With the recent server problems I upped the cache limit from 1+1 day to 4+1, this made no difference to how many WUs were downloaded! (approx. 52)

I've tried removing then re-adding A@H to BOINC, detaching A@H, un/re-installing BOINC, no change.

My old quad core Core 2 rig crunches WUs in ~39mins (0.65 hr), all 4 cores are dedicated to A@H (temps allowing). So 52 WUs would last about 20 hrs.
Running Win 7 64bit, BOINC v7.0.64(x64).

Any ideas?

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

One of BOINC's primary goals is to not download more work than your host can process and return before the deadline. Sometimes it is overly cautious. BOINC has some ability to learn about your computing habits and the time required to process tasks so that it can do a better job of achieving that primary goal but it takes some time to learn. If you reinstall BOINC and detach/reattach then the data BOINC collects in its learning process is lost and it it has to start learning all over again.

Another reason it's not ramping up your cache is that the project has been updating their apps and issuing new versions of them. When BOINC sees a new app version it exercises caution regarding task completion times.

Another reason could be that the project's estimates of how many flops are required to complete a task have been revised upward which again causes BOINC to exercise caution.

The best advice is to stop re-attaching projects and re-installing BOINC. Just patiently wait for BOINC to do its job and to adapt to changes on the server side. It's far from perfect and sometimes it takes a while for BOINC to learn what it needs to learn. Also, I doubt there is going to be another prolonged server shutdown so I would advise against a 4 day cache if you have 24/7 connectivity to the 'net.

Some volunteers become over-zealous in their attempt to support their favorite projects and they make the mistake of trying to ensure they will always have tasks from their fav project by setting a large cache. Or maybe they have a need to be perceived as "hard core". While their cause may be noble (or stupid, depending on ones outlook), a large cache can cause problems. Usually it's better to set a small cache and have one or more backup projects to keep your host busy if your fav project goes offline. I set my cache to .1 and 0 and it works great. I recommend no higher than .5 for the first figure as it's a good idea to have your host contact the server frequently to check see if any tasks in your cache have been cancelled by the server.

Re paragraph 1, point taken :)
Btw my rig runs 24/7 so WUs get nowhere near the deadline.

Re para 2, other peoples cache (in my team) are not affected to the same level, but maybe it's because of the un/re-install issues you mentioned.

Re para. 4, none of us know how often or for how long project servers will go down for, having only a .1 day cache would be woefully inadequate if you want to run just 1 project, you would at the mercy of server outages or your net going down.
I've been doing DC for nearly 14yrs now (in 2wks time :)) & server outages of 1-2 days are relatively common. I remember in the pre BOINC days SETI was down for about a week thanks to some workmen cutting through the Uni's main pipe!
My SETIQ just barely had enough WUs for that one! ;).

I agree that some people are over zealous on cache limits but 3-4 days is not over zealous but prudent. Also I currently don't want to run any other project on my main rig so I need enough WUs to keep it going through the occasional server outages. With the recent struggles A@H has had in dealing out WUs I've lowered the cache to 3 days, not that it matters as BOINC is just completely ignoring it! Currently it only has 14 WUs, about 5 1/2hrs worth :/

Rather than waiting for BOINC to learn the right cache limit (& leave me short of WUs through the recent server outages!) can't I just over-ride it with a config file edit?

[update] I can't believe I missed this, I've just found in the network usage tab the cache limits! :o, previously I had been altering the BOINC cache limits through the online preferences (to no effect), which makes me wonder, what's the point in it if where the client takes priority?

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

Well, see, thinking that this project is the only project worthy of your donation is what I call over zealous but to each his own.

If I miss 4 days of production out of 365 I don't look on it as a major calamity. Life is much too short to worry about potatoes that small. For me there are more important things. For example, earlier this year the admins made a typo when setting the number our hosts use to estimate task duration. As a result task duration estimates were 10 times smaller than they should have been so hosts downloaded 10 times more tasks than they should have. Volunteers with large caches downloaded way more tasks than they could return by deadline and ended up crunching days worth of tasks that missed the deadline and were rejected. That's says of lost production. My hosts also downloaded 10 times more tasks than they should have but since my cache was minimal I was able to complete all tasks well within deadline. The point is there is more than one way to set yourself up for a loss of production.

