-[ snip ]-

-[ snip ]-

Greetings Hal,

I too am running an i7 860 processor. I was wondering if you have an overheating problem with yours? My 860 likes to overheat really easy, right from the build. It will not run with the stock fan. I have a Corsair cooler installed to keep the temp down. I have to suck the dust accumulation from the radiator about every 6 weeks or so. I also use TThrottle to help keep the temp down. I can only run on 6 cores safely. If memory serves from a few years ago when I researched the 860, I found that the temp threshold was at 72 deg C. Any thoughts? I do plan on doing a new build within the next 6 months, so this question may become moot. ;)

Keep on BOINCing...! :)

In warmer months, as my central AC is a bit rubbish & ~30 years old, it would have issues. So I bought a Thermalright Venomous X RT for it. I only just cleaned the dust out of it a few weeks ago after 3-4 years of running 24/7.

Greetings Hal,

3-4 years without dust removal? :O What part of the US are you in? I just reread your post and saw that you only run on 4 cores. As I mentioned, I run SETI on 6 cores. In cooler weather and with TThrottle I can run on 8. But, we're getting near to Summer and here in Iowa it gets quite warm.

I have my monitoring utility set to sound the alarm if the CPU gets to 63 deg C. That will tell me it's time to blow out dust again. :)

Keep on BOINCing...! :)

---

CAPT Siran d'Vel'nahr - L L & P _\\//
Winders 11 OS? "What a piece of junk!" - L. Skywalker
"Logic is the cement of our civilization with which we ascend from chaos using reason as our guide." - T'Plana-hath

-[ snip ]-

-[ snip ]-

Greetings Hal,

I too am running an i7 860 processor. I was wondering if you have an overheating problem with yours? My 860 likes to overheat really easy, right from the build. It will not run with the stock fan. I have a Corsair cooler installed to keep the temp down. I have to suck the dust accumulation from the radiator about every 6 weeks or so. I also use TThrottle to help keep the temp down. I can only run on 6 cores safely. If memory serves from a few years ago when I researched the 860, I found that the temp threshold was at 72 deg C. Any thoughts? I do plan on doing a new build within the next 6 months, so this question may become moot. ;)

Keep on BOINCing...! :)

In warmer months, as my central AC is a bit rubbish & ~30 years old, it would have issues. So I bought a Thermalright Venomous X RT for it. I only just cleaned the dust out of it a few weeks ago after 3-4 years of running 24/7.

Greetings Hal,

3-4 years without dust removal? :O What part of the US are you in? I just reread your post and saw that you only run on 4 cores. As I mentioned, I run SETI on 6 cores. In cooler weather and with TThrottle I can run on 8. But, we're getting near to Summer and here in Iowa it gets quite warm.

I have my monitoring utility set to sound the alarm if the CPU gets to 63 deg C. That will tell me it's time to blow out dust again. :)

Keep on BOINCing...! :)

Even with 8 cores I don't recall it running much warmer with that cooler. The fan runs 1000-1200RPM & I use a bit of a 3M furnace filter, which I use in my central system, crammed in the front of the case where the air gets sucked in. So I just have to clean that about once a year. Instead of dealing with cleaning a heatsink.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

-[ snip ]-

-[ snip ]-

-[ snip ]-

-[ snip ]-

Greetings Hal,

3-4 years without dust removal? :O What part of the US are you in? I just reread your post and saw that you only run on 4 cores. As I mentioned, I run SETI on 6 cores. In cooler weather and with TThrottle I can run on 8. But, we're getting near to Summer and here in Iowa it gets quite warm.

I have my monitoring utility set to sound the alarm if the CPU gets to 63 deg C. That will tell me it's time to blow out dust again. :)

Keep on BOINCing...! :)

Even with 8 cores I don't recall it running much warmer with that cooler. The fan runs 1000-1200RPM & I use a bit of a 3M furnace filter, which I use in my central system, crammed in the front of the case where the air gets sucked in. So I just have to clean that about once a year. Instead of dealing with cleaning a heatsink.

