| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project Forum
Thread: CEP is experiencing a huge PF Delta rate |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 27
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Benchmarks have nothing at all to do with a particular task on hand. It's the efficiency of the task to work according the benchmark expectation as stored in the whetstone/dhrystone values which are updated every 5th day. I must have just imagined that then. :-) If you get a poor benchmark the "Expectation" will last for those 5 days until it runs again. Try running 4 HPF2 jobs together Sek and during that run the benchmarks, then do the same with 4 HCC and you will see what I mean. Now RICE is unique. It's runs each job a internal benchmark of 5 minutes through a reference mini wu. Most devices seem to do very good on this, at least I always seem to get 5 to 10% more than claimed. Yeah I know. I dont like it. I want that one. :-)  [Edit 1 times, last edit by Former Member at Dec 16, 2008 1:56:44 PM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
That's a miraculous event. My client suspends the tasks prior to benchmarking. Who'd have thought 4 suspended HPF2 would impact a benchmark differently from 4 HCC. Something to revalidate on my device.
----------------------------------------One possible variable: If LiM is not used, the jobs unload during benchmark. 4 big HPF2 v 4 small HCC could have disks whirring effects, such as Swap file resizing/cleaning to go with that. I run with LiM always, also, but mostly because I don't want to loose 4x the time from last checkpoint to present. Sadly so. [added: and then, I have the very secret secret little tool that lassos BOINC.exe and throws it 1 priority level above normal which causes the bench to be next to rock solid. This thread after all might still end up being a little educational ;>)]
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Dec 16, 2008 2:29:00 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I too have been experiencing huge PF Delta rates when running this project. However, I'm not sure how significant this is. Here's a description of my computer:
Intel Core 2 Quad Q9450 Vista Business SP1 4 GB DRAM Configured to run 4 cores at 100% No overclocking Yesterday once I was running three CEP work units and one HCC. When I looked at the Task Manager graphs with "Show Kernel Times" enabled, each of four graphs shows a slightly wiggly red line (kernel time) hugging the bottom of the graph. There was no difference between any of the four work units. Apparently less than 10% of the CPU time is being spent in the kernel. This is typical of what I generally see on this machine. Maybe this small inefficiency could be improved, but it doesn't look anything like what we saw with HCC project before that was fixed. Of course, this is just my experience. Steve |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Its not as simple as emptying the bucket and dumping to VM while you use the bucket. You know that Sek.
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Well, then share how you measured it. Now there's nothing that allows me to replicate this big benchmark(?) variation, Whl, OS, RAM, CPU, BOINC parms, BOINC version, stock, home compile. Never observed it, first time heard of it.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1408 Status: Offline Project Badges: 
|
Back to threads topic:
E000027_ 324A_ 00035r00u_ 0-- In Progress 14-12-08 00:29:13 21-12-08 00:29:13 0.00 0.0 / 0.0 E000027_ 324A_ 00035r00u_ 1-- Pending Validation 14-12-08 00:25:05 16-12-08 20:10:22 45.35 637.1 / 0.0 Waiting and hoping. But whilst 0-- not returned yet a good change for a long runtime too. Some specs of this WU: Vista 64bit, Quad, 2.1GHz; 1/4 of the time running with another CEP on another core. Uploaded 5 files: 1450kB Peak WS 90,148; VM 357,712; Pagefaults 1,404,000,000 this is 8600 PF's/sec. |
||
|
|
jay_Orlando
Senior Cruncher USA Joined: Jan 4, 2006 Post Count: 189 Status: Offline Project Badges: 
|
Hi,
----------------------------------------Thank you all for the discussions and explanations of page faults. (especially hard and soft faults) I was concerned with over 3,000 page faults in a two second period - mostly continually. BUT - the disk activity light was not lit. (Running XP with 2GB of memory with a an Intel duo) The Windows task manager said I had over 1 GB of physical memory available. The explanations of the o/s handling multiple mem requests fits what I saw. Thanks again,  Jay E.  |
||
|
|
|