Message boards : Number crunching : Rosetta 4.1+ and 4.2+
Previous · 1 . . . 20 · 21 · 22 · 23 · 24 · 25 · 26 . . . 34 · Next
Author | Message |
---|---|
Brian Nixon Send message Joined: 12 Apr 20 Posts: 293 Credit: 8,432,366 RAC: 0 |
|
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
Immersion cooling That's the one. I'm surprised it's not done more often by gamers etc. Perhaps the hassle of changing a component? You'd have to clean the oil off the contacts in the PCI express socket for a start. |
Brian Nixon Send message Joined: 12 Apr 20 Posts: 293 Credit: 8,432,366 RAC: 0 |
From what I can gather, most setups still need a circulating pump and heat exchanger. Convection through static oil is probably not sufficient to remove heat from a CPU quickly enough. So a rig is going to be heavy, and cumbersome, and messy when it all goes wrong. And the oil itself is expensive. (I’ve seen $1000 a gallon mentioned for top-tier pro-grade stuff. I dare say there are alternatives, but I don‘t know how suitable used chip fat is for the job…) |
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
From what I can gather, most setups still need a circulating pump and heat exchanger. Convection through static oil is probably not sufficient to remove heat from a CPU quickly enough. So a rig is going to be heavy, and cumbersome, and messy when it all goes wrong. And the oil itself is expensive. (I’ve seen $1000 a gallon mentioned for top-tier pro-grade stuff. I dare say there are alternatives, but I don‘t know how suitable used chip fat is for the job…) If you can put used chip fat in a car engine, it should work for a computer :-) You could always use new chipfat! As for going wrong, I've got a motherboard wet and made a power supply emit fireworks when using watercooling. I'm sticking to fans. |
mikey Send message Joined: 5 Jan 06 Posts: 1895 Credit: 9,168,899 RAC: 3,916 |
From what I can gather, most setups still need a circulating pump and heat exchanger. Convection through static oil is probably not sufficient to remove heat from a CPU quickly enough. So a rig is going to be heavy, and cumbersome, and messy when it all goes wrong. And the oil itself is expensive. (I’ve seen $1000 a gallon mentioned for top-tier pro-grade stuff. I dare say there are alternatives, but I don‘t know how suitable used chip fat is for the job…) Immersion is the key to that...sprinkling oil on a MB will also cause sparks, but being immersed there is not enough heat or spark generated to cause them to spark. I've also heard of people putting mb's in freezers but the problem is the moisture they retain. Peter where you live a simple radiator hung outside a shaded window will keep things fairly cool if you went liquid cooling but again it's the leaks that cause problems. A closed loop system will help with that but then there's no radiator outside the window to keep things cool. |
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
Immersion is the key to that...sprinkling oil on a MB will also cause sparks, but being immersed there is not enough heat or spark generated to cause them to spark. I've also heard of people putting mb's in freezers but the problem is the moisture they retain. How can oil cause sparks? It's not electrically conductive. I don't like water cooling because of the leaks. And I did have a closed loop system, the pump wore out. Fan fails - replace, easy. Pump gone - change the whole expensive contraption. And it still had fans to cool the radiator anyway! |
Sid Celery Send message Joined: 11 Feb 08 Posts: 2124 Credit: 41,224,965 RAC: 11,021 |
I don't like water cooling because of the leaks. And I did have a closed-loop system, the pump wore out. Fan fails - replace, easy. Pump gone - change the whole expensive contraption. And it still had fans to cool the radiator anyway! I don't know if I'm just lucky, but I've used 3 AIO water coolers in 2 PCs over the last several years and none have failed or leaked - pump or fans. I know it sounds a terrible idea but once I saw my first one I was convinced for ever more - and my faith hasn't been in vain |
Keith Myers Send message Joined: 29 Mar 20 Posts: 97 Credit: 332,619 RAC: 298 |
I'm a firm convert to water cooling. Never had any one leak. All my cpus are water cooled and most of my graphics cards with hybrid designs. Have had several hybrid AIO pumps fail at usually 18-24 months of operation. Always under warranty so no issues getting them replaced quickly. I have older air cooled cards on the shelf I can substitute while awaiting the hybrid replacement. |
[VENETO] boboviz Send message Joined: 1 Dec 05 Posts: 1994 Credit: 9,620,547 RAC: 9,436 |
This thread is about issues/bugs/other things on 4.2 app version. It's not a thread about hw configuration. Please, on topic |
Tomcat雄猫 Send message Joined: 20 Dec 14 Posts: 180 Credit: 5,386,173 RAC: 0 |
Suffered a second-long power outage (I probably should get a UPS due to how frequent these have become) and got this. rb_08_19_34924_34163_ab_t000__h002_robetta_IGNORE_THE_REST_06_15_1007200_6_1 <core_client_version>7.16.7</core_client_version> <![CDATA[ <stderr_txt> command: projects/boinc.bakerlab.org_rosetta/rosetta_4.20_windows_x86_64.exe @rb_08_19_34924_34163_ab_t000__h002_robetta_FLAGS -in::file::fasta t000__h002.fasta -in:file:boinc_wu_zip rb_08_19_34924_34163_ab_t000__h002_robetta.zip -frag3 rb_08_19_34924_34163_ab_t000__h002_robetta.200.3mers.index.gz -fragA rb_08_19_34924_34163_ab_t000__h002_robetta.200.15mers.index.gz -fragB rb_08_19_34924_34163_ab_t000__h002_robetta.200.6mers.index.gz -nstruct 10000 -cpu_run_time 28800 -boinc:max_nstruct 20000 -checkpoint_interval 120 -mute all -database minirosetta_database -in::file::zip minirosetta_database.zip -boinc::watchdog -boinc::cpu_run_timeout 36000 -run::rng mt19937 -constant_seed -jran 1829116 Using database: database_357d5d93529_n_methylminirosetta_database ERROR: Fullatom mismatch in checkpointer. ERROR:: Exit from: ......srcprotocolscheckpointCheckPointer.cc line: 380 06:54:58 (11364): called boinc_finish(0) </stderr_txt> ]]> It validated as valid instead of error? Is that normal? I also got the same amount of credits for this task as my wing-man despite having half their run-time, somehow. Now we are back on topic, I hope. Let's not steer the topic towards UPSes. |
