Many moons ago I learned that hyperthreading was causing some of my heat problems. I dealt with the overheating problems, but now I would really rather just turn off hyperthreading so I don't have to worry about my preferences turning it back on.

Is turning off HT a bad thing for a dedicated cruncher, and if it is not, how do I turn it off? Will I be able to turn it back on if I change my mind later?

---

(Click for detailed stats)

There are two ideas. One is to limit BOINC to one CPU, or to limit it to 50% of CPU. This you control via your General Preferences.

The other is to change the PC to not use hyperthreading. I believe this is done in your BIOS settings... which you get to by hitting that magic key during boot up (is it F1?).

Since the floating point operations are the key to throughput, hyperthreading really only gives you 10-15% more work completed. Least that's what I've seen posted elsewhere. So, turning it off is not a "bad thing". But, will reduce your heat problems, and overall throughput (and generally the only way to retain the throughput is to find a way to deal with the heat). So, no, not a bad thing. Turning off HT will make the PC report as a single CPU system and only crunch one Rosetta task at a time.

---

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/

One of the Athlon 64 motherboards I've used has DEL being the key to get into the BIOS, and I've seen F10 as the BIOS entry key in the past as well. It'd be nice if everyone standardized on F1..

---

One of the Athlon 64 motherboards I've used has DEL being the key to get into the BIOS, and I've seen F10 as the BIOS entry key in the past as well. It'd be nice if everyone standardized on F1..

Looked it up on a google search and found a site that had F2 as the most common. It worked on my box, but I am certain I used another button on my last main windows box. I will have to put that in my back pocket for future reference.

---

(Click for detailed stats)

Many moons ago I learned that hyperthreading was causing some of my heat problems. I dealt with the overheating problems, but now I would really rather just turn off hyperthreading so I don't have to worry about my preferences turning it back on.

Is turning off HT a bad thing for a dedicated cruncher, and if it is not, how do I turn it off? Will I be able to turn it back on if I change my mind later?

Hyperhthreading optimized the utilization of your execution units, turning it OFF will reduce your number of units per day.

if you have thermal problem, concidere contacting your PC maker, or upgrade to a better heat think and Fan.

Hyperthreading greatly improve the responce time of your machine too, it is sad to turn it off.

who?
Top 1 and Top 2 computers are mine! na!

I can't believe that people still think that they can produce more with HT turned off.

Sad.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

I can't believe that people still think that they can produce more with HT turned off.

Sad.

No one in this thread said it did :-P

---

Team mauisun.org

just for information of anyone interested . . .

one of my machines has an AMD 4200+ cpu overclocked by 10%. I'm using a Zalman CNPS9500 AM2 AMD heatsink/fan and cpu runs at 30degrees Celsius. Memory is Patriot DDR2-800 with 4-4-4-12 timing and built in heatsinks. HT is on. The Antec case has 4 120mm fans. M/b temperature is consistently 32degrees Celsius - yes, cpu actually registers lower temp than m/b.

So, no heat problems with ht or overclocking if you have adequate cooling.

I can't believe that people still think that they can produce more with HT turned off.

Sad.

Hmm... If it's such a good thing, why wasn't it included with the C2D's???

Confused.

I can't believe that people still think that they can produce more with HT turned off.

Sad.

No one in this thread said it did :-P

Then why would anyone even bring it up in this context? If you're here to crunch, then ya need everything ya can bring!

Fix the cooling issues, and leave HT on, and run as many CPUs (virtual or otherwise) as ya got!

sheesh.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

--

Hyperthreading on will effectively split your CPU into two execution units.. more accurately it splits the one executuion unit more or less in half.. (a bit more than half by the way of some additional circuitry).. This allows operating systems that understand how to utilize SMP to use each half as a seperate processor... It does, however, split the amout of credits per hour approximately in half...

So, it depends what you think is better... twice as many WU's completed in a given time at 1/2 credit each... or the normal amount that one P4 will complete at a normal amount of credit...

I personally run HT on, because I feel that it gives me better control of CPU utilization... but most of my HT systems are large (more than two XEONS) and use L4 cache, so it may be that these systems actually produce a bit more with HT on... As with all computers your mileage may vary as we say...

---

Looking for a team ??? Join BoincSynergy!!

I can't believe that people still think that they can produce more with HT turned off.

Sad.

It is possible, at least in theory, and the fact that there are reports where others claim to have ssen this, and the fact that WCG recommend to turn it off, all make the theory plausible.

One example, if a single WU uses a lot more than half the cache, at any level, then turning HT off will reduce cache misses. The increase in speed due to better use of the cache will offset the increased productivity of HT in all cases, and in extremecases may lose more than HT has saved.

Another example, is if a single WU can run in main memory without using the swapfile, but the addition of a second WU brings the swapfile into use.

In either of these cases, and in several other examples, turning HT off would increase speed. For that reason, my advice is to try it for a few days, or at most for a week, and see what the actual throughput is with HT on and HT off, and with your typical mix of projects and your typical use of the box for non-BOINC work.

Interestingly, the new memory management features of the client will step the number of tasks down when the swapfile comes into play, meaning that if it was swapfile issues that stopped you using HT before, you should try it again and let BOINC choose whether to use one thread or two.

River~~

---

Hmm... If it's such a good thing, why wasn't it included with the C2D's???

Confused.

I believe some form of HT is going to be re-introduced with the nehalem core.

My guess is that rather than having two virtual cores per physical core, it'll swap threads between two or more physical cores to make best use of them... just a mildly educated guess though!

---

Hi everybody,

here is an article on the effect of hyperthreading under a number of different conditions, although it describes its effect in the special MATLAB environment. I thought after there has been so much discussion on the topic it might be interesting for some here to see a light shed on it from various perspectives.

Somewhen in the next week or so there will be a German version toohere, for now it is also in English until the final edits have been made.

Regards,

Christoph

---

"I know that you believe you understand what you think I said, but I'm not sure you realize that what you heard is not what I meant." R.M. Nixon

How much you do (or do not) get out of hyperthreading can also be affected by cache size. The newer Xeons come with up to 4 MB of on-chip cache per two cores (8 MB with 4 cores at the rate of 4 MB per two cores). I don't get as much out of hyperthreading on my old Xeons which had only 512 KB of cache. It is all a matter of the "hit rate" before the CPU has to stop and wait for some RAM to get fetched.

How much the cache size affects throughput is very task-specific. If you are running a computational task that takes very small amounts of memory to contain both the program and data values, the entire task can be sucked up into cache and execute without interruption if the cache is large enough. If the task is very memory-intensive, even a 4 MB cache will not be enough to make the task perform very well in a hyperthreaded environment, and in that case you would be better off disabling hyperthreading and using all the cache for a single task instead of trying to timeshare the cache between two hyperthreaded tasks (or, in the case of the newer CPUs, between four tasks on two cores).

Anyway, if heat is your motivation, then you probably would be better off just limiting your CPU utilization in your user preferences. Keep hyperthreading on and let one virtual CPU churn away on whatever task(s) you please. That will have a bigger impact on heat dissapation than just turning hyperthreading off, and will still leave the other thread(s) to do "regular work" for you.

== Bill

---