Posts by FalconFly
|
81)
Message boards :
Number crunching :
"Can't Merge Hosts - They're Incompatible"
(Message 901613)
Posted 30 May 2009 by  FalconFly
Post: This line looks weird in one System : Microsoft Windows XP Note the "Editon" vs. the correct "Edition", this may be sufficient to block the Merge (not sure though) Try temporarily removing the 2nd Video Card for a quick test (launch BOINC, Update SETI, then attempt manual merging). Also, see if the Host Name (only visible to you) and IP address are really the same. Other than that, looks like you used a different BOINC Version on the old Host (which did not report Driver Version). If all else fails, you could temporarily give the new Host the old Host ID (requires manual edit of the client_state.xml - make a backup of it first). Look for : <project> <master_url>http://setiathome.berkeley.edu/</master_url> <project_name>SETI@home</project_name> further down..... <hostid>4734505</hostid> After re-launching BOINC, updating Project SETI, then exiting BOINC again, it "should" be possible to manually merge the Hosts. If even that doesn't work, I wouldn't know what would... -------------------------------------------------- Short version : 0) Preparation : set BOINC run mode to "Suspend" ...otherwise the later write-protected client_state.xml might screw up WorkUnits just finishing or downloading, as BOINC can't update their status for the moment 1) exit BOINC 2) make backup of client_state.xml (BOINC data Directory) 3) edit Host ID into 4734505 (your old host's ID) 4) write-protect client_state.xml 5) launch BOINC, hit Update Project on SETI 6) after confirming your old Host now holding updated Info, Quit BOINC again 7) attempt to Merge Hosts again (try both "Merge with the same Name" and manual function "Merge this Host" with your old Host, bottom of screen in Host Details 8) un-writeprotect client_state.xml 9) re-edit client_state.xml for Host ID listed in SETI Computers Page (should be your current ID again) 10) done (hopefully) |
|
82)
Message boards :
Number crunching :
"Can't Merge Hosts - They're Incompatible"
(Message 900362)
Posted 28 May 2009 by FalconFly
Post: I recently was able to merge an abandoned host that the "Merge Hosts with the same name" function also stated was incompatible (create date apparently was a problem). I went to the affected Host and manually selected "Merge Hosts" from there, at least it offered two viable other hosts that I could merge it onto. May not help you in your case but worth a shot maybe... |
|
83)
Message boards :
Number crunching :
Replica database up and running again
(Message 899145)
Posted 24 May 2009 by FalconFly
Post: I assume they have Flooding filters in place since years, since the Servers have always been under high default load - would be too easy for some bad guys to knock it over the limits. Other than that : It a Webserver is hosting for some ~145000 active Users, a fraction of which frequents the Website... The server setup simply has to handle it. Period. If that becomes questionable, Staff would do what they already have done multiple times - remove too I/O-intensive features until more server (or database) capacity is available. When it comes to scrapping realtime data off the online database, I assume they do run regular logfile analysis and block off-site services that cross their personal limits. So if it still works, it's safe to assume that they have no complaints about it (yet). Arguably though, all large Stats site Admins I know are actually very sensible towards the topic since years (there've been worse times than today) and long adapted "fair use" policies concerning their stats gathering. |
|
84)
Message boards :
Number crunching :
10th Anniversary Donation Challenge!
(Message 899100)
Posted 24 May 2009 by FalconFly
Post: Not going well this month I guess... Anyway, Blurf calls so I'm there ;) Thank you for your gift of $500.00 for: SETI@home. |
|
85)
Message boards :
Number crunching :
How much power!
(Message 898908)
Posted 24 May 2009 by FalconFly
Post: Do you think this will cost me monthly. It has a 1600 Watt PSU. Going to have seti Running 24/7 As soon as i get my GPU'S Cooled down. Unless you plan to add 2 or 3 more GTX295 to the setup, the PSU is absolute overkill (?) |
|
86)
Message boards :
Number crunching :
CLOSED** SETI/BOINC Milestones (tm) XVII **CLOSED
(Message 898726)
Posted 23 May 2009 by FalconFly
Post: Just about as I type going past 6000000 on SETI for the 10-year anniversary race  Having installed an "Afterburner" which I'll use for another 2-3 weeks at maximum power, I'm still waiting and accelerating to pass my personal sound barrier : 100k RAC. After that, let's see how fast it will go 
|
|
87)
Message boards :
Number crunching :
your opinion-best spyware & antivirus.
