Message boards :
Number crunching :
Panic Mode On (101) Server Problems?
Message board moderation
Previous · 1 . . . 8 · 9 · 10 · 11 · 12 · 13 · 14 . . . 27 · Next
Author | Message |
---|---|
betreger Send message Joined: 29 Jun 99 Posts: 11361 Credit: 29,581,041 RAC: 66 |
Ohh great ... MORE Splitter Errors ! That's why we crunch Seti, it's exciting. |
Darth Beaver Send message Joined: 20 Aug 99 Posts: 6728 Credit: 21,443,075 RAC: 3 |
Can some one please tell me what this error is in my event log . 7/11/2015 8:54:35 AM | | app version refers to missing GPU type C0.5DA I redid Lunatic and changed from AVX on the CPU to sse3 and change GPU to sse4.1 from AVX so I'm shore that's what caused it so I'm not to worried. Bionic seems to be working ok but I would like to clean it up and have no errors any advise most help full thanks |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
Your Intel system is a Core 2 Quad, so SSSE3 would be best application (notice the 3 S es). For both systems CUDA50 would be the best application for the video cards. Grant Darwin NT |
Darth Beaver Send message Joined: 20 Aug 99 Posts: 6728 Credit: 21,443,075 RAC: 3 |
Your Intel system is a Core 2 Quad, so SSSE3 would be best application (notice the 3 S es). For both systems CUDA50 would be the best application for the video cards. Thanks Grant not shore if you where talking to me but it's the AMD that has the error , I did load CUDA 50 with Lunatics witch is why I ask if anybody knows what it means or why it's saying the error . It says I have CUDA 50 in the client and is doing them but thanks for the advise . |
WezH Send message Joined: 19 Aug 99 Posts: 576 Credit: 67,033,957 RAC: 95 |
Now it seems that RTS is recovering. Maybe those many resends of those "corrupted" tasks helps a little bit. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Now it seems that RTS is recovering. There have been many fewer shorties in recent tapes. Because my work fetch is constrained by the 100 WU quota limit (rather than the 1 day cache setting), I drain far fewer tasks from RTS while working on the chewy ones. |
WezH Send message Joined: 19 Aug 99 Posts: 576 Credit: 67,033,957 RAC: 95 |
Now it seems that RTS is recovering. Yes, You are right, that is also helping RTS recovery. EDIT: Downside of RTS, now those "corrupted" wu's stay longer in action... Previously when I did abort one of those, it was sent to next host at once. Now they stay in buffer for some time. Like this WU |
rob smith Send message Joined: 7 Mar 03 Posts: 22199 Credit: 416,307,556 RAC: 380 |
Richard, I'm fairly certain you've explained how to decode the job_log_setiathome.berkeley.edu.txtfile - I can so far into a typical line, but how does one get the run date? 1241909224.687500 ue 40434.450553 ct 17077.450000 fe 80360000000000.000000 nm 21fe09ae.26657.73409.6.8.83_1 et 18191.468750 The bits I've identified: 21fe09ae- the "tape" 26657- splitter process id 80360000000000.000000 nm- looks to be something to do with the observation frequency 17077.450000 fe- is this the bit to do with the observation date & time stamp? I've got four large files of this data (one for each of my crunchers), and it would be interesting to decode them and see how it stacks with your published records. