[QUOTE=kladner;531081]Condolences on _our loss, whatever the name connotes.


I find that I have lost the function of the ke_ between "u" and "t". Have to find out if one of the other boards around here will fire up.[/QUOTE]
As a temporary workaround (absent a properly functioning "Y" key), you can perhaps hold down ALT and enter the ASCII character codes (89 for Y and 121 for y) via the numeric keypad.


You can also try to restore function to the key. With the power off, turn the keyboard upside-down (over a sheet of newspaper is a good idea) and whup it top, bottom, and sides. If the problem is crud inside the keyboard, that might do the trick. If that doesn't work, there is a possible remedy involving getting inside the keyboard.


However, in my experience it might be easier to go to a Goodwill or other used-good store and invest a couple of dollars in a keyboard.

Doing ASCII had not occurred to me. "Turn it over and bang it on the desk" is my first response to just about anything keyboard related. In this case all it yielded was crumbs. :smile:

Been finding my haswell system mysteriously resetting itself... after a few times of that I decided to run a memory stress test... didn't take long for it to find problems... turned off to XMP from 2400 to the system default 1600 MHz and all was fine... :bangheadonwall:

:whistle:

[QUOTE=kracker;531098]Been finding my haswell system mysteriously resetting itself... after a few times of that I decided to run a memory stress test... didn't take long for it to find problems... turned off to XMP from 2400 to the system default 1600 MHz and all was fine... :bangheadonwall:

:whistle:[/QUOTE]


My Kaby Lake system has XMP mode, but I do not use it. It makes my CPU run really hot. Not using it does not change my RAM speed. It stays at 2400.

[QUOTE=kladner;531081]I will venture that "burrow" has something to do with rabbits.[/QUOTE]

Yup... :smile: Burrow, Fur, Hutch, Hole... Have fun with your machine names...

And, sorry LaurV, but you were totally off. Burrow was a reasonably fast machine, with 32GB of RAM. I'm now down to half the speed and half the RAM, but at least I'm working!

[QUOTE=chalsall;531114]Yup... :smile: Burrow, Fur, Hutch, Hole... Have fun with your machine names...

And, sorry LaurV, but you were totally off. Burrow was a reasonably fast machine, with 32GB of RAM. I'm now down to half the speed and half the RAM, but at least I'm working![/QUOTE]

I suspect it's just a dead power supply, but power supplies are sometimes hard to come by (I have three dual-octacore-SNB 1U servers which would probably work if only their power supplies could be replaced, but such power supplies appear to cost more than I paid for the second-hand servers)

[QUOTE=chalsall;531114]And, sorry LaurV, but you were totally off. Burrow was a reasonably fast machine, with 32GB of RAM. I'm now down to half the speed and half the RAM, but at least I'm working![/QUOTE]
You are quite easy to satisfy... My 10 cores/20 threads, 4GHz, 64GB RAM, 2x2080Ti+2x1080Ti is still very slow for me, I call it "wheelbarrow" and I would like it to be at least M82589933 times faster... :razz:

I’m safe, was not in London today. Two members of the public have died. Terrorist incident in London Bridge.

[QUOTE=kracker;531098]Been finding my haswell system mysteriously resetting itself... after a few times of that I decided to run a memory stress test... didn't take long for it to find problems... turned off to XMP from 2400 to the system default 1600 MHz and all was fine... :bangheadonwall:

:whistle:[/QUOTE]

My haswell has always been a very unstable system, even though I run everything at stock. Running Mlucas I recently found that said instability appears to be FFT-length dependent ... exponents at 4608K, last of a batch of which I recently finished, were OK, needing perhaps ~2 resets per week. At 5120K, however, I was lucky to get even 6 *hours* without some kind of data-corruption error or outright system crash - ended up adding a crontab entry to check "is job running?" every 15 monutes and restart if if not, that allows me to get back to the approximate 2-crashes-per-week level. So been doing final pre-release debug of Mlucas v19, wanted to run a PRP-DC with Gerbicz checking, so fired up one, but kept the regular LL-test @5120K running at same time, in case the PRP run crashed for any reason - lots of new code logic, etc. Voila! With two jobs running side-by-side, each 4-threaded (system is non-HT), the beast has been rock-solid stable, and certainly not because e.g. it's running cooler. Weird, but I'll take it! So this will be my normal operating mode, going forward.

My haswell needs 1.1v in order to be stable -- the mother:censored: will not boot Linux properly at 1v.

My pinball machine failed again today. It was fixed a couple of months ago. It took them a few weeks and her $200 of my money to find a cracked diode. The issue today is different, but will cost me at least $50 to find out.