Posts by HAL9000


log in
21) Message boards : Number crunching : CPU Benchmarks Did Not Run on Z800 (Message 1683557)
Posted 6 days ago by Profile HAL9000
That is truly odd, then, because this was a brand new computer based on an HP Z800 motherboard (I love *cheap* multiple threads (in this case, 16) - the whole machine (except for the video cards) cost me only about $100 net) and had never had BOINC run on it before; this was the first time BOINC was started IIRC.

In fact, I had come to the conclusion that the executables that actually run the benchmarks had not yet been downloaded (is that possible?).

The benchmark is performed by BOINC not the project. So there wouldn't be anything to download.
22) Message boards : Number crunching : Panic Mode On (97) Server Problems? (Message 1682721)
Posted 7 days ago by Profile HAL9000
I'm sure more will come soon. Probably not too many, and only for a short feeding cycle.. but hopefully we get some more soon.

Not to toot my own horn.. but there is some minor splitting activity afoot. I've picked up five so far. It's not much.. but it's something.

Looks like I've been picking up a few as well over the past 2-3 hours. I even managed a few for the NV GPU on an old iMAC I'm using. Should be interesting to see how it manages the GPU work for the first time.
23) Message boards : Number crunching : Why use CPU on SETi@home? (Message 1682569)
Posted 7 days ago by Profile HAL9000
I'm using two of my laptops at home for crunching. An old Pentium M. Which is overclocked from 1.5GHz to 2.4GHz & runs 100% 24/7. I have actually been trying to kill the things for the past 2 or 3 years, but it just keep going!
Them my newer i3-390m I am currently only using the GPU. As I can run either run 4 threads on the CPU at 100% or I can use the GPU at 100%, but the cooling in it is not sufficient to run both at 100%.

you got me an idea going on 4 my fathers PentiumM laptop T41...btw, how did you change the multiplier?

That is why you should use TThrottle!
This way either CPU gets heat from GPU on laptop...or GPU gets it heat from CPU...
running both CPU & GPU is to hot...but using the TThrottle you equalize the top temps...& it doesn't overheat! ;)

but you also, have a faster data release with your GPU, than with your CPU...did you choose GPU crunching beacuase of that fact?

For the Pentium M overclocking I did a jumper trick on the CPU to increase the FSB from 100MHZ to 133MHz. Then used a software tool that let me overclock the FSB further. I think I still have the links for that info somewhere at home, but I am current at work.

Since I actually use the i3 laptop it works out better if I don't use the CPU for crunching. It has weird lag issues when the CPU is fully loaded. Which I don't see on any other system I use for crunching. Plus I prefer hardware cooling solutions. So if I wanted to run both I would likely just engineer something to help cooler it better. I do have a 50w peltier around somewhere from the days of overclocking old slot1 celerons.
When I get a new laptop I'll retire the i3 to do nothing but run BOINC on the CPU. Running 4 CPU threads gets more work done in a day than the GPU alone.
24) Message boards : Number crunching : CPU Benchmarks Did Not Run on Z800 (Message 1682465)
Posted 8 days ago by Profile HAL9000
I can't say I have ever seen BOINC have any "default values" other than 0's for the benchmarks.
With no benchmark values it does make it rather impossible to fetch work for any projects. Which I have experienced a few times setting up a new machines when forget to disable the skip benchmarks flag in my default cc_config.xml.
25) Message boards : Number crunching : Why use CPU on SETi@home? (Message 1682462)
Posted 8 days ago by Profile HAL9000
Everybody chooses their own projects...I will not comment on other people choises, 'cause it's all about personal preferences...and we know all that in those we do disagree... ;)

With this tool: http://www.extreme.outervision.com/PSUEngine
it gives that you use 515W in peak power - without WATER PUMP...so add that up! ;)
so I would on your place check the RAM...my one system has worked OK, but occcionally give me BSoD or "black screen" or stoped at some given time -> MEMtest86(+) gave me a clue, it was a RAM... :/

Sorry to hear that...did it go out in the 2y warranty?

