I smelled some uptime's being thrown around in here and thought I may be able to compete.
I have a little P2-266 running linux and crunching RC5 for the federation that has been running for a little while:
[Dec 11 16:25:13 UTC] Completed RC5 packet D471DB5A:40000000 (2*2^28 keys)
0.00:12:36.07 - [710,080.98 keys/sec]
[Dec 11 16:25:13 UTC] Loaded RC5 1*2^28 packet D471DB5D:00000000
[Dec 11 16:25:13 UTC] Summary: 9261 RC5 packets (54813*2^28 keys)
260.07:44:38.07 - [653.59 kkeys/s]
[Dec 11 16:25:13 UTC] 7 RC5 packets (19 work units) remain in buff-in.rc5
[Dec 11 16:25:13 UTC] Projected ideal time to completion: 0.01:53:22.00
[Dec 11 16:25:13 UTC] 10 RC5 packets (29 work units) are in buff-out.rc5
.....10%.....20%.....30%.
That same box runs a pproxy too:
rc564 r=84983/85000, d=11/100, 22.2 Mkeys/sec, tot=5317796
ogr r=0/0, d=0/0, 4.0 Mnodes/sec, tot=1049 stubs
Status: Uptime:
190.02:02:29, 2 listeners, 0 uplinks
Status: 0 active clients (peak: 8, mean: 0.08)
It also has an uptime of:
[root@stats heavyiron]# uptime
8:35am up 306 days, 20:29, 2 users, load average: 1.21, 1.48, 1.31
I have another linux box with 186 days and running.
Here's a line out of my LAN's setiq:
SetiQueue 3.03.1.7 was last restarted on Mon 2001 Aug 13 8:59:59am (120 days 0 hr 26 min ago)
That's on an NT4 box.
This same box is also crunching for the federation:
[Dec 11 16:10:58 UTC] Completed RC5 packet 9F729D12:C0000000 (1*2^28 keys)
0.00:06:27.96 - [691,901.04 keys/sec]
[Dec 11 16:10:58 UTC] Loaded RC5 6*2^28 packet 9FF23DC8:A0000000
[Dec 11 16:10:58 UTC] Summary: 8143 RC5 packets (26639*2^28 keys)
119.23:03:26.70 - [689.80 kkeys/s]
[Dec 11 16:10:58 UTC] 2 RC5 packets (3 work units) remain in buff-in.rc5
[Dec 11 16:10:58 UTC] Projected ideal time to completion: 0.00:19:15.00
[Dec 11 16:10:58 UTC] 2 RC5 packets (3 work units) are in buff-out.rc5
It is a P2-266 also.
Notice how much better Linux does RC5 over NT4 on these two P2-266's?
Linux = [710,080.98 keys/sec]
NT4 = [691,901.04 keys/sec]
That's +18kkeys/sec
The only reason the overall rate is below the NT's box is because the linux box processes over a years worth of ppstats logs every 15 minutes, kinda pulls down on the available CPU cycles
I remember in IRC some time back, SyZ pulled out one or two linux boxes with 400+ days of uptime on 'em