Brian Nixon Send message Joined: 12 Apr 20 Posts: 293 Credit: 8,432,366 RAC: 0 |
It validated as valid instead of errorIf I had to guess: the task failed to resume from a checkpoint after the power failure, but was able to report the valid results it had calculated up to that point. I also got the same amount of credits for this task as my wing-manDespite causing disparities on projects that use fixed duration / variable work tasks, that is how BOINC works. Rosetta tasks are generally not sent to more than one computer. You only got that one because the other machine had failed to return it by its deadline. |
James W Send message Joined: 25 Nov 12 Posts: 130 Credit: 1,766,254 RAC: 0 |
Device: 1759960 Task: 1245719883. WU: 1115277497 Puzzling credit amount. I was assigned this task after the original host did not complete by deadline. The original host and I both rec'd 26.22 credits. My host worked almost 8 hrs (CPU time) whereas original host only worked less than 1 hr., with same results. Unsure of logic of this. Computer ID 5135728And my Stderr output, etc.: Computer ID 1759960 |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1680 Credit: 17,848,992 RAC: 22,990 |
I was assigned this task after the original host did not complete by deadline. The original host and I both rec'd 26.22 credits. My host worked almost 8 hrs (CPU time) whereas original host only worked less than 1 hr., with same results. Unsure of logic of this.My WAG (Wild Arse Guess)- It did return it's result before you did, and Credit for work done hasn't yet settled down on that particular host, and it may be set for a 2 hour Target CP U time, so in this case it only ran for 1 hour and then ended as it would have gone past it's Target CPU time. So the Credit for the Tasks was set by that host. However (to my way of thinking), while still low, you should at least have got 209.76 (8 * 26.22) and not the same 26.22 it got. Grant Darwin NT |
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
It validated as valid instead of error? Is that normal? I also got the same amount of credits for this task as my wing-man despite having half their run-time, somehow. God forbid anyone mentions anything that is required to run the computer better. [Raises right arm to 30 degrees] |
Bryn Mawr Send message Joined: 26 Dec 18 Posts: 393 Credit: 12,110,248 RAC: 6,015 |
It validated as valid instead of error? Is that normal? I also got the same amount of credits for this task as my wing-man despite having half their run-time, somehow. Out of order and totally unnecessary :-( |
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
It validated as valid instead of error? Is that normal? I also got the same amount of credits for this task as my wing-man despite having half their run-time, somehow. How is a UPS irrelevant to running Boinc 24/7? |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1680 Credit: 17,848,992 RAC: 22,990 |
How is that relevant to the thread? Just a reminder- Rosetta 4.1+ and 4.2+How is a UPS irrelevant to running Boinc 24/7?God forbid anyone mentions anything that is required to run the computer better. [Raises right arm to 30 degrees]Out of order and totally unnecessary :-( Keeping things on topic makes life easier for everyone. Grant Darwin NT |
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
How is a UPS irrelevant to running Boinc 24/7?How is that relevant to the thread? Just a reminder- Rosetta 4.1+ and 4.2+ How is it easier to have to constantly jump around to different threads just because we dare to go off on a tangent? The art of conversation..... And the tangents are usually far more interesting than people moaning they didn't get precisely the right amount of credits, or that 1 in a thousand tasks fail. There are more important things in life.... |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1680 Credit: 17,848,992 RAC: 22,990 |
How is it easier to have to constantly jump around to different threads just because we dare to go off on a tangent? The art of conversation.....Then use a conversation/chat thread if that's what you want to do. And the tangents are usually far more interesting than people moaning they didn't get precisely the right amount of credits, or that 1 in a thousand tasks fail. There are more important things in life....Then don't bother with those threads if they don't interest you. No need to trash them just to make them appealing to you. That way other people can find the information they need, without having to wade through heaps of noise to find the help that they are looking for, without having to start another thread because the noise has swamped out the useful responses where the question has been asked and answered previously. If you like noise, go to the Cafe and start your own thread there. Grant Darwin NT |
Mr P Hucker Send message Joined: 12 Aug 06 Posts: 1600 Credit: 11,839,463 RAC: 13,224 |
Then use a conversation/chat thread if that's what you want to do. But it's not, I was following this thread and something relevant to it came up in conversation. No point in starting a new thread for that, since everyone's here anyway. And it's not just me or there wouldn't be a conversation. Then don't bother with those threads if they don't interest you. No need to trash them just to make them appealing to you. I'm not trashing anything, the posts about the main topic are still here. You don't have to read mine. That way other people can find the information they need, without having to wade through heaps of noise to find the help that they are looking for, without having to start another thread because the noise has swamped out the useful responses where the question has been asked and answered previously. Once a thread goes off topic, the original question has most likely been answered anyway. If you like noise, go to the Cafe and start your own thread there. I like more relevant noise :-) |
Message boards :
Number crunching :
Rosetta 4.1+ and 4.2+
©2024 University of Washington
https://www.bakerlab.org