As I've mentioned - I did...when using the SETi@home on a laptop, I always use TThrottle (but you have to have a newer proc than P4)...on my laptops TThrottle is set to 80-85% of Tjunc proc temp...on other peoples laptops it's set on 65% of Tjunc proc temp -> to keep them cooler! It really works great...so my suggestion is to use that program with using 100% of BOINC time -> gives best results! ;)

So what preferences do you run on your laptops? ;)

I'm using two of my laptops at home for crunching. An old Pentium M. Which is overclocked from 1.5GHz to 2.4GHz & runs 100% 24/7. I have actually been trying to kill the things for the past 2 or 3 years, but it just keep going!
Them my newer i3-390m I am currently only using the GPU. As I can run either run 4 threads on the CPU at 100% or I can use the GPU at 100%, but the cooling in it is not sufficient to run both at 100%.
26) Message boards : Number crunching : Panic Mode On (97) Server Problems? (Message 1682333)
Posted 8 days ago by Profile HAL9000
Crap, or something similar. Almost 10-days now, since the last good AP split/splat/splut.

It just has to be millions of AP's ready to split, since AP was started years after MB.

Where are they? In storage on another planet?

The aliens only let them make so many at a time. :/
27) Message boards : Number crunching : If I am going to start crunching with my car, it has a radiator. (Message 1682283)
Posted 8 days ago by Profile HAL9000
I wouldn't crunch with my car as I don't run it 24/7.

if it's electric or plug-in hybrid...then you charge it in your garage, right?

so, why not run it? on idle or parked (charging)...


The place I'm staying at right now doesn't have a garage suitable for parking/charging so I park on the street.

A report 2-3 years ago stated something like 60% of US residents live in multi-family housing (condos, apartments, & the like) without access to a location to plug in an electric/hybrid car.
Of course with my old commute, of at least 2.5 hours each way, running something while driving would have been more practical. Now I may drive about 2.5 hours total a week commuting.
28) Message boards : Number crunching : CPU Benchmarks Did Not Run on Z800 (Message 1682253)
Posted 8 days ago by Profile HAL9000
From the looks of that host it certainly looks like the benchmarks completed.
Measured floating point speed 2874.18 million ops/sec
Measured integer speed 8715.46 million ops/sec
29) Message boards : Number crunching : Strange Memory Phenomenon When Starting BOINC (Message 1681840)
Posted 9 days ago by Profile HAL9000
Here's something to look for:

In your task manager, go to the processes tab and check the box at the bottom that says show processes from all users. Click on the memory column so that you can see the process that is using the most memory at the top. On my work machine the top process right now is FileMaker Pro at about 200MB. Look for svchost.exe and note how much memory it's using. For me that's about 70MB. Now check for windows updates. On my computer, svchost.exe will peak at over 1.8GB before settling down.

I did the same thing on Windows 8.1 and svchost.exe only goes from about 25MB to about 70MB. I would expect Windows 10 to behave similarly.

Most services & background functions run under svchost.exe I have 13 instances of it right now.
Normally for me the item using most of memory is my internet browser. That probably doesn't apply to their system. As it is not common to put that in a machines startup.

If the OP wants to spend the time to try and find out what is going on it can be done. It just requires further investigation. Perhaps using more detailed tools. Such as Microsoft's Process Explorer.
30) Message boards : Number crunching : Why use CPU on SETi@home? (Message 1681819)
Posted 9 days ago by Profile HAL9000
I was just wondering how EFFICIENT is the crunching, 'cause we all spend electricity for it. Mostly, when the last month a bill came for half year change of electricity!

And, btw, my 3 cards are:
1. nVidia Quadro 2000 - listed as 30% of the speed of the Titan, on an office PC
2. intel HD 2500 - listed as 55% of the fastest intel HD 4600, also on an office PC
3. nVidia 730GT - a PCIe 8x card for a server board that I have in my computer.
Not so much, but works!

So I would like to hear from bronze users or less...what is their opinion?
;)

Also, keep in mind that most future users will use CUDA or OpenCL cards in their computers...so that is the future, obviously! ;)

GPU computing is often more efficient in work done per KWh. However there are some kinds of data that GPUs still can not processes as well as CPU. So a GPU may take longer than the CPU in the system to do the same work. As hardware, drivers, and software develop that will likely not always be the case.
SETI@home lets volunteers choose to use whichever resources they like. So you can choose to not use your CPUs for SETI@home if you like. Just change "Use CPU" in your SETI@home preferences.