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Not quite. Remember that this is a BOINC log, and it is kept in the same format for all projects - so it only contains processing and administration data, no science (forget the observation frequency). I would break it down as follows: () 1241909224.687500 Processing time (task completion), in Unix timestamp format. http://www.onlineconversion.com/unix_time.htm makes your example Sat, 09 May 2009 22:47:04 GMT Using Microsoft tools (Access, in particular), I use Format$(Int(Min([Datestamp])/86400)+25569,"dd-mmm-yyyy") to get the first ('minimum') processing date for each tape: 25569 accounts for the difference between the start of the Microsoft epoch (31 December 1899) and the start of the Unix epoch (1 January 1970) (ue) 40434.450553 I believe this is the original estimated runtime for the task. (ct) 17077.450000 CPU time used during processing. (fe) 80360000000000.000000 FLOPs estimation (<rsc_fpops_est>). (nm) 21fe09ae.26657.73409.6.8.83_1 Task name - project-specific elements. (et) 18191.468750 Elapsed time for processing. To confirm the processing time - this machine finished a task at 1446904926 (no microseconds logged by more recent clients), or Sat, 07 Nov 2015 14:02:06 GMT, just as I was proofreading this post. |
rob smith Send message Joined: 7 Mar 03 Posts: 22199 Credit: 416,307,556 RAC: 380 |
Thanks Richard - I can have a play now :-) Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Have fun. Just confirmed my association of ue with initial estimated runtime - fe divided by APR. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Seems the GPU models page is Still Broken, Still showing; NVIDIA Mac No GPU tasks reported The latest stats from the Applications page shows; Linux/x86_64 7.08 (opencl_ati5_cat132) 18 Sep 2015, 18:34:25 UTC 141 GigaFLOPS Linux/x86_64 7.08 (opencl_ati5_nocal) 18 Sep 2015, 18:34:25 UTC 194 GigaFLOPS Linux/x86_64 7.08 (opencl_ati5_sah) 18 Sep 2015, 18:34:25 UTC 0 GigaFLOPS Linux/x86_64 7.08 (opencl_ati_cat132) 18 Sep 2015, 18:34:25 UTC 92 GigaFLOPS Linux/x86_64 7.08 (opencl_ati_nocal) 18 Sep 2015, 18:34:25 UTC 20 GigaFLOPS Linux/x86_64 7.08 (opencl_ati_sah) 18 Sep 2015, 18:34:25 UTC 10 GigaFLOPS Linux/x86_64 7.08 (opencl_nvidia_sah) 18 Sep 2015, 18:34:25 UTC 3,213 GigaFLOPS Linux/x86_64 7.08 (cuda_opencl_100) 21 May 2015, 23:56:05 UTC 17 GigaFLOPS Linux/x86_64 7.08 (cuda_opencl_cc1) 21 May 2015, 23:56:05 UTC 4 GigaFLOPS Linux/x86_64 7.08 (opencl_ati_100) 21 May 2015, 23:56:05 UTC 78 GigaFLOPS Linux/x86_64 7.08 (opencl_nvidia_100) 21 May 2015, 23:56:05 UTC 317 GigaFLOPS Linux/x86_64 7.08 (opencl_nvidia_cc1) 21 May 2015, 23:56:05 UTC 165 GigaFLOPS Total GPU GF = 4,251 Mac OS X/64-bit Intel 7.08 (opencl_nvidia_mac) 16 Oct 2015, 18:45:36 UTC 8,967 GigaFLOPS Mac OS X/64-bit Intel 7.07 (opencl_nvidia_mac) 23 Apr 2015, 18:50:41 UTC 1,684 GigaFLOPS Total nVidia GF = 10,651 There are Over Twice as many Mac nVidia Results as Linux Combined GPU Results, there are Numerous Linux GPUs listed, yet Not a Single Mac nVidia GPU is said to be returning Tasks.. |
WezH Send message Joined: 19 Aug 99 Posts: 576 Credit: 67,033,957 RAC: 95 |
TBar, just report this problem to BOINC server code developers. Not to this Seti@Home Server Problems -thread. IMHO. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
TBar, just report this problem to BOINC server code developers. Both pages have this on them; They Are Hosted on a SETI Server. Why you think it should be posted on a BOINC page is a Mystery. The Problem is Most Definitely a SETI Server problem and has More relevance than many of the Posts here. ??? |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
There are Over Twice as many Mac nVidia Results as Linux Combined GPU Results, there are Numerous Linux GPUs listed, yet Not a Single Mac nVidia GPU is said to be returning Tasks.. No, Those stats are not the number of results, But the amount of compute power for each app version, in GigaFLOPS. Claggy |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