Of course you can argue that the volunteers who lost 5 days production could have avoided that loss by checking their hosts daily and deleting all those extra tasks but for some of us that's a major nuisance. We have better things to do. The bottom line is everyone is allowed to define their priorities their way. Nobody gets exactly what they want all the time, we all make trade-offs, pick your favorite poison and enjoy it. Just remember if your largish cache bites you in the ass one day and you whine about it I'll be the first to ROFLMAO and troll on ya with every variation of "told ya so" I can think of :-)

Rather than waiting for BOINC to learn the right cache limit (& leave me short of WUs through the recent server outages!) can't I just over-ride it with a config file edit?

If you mean cc_config.xml then the answer is no. You could edit the project's DCF and/or fpops estimate to effect an override (if those numbers are in fact the problem) but those are in client_state.xml.

[update] I can't believe I missed this, I've just found in the network usage tab the cache limits! :o, previously I had been altering the BOINC cache limits through the online preferences (to no effect), which makes me wonder, what's the point in it if where the client takes priority?

Doh! Hey don't feel bad. If I had been on the ball I would have mentioned that scenario as a possible cause for your problem. The thing is I've been using the hostside prefs for so long I'd forgotten the website prefs even exist until you reminded me.

The website settings take priority by default. The settings in BOINC manager don't take priority unless you open the manager's "prefs edit window" and click OK at the bottom of the window. Obviously you did just that at one time or another. If you want to return to having the website prefs take priority then click the Clear button near the top of the manager's "prefs edit window".

Well, see, thinking that this project is the only project worthy of your donation is what I call over zealous but to each his own.

As you can see from my sig I've run a few other projects ;), but as far as each rig goes I like to stick to one project at a time (unless the GPU is crunching too, when temps allow I'll run MW@H alongside other projects). My 2nd rig still runs DPAD as does a friends rig, my father crunches SETI for me, so I'm not solely crunching 1 project when you consider all the PCs I crunch on.

If I miss 4 days of production out of 365 I don't look on it as a major calamity. Life is much too short to worry about potatoes that small.

I don't worry about it either, I just set the appropriate cache, job done! :P ;)

As for the typo error you mention, that is a highly unusual (I don't recall hearing that 1 before, although I'm sure it has on the rare occasion), were you to run just 1 project then you would be far more likely to come unstuck from a server or net outage than a typo that screws up crunching estimates. Your backup is a 2nd (or more) project, you're very unlikely to have 2 projects go down at the same time ;).
10x too many WUs is rather a massive amount OTT! lol, ouch! I'd look out for BOINC notices regarding that (I usually quickly check my client daily anyway so it's no issue & deleting dead WUs only takes a minute), but if that didn't work I'd just live with it, like I said we are far more likely to be hit by a server outage than that kind of error.

But yes your right it is a trade off, but my modest size cache (3 days) is by far a better trade off than virtually no cache when running 1 project.

No worries on the config edit, I don't need to now anyway since I sorted the cache levels :), BOINC very quickly got the right runtimes for my rig, well within a day.

The website settings take priority by default. The settings in BOINC manager don't take priority unless you open the manager's "prefs edit window" and click OK at the bottom of the window. Obviously you did just that at one time or another.

Ahhh now it makes sense! thank you :), I couldn't understand how I hadn't altered the clients cache levels yet it had been over-riding the web preferences, yes I had hit ok after altering options in processor usage.

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

As you can see from my sig I've run a few other projects ;), but as far as each rig goes I like to stick to one project at a time (unless the GPU is crunching too, when temps allow I'll run MW@H alongside other projects). My 2nd rig still runs DPAD as does a friends rig, my father crunches SETI for me, so I'm not solely crunching 1 project when you consider all the PCs I crunch on.

I recognize all the acronyms in your sig except DPAD. Googling it returns game related references so I still don't understand what it is if it's a DC thing. Please enlighten me as I like to keep up with DC stuff though I suppose I should not assume it is not DC.

If DPAD is a DC project then I can see why you don't want to run it alongside BOINC on the same machine. And I can see your temperature rationale with Milkyway. What I don't understand is why only SETI on one machine and only Asteroids on another (or have I missed something?). Not to mock or criticise, but what I would do is put SETI and Asteroids on both machines and if I wanted to run primarily SETI on machine A I would give SETI a resource share > 0 and Asteroids a share = 0 which causes the host to request only SETI tasks unless it runs out of SETI (ie SETI goes down) at which time it would request Asteroids. Thus Asteroids would be solely a backup project on that host. I would setup shares on host B exactly opposite to A with Asteroids share > 0 and SETI share = 0 to make SETI solely a backup project and Asteroids the primary project.