Personally I only use 3 or 4 GPUs for SETI@home. One of the GPUs is nearly the oldest GPU you can even use for SETI@home. An ancient Nvidia GT 8500. The rest of my crunching comes from the vast number of CPUs I have on hand.
31) Message boards : Number crunching : Panic Mode On (97) Server Problems? (Message 1681610)
Posted 10 days ago by Profile HAL9000
Since we are now back from the outage I checked to SSP to see how many APs were being split, I'm disappointed.

Yeah, I've just had to let my main rig's CPU have another feed of MB's to keep it happy.

Cheers.


Yea, I REALLY wish there was a setting for "Store X days of MB, and Y days of AP" without having to do it manually. I love AP's but am happy to crunch MB's when they're not available.

But since it is a combined project, we can't do that. :(

Just run 2 clients. Set on for a venue with MB only & the other with AP only. Easy as cake. Piece of pie.
32) Message boards : Number crunching : Strange Memory Phenomenon When Starting BOINC (Message 1681609)
Posted 10 days ago by Profile HAL9000

Greetings BilBg,

Obviously you did not understand what I was saying in my post, so I will reiterate: :(

The thing I find strange is that since I updated to the latest version of BOINC, when I restart my PC an abnormally larger amount of RAM is being used initially. Then after a few minutes, or so, the RAM usage settles into what the machine normally uses when BOINC is running.

I have done scans for malware and viruses and have checked Task Manager often to see what may be causing this. I have found nothing amiss so I came here to find if there is something else I can do to figure this 'problem' out. I quote the word "problem" because I don't know that it IS a problem. I only know that this has only started since updating BOINC... or perhaps since doing some Microsoft updates... Come to think of it, this may have started after doing MS updates. Ok, I don't have a restore point before those updates since I have done 2 since then.

Ah HAHHHH! It's Microsoft's fault. ;)

Keep on BOINCing...! :)

The signs of RAM going bad are generally not increased memory usage. You would be looking at a system lockup, random reboots, or apps crashing.

You said you have checked task manager often. Alright. What is it you are actually looking at in task manager? You simply listed that your system was going from 50% to ~35% memory usage. That leads me to think you are only looking at the Performance tab & not the Processes tab. Where you can enabled various memory options for programs & sort by the size.
Another handy took is the Resource Monitor. Which you can even open from the Task Manager Performance tab.

Given you have stated you have 8GB of RAM & memory usage goes from about 4GB down to about 2.8GB. Then finding something eating up over 1GB of RAM should take only a moment.
33) Message boards : News : Network Outage - Sunday May 17th (Message 1681499)
Posted 10 days ago by Profile HAL9000
Thanks HAL - actually just planting "seed"
Re: Pre and Post Debunking/Troubleshooting potentially "Bad" data from tapes/images; possibly correlating to these highly-magnetic events.
=
My thoughts are, to correlate "significant events such as Xclass Solar Flares and CMEs or EMPs or significant Solar winds" Calendar with Recordings Dates and Times, in order to cull these from our precious Time and Efforts.

Or, build-in a specific algorithm or filter(s) to try and salvage the potentially compromised data from these correlated recordings.

I would guess that they would shutdown activity at Arecibo if to much interference would be caused by an event. It would be interesting to see if any data we have processed correlates with any specific events & what the outcome of the data we have processed happens to be.
34) Message boards : Number crunching : Loading APU to the limit: performance considerations - ongoing research (Message 1681405)
Posted 10 days ago by Profile HAL9000

I am not sure is the iGPU running ~39% & the CPU times still being that high is a bad sign or helps point to something else to try. I plan to try more configs to see if there is one that will make everything magically work together as well as BayTrail in the meantime.
-unroll 4 -ffa_block 64 -ffa_block_fetch 16
-unroll 4 -ffa_block 128 -ffa_block_fetch 32
Then play around with unroll again.