Greetings Hal,

I'm glad you go an 860 the doesn't get too hot. :)

Yeah, filters would make a big difference. When I started my wishlist for my next new box I was looking at new cases also. I found one on TigerDirect that has removable filters for cleaning. I believe it has 5 or 6 filters throughout the case.

Keep on BOINCing...! :)

---

CAPT Siran d'Vel'nahr - L L & P _\\//
Winders 11 OS? "What a piece of junk!" - L. Skywalker
"Logic is the cement of our civilization with which we ascend from chaos using reason as our guide." - T'Plana-hath

My Antec 902 cases have filters. The problem is you have to take a side panel off to get to one. The biggest hassel is undoing 16 thumbscrews per filter for the other two. My next case will have filters on the outside. I clean thenm on average once a month.

RAC is at 2,942 and falling still. My pendings are at 32 for lunatic apps.

---

[/quote]

Old James

Rac went up a tad to 2,950. Pendings for OP apps are now at 25.

My plan at the start was this.
Here is what I plan on doing.
1. Run stock apps on CPU doing only MB work. Will let RAC get stable.
2. Run stock AP on CPU only. Let RAC get stable.
3. Run stock MB and AP on CPU. And run stock MB on gpu. Until stable.
Im wondering that once my pendings are gone and RAC is stable if I should let stock run on the CPU and the GPU. Just to see what increase happens before I go to AP only on the CPU.

Any suggestions?

---

[/quote]

Old James

Rac went up a tad to 2,950. Pendings for OP apps are now at 25.

My plan at the start was this.
Here is what I plan on doing.
1. Run stock apps on CPU doing only MB work. Will let RAC get stable.
2. Run stock AP on CPU only. Let RAC get stable.
3. Run stock MB and AP on CPU. And run stock MB on gpu. Until stable.
Im wondering that once my pendings are gone and RAC is stable if I should let stock run on the CPU and the GPU. Just to see what increase happens before I go to AP only on the CPU.

Any suggestions?

The AP wells are dry at the moment. So you have a few options of what you could try next
MB: GPU only
MB: CPU & GPU
MB: CPU only, but with fewer tasks. However you would run if reserving for GPU
MB: CPU only HT off
MB: CPU only HT on but only # of tasks = to real cores

You could also take an average of like 1-2 weeks from the daily numbers instead of waiting for the RAC to even out. I'm considering doing this for my 860 I'm playing around with at the moment. Average 2 weeks worth of numbers. Then change config Ignore the numbers for the next two days and then average numbers for the next two weeks.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Rac went up a tad to 2,950. Pendings for OP apps are now at 25.

My plan at the start was this.
Here is what I plan on doing.
1. Run stock apps on CPU doing only MB work. Will let RAC get stable.
2. Run stock AP on CPU only. Let RAC get stable.
3. Run stock MB and AP on CPU. And run stock MB on gpu. Until stable.
Im wondering that once my pendings are gone and RAC is stable if I should let stock run on the CPU and the GPU. Just to see what increase happens before I go to AP only on the CPU.

Any suggestions?

The AP wells are dry at the moment. So you have a few options of what you could try next
MB: GPU only
MB: CPU & GPU
MB: CPU only, but with fewer tasks. However you would run if reserving for GPU
MB: CPU only HT off
MB: CPU only HT on but only # of tasks = to real cores

You could also take an average of like 1-2 weeks from the daily numbers instead of waiting for the RAC to even out. I'm considering doing this for my 860 I'm playing around with at the moment. Average 2 weeks worth of numbers. Then change config Ignore the numbers for the next two days and then average numbers for the next two weeks.

Seeing as AP work is scarce I will run the CPU and GPU with stock MB on HT with a free core and see what improvement I get. As I have no set time for this test I will let RAC even out before going to the next phase.
So after the CPU and GPU stock test will be AP only on CPU with 8 cores. Now the question is Do I just let AP runout and do no work, Or do I run MB to fill in the gaps. Im thinking run till empty and do no work.

I will edit in this hostin a few minutes.

RAc is now 2919 with 24 non stock pendings.