10x too many WUs is rather a massive amount OTT! lol, ouch! I'd look out for BOINC notices regarding that (I usually quickly check my client daily anyway so it's no issue & deleting dead WUs only takes a minute), but if that didn't work I'd just live with it, like I said we are far more likely to be hit by a server outage than that kind of error.

I've seen the duration estimate error happen a few times but you are right, the more frequent problem is that project servers go down. As for one's ISP going down, I think I see that differently than other users. I am blessed with what seems to be a top notch, ultra reliable ISP named Shaw Communications. In 10 years they've been down for more than 5 minutes on only 1 occasion when some silly bugger cut the cable with his trenching machine which of course is not Shaw's fault. Shaw's principle competitor in this area is Telus and their service is quite the opposite, somewhere between pathetic and disgusting.

But yes your right it is a trade off, but my modest size cache (3 days) is by far a better trade off than virtually no cache when running 1 project.

I agree, modest is the appropriate adjective and you'll probably not run into trouble with a 3 day cache.

I think DPAD possibly refers to Distributed Particle Accelerator Design now more commonly known as Muon.

There is original Stephen Brook Muon and a yoyo@home BOINC wrapper Muon.

Re DPAD, in Google it's the 7th result (not including image links), Kashi is right it's also known as Muon1.
Here's an exert from Muon1DPADs site:-
The experiment is called the Neutrino Factory, scheduled for construction some time from 2015. Its primary aim is to fire beams of neutrinos (fundamental particles) through the Earth's interior to detector stations on different continents. They're doing this to measure whether they change type en route (there are 3 types of neutrino) and data from this in turn will allow them to determine the neutrino's mass more accurately

More info here http://stephenbrooks.org/muon1/info/

I ran it on my main rig exclusively for a few years after SETI classic was shut off, followed by F@H until I started to run A@H recently.
I'm still interested in physics so I will probably run it again at some point (although I vaguely recall being a bit miffed about Stephens sparse feed back at the time) or maybe I'll run LHC instead.
Anyway it's a good project with a pretty reliable client & no issue with WUs serve outages as the client generates it's own WUs with the servers dishing out 'seed' files.

A@H is my main focus now, hence giving it 100% of my quad cores time. That & I'm racing a team mate ;).
SETI & DPAD remain on friends & family rigs (& my 2nd rig) to maintain a little input there. My 2nd rig is rarely on btw, generally sunday afternoons, as mentioned it runs DPAD & so no issues with WU availability.

Re ISPs, wow your 1 is reliable! :)
I'm with Virgin (was NTL), their pretty reliable nowadays but nowhere near as good as yours! I had an outage last week which lasted most of a day, I ran F@H alongside A@H as I knew A@H would run out of WUs (that was before I sorted the cache issue). When my ISP was NTL is was less reliable.

Now with A@H servers running well, currently!, my cache sorted out & ISP running ok the only problem I have now is that A@H is so well optimised is that it's running the CPU a good 4-5C hotter than F@H & running it close to it's limit! Currently its at 69C, I set the alarm at 70C at which point I reduce BOINC to 3 cores which is frustrating regarding the 100% time I want to give A@H! (ambient temp's 24C atm).
My Q6600 is o/ced to 3.24 GHz so I may have to look at dropping its speed & voltages again but it's a right rigmarole to stability test it all over again! Or I'll have to improve cooling.

What do you run A@H on?

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

Kashi, thanks for your response and info. My apologies for the late reply.

I am very interested in physics related projects too which is why I crunch LHC (renamed to Sixtrack) and it's sister project Test4Theory (T4T). T4T does very interesting work for CERN's LHC by crunching theoretical models of the Higgs boson. If you're interested in neutrinos and know that much about them you probably already know about the Higgs boson and might be interested in helping T4T.

I am definitely going to hook up with Muon-DPAD. I think inside the neutrino there are some very important "secrets" that we need to unravel if we want to keep physics and technology moving forward.

I am running A@H on some old junk I picked up a few months back: a slow laptop and an old C2D desktop. Click on my name at the left for the details. The desktop runs on Linux, the lappy on Win 8 which is scheduled to be replaced with Linux soon. I run about 6 projects on each machine,not too many because it takes too much time to keep track of them all.

