Monthly Archives: August 2013

Will Raspberry Pi hold the road?

We are now moving into the second year of Raspberry pi. It is still going strong and for good reason, it is inexpensive. But the market is seeing some up and coming products that are also inexpensive and feature rich. These boards are not as inexpensive as the Raspberry Pi, in the $45 to $75 region. They have newer version ARM processors than the RasPi and faster memory.

The one claim that the RasPi has is its VCIV GPU. I wonder how much of an advantage that really is. Many projects that I see have to do with being controllers or servers. Video performance is pretty much a moot point for either of those uses. For example, The Cubieboard has a SATA connector and a power connector for SATA (2.5″ 5V). For a server it has the right features and many people have asked for that on a future RasPi. It also has 512MB and 1GB DDR3 memory options. With the faster memory and A8 1GHz processor it should be fast. There is also a dual core (A20) A7 based Cubieboard2 available.

I do not know how much of an advantage more memory is. In my own tests of WordPress on RasPii, a 256MB and 512MB, I found almost no difference at all with more memory, at least when every thing else was the same.

Another example, the Beagle Board Black. This board has 512MB DDR3 and a 1GHz (A8) processor. Overall it should have a decent performance edge over the RasPi. The main items being the A8 and faster memory. The ARM v7 instruction sets are supposed to be faster than the ARM v6 on the RasPi. The cortex A8 is superscalar so it can execute almost double the instructions at the same clock speed as the RasPi. The memory is faster and that alone should provide some performance gains. Along with that an extra 300MHz can not hurt.

So the question is when will the RasPi see an upgrade? The answer seems to be sometime before 2016. No clear timeline at all. And as to what it will be like, even less info. The Foundation is very leery of the “Osborne effect”, where you announce a newer product and everyone sits back and waits to buy the newer product. I think they are being just a little too “leery” of the Osborne effect. Having open discussions about upgrading a product would not have an effect on current sales. An honest discussion of the future path would probably act more to stimulate the RPi community. Community discussion is what led to the adoption of cell phone chargers as the power source for the RasPi. I was against this but that is where it came from. I think the community also helped decide the fate of the polyfuses at the USB ports. So community involvement may make a future board a greater success. The threads speculating on future features is supposedly being watched by the Foundation, so why not open the process up a little?

An updated RasPi with the BRCM 11311 processor, which is a dual core, native Ethernet USB 2.0, GPU CoreIV product has been postulated. Many people want to see USB 3.0. The fact is USB 3.0 relies too much on the processor for its operation. You will not realize true USB 3.0 speeds in a single, or even a dual core, design. Most of these SoC’s have a USB 2.0 interface which limits the total speed to 480Mb/s, well below the 5000Mb/s of USB 3.0. Built in WiFi and/or Bluetooth is also high on most peoples list. These features are just too expensive to add when your goal is to keep the board inexpensive. Comparable boards like the Cubieboard and the Beagle Bone Black do not have these features because they are expensive. A board like the Panda Board does have these features and also is expensive because of them. Certainly there will need to be a decent GPU on a RasPi replacement. We do not want to see the media player crowd slighted. And it is nice to have good multimedia power for teaching purposes.

I have been for the next product being a server version. Little GPU, faster memory and faster CPU in the ARM v7 family. As stated before many of the projects that are out there have to do with “servers”. Websites (server), Robotics, controllers and such. Most of these need little or no GPU capabilities, other than for initial setup, to operate. I think a server version would be a good board for higher education. leaving out a high performance GPU should also help keep the SoC price down so other features can be improved.

In all it will be interesting to see where, if anywhere, the foundation goes from here. I am definitely one of the restless masses that would like to see at least a road map of where the Foundation wants to be in the near future. Will they rest on their laurels or will they move to a new step in their goal of better computer education and programming?

Richard

Back to a 256MB RasPi.

I found there was no difference in performance between the 256 MB and 512MB models for WordPress. So I have put the website back onto the 256MB RasPi. I found more difference in removing useless plugins and installing WP Fast Cache. I also have Ban Hammer and Revision Control installed.

A real performance gain was by using the Panda Board. It was 2/3rd faster at everything. Even before it was fully finished. Power usage was very similar. I was able to get WP Super Cache working on the Panda Board. So now I know the path to get it working on the website. I will be doing that soon.

It is about time for another backup of the website so I will try to get super cache working after I get it backed up. The last time I tried to get Super Cache working it did the website in and I had to restore from backup!