---

[/quote]

Old James

I have done a few experiments of my own from time to time to quantify the effect of, for example, 3 GPU MB tasks vs. 2 GPU MB tasks. (Lunatics x41g for Linux/AMD64 platform with nVidia GTX650.) I take a different approach to the problem of waiting for the RAC to reach an equilibrium. I just look at the reported elapsed times and credits in my account/tasks table - take an average of 5 or 10 work units - and from those numbers calculate what the RAC would be. I think the statistical variation in this approach is much less than the unsteady RAC numbers due to pending credits coming in at an irregular pace - not to mention the time it takes for the RAC to reach any kind of "steady" state. What am I overlooking here?
On a related topic -- back on April 9 in this thread TBar asked about performance of a "mid-range nVidia 6xx" GPU card. Since I do have a GTX650 (but maybe that's considered a low-end 6xx) I could contribute my own numbers if I had a better idea just what would be most useful. There are a bunch of AP work units now coming through the pipe so I could easily set up a controlled configuration for them to run in. And get the comparison MB work units later.

For this host Rac is at 2,702 and dropping slow. Still have 20 pendings.
@Gene, Why not give TBar a PM and ask him.

---

[/quote]

Old James

>James
I have taken a look at recent completed tasks for host 5065145. I averaged the last 10 (excluding one obvious outlier) and by my estimate that host should give you a "RAC" of 2110. As you observe a "dropping slow" passing the 2700 mark I predict it will level out real soon.

If TBar is seeing this, I've started a test run with just a single V7 GPU task (x41g optimized) running. With some AP V6 GPU tasks in the buffer they will get their chance in due time, running just the single V6 GPU task. When I have maybe 8 or 10 of each, to give decent averages, I'll post here and PM to TBar. The objective being to compare V7-MB vs. V6-AP in an nvidia 650 GPU.

Gene;

I think it was Richard who gave a guesstiment of near 2000 give or take. I have 19 pendings left. I know one is an AP that will never clear until its manually done.

Im in no hurry. When the pendings are gone I will fire up the GPU for stock MB with one core free. Just to see what my GTS 250 can do.

---

[/quote]

Old James

Rac is 2516 and pendings are down to 17.

I had a thought, I was going to let the GPU run MB stock when My rac was stable. But Im wondering if I should turn off HT and just run 4 real cores? And see if that really makes a differance. We have all read that 4 cores can beat 8 HT cores. Seeing as how im doing a test would it be worth a try before I run a GPU.
I would then run 3 cores and a GPU and see what happens before I went back to 7 cores HT and a GPU. All this would be stock MB.

Edit-this host

---

[/quote]

Old James

Seeing as you're only going to be doing MB on that GTS 250 you really shouldn't need to reserve any cores.

Cheers.

Seeing as you're only going to be doing MB on that GTS 250 you really shouldn't need to reserve any cores.

Cheers.

Your most likely right. Ive run all 8 with HT on for years.

---

[/quote]

Old James

Rac is 2516 and pendings are down to 17.

I had a thought, I was going to let the GPU run MB stock when My rac was stable. But Im wondering if I should turn off HT and just run 4 real cores? And see if that really makes a differance. We have all read that 4 cores can beat 8 HT cores. Seeing as how im doing a test would it be worth a try before I run a GPU.
I would then run 3 cores and a GPU and see what happens before I went back to 7 cores HT and a GPU. All this would be stock MB.

Edit-this host

You can also run 4 tasks with HT on. Which is what I am going at the moment. Then there are two different options when doing 4 tasks & HT on. Let the tasks run across all the cores at 50% or specify which cores BOINC may use. To specify which cores you can start BOINC using the built in windows affinity commands. I did a little write up on how to do that here if you want to give it a go.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Rac is 2516 and pendings are down to 17.

I had a thought, I was going to let the GPU run MB stock when My rac was stable. But Im wondering if I should turn off HT and just run 4 real cores? And see if that really makes a differance. We have all read that 4 cores can beat 8 HT cores. Seeing as how im doing a test would it be worth a try before I run a GPU.
I would then run 3 cores and a GPU and see what happens before I went back to 7 cores HT and a GPU. All this would be stock MB.