(Message 898367)
Posted 22 May 2009 by FalconFly
Post: I can easily beat that, personal record witnessed on someone installing WinXP (directly connected to the Internet) getting infected was 3 Seconds after System startup. Security and Windows just doesn't mix, never meant to. |
|
88)
Message boards :
Number crunching :
PCI CUDA Cards Now Available In Australia
(Message 898178)
Posted 22 May 2009 by FalconFly
Post: I'm not sure if you would want to run CUDA on a PCI card. The theoretical 133MB/sec peak bandwidth (assuming you won't find a Motherboard/Card combination that actually allows for the theoretically possible 66MHz / 266MB/sec) will seriously cripple the performance. Practically, the actual bandwidth will be below these figures (100-110MB/sec). Even 8x AGP was found not really feasible due to its limited downlink polling results back from the card. |
|
89)
Message boards :
SETI@home Science :
Do you really think SETI(@home) will succeed?
(Message 897810)
Posted 21 May 2009 by FalconFly
Post: Quote from Calvin & Hobbes Sometimes I think the surest sign that intelligent life exists elsewhere in the universe is that none of it has tried to contact us. When I look at what's happening on this planet, there may be more truth to it than we'd like to hear about... |
|
90)
Message boards :
Number crunching :
your opinion-best spyware & antivirus.
(Message 897801)
Posted 21 May 2009 by FalconFly
Post: After my battle with sys defender, I realize I need better protection. Please tell me what you recommend for both spyware and antivirus protection. Sounds cheesy but here's my preferred ones : Common Sense(tm) V1.0 Experience(tm) V36.1 Honestly, I found no Software Package could ever beat above combo and essentially, they're the only ones who actually work. For everything else, I recommend a blend mix of the following : - Anti-Rootkit (multiple) - Anti-Spyware (multiple) - Anti-Virus (multiple) - Anti-Malware (multiple) - Anti-Trojan (multiple) Since all existing Antivirus Programs utterly fail at doing their job right (fully cleaning and restoring a functional system) since many years now, salvation of Data and a following new Windows Installation is often the quicker way. When typical AntiVirus software reports a problem, it's usually too late as the system is already infected and therefor compromised. The times where a user would read "Virus found and destroyed, have a nice day" are long over; modern Antivirus suites more and more often can't even quarantine and contain the threat anymore. To add to injury, a system left "clean" by any AntiVirus scan typically is not clean and few Antivirus programs even work - let alone excel - in all six required categories (Rootkits, Worms/Viruses, Spyware, Malware, Trojans, Exploits). With recognition rates far below 100% accross the board, they're effectively useless except for signaling "Congratulations, your System is ready to be re-installed" (compared to what their Advertising claims it could do). Everything else is manual hand-work, i.e. identifying the premier problem, downloading and running dedicated removal tools etc. Still the system may easily be still compromised and no trust should be given in it until a known clean backup is restored or it is reinstalled from scratch. |
|
91)
Message boards :
Number crunching :
Not requesting tasks
(Message 897219)
Posted 20 May 2009 by FalconFly
Post: Based on my experiences, here's a list of things to look after and possible tweaks : Average turnaround time : x.xx days Displayed in the SETI Computer details tables, this value normally corresponds to whatever cache setting a User has set. However, losing a chunk of Workunits (i.e. system failure, HD crash or alike) apparently negatively affects the value. It's only my suspicion, but having lost a whole bunch of Astropulse workunits resulted in one of my systems Average turnaround time to rocket up beyond 10 days - resulting in almost no cache being maintained anymore and single cores beginning to run empty. I assume that for whatever reason the entire lost workunits deadlines were counted into the previous Average turnaround time, significantly increasing the value. Symptom : BOINC states "requesting 0 seconds of work" despite CPU cores starting