I've been running S41.07(HT) - based on little more than I thought it was probably the most appropriate version, though I don't know whether there was any real benefit of using it instead of the standard S41.07 client.
It depends on how your are running.
If you are running your host in HT mode, and are running Einstein in both virtual CPUs, then S41.07HT will give considerably more total output than S41.07 in the same configuration.
However if you are running Einstein on one virtual CPU, and SETI on the other, then S41.07HT offers about the same performance as S41.07, I think. If you are running with HT disabled, then I think S41.07HT is significantly slower than S41.07.
When I was still running SETI, I gave it exactly half of my (one) hyperthreaded P4 (Gallatin) using a tag in Trux's tx36 client. As I've backed away from SETI, I swtiched this host to single thread operation, and the Einstein science ap (not client) to S41.07--upgraded today to S41.08. While I did not do a captive Work unit controlled comparison, it appears that S41.08 is little if any changed compared to S41.07 in execution time, so I might as well enjoy its bug fix.
quoting archae86:
If you are running your host in HT mode, and are running Einstein in both virtual CPUs, then S41.07HT will give considerably more total output than S41.07 in the same configuration.
That's what I suspected. I was also splitting virtual CPU's (E@H/S@H) using the trux client, but I've recently started using BoincStudio which doesn't provide that capability, at least at present.
Quote:
quoting nevermorestr:
I've found that on my P4 3.0 HT(on) that the fastest version was the S40.04.
Interesting - thanks for that, I'd missed it previously. I suppose I'll have to give that one a try here, although I'll probably limit to a machine or two untill we know what's going to shake out with the new project version and whether it will still be compatible.
RE: I've been running
)
It depends on how your are running.
If you are running your host in HT mode, and are running Einstein in both virtual CPUs, then S41.07HT will give considerably more total output than S41.07 in the same configuration.
However if you are running Einstein on one virtual CPU, and SETI on the other, then S41.07HT offers about the same performance as S41.07, I think. If you are running with HT disabled, then I think S41.07HT is significantly slower than S41.07.
When I was still running SETI, I gave it exactly half of my (one) hyperthreaded P4 (Gallatin) using a tag in Trux's tx36 client. As I've backed away from SETI, I swtiched this host to single thread operation, and the Einstein science ap (not client) to S41.07--upgraded today to S41.08. While I did not do a captive Work unit controlled comparison, it appears that S41.08 is little if any changed compared to S41.07 in execution time, so I might as well enjoy its bug fix.
RE: quoting archae86: If
)
That's what I suspected. I was also splitting virtual CPU's (E@H/S@H) using the trux client, but I've recently started using BoincStudio which doesn't provide that capability, at least at present.
Interesting - thanks for that, I'd missed it previously. I suppose I'll have to give that one a try here, although I'll probably limit to a machine or two untill we know what's going to shake out with the new project version and whether it will still be compatible.
.