TBar, just report this problem to BOINC server code developers. The project runs Boinc server software, maintained by the BOINC developers: boinc / html / user / gpu_list.php 1 <?php 2 3 // This file is part of BOINC. 4 // http://boinc.berkeley.edu 5 // Copyright (C) 2011 University of California 6 // 7 // BOINC is free software; you can redistribute it and/or modify it 8 // under the terms of the GNU Lesser General Public License 9 // as published by the Free Software Foundation, 10 // either version 3 of the License, or (at your option) any later version. 11 // 12 // BOINC is distributed in the hope that it will be useful, 13 // but WITHOUT ANY WARRANTY; without even the implied warranty of 14 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. 15 // See the GNU Lesser General Public License for more details. 16 // 17 // You should have received a copy of the GNU Lesser General Public License 18 // along with BOINC. If not, see <http://www.gnu.org/licenses/>. 19 20 // generate a page of the best-performing GPU models. 21 // 22 // "best-performing" is defined as minimizing the average of 23 // 24 // elapsed_time(J)/rsc_fpops_est(J) 25 // over completed jobs J currently in the DB 26 27 require_once("../inc/util.inc"); 28 29 // take a host.serialnum field (which may encode several GPUs) 30 // and extract the model name for the given vendor 31 // 32 function get_gpu_model($x, $vendor) { 33 $descs = explode("]", $x); 34 array_pop($descs); 35 foreach ($descs as $desc) { 36 $desc = trim($desc, "["); 37 $d = explode("|", $desc); 38 if ($d[0] == "BOINC") continue; 39 if ($d[0] != $vendor) continue; 40 return $d[1]; 41 } 42 return null; 43 } 44 45 function add_model($model, $r, $wu, &$models) { 46 if (array_key_exists($model, $models)) { 47 $models[$model]->count++; 48 $models[$model]->time += $r->elapsed_time/$wu->rsc_fpops_est; 49 } else { 50 $x = new StdClass; 51 $x->count = 1; 52 $x->time = $r->elapsed_time/$wu->rsc_fpops_est; 53 $models[$model] = $x; 54 } 55 } 56 57 // return a data structure containing GPU usage info for a vendor 58 // $x->total: combined list 59 // $x->windows 60 // $x->linux 61 // $x->mac 62 // 63 function get_gpu_list($vendor, $alt_vendor=null) { 64 $clause = "plan_class like '%$vendor%'"; 65 if ($alt_vendor) { 66 $clause .= " or plan_class like '%$alt_vendor%'"; 67 } 68 $avs = BoincAppVersion::enum($clause); 69 if (count($avs) == 0) { 70 $x = new StdClass; 71 $x->total = array(); 72 return $x; 73 } 74 75 $av_ids = ""; 76 foreach($avs as $av) { 77 $av_ids .= "$av->id, "; 78 } 79 if ($vendor == "cuda") { 80 $av_ids .= "-3"; 81 } else if ($vendor == "ati") { 82 $av_ids .= "-4"; 83 } else if ($vendor == "intel_gpu") { 84 $av_ids .= "-5"; 85 } else { 86 $av_ids .= "0"; 87 } 88 89 $t = time() - 30*86400; 90 //echo "start enum $vendor $av_ids\n"; 91 $results = BoincResult::enum( 92 "app_version_id in ($av_ids) and create_time > $t and elapsed_time>100 limit 500" 93 ); 94 //echo "end enum\n"; 95 $total = array(); 96 $win = array(); 97 $linux = array(); 98 $mac = array(); 99 foreach ($results as $r) { 100 $h = BoincHost::lookup_id($r->hostid); 101 if (!$h) continue; 102 $wu = BoincWorkunit::lookup_id($r->workunitid); 103 if (!$wu) continue; 104 $v = ""; 105 if ($vendor == "cuda") { 106 $v = "CUDA"; 107 } else if ($vendor == "intel_gpu") { 108 $v = "INTEL"; 109 } else { 110 $v = "CAL"; 111 } 112 $model = get_gpu_model($h->serialnum, $v); 113 if (!$model) continue; 114 add_model($model, $r, $wu, $total); 115 if (strstr($h->os_name, "Windows")) { 116 add_model($model, $r, $wu, $win); 117 } 118 if (strstr($h->os_name, "Linux")) { 119 add_model($model, $r, $wu, $linux); 120 } 121 if (strstr($h->os_name, "Darwin")) { 122 add_model($model, $r, $wu, $mac); 123 } 124 125 } 126 $x = new StdClass; 127 $x->total = $total; 128 $x->win = $win; 129 $x->linux = $linux; 130 $x->mac = $mac; 131 return $x; 132 } 133 134 function get_gpu_lists() { 135 $x = new StdClass; 136 $x->cuda = get_gpu_list("cuda", "nvidia"); 137 $x->ati = get_gpu_list("ati"); 138 $x->intel_gpu = get_gpu_list("intel_gpu"); 139 $x->time = time(); 140 return $x; 141 } 142 143 function gpucmp($x1, $x2) { 144 return $x1->avg_time > $x2->avg_time; 145 } 146 147 function show_list($models, $name) { 148 echo "<td><h2>$name</h2>\n"; 149 if (!count($models)) { 150 echo tra("No GPU tasks reported")."