SoylentNews is people
SoylentNews
For those not following this project it is a FOSS reimplementation of the Win32 interface, which supports a great deal of humanity's historical computational effort. The new ReactOS release has reached 0.42 and the filesystems ext, btrfs are apparently RW, though Reiserfs and UFS are readonly mounts, successful systems have been shown running.
A nice gallery of some successfully run high profile applications is here (e.g. SimCity and PhotoshopCS2 !!), although interesting, not why I am reporting this.
There are an *enormous* number of scientific instruments (not just microscopes, but various scanners, PCR decks , robots) which originally came with a Win32 driver disk, and have since gone out of business or stopped support. There might only be a single run instance on a crusty old i386 (yes, I've seen that!!).
This is an ambitious project and of course depends on the effective WINE project. It deserves some specific credit and visibility, for providing a possible threshold in the future that sufficient OLD applications can be run independent of the new Microsoft "One OS to rule them All", that it may be possible to construct hybrid machines running Linux, and sufficient driver support from ReactOS to manage the old device drivers that WINE may find difficult to reverse engineer.
But in general, more OS choice's are a good thing!
(Score: 2, Interesting) by Scruffy Beard 2 on Monday August 29 2016, @04:00AM
I have recently been memory testing with a Pentium D and Athlon 64 X2.
AMD beat the pants off of Intel back in the day.
The Pentium D, running at 3Ghz with a memory bandwidth of 2.2GB/s (DDR2) runs Memtest86+ 3x slower than the Athlon 64 at 2Ghz with a memory bandwidth of 1.5GB/s (DDR1).
The Athlon also supports ECC memory. At this point, I think buying the memory will be cheaper than buying a used server board+used memory (and would not save on power draw).
I actually got the machine from the trash. The reason it was there: Video card + resulting PSU and memory failure. I checked e-bay for replacement boards: they all appeared to use an added DDR2 memory controller that does not support ECC. The PCI-E slots appear to be damaged, but I can work around them. I plan to do more extensive testing before buying ECC memory, but I currently would not trade my Athlon 64 for newer crap.
TL;DR: For my use-case of a back-up server made from salvaged parts, the Athlon 64 should be good. ECC memory is essential since I want to use ZFS to correct any disk errors from redundant disks. For intermittent use, 70-100W of power draw is not that big a deal.
PS: I am aware that there is an economic theory that capital costs don't matter (possibly related to the sunk-cost fallacy), but they do.
I am also planning on building a power-hungry computing machine: as a proof of concept until I can afford more efficient hardware. The proof of concept would let me get all the software working: preventing new, expensive, hardware from sitting idle.
(Score: 1) by Scruffy Beard 2 on Monday August 29 2016, @04:19AM
Yes, the video card failure was caused by blown caps.
I opted to recover components like inductors from my old Pentium 4 boards (rather than repair them).
I mean: even if I replace the caps...Pentium 4's are simply too power hungry.