Posts by HAL9000


log in
1) Message boards : Number crunching : GPU Wars 2016: NVIDIA GTX 1080/1070 (Message 1792200)
Posted 7 hours ago by Profile HAL9000
I'm very curious as to what the 1060 and even the 1050 will perform like, as well as power consumption. In the blurbs I've been reading about pascal, best rumour I've heard is late summer/early fall, anyone see anything else mentioned about them?

The 1070 is to be released June 10th. The 1060 is said to be released in "the fall". Which looks to fit their normal schedule of releasing the first GPUs a few weeks apart. Then staggering the rest at 3+ month intervals.
2) Message boards : Number crunching : LotzaCores and a GTX 1080 FTW (Message 1791746)
Posted 1 day ago by Profile HAL9000
Well, just got up, so I went down, paused and then exited BOINC, uninstalled and then reinstalled Lunatics, and it seemed to start right where it left off, with no drama. Only thing slightly unusual was that windows security asked if it was alright to allow BOINC thru the firewall, I've never seen that one before.

I am running Hynix 1866 memory, in singles per bank to allow the system to utilize it at the full speed, as I read that more sticks = slower speeds. And 32 gig is more than enough for what I am running on this.

So far, looking at temps, it appears that they may have crept up a few degrees, maybe an average of 3-5, but it looks like they are still for the most part at 50 or below except on 3-4 cores out of 24 on each CPU. But, I suppose that will vary depending on they type of WU is being processed.

I have seen several benchmarks over the years that have shown an increased latency as the number of DIMMs per memory channel is increased. So I try to stick to one DIMM per channel as well.

Normal AR tasks with the AVX app appear to be running slightly faster. With run times as low as 2hr50min. Once you start using GPUs that could change. The GPUs could either choke the CPU tasks or be restricted by already saturated memory i/o. Then again maybe nothing will change. This system system has been defying all previous attempts at logical behavior thus far.

Is the increase in temp you mentioned after using the AVX app? That would normally be reasonable behavior.
3) Message boards : Number crunching : LotzaCores and a GTX 1080 FTW (Message 1791730)
Posted 1 day ago by Profile HAL9000
Good info to know, thanks. Last thought, is the problem the size of the database in terms of it's software abilities, it's hardware limitations, storage limitations? Do you happen to know what the bottleneck is and any thoughts on what would be needed to address and correct it, so we wouldn't run into problems with it again for many years? Money, obviously, but on what? Thanks!

I think All of the Above is probably the most accurate answer regarding the db limitations. I don't believe anything has changed since you were asking about the limits a few weeks ago.
It was mentioned, maybe last year, that other database options were being explored. Breakthrough Listen may have interrupted that process. It was also mentioned they were looking to reduce their server footprint. So smaller more powerful servers may go hand in hand with new db software.
4) Message boards : Number crunching : LotzaCores and a GTX 1080 FTW (Message 1791637)
Posted 1 day ago by Profile HAL9000
Good metric to gauge optimal loading might be temperature, or power from the wall. With the beanbag analogy it'd be firmness.

Keep an eye on temperatures when running the AVX application.
On my i7 system it was (barely) able to run the SSE3 application without getting too hot. With the AVX application I had to replace the stock cooler- it worked the CPU that much harder. And the crunching times came down, a lot.


Yeah, don't burst the beanbag. AVX shaped beans are larger and denser, lol.

I suspect that AVX may prove to be slower on that system. With 48 tasks at once that is a lot to stuff down the memory pipeline all at once.
It may not be the most correct way to say it, but I think higher level SIMD instructions tend to be more memory intensive.
I was already very surprised by the performance of the E5 v2 CPUs versus the E5 previous generation. So I'm split 50/50 on how AVX will compare to SSE3 & will have to find out if they are using DDR3 1600 or 1866 memory.

AVX apps proved to be the most efficient on my i5-4670K systems with DDR3 1600 memory.
5) Message boards : Number crunching : LotzaCores and a GTX 1080 FTW (Message 1791550)
Posted 1 day ago by Profile HAL9000
Running all 48 at once. I'll head downstairs now and take a look at it to see if the temps are still in the same range.

I would have expected to see a much larger difference between the Run Time & CPU Time with 48 running. Looks like normal AR tasks are pretty solid between 2hr59min & 3hr6min. You could switch to the AVX app tomorrow to see how it runs. Unless you want to get more data on how VLARs run, but they normally track the same between the CPU apps.
6) Message boards : Number crunching : LotzaCores and a GTX 1080 FTW (Message 1791527)
Posted 2 days ago by Profile HAL9000
Well, instead of over-running the GPU Wars 2016: GTX1080 thread with my new build info and questions, I thought it would be more proper to start a new one with my experiences with both a large number of cores (48) and the new EVGA 1080 FTW. As I just put the order in to EVGA last night, and honestly don't know exactly when they will be shipping it, initially this thread will be about the 48 core build, and the things I run into in the process.