to run idle even with no other projects to run. It took a week of normal system operation to return the Average turnaround time back to its normal value (~1 day). Ever since, the Cache problems have vanished again. Letting the system return a high number of Mainbeam workunits only for a day should reduce the value and resulting symptoms sufficiently to allow processing of a full Astropulse cache again the next day. But other than quickly returning series after series of results, nothing can be done about it. client_state.xml <host_info> <p_fpops>2228509013.328131</p_fpops> <p_iops>4833206207.789764</p_iops> See if the Benchmark values are correct. If for whatever reason the CPU remained in a power saving mode during the benchmark, it will only request new work according to the apparently lower performance. Manually triggering a benchmark should solve the problem. If need be, manually increasing the values are a functional way to increase work cache - at least until the next automatic benchmark runs. <time_stats> <on_frac>0.950683</on_frac> <active_frac>0.999722</active_frac> These numbers are muliplied with the desired cache size. If they are too low for whatever reasons (longer vacation with the Host turned off or BOINC simply not running), can significantly reduce Cache. Values close to 1.0 are ideal. <project> <master_url>http://setiathome.berkeley.edu/</master_url> <duration_correction_factor>0.177394</duration_correction_factor> In case BOINC returns workunits with erroneous (or even negative) cpu_time i.e. due to a series of faulty workunits, this value can easily go havoc very fast. Value around 1.0 should be normal, with modern hosts or when using optimized Apps, should be significantly below 1.0. <long_term_debt>-2229.060553</long_term_debt> This can be a naughty one. One should recheck that no Project a Host is attached to contains critical values (2000000 or higher). Unless this year-old bug is fixed in BOINC 6.x, will effectively cripple the Client Scheduler when exceeding approx. 2500000s (BOINC 5.x, BOINC 4.x experienced problems exceeding approx. 1500000s) for any attached project. Initially the LTD bug will reduce total cache in a frenzy attempt to favor refilling work for a work-starved project (i.e. LHC), eventually this will essentially prevent work being downloaded for any but the LTD starved project when left unattended. Setting LTD (and short term debt) to 0 for all projects will solve potential, related deadlocks. cc_config.xml Trick <cc_config> <options> <ncpus>10</ncpus> </options> </cc_config> Manually boosting the number of CPUs for a short time (i.e. allow BOINC to download, then reset back to normal value or delete file) is a dirty workaround but very useful to refill an empty cache. Remember that BOINC needs to be shutdown before editing any config Files --------------------------------------------------------------- That's all the potential troublemakers I'm aware of and all tweaks I know to restore a normal cache size on a troubled host. Hope that helps... |
|
92)
Questions and Answers :
GPU applications :
CUDA on x2 GTX295?
(Message 895364)
Posted 16 May 2009 by FalconFly
Post: Really don't understand what's happening here. I'm running Vista Home Premium x64 with one GTX295 without any problem. The only thing I can think of is that I found the monitors (or dummy plug) do need to be attached to different connector types on the GTX i.e one DVI and one HDMI. Running one single GTX295 never has been a big Problem for me (except under Windows 7 RC), it's two of those combined that cause all the Troubles. For one GTX295, I also did not need any monitor plugs or alike - it just ran as advertised out of the box (at least with the Palit brand cards I got). @Bad Actor I know and that's what motivated me to keep trying. But eventually, my time for experiments and failures simply ran out. Now I need reliable working configs and a single GTX295 is all I can get to run, despite many years of experience with BOINC and its .xml tweaks. Bottom line : One dual GPU card - no sweat. It becomes interesting when 3 or more GPUs come into play and that's what caused me more than one grey hair ;) |
|
93)
Questions and Answers :
GPU applications :
CUDA on x2 GTX295?
