2 M 2000 1092.56 121.25Core 2 Duo 1 CP 2400 1315.42 195.13Phenom II 3000 1412.83 244.43Core i7 930 **** 1764.75 428.00Core i7 860 #### 2004.31 381.97Core i7 3930K &&&& 2529.73 746.01Core i7 4820K $$$1 2671.15 892.04Core i7 4820K $$$2 2684.05 895.54Core i7 3930K OC 3112.94 926.92
The Raspberry Pi is not designed for high-performance environments, but in a different scenario, a server based on the arm scenario is a promising
Forward
1. Raspberry Pi 2 B Cluster
The original PO on the microblog called the Tang Prairie dry giant Many
If it's a floating-point calculation, the Raspberry Pi will fail.
If it is playing IO throughput the Raspberry Pi is very hopeful
We used Cubieboard to do the test study, the cost is limited when the practiced hand Raspberry Pi performance is not as good as it advertised, my own test result is: Raspberry Pi run a phpinfo, can feel the obvious delay, and on my previous abandoned Android phone, the same 512M memory, Running WordPress is very smooth.
The sense of CPU frequency and disk I/O is flawed. TF card This kind of thing, really can't get up. The advantage of clustering is when dealing with large numbers of operations. Simple HTTP requests do not come out of the cluster, but will become slower. It's not as fast as a Raspberry Pi. Compare floating-point performance is bullying--not supercomputer
Want to achieve the same performance, a few of the--rpi IO performance is too bad