I'd expect the 1080 to hopefully ship sometime next week, but it will get here when it gets here, I'm not going to sweat it. I'm interested in how this system runs when running 48 cores at once, as it is only a 2.2ghz proc, though it speeds up to 3.0 in certain situations. I just couldn't justify the cost of the 2.7/3.6ghz version, they were like 4x the cost if I remember correctly. And this is just a V2 version, I wouldn't even want to think about what the V3 or V4's would cost...

Well, this is my newest addition, and I'm going to see how it produces for me with just the optimized app (using SSE3 this time, and AVX when I install the new card), feel free to check on in on it's progress as time goes on. Oh, and one thing I noticed right now, as it is using SuperMicro's OEM coolers, is that the temps are running between 43 and 54, with the vast majority running between 45-50. That is one of the benefits of running the 2.2 CPU (Coretemp says it is running at 99.4 watts currently, while crunching on all cores), it is 10-15w lower power draw than the 2.4, and I believe 35 or so watts lower than the 2.7. This is good, because I tried installing Speedfan, and it installed properly, it just doesn't allow me to control _anything_ unfortunately. I may ping SuperMicro next week, to see if they have any software controls which allow me to optimize the temps a bit by spinning up the fans.

Looks like normal AR tasks are running just a hair over 3 hours with the SSE3 app. Is that running all 48 threads or have you limited it to running fewer?
7) Message boards : Number crunching : Using Android mobiles to crunch BOINC stuff (Message 1791520)
Posted 2 days ago by Profile HAL9000
Grant, that is very good info to know. I go thru laptop batteries much too quickly, due to the fact that they are plugged in 90%+ of the time, used as a desktop replacement. The mfg's must know this, they should have much smarter chargers, and not allow _any_ charging after 100% (and allow the option of slow charging) when the battery is in a docked situation, but they know this, and make a very good profit off of selling replacements. As well as the aftermarket, but those can _definately_ be hit or miss... And I've found that if my laptop has no battery, it won't boot. Such a deal, eh? That one has bugged me for Years.

Most computer manufactures put somewhere in the documentation that is the notebook will be plugged in for extended periods to remove the battery. I know for some devices, Apple notebooks & many mobile devices, battery removal isn't an option. So the user has to be knowledgeable if they want to have the greatest battery life expectancy & avoid micro charge cycles killing a battery.
8) Message boards : Number crunching : Decided to Bail Out on SETI for a Short While (Message 1791419)
Posted 2 days ago by Profile HAL9000
If you read between the lines of Matt's Technical News post a couple of weeks ago, I'd say that the benefactor's contribution to SETI@Home's general funding so far has been negative - and substantially so. Much extra work, but no extra bodies or brains to do it.

To my mind, that would violate the UK concept of Full Cost Recovery (FCR) - but that's a local funding standard, not a global one.

I read that mostly as "Breakthrough Listen is funding Matt's paycheck so we haven't lost him" at the time.
9) Message boards : Number crunching : GPU Wars 2016: NVIDIA GTX 1080/1070 (Message 1791417)
Posted 2 days ago by Profile HAL9000
Gotcha. So, does it make sense to install them both, trying one for (how long, days, weeks?) and then uninstalling and installing the other? Do you think there would be a significant enough difference to go thru the trouble? If so, I guess it would probably make sense to try it before the card arrives, but if it arrives the 1st week or so of June, that is going to be SO hard to just let it lay there while the testing finishes.

I would start with the AVX app. Then once you get the system fully built out with the GPUs you want and it is stable. Then give the SSE3 app a try.
10) Message boards : Number crunching : No work being done by one of my laptops??? (Message 1791410)
Posted 2 days ago by Profile HAL9000
Is one better than the other? I can relate, I preferred the old version of EVGA Precision, but the X and X16 came out, all bells & whistley, and found that I liked the older one better. Unfortunately, unlike BoincView, it doesn't play well with the latest hardware. But I can relate.

Even the last (v1.4.2 - 2007) version of BoincView predates the introduction of GPU computing to BOINC by about 18 months. Many readings from GPU tasks are, shall we say, quirky.

I don't think I could recommend it to anyone who hadn't grown up with the changes, but I have still got the original distribution .zip file if anyone wants a copy.