</td>\n"; 151 return; 152 } 153 $max_count = 0; 154 foreach ($models as $model=>$x) { 155 if ($x->count > $max_count) $max_count = $x->count; 156 $x->avg_time = $x->time/$x->count; 157 } 158 $min_time = 1e9; 159 foreach ($models as $model=>$x) { 160 if ($x->count < $max_count/10) continue; 161 if ($x->avg_time < $min_time) $min_time = $x->avg_time; 162 } 163 uasort($models, 'gpucmp'); 164 echo "<ol>\n"; 165 foreach ($models as $model=>$x) { 166 if ($x->count < $max_count/10) continue; 167 $s = number_format($min_time/$x->avg_time, 3); 168 echo "<li>($s) $model\n"; 169 } 170 echo "</ol></td>\n"; 171 } 172 173 function show_vendor($vendor, $x) { 174 echo "<h2>$vendor</h2>\n"; 175 if (!count($x->total)) { 176 echo tra("No GPU tasks reported"); 177 return; 178 } 179 $have_win = count($x->win)>0; 180 $have_mac = count($x->mac)>0; 181 $have_linux = count($x->linux)>0; 182 $n = 0; 183 if ($have_win) $n++; 184 if ($have_mac) $n++; 185 if ($have_linux) $n++; 186 $show_total = $n>1; 187 start_table(); 188 echo "<tr>"; 189 if ($show_total) { 190 show_list($x->total, "Total"); 191 } 192 show_list($x->win, "Windows"); 193 show_list($x->linux, "Linux"); 194 show_list($x->mac, "Mac"); 195 echo "</tr></table>\n"; 196 } 197 198 $d = get_cached_data(86400); 199 if ($d) { 200 $data = unserialize($d); 201 } else { 202 $data = get_gpu_lists(); 203 set_cached_data(86400, serialize($data)); 204 } 205 206 page_head(tra("Top GPU models")); 207 echo tra("The following lists show the most productive GPU models on different platforms. Relative speeds, measured by average elapsed time of tasks, are shown in parentheses."); 208 show_vendor("NVIDIA", $data->cuda); 209 show_vendor("ATI/AMD", $data->ati); 210 show_vendor("Intel", $data->intel_gpu); 211 echo "<p>Generated ".time_str($data->time); 212 page_tail(); 213 214 ?> Claggy |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
They Are Hosted on a SETI Server. Why you think it should be posted on a BOINC page is a Mystery. Because the operational code is written and maintained by BOINC: https://github.com/BOINC/boinc/blob/master/html/user/gpu_list.php Only the 'skinning' - style sheet carrying logo and navigation bars - is provided by SETI. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
Keep arguing Semantics. We are watching you dance around an Obvious problem with the SETI Stats on a SETI webpage from a SETI Server. Obviously if you don't return results you will Not receive a GF reading. Likewise, the more results the higher that GF reading will be on that page. It's all very interesting... |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
Seems the GPU models page is Still Broken, Still showing; My feeling is that while there may be more compute power produced with the mac apps, If the elapsed times of the devices producing that compute power fall out side the time limits then the results won't be shown in the gpu_list (As HAL9000 already hinted): 91 $results = BoincResult::enum( 92 "app_version_id in ($av_ids) and create_time > $t and elapsed_time>100 limit 500" 93 ); Message 1736677 Claggy |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
I think I see the bug ...
On the application page it says MAC, not Darwin. So it's probably not counting them. |
©2024 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.