[RC5] RC5 Stream client on Centos Linux

Daniel Andrea II ka0mvk at arrl.net
Sat Jun 22 02:37:17 EDT 2013


And now even more weirdness.

Stopping and starting a few times was no problem, then it did it and
kept failing every time I tried to start it, even after a reboot.

Not sure but I'm wondering if it's a corrupted output buffer file?
When I clear the buffer files out, I clear them all (after copying the
input buffer to another machine as a chkpoint file so I don't lose
them), and it works again.  Next time I'll try just deleting the
output buffer file and see if that fixes it.

On Fri, Jun 21, 2013 at 9:11 PM, Daniel Andrea II <ka0mvk at arrl.net> wrote:
> Now it's even stranger.  Every time the client got interrupted and had
> to restart for any reason the other day, it did that.  Now I can stop
> it and restart it with no issues.  What the heck????  Not making any
> sense now.
>
> On Fri, Jun 21, 2013 at 3:25 PM, Daniel Andrea II <ka0mvk at arrl.net> wrote:
>> I've got a Centos 6.4 Linux install that I'm running the Linux Stream
>> client on, giving me some weird issues on occasion.
>>
>> The box is an older Del Precision 390 workstation, which has a Core2
>> Quad Q6600 CPU and a Radeon HD 4850 PCI-e card in it.  I've got the
>> 64-bit linux client running on the CPU doing OGR.  I have Catalyst
>> fglrx legacy driver version 13.1 installed, along with the AMD APP SDK
>> 2.8.
>>
>> I've tried both the 32-bit and 64-bit Stream clients, both do the same thing.
>>
>> Here's what happens:
>> If I stop and restart the client with a partially completed block
>> stored in the buffer files, it throws some errors about values not
>> matching (including something about Iters and Maxiters) and quits.  It
>> will not work again until I delete the buffer files and have it start
>> on fresh blocks.
>>
>> For reference, I recently had been using this setup to mine bitcoins,
>> which worked flawlessly (even slightly overclocked) for several months
>> now.  No problems stopping/starting that, but that used OpenCL to do
>> the calculations.
>>
>> So, any ideas on what could be going on with this?


More information about the rc5 mailing list