To make your search more targeted:
main loop intervenes FFAs in AP. And time to time, additionally to short FFA much bigger long FFA are performed. So, with Clean01 you could see periodical load in GPU-Z. Also, allocated memory will be change periodically. To separate main loop time from FFA time one can vary corresponding memory buffers size and look for GPU memory usage in GPU-Z. -unroll changes mainloop. -ffa_block_size changes FFA buffers.


With various configurations I have found that I can reduce iGPU usage, as I noted previously. However, I still have not found a config that reduces iGPU load on CPU any further. Maybe it is more correct to say "config to reduce iGPU load on memory" rather than CPU? Since I think focus is on memory contention right now.
35) Message boards : Number crunching : Disk space error message (Message 1681396)
Posted 11 days ago by Profile HAL9000
Hiya, Thanks

It took a bit of spotting, I had set max disk usage to 10Gb, leave at least 10% free, and leave at least 100GB free in the BOINC website defaults. This is fine on my old system as BOINC is installed on a 1Tb disk.

The system I've just resurrcted has a 80Gb SSD C drive with everything on it (it'll only run BOINC/ S@H). The leave 100Gb free was the killer :-).

D'OH.

Heh that will do it. While there are some projects that can use huge amounts of disk space. I don't happen to run any of them so I set my disk settings for BOINC a bit more... freely.
Disk Use no more than 5000 GB Leave at least 0.001 GB free Use no more than 100% of total

With the current SETI@home cache limits. Only running AP tasks would give a max disk usage of ~800MB for each device used, or ~2.4GB for a dual GPU system also running CPU.

I don't happen to run BOINC on any drives with 5TB or more of storage. My 24 core system only uses a 2GB RAM drive in fact. I figured this way I wouldn't have to ever worry about any space requirements while running projects. I imagine if I ever run into a situation where I'm out of disk space from BOINC using it all I probably need to upgrade my storage anyway.
36) Message boards : Number crunching : If I am going to start crunching with my car, it has a radiator. (Message 1681388)
Posted 11 days ago by Profile HAL9000
Most car computers are still closed systems, crunching with a car might never be possible, nvidia or not..


Ten years ago that was true, but between smartphone and USB connectivity and built-in OnStar-type cell systems, a very large and growing number of car networks are not closed. Bluetooth, for instance, is standard in almost all new cars, although crunching is not.

Well since I didn't mention a network, a computer system in a car doesn't need to have a network to be a computer, I used a few computers that were programmable that did not have what We today would consider a network, still the cpu I'd think is still closed, usb/bluetooth connections I think just go to the stereos, not to something that is meant to manage the cars engine and transmission for emission purposes, and not for crunching. I doubt the cars cpu will ever be allowed to crunch, since smog emissions are heavily regulated, unless you like dirty air, anything that interfaces with the car's cpu, needs to meet CARB approval, that's the California Air Resources Board, cause that deals with emissions controls. Now Stereos are a different animal altogether, though I doubt you'll see any crunching there either, since Car Stereos are also regulated, by the FCC(the Federal Communications Commission). And even if you could crunch with a cars computer, who in the automotive world is going to let you know how to do that? If that is even possible to do and it may not be. This was an example of cars in California and more than a few states use CA as their standard for smog equipment.


Since I drive VWs I have a tool from http://www.ross-tech.com/ that allows me to change numerous things in my car. Also it is quite a bit more than a simple ODB code reader. While I can change engine parameters to run outside of federal emissions guidelines. I would need to change them back for state inspection & I don't feel the need to do anything like that.
I do adjust several other non engine related settings. Some of which are just enabling features that are not offered in the US. Like opening/closing the windows/sunroof with the key fob or having the option to disable the daytime running lights from the menu.

