On Thu, Nov 12, 2009 at 1:31 PM, P.H.Welch <P.H.Welch@xxxxxxxxxx> wrote: > Thanks! Not too bad ... but the times for occam-pi on the same machine > (what was the machine?) are? ;) Since you asked and I was working on this anyway for unrelated reasons, here are the timings on my Intel Core 2 Quad CPU for both sequential and parallel deltas: With sequential delta: Commstime(go) starting... ============================= Last value received = 1000015 Time = 1580553 microsecs Time per loop = 1580 nanosec Context switch = 197 nanosec Commstime(occam) starting ... ============================= Last value received = 1000015 Time = 147243 microsecs Time per loop = 147 nanosecs Context switch = 18 nanosecs With a parallel delta: Commstime(go) starting... ============================= Last value received = 1000015 Time = 2864646 microsecs Time per loop = 2864 nanosec Context switch = 358 nanosec Commstime(occam) starting ... ============================= Last value received = 1000015 Time = 208679 microsecs Time per loop = 208 nanosecs Context switch = 26 nanosecs KRoC continues to win out against the competition, but these times aren't that bad for Go, considering its maturity. This is using the 8g compiler, for anyone that wants to try and replicate the tests. I've attached the source I used for both languages if anyone is interested in them or can note that I've done something terribly wrong. - Jim
Attachment:
commstime.go
Description: Binary data
Attachment:
commstime.occ
Description: Binary data