(Message 895338)
Posted 16 May 2009 by FalconFly
Post:
Enermax Revolution85+ 1050W... that should be more than sufficient. I restored the display configuration via Safe Mode and again attached a second monitor (this time lowering the Resolution and more carefully setting things up). Result : no change BOINC 6.6.20 (both 32bit and 64bit) still sees only 2 GPUs after being point blank installed from scratch. At this time I have to assume that there is a significant bug in the 64bit 185.85WHQL Driver (or elsewhere) that prevents BOINC from properly detecting all GPUs on dual GPU cards (the Driver apparently reports only one of the GPUs per card via CUDA). Installation of the lastest CUDA 2.2 Toolkit did not change anything either. The only thing left for me to do is to remove the 2nd GTX295 from all Systems (once again) and accept that some 1500$ were a wasted investment. I'm better off putting each GTX295 into a 32bit WinXP system due to these issues (I'm way behind timeline already so any further fiddling with this major pain in the arse is out of the question :p ) For me, the Dual GTX295 has proven to be a complete No-Go at this time and I don't have the time to wait for a new Driver. |
|
94)
Questions and Answers :
GPU applications :
CUDA on x2 GTX295?
(Message 894852)
Posted 15 May 2009 by FalconFly
Post: I didn't have the chance anymore, as the Driver did something upon detecting the 2nd Monitor automatically. Whatever it was, it went black screen and couldn't be revived to show the Desktop anymore, regardless of Monitor and where it was plugged into (even after re-enabling the IGP; after installing its driver - the System gave a black screen again). I may try Safe mode and other stuff tomorrow but as usual - it costs time again and prevents advancing (something that happened way too often the last week) |
|
95)
Questions and Answers :
GPU applications :
CUDA on x2 GTX295?
(Message 894750)
Posted 14 May 2009 by FalconFly
Post: Since I'm seeing the exact same Problems with the 185.85WHQL both under Win 7 RC 64bit and Vista 64bit, I guess it's a driver Problem. I'll try attaching 2 Monitors now, because otherwise I'm completely out of Ideas. Those dummy plugs or Monitors attached appear to be the only reliable workaround - and I tried hard over several days to get BOINC to recognize my Dual GTX295 my 4 GPUs. Ironically, WinXP 32bit and BOINC 6.6.20 32bit does recognize all 4 GPUs right from the start with the 185.85WHQL Driver. Unfortunately, having some 3.5GB of Video RAM stealing RAM for the upper PCI address area badly screws this 32bit OS, thus it can only run one GTX295. The closest I ever got to get 4 GPUs running with Win 7 64bit or Vista 64bit was the V10a Team Mod - unfortunately is isn't smart enough to limit launching Tasks i.e. for 4 CPU cores and 4 GPUs; instead it can easily end up running 6 or more tasks on the CPU while running only 2 (or less) Tasks on the GPUs which may end up running idle, all by pure unlucky draw. This whole thing has been literally driving me nuts, having some quite costly hardware sitting around unused :( -- edit -- Oh well, after installing a 2nd Monitor onto the other Card, the screen now stays black unter Vista64, even switching to onboard IGP doesn't help once Vista64 is up and running (which it does in the background). I think I'll scrap Win7 RC 64bit & Vista64 and go with one card per System with XP32, at least that's working without the incredible pain... |
|
96)
Questions and Answers :
GPU applications :
CUDA app_info related Questions
(Message 893173)
Posted 9 May 2009 by FalconFly
Post: There is an important Update at Lunatics. It appears the V10a Team Mod Package is obsolete and much of the hassles not required anymore. A new Package tailored for BOINC 6.6.20 capabilities is in the making right now. |
|
97)
Questions and Answers :
GPU applications :
CUDA app_info related Questions
(Message 893097)
Posted 9 May 2009 by FalconFly