As you mentioned. I imagine the computer we would be most interested in would be the cars infotainment system rather than the systems that manage important functions such as the engine. I would hazard a guess there are a few people out there that have looked into getting into the infotainment system to have it to their bidding. Once we start getting ones that basically run Android or something something that has the ability to install apps it will probably just be a matter of click, click, install.
Even if the factory systems are not usable right now automotive PCs have been around for more than a decade. In some cases you can have a windows PC in your dash replacing the factory system, or use the backup camera input to connect a PC display. I was told when I got my '15 Golf in December that starting in 2018 backup cameras will be mandatory safety equipment. So that at least gives a video in to use.
37) Message boards : Number crunching : Panic Mode On (97) Server Problems? (Message 1681226)
Posted 11 days ago by Profile HAL9000
Before we all forget, this website setiathome.berkeley.edu is not where the upload/download servers are located. Open your server response file in the BOINC directory and you will see those are at setiboinc.ssl.berkeley.edu

The website traffic is carried on the normal campus data paths.
The data is carried on the SETI paid for Hurricane Electric path.

Yes the data graph we monitored was the traffic over the Hurricane line. However it is routed though the same hardware as the web pages.
Until recently setiathome.berkeley.edu traced to
t1-3.inr-201-sut.Berkeley.edu [128.32.0.65]
t5-4.inr-211-srb.Berkeley.edu [128.32.255.41]
Now with it tracing to
t2-3.inr-202-reccev.Berkeley.EDU [128.32.0.39]
et3-47.inr-311-ewdc.Berkeley.EDU [128.32.0.103]
It does seem very likely that the Hurricane line will also be routed through the new hardware.
38) Message boards : Number crunching : Panic Mode On (97) Server Problems? (Message 1681200)
Posted 11 days ago by Profile HAL9000
Well, I think we're closer to figuring out where the new cricket is going to be.
15 234 ms 240 ms * et3-47.inr-311-ewdc.Berkeley.EDU [128.32.0.103]

I just ran a tracert myself, and I came up with et3-47.inr-310.. so I'm not sure if 310 or 311 is going to be the primary, or if it does round-robin for load-balancing.

Regardless, "inr-310/11, 3_47" looks like probably maybe our new home, or really close to it. Unfortunately, the cricket summary pages for inr-310/11 are "temporarily out of service" at present, but based on the DNS name, those are definitely in the colo (ewdc - Earl Warren Data Center), which on the main page for the crickets, all of the entire EWDC section is gone. So.. that's progress, I think.

Thinking they used a logical numbering system for the routers around campus. I first tried cricket pages for 310 & 311 yesterday with the same results you have found. I later tried 200 to 399 also without any luck.

Given these graphs are provided for the "customers" of UC Berkeley Information Services and Technology (IST). I would imagine generating the new pages is on someones "to do" list, but as the priority of a heater repair person in the middle of summer.
39) Message boards : Number crunching : If I am going to start crunching with my car, it has a radiator. (Message 1680900)
Posted 12 days ago by Profile HAL9000
There was a show I was watching about cars on Discovery ~15ish years ago. Where they made a statement along the lines of "Computers are becoming an important part of how cars work & one day you will measure your cars performance in gigaflops instead of horsepower".
Seems they were not really that far off of the mark.
40) Message boards : News : Network Outage - Sunday May 17th (Message 1680770)
Posted 12 days ago by Profile HAL9000
Expect the unexpected - an FYI - headsup
NASA Spaceweather.com News: 2 items -- coinciding with "our" Network update

GLANCING BLOW ON MAY 17: NOAA forecasters estimate a 45% chance of polar geomagnetic storms on May 17th when a CME is expected to deliver a glancing blow to Earth's magnetic field. The storm cloud was hurled in our direction by a magnetic filament, which erupted away from the sun on May 13th
=
Coronal Holes: 15 May 15
A stream of solar wind flowing from the indicated coronal hole should reach Earth on May 18-19. Credit: SDO/AIA.
=
Anytime either of these type events occur, otherwise unexplained phenomena tag along ie geo-magnetically influenced phenomena; like cell phones / Wi-Fi and other radio wave/microwave stuff, TV signals and internet connectivity often affected. Usually temporarily but noticeable.

I would venture to say, it's possible that even Arecibo et al recordings could be affected ( by severe "noise" and interference during these events ).

You can always check the Arecibo Observatory Schedule to see if anyone is using the telescope for something that would allow SETI@home to record data.


Previous 20 · Next 20

Copyright © 2015 University of California