Feet1st mentioned having two identical machines running Rosetta; one with HT turned on and showing 2 virtual cpus, and the other system with HT turned off, and showing just 1 cpu.

With a week's results - for each, here's the comparison.

HT on: 1039576.48 seconds of work, score: 1496.37

average score for 86032 (times 2) seconds: 247.66
- i.e. this machine produces roughly 247.66 credits a day.

HT off: 594774.93 seconds of work, score: 1528.11

average score for 86032 seconds: 221.04

247.66/221.04 = 1.12 so having HT on seems to create a 12 percent boost. So if you have the ram, and a stable machine - then you should put it to use.

---

Yes... now if only I knew how to turn it on! :) The single threaded machine is scheduled to be replaced today :):)

By posting here, I've now made it easier if anyone wants to see these machines, click my "feet1st" handle at left. He's referring to host 17286 and 181314. Both 3ghz P4s.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

I found with the older Northwood the opposite to be true, I gained by having HT off.

---

Not all Czech`s bounce but I`d like to try with Barbar ;-)

Make no mistake This IS the TEDDIES TEAM.

Yes... now if only I knew how to turn it on! :)

It is usually just a matter of turning it on in the BIOS.

---

It is usually just a matter of turning it on in the BIOS.

...which I would have to load a different BIOS version... right?? I mean I checked the settings while it boots and it looked the same between the two machines.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

Feet1st what is the motherboard ? If you can tell us what model you have we can look up exactly what you (may) have to do !

---

Not all Czech`s bounce but I`d like to try with Barbar ;-)

Make no mistake This IS the TEDDIES TEAM.

Thanks for the offer... that's the PC that's being replaced today though, so it was more a curiousity question. I presume the new box will be configured as a dual.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

I`ve no idea what you`ve got to replace it so can`t comment on the dual thing !

Default with Pentiums tends to be HT ON !

Replacing a 3Ghz ? With what ?

---

Not all Czech`s bounce but I`d like to try with Barbar ;-)

Make no mistake This IS the TEDDIES TEAM.

Feet1st mentioned having two identical machines running Rosetta; one with HT turned on and showing 2 virtual cpus, and the other system with HT turned off, and showing just 1 cpu.

With a week's results - for each, here's the comparison.

HT on: 1039576.48 seconds of work, score: 1496.37

average score for 86032 (times 2) seconds: 247.66
- i.e. this machine produces roughly 247.66 credits a day.

HT off: 594774.93 seconds of work, score: 1528.11

average score for 86032 seconds: 221.04

247.66/221.04 = 1.12 so having HT on seems to create a 12 percent boost. So if you have the ram, and a stable machine - then you should put it to use.

Did you check which machine gave the most decoys, rather than the most credit?

That would be a more useful measure for the project - as opposed to the individual user's score, which for the individual may be more important. An analogy would be the Swedish round-the-world sailing team that had two crews for the boat, each team tried to log the most number of (nautical) miles per shift, but sometimes this lead to the first team logging lots of miles "in the wrong direction", and the second team having to sail against the best direction to get back on course - not the best way to win a sailing competition!!

In the Rosetta case, individual people getting good scores could be achieved by reducing the RAM settings to run the memory slower, and thus letting the machine take longer to produce a given result - but the benchmark score isn't affected since it's not using RAM at all (basides the first iteration, which is a few microseconds, out of a minute long run!).

--
Mats

---

Replacing a 3Ghz ? With what ?

Corporate replaced it with another 3Ghz... brilliant! ...actually the box is much smaller, and presumably uses less power. The hard drive seems a bit peppier, and they thought they were upgrading my memory :) don't tell 'em I already had an extra GB in there ;) So went from 256MB (+1GB) to 1GB(+0 'cuz the memory's different :( ). Had to put the 1GB stick in my other PC, which they didn't replace.

You're right, checking number of decoys would be better. If you wish to do so, make it quick, cuz the non-HT box is no longer crunching.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

Mats:

During the same week, my system run WUs for around 86500 seconds and on one WU created 11 decoys and on another WU had a high of 53 decoys.

You'll need to find the same exact WU type finishing twice at about the same time on the Hyperthreaded machine and the same exact WU type on the second machine with HT turned off.

---

I've already brought up the idea of posting some specific WUs which can be explicitly downloaded and dropped in to a slots directory and run for the purpose of studying such comparables. Such a WU would never be accepted back to the server, but it would give an identical comparable to run and measure and discuss and compare. Hopefully they will adopt the idea in the near future.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/