Given the right uses, like an elementary school computer lab for learning Scratch, Logo, etc; they would be fine.
The problem is they are right on the cusp of the next cycle of Moore's Law, Raspberry Pi is now quad core and 1GB, it puts it in the same class as this machine. The cost of "good enough" compute hardware is going to zero (shipping, financial trans cost will dominate).
The main power draws of these systems will probably be the monitor they are attached to, the older PSU inside, and the hard disk. Keep in mind that for a RPi based solution, you still need to provide all of these. The tiny PSU needed for a RPi will probably be quite efficient, so maybe count that one out.
Also, as great as the RPi 2 is, I still don't think it is the same class as these machines, which have a 3x clock cycle advantage, plus much wider CPUs. For web browsing, this is going to matter way more than 4 cores.
P4s have a horrible stall happy pipeline, the RaspPi has a decent GPU, I'd call it even. For a school environment and a good minimal window manager, the new RaspPi would be excellent for web browsing.
> I haven’t tried any normal desktop applications (other than Chromium), but I’d say the Pi 2 is finally worth considering as an full-fledged educational computer (i.e., beyond the introductory toy stage).
With a newer monitor and the current generation raspberrypi, it should powered directly from the USB hub in the monitor. With a bolt through VESA mount, a wifi dongle and netbooting off an NFS, a rapidly deployable Raspberry Pi school lab should be easily doable.
Finding the perfect monitor (internal powered usb hub) for the right price point is difficult, the device that needs to get built is a VESA case for the raspberry pi that has a NEMA 5-15R input and AC pass through, a USB powersupply for the RaspPi and a power switch that turns both the rasp and the monitor off.
DC in is nice, could run the lab off batteries or direct solar for off grid installations.
The problem is the software configuration for the NFS bootserver and making the netboot images.
Someone should make a VM image that can netboot a cluster of Raspberry Pis with a choice of roles (scratch, web browsing, quake 3, etc). With remote mounted home directories (served from ZFS, for snapshots).
One modification I'd like to see for the raspberry pi is better sleep support. I don't know if the SoC could do it, but if it could go into extra low power mode, possibly powered by a supercap, it could wakeup on external events (like an RTC) and not take any power from the USB.
They did, with the B+ in mid 2014, and the current Pi 2 with the same power design. The B+ and newer boards don't suffer the odd power supply issues that the older boards (even the B rev2) suffered.
And I'm not just rehashing what someone else said; I've had one each of the original B, a rev 2 B, a B+, and now a Pi 2. I've used the same power supply on all four, and on the B+ and Pi 2 I don't have any issues with hotplugging flash drives and wifi adapters whereas I did on the original and rev 2 models.
I've had my B+ running 24/7 since August as a webcam server to watch my dog in the back yard, and I've only had to reboot it once for an update. It's running a Microsoft USB-powered 720p webcam and an Edimax wifi adapter directly off of the built in USB, and it hasn't once had a power issue. I've had my Pi 2 running 24/7 since I got it a couple of weeks ago, as a "light" GNU/Linux workstation I can switch to and use when my main workstation is otherwise occupied. Again, no power issues on that board.
This reflects my experience as well. Not only did the B+ add a switching power supply, but there is proper USB power protection and soft-start as well.
Unless GP is referring to the flash problem of the RPi 2.0, which is readily solved by using a case, piece of tape, or an epoxy blob (which I imagine is what will be added to the manufacturing line eventually)
Many DC devices have capacitors on their power rails. When they are plugged in, they generate a huge current spike as the capacitors charge up (limited only by the resistance and inductance of the capacitors/wires). This is enough to overwhelm the power supply and cause a reset.
The solution is a current limiter of some sort. Sometimes this is as simple as initially powering the USB port through a resistor, watching until the USB voltage gets high enough, then switching over to a direct connection.
First I've heard of this; it does need a decent USB power supply, but that's a consequence of not shipping one and expecting the user to supply one of suitable quality.
I found the first model would sometimes brownout when trying to maximally utilize the ethernet. Which was not a big deal for a hobby device, but it did eliminate a number of possibilities.
I was wrong.
Given the right uses, like an elementary school computer lab for learning Scratch, Logo, etc; they would be fine.
The problem is they are right on the cusp of the next cycle of Moore's Law, Raspberry Pi is now quad core and 1GB, it puts it in the same class as this machine. The cost of "good enough" compute hardware is going to zero (shipping, financial trans cost will dominate).