Well that is newer than the version hosted by SETI@home, v1.2.5, & the one I have.
11) Message boards : Number crunching : GPU Wars 2016: NVIDIA GTX 1080/1070 (Message 1791406)
Posted 2 days ago by Profile HAL9000
So, is that SSE3, SSSE3, or SSSSSSSSE3? ;-) lol This is a Xeon V2 proc, I thought that SSE3 (or whatever) was for older procs, a few more gens older than this one? Or does this proc fall under that version as well? Or, is it just simpler code that might run more efficiently? Too much to know! ;-)

For MBv8 there are only AVX, SSE3, & SSE2 applications in the Lunatics installer 0.44. The other options are for the MBv7 app where there were more options. Logically whatever selection is made that is not one of the MBv8 options would select the next lower options for MBv8. Such as selecting SSSE3, SSE4.1, or SSE4.2 should install that version of MBv7 & the SSE3 version of MBv8.

SIMD instructions are normally additive. So if you have a SSE 4.2 CPU it supports all previous versions as well.
SSE3 might just be more efficient than AVX on your 24c/48t configuration. The highest level SIMD instruction set is not always the most efficient option.
12) Message boards : Number crunching : No work being done by one of my laptops??? (Message 1791397)
Posted 2 days ago by Profile HAL9000
Thanks Hal, that is great to know.

The official instructions on Controlling BOINC remotely.
For my machines I use the cc_config.xml option <allow_remote_gui_rpc> instead of using a remote_hosts.cfg file. That way I don't have to configure the IP of each machine I might be connecting from.
13) Message boards : Number crunching : GPU Wars 2016: NVIDIA GTX 1080/1070 (Message 1791394)
Posted 2 days ago by Profile HAL9000
Hey guys, as I mentioned in this post above, since I am getting one, and will be installing it on the 48 core system that I am configuring right now, any thoughts you might have about proper apps to use, since it is a fresh load? Won't be installing vid drivers till I get the card, so will probably have to reinstall the optimized apps if that is suggested to install today as the CPU cruncher it will start out life as, till the vid card arrives. Or just run it stock? And is there a specific version I should install now, to be ready for when the card arrives? I am open to any and all suggestions.

I typically find the best performance with the AVX app, but with that many tasks running at once SSE3 might be more efficient. It could take running both for a while to see. Either option should be more efficient than the base CPU app.
14) Message boards : Number crunching : No work being done by one of my laptops??? (Message 1791390)
Posted 2 days ago by Profile HAL9000
I think I'll be installing the client on all of my computers in the near future. Are there any security holes opened up by installing it that you are aware of, that I should make sure are closed by proper configuration? Or is it just run it and use it?

The believe the only security related issues are the configuration changes to BOINC itself. By default it only allows connections from the localhost. However unless you are opening port 31416 to the internet it shouldn't be a problem.
15) Message boards : Number crunching : Decided to Bail Out on SETI for a Short While (Message 1791388)
Posted 2 days ago by Profile HAL9000
HAL, question for you. Since our 'benefactor' made his (very large) contribution to this science en devour, why hasn't (at least as far as my limited knowledge of it all goes) any of those resources migrated toward where they can do the most good long term for everyone, and the project as a whole - the back end (meaning the software that we all use to process all the work that the donation has provided)?

I'd think that with the kind of dollars I have heard mentioned previously, there would be a percentage, let's say maybe 3-5% of it that could be strictly dedicated to making the basic program be the best it can be? Which may include hiring some of the people who have been so kind to volunteer their time to this project so generously (if they are interested, of course), otherwise hiring full time, highly qualified programming staff for it (maybe poach a programmer or 2 from Nvidia and ATI? Bwhahaha!). This would enable us to be able to efficiently run the latest work units that said contribution has provided us in such large quantities?

From that would flow much good science, as well as long term (relative, as of course hardware and O/S'es change over time) stability, along with the ability to maybe work with said vendors in a serious way (ATI/Nvidia) so we might be able to plan somewhat our path, and not be blindsided by the changes and have to scramble madly to get the product to work properly. Maybe this is all pie in the sky, but with the proper resources, directed the proper way, I think it should be doable?


I haven't really been following everything related to this large funding contribution.

However I know there is a lot more going on behind the scenes than we observe directly or get notifications.
The amount that may be directed to SETI@home specifically is likely VERY small. I'm not sure that SETI institute itself even had that large of a chunk.
I think most of the money was said to be going to dedicated telescope time globally.
Also there may be limitations to how the money can be allocated.
16) Message boards : Number crunching : No work being done by one of my laptops??? (Message 1791380)
Posted 2 days ago by Profile HAL9000
Huh, ok, thanks Richard. I guess that makes sense, but it is strange that on all (as best as I can recall) the 30 some systems I've set up in the last couple months, this one was the only one that didn't have the countdown timer running continuously. In fact, when I brought up that screen on that computer, I had to look at it for a second to see what wasn't right about it, as I knew something was off, and then noticed that there was no countdown on this one unlike all the others.