Post: Alright, I finally got the setup running. (reasons behind the exact steps I took are colored italic) Steps I took : - changed Network config from my existing LAN Proxy settings to Direct Internet Connection This may not have changed much, but I feared the V10 CUDA app may download behind BOINC's back. I've been told this is not the case, but in the past I experienced that Proxy settings required special attention via cc_config.xml tweak when used with BOINC 6.x. Just needed to eliminate any potential sources of Problems. - set MultiGPU Rendering (SLI) to Disabled in the NVidia System Controls Reportedly is required to use both GPUs for CUDA under BOINC. Under Windows 7 RC x86 no restart was required with 185.85 WHQL Driver. - created new Venue in User preferences "Work" - in Venue "Work" accepted only SETI Enhanced WorkUnits This is to be used for dual GPU machines only, thus not affecting existing Network including single GPU setups, which of course continue to run SETI Enhanced and Astropulse WorkUnits without troubles. Otherwise, I repeatedly experienced that a 4CPU core/2GPU setup was ending up running 6x AP Workunits on the CPU, while the GPUs were actually idle - with some 20-30h before getting another chance to kick in on new WorkUnits (which actually never came, System downloaded only more AP WorkUnits :p ) - Installed BOINC Win64 6.4.5 This because it was the first Version I could confirm in the Lunatics Forum that worked for another User with the V10 mod. Others likely do as well. BOINC 6.4.5 also works fine to monitor with BOINCview 1.4.2 - before attaching to Project, set Run mode to "Suspend" Prevents filling the Cache with Astropulse WorkUnits already, as the Venue for the Host initially will still be Default or Home - attached to SETI - after downloading stock client and a single WorkUnit, quit BOINC The Run mode Suspend will make BOINC request only 1 sec worth of work, which will be max. a single WorkUnit - changed Host Venue from "Home" to "Work" The Host is now in SETI's 'Computers on this Account' listing, now allowing to set the correct venue at bottom of computer Details - Prepared the V10 Team Mod (set value in "number_of_GPUs" File and created&set cc_config.xml to reflect total number of CPUs+GPUs) For a QuadCore CPU and a dual GPU card, number_of_GPUs set to 2, cc_config.xml set to <numcpus>6</numcpus> , as written in the Instructions at Lunatics - Installed the V10 Team Mod - Restarted BOINC and set Run Mode back to Normal Within a minute downloaded SETI Enhanced WorkUnits and now correctly - for me for the first time - fed 4 CPU cores and 2 GPUs on the GTX295 with work :) Voila, at worst there is one single WorkUnit in cache from the initial download (BOINC will request 1 sec of Work), the remaining cores now download and begin processing SETI Enhanced WorkUnits as planned. At least that worked for me, after many many hours of frustration and reading countless instructions but nothing really helping right-on. Maybe it helps others as well. |
|
98)
Questions and Answers :
GPU applications :
CUDA app_info related Questions
(Message 893028)
Posted 9 May 2009 by FalconFly
Post: Hm, okay... I've read some notes about useing the 2nd display but it had no details so that didn't make sense to me at the time. I'll keep that in mind (right now the system is kinda screwed and CUDA doesn't run at all anymore). I hope there's no general incompatibility with Windows 7 that throws sticks between my legs. Anyway, thanks for the advices, now I have a few Options to try :) |
|
99)
Questions and Answers :
GPU applications :
Win7 and CUDA - 9800GX2
(Message 892881)
Posted 9 May 2009 by FalconFly
Post: Yes, I just built a Win 7 x86_64 with a GTX285, using BOINC 6.6.20. I see exactly the same problem and still look for a solution. I think I will revert to an older Version of BOINC and attepmt this one again : http://lunatics.kwsn.net/12-gpu-crunching/v10-of-modified-seti-mb-cuda-opt-ap-package-for-full-multi-gpucpu-use.0.html Note that it will not work with BOINC 6.6.20, thus an older Version is needed. I still have to read myself which version exactly do run with this tweaked version. |
|
100)
Questions and Answers :
GPU applications :
CUDA app_info related Questions
(Message 892839)
Posted 8 May 2009 by FalconFly
Post: Okdiok, sounds good. I have the System up and running but one Problem remains : I can't get BOINC 6.6.20 (Win64) to recognize the 2nd GPU of the GTX295. So far I used the fresh installation, latest NVidia Driver and attached to SETI; it sees one of the GPUs and started crunching away, at this time all with the normal stock apps. SLI is disabled in the Driver Preferences and confirmed that with GPU-Z. How do I enable the 2nd GPU for SETI CUDA ? (When first launched, I still had the IGP enabled. That caused SETI only to see this IGP and denied it work due to too low VRAM. Already there, it failed to see the 2 GTX295 GPUs altogether. Now I've disabled the IGP run only on the GTX295, unfortunately only at half load...) Is that the solution when implemented into an app_info.xml ? <coproc> <type>CUDA</type> <count>2</count> </coproc> |
Previous 20 · Next 20
©2026 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.