Edit-this host

You can also run 4 tasks with HT on. Which is what I am going at the moment. Then there are two different options when doing 4 tasks & HT on. Let the tasks run across all the cores at 50% or specify which cores BOINC may use. To specify which cores you can start BOINC using the built in windows affinity commands. I did a little write up on how to do that here if you want to give it a go.

Im curious about what that would mean HAL? Faster crunching?

---

[/quote]

Old James

Rac is 2516 and pendings are down to 17.

I had a thought, I was going to let the GPU run MB stock when My rac was stable. But Im wondering if I should turn off HT and just run 4 real cores? And see if that really makes a differance. We have all read that 4 cores can beat 8 HT cores. Seeing as how im doing a test would it be worth a try before I run a GPU.
I would then run 3 cores and a GPU and see what happens before I went back to 7 cores HT and a GPU. All this would be stock MB.

Edit-this host

You can also run 4 tasks with HT on. Which is what I am going at the moment. Then there are two different options when doing 4 tasks & HT on. Let the tasks run across all the cores at 50% or specify which cores BOINC may use. To specify which cores you can start BOINC using the built in windows affinity commands. I did a little write up on how to do that here if you want to give it a go.

Im curious about what that would mean HAL? Faster crunching?

Some people think that having BOINC run only on the "real" cores vs running across all cores makes things faster. In either case the processing times when running 4 instead of 8 tasks will be shorter. However the total amount of work done, or RAC, will be less when running 4 instead of 8. The last time I tested 4 tasks vs 8 was with MB6 where 4 tasks ended up giving something like 10-20% less RAC IIRC. MB7 could prove to be different.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Rac is 2516 and pendings are down to 17.

I had a thought, I was going to let the GPU run MB stock when My rac was stable. But Im wondering if I should turn off HT and just run 4 real cores? And see if that really makes a differance. We have all read that 4 cores can beat 8 HT cores. Seeing as how im doing a test would it be worth a try before I run a GPU.
I would then run 3 cores and a GPU and see what happens before I went back to 7 cores HT and a GPU. All this would be stock MB.

Edit-this host

You can also run 4 tasks with HT on. Which is what I am going at the moment. Then there are two different options when doing 4 tasks & HT on. Let the tasks run across all the cores at 50% or specify which cores BOINC may use. To specify which cores you can start BOINC using the built in windows affinity commands. I did a little write up on how to do that here if you want to give it a go.

Im curious about what that would mean HAL? Faster crunching?

Some people think that having BOINC run only on the "real" cores vs running across all cores makes things faster. In either case the processing times when running 4 instead of 8 tasks will be shorter. However the total amount of work done, or RAC, will be less when running 4 instead of 8. The last time I tested 4 tasks vs 8 was with MB6 where 4 tasks ended up giving something like 10-20% less RAC IIRC. MB7 could prove to be different.

Thats why I should run with 4 cores and see what happens. It should be intersting.

---

[/quote]

Old James

I did experimenting a couple of years ago on my i7-3930K.
Running 6 MB was more productive compared to running 12. Execution times more than doubled when then number of processes doubled.

When AP GPU AVX was released on Lunatics I tried again with same results.

At that time I had no GPU's. I suspect it is heavily dependent on what kind of hardware you run on (CPU + cache, RAM).

Nowadays I run HT enabled using 5-6 cores for CPU tasks and the 6 extra cores are used to feed the four GTX 780's running either 4 AP or 8 MB or a mixture of them each.

YMMV.

---

To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones

Some people think that having BOINC run only on the "real" cores vs running across all cores makes things faster. In either case the processing times when running 4 instead of 8 tasks will be shorter. However the total amount of work done, or RAC, will be less when running 4 instead of 8. The last time I tested 4 tasks vs 8 was with MB6 where 4 tasks ended up giving something like 10-20% less RAC IIRC. MB7 could prove to be different.

Thats why I should run with 4 cores and see what happens. It should be intersting.

I am also going to get out my meter and check the power usages of 4 vs 8 tasks. Then I'll see if the % increase in RAC is along the same lines at the $ increase in power usage.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[