Well, I'll just chalk it up to weirdness, and keep an eye on it when I get them all up and running. Oh, one question. I remember mentioned to me in the last month or two that there was a program that I could put onto each machine, and be able to monitor all of the systems running BOINC as if I was sitting right in front of it, it gives the same info that I see when I pull up BOINC. Do you happen to know what that is, and how well it works? Thanks!

You can use BOINC Manager to view any system on your network. Once you have configured BOINC to allow remote connections. However if you want to see all of them at once you are probably thinking of BOINC Tasks or BOINC View
I prefer BOINC View. I use it to: 1) Make sure all the hosts are running. 2) Make sure they all have work. I liek the way it does this better than BOINC Tasks. YMMV
17) Message boards : Number crunching : Using Android mobiles to crunch BOINC stuff (Message 1791367)
Posted 2 days ago by Profile HAL9000
As it has been said. Without a SIM card an Android phone will function basically like a tablet. The mobile device only needs an internet connection so that BOINC can download/upload work. Wifi is normally the easiest option, but sometimes, depending on the device, it can be configured to access the internet via an USB connection from a PC.

You can run your mobile device several different ways.
-If you set BOINC to only run 1 task at a time it will run the coolest, use the least amount of power, & likely complete work the fastest. Then you might be able to go a normal day letting BOINC run while on the battery & only charging it overnight. On my Galaxy S4 went for this configuration & set BOINC to stop when the battery got to 20-30%.
-You could use your device normally and let BOINC only run when plugged in to charge. If you had it plugged in overnight 7-8 hours it would take a few days for work to complete, but that isn't a problem.
-You could leave the device plugged to a power source with the battery removed to run 24/7.

If you have an interest in running several mobile devices without batteries this article might be of interest to you.
http://www.instructables.com/id/How-to-use-phone-without-battery/

Leaving a lithium battery connected to a charger 24/7 will cause it to wear faster then with normal use. Some have sound that doing so caused the batter to swell and damage the device. If a swelling battery isn't detected soon enough it could be a fire hazard.

My observations have been that most of the heat seems to come from the battery charging/discharging rather than the device processor. Even while setting in direct sunlight charging the CPU temp, while running a single task, didn't reach the point where the clock was throttled.
18) Message boards : Number crunching : Decided to Bail Out on SETI for a Short While (Message 1791350)
Posted 2 days ago by Profile HAL9000
And Kittes@home just look at mices...........
My kitties just keep on looking at the stars.

My fuzzy minions waste their time watching movies all day.
http://i.imgur.com/IvWLL9K.jpg

So, can you explain better what a 6-month GBT task would look like (download data size, perhaps), what mathematics would be performed, and why it can't be broken down into smaller sub-sections? That's asked out of genuine curiosity.

Would I be project scientist - perhaps. I will know design decisions when they will be formed. From common sense to get improvement over Arecibo data longer time periods can be used for signal accumulation so longer initial arrays for FFA. Smth like large FFA or bigger. Targeted search in general allows longer time domain for signal accumulation that results in better signal/noise ratio => increased sensitivity on the same hardware.

If longer tasks increase our chance of finding our needle in the cosmic haystack, I am all for it.
I do not care if the apparent result is me crunching one WU per day.

If that one WU is more productive than tossing ten a minute out the window, so be it.

I have long ago said that I luv my credits, but that is NOT why I am here.

I am here for the freaking SCIENCE of it, and that is what attracted me to the project in the first place, and that is what keeps me here.

And the fact that things are progressing here, rather that the same old...
Makes me even more determined to participate in this amazing project.

Meow!

I believe one of the issues the project tries to balance is allowing as many host to participate as possible.
Only having much larger, or complex, workunits could raise the minimum machine specifications. It might even cause a significant portion to no longer be able to contribute. Offering both types of work would allow a large number of hosts to participate & have more complex data analyzed by hosts that are capable. However that is more work on the back end.
19) Message boards : Number crunching : Kitty cooling off thread................... (Message 1790939)
Posted 3 days ago by Profile HAL9000
Just for a cooling off period..................
Might do well to watch How about a kitty little vid for a rest stop?

Meow.

While your kittehs are cooling off you might find this helpful. To better optimize them for the warmer weather.
20) Message boards : Number crunching : Optimisations for NUMA (Message 1790564)
Posted 4 days ago by Profile HAL9000
I had a server with dual Xeon E5645 CPUs. I could enable or disable NUMA in the BIOS & doing so had no apparent effect on BOINC or the project applications I was running.


Next 20

Copyright © 2016 University of California