I had a couple of very nice rigs (an i7 2600K with GTX 570 GPU and an i3 with AMD 7990 GPU) but I gave them away to my friend's kids this spring. They came for a visit and were so fascinated with BOINC and especially crunching on GPUs I decided the right thing to do was give them those rigs. They're good kids and I want them to stay that way. My hope is that my gifts to them will keep them interested in science and provide tangible proof to them that they can play an important role in our society. I hope that gives them a sense of belonging they hold onto rather than grabbing hold of drugs, alcohol and so many other self-destructive things kids turn to when they lose their sense of belonging and have no handles to hold on to. For their birthdays I gave them some old parallel port expansion cards I had laying around and am showing them how to program the port and control simple external devices connected to the port (eg. 8 LEDs wired up on some perf board). The look of awe on their faces when they first saw the LEDs flashing on and off under the control of a program we wrote is the kind of reward I live for. Now, the minute they get their homework done they're building their own stuff and coding their own programs, with guidance of course. They fried one parallel port but that's all part of the learning process. One day, when they're interested, I'll show them how to locate the fried component on the board, de-solder it and replace it.

I'm moving away from CPU computing and migrating toward GPU. For me it provides the best bang for my hardware and electrical investment. I want to get into it in a big, inexpensive, green way. I don't want to just put a bunch of computers with GPUs in a room and have at it. I already know that causes heat/cooling problems and you soon run into a wall. I'm taking my time, doing the research and engineering a scalable solution that works well rather than just good enough.

For example you mentioned you have a bit of a temperature problem and the ambient is 26C. As you know you can buy a better "cooling solution" off the shelf and bolt it on but that doesn't provide a scalable solution IMHO. Of course you might not intend to scale up but if you do then putting better (say more expensive) fans and heat sinks on more and more CPUs isn't the best answer because all those "solutions" do is dump the heat into the room which raises ambient temp which makes it harder for your expensive "solutions" to do their job. Eventually you hit a wall most people can't get over/around except by adding refrigeration type air conditioning to the room. That's an investment/expense I would rather spend on more crunching power so for me refrigeration style air conditioning isn't acceptable. I've found a much cheaper solution that works just as good is to vent the hot air to the outdoors rather than into the room. There's lots of inexpensive but very effective ways to do that.

Yea I know a little about the Higgs, mainly from New Scientist & from the news. Cool that they've finally proved it :).

When temps cool down a bit I think I'll have to fire up my 2nd rig more to crunch some DPAD, it's been quite a while since I've had any serious output there.

I hope you got those kids crunching for you with those nice rigs you generously gave away ;)

I'm moving away from CPU computing and migrating toward GPU. For me it provides the best bang for my hardware and electrical investment. I want to get into it in a big, inexpensive, green way
I think the only really green way to do that is to generate your own renewable electricity (unless you happen to live on a farm or something else that produces waste you could use as energy).

Re cooling on my quad, at the time the cooler I got for it was considered one of the better air coolers you could get (Noctua NH U12, I think!), I suppose I could get the Thermalright Ultra-120 eXtreme (found some on ebay for £36-38), otherwise I think the next step would be water cooling! Which I don't want to get into, & would be a waste for a rather old rig. Not sure it'd be worth getting the TR U120E for that matter!

I've found a much cheaper solution that works just as good is to vent the hot air to the outdoors rather than into the room. There's lots of inexpensive but very effective ways to do that.

I've often thought about doing that but in my room it's just not practical, I had even wondered about ducting heat into the then unused chimney stack before the landlady installed a wood burner ;).
Hmm taking the side off my PC has lowered temps by 4C, maybe theirs something I can do about that?.....
So how do you duct heat out then?

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

Those rigs crunch for science. The kids can have the credits and badges, I don't need them.

I'm not surprised removing the side reduces the temperature. The average case isn't designed to provide the air flow required for crunching. The first thing many crunchers do when they get a new computer is take the side off. It's standard procedure here. If you have curious pets/kids and don't want to leave the side off you can buy window screen fabric and cover the side with it. It's an inexpensive fix that you can do with a pair of scissors to cut the screen and a needle and thread to sew the screen on.

One of the reasons the average case has air flow problems is the manufacdturers put cheap vents in them. A vent that consists of a bunch of small holes punched/drilled through the metal hinders airflow considerably compared to a grill type vent. If you have the tools you can cut the punched/drilled vent out to produce one big hole then cover that with a grill type vent. If you don't have the tools then consider using window screen instead. Or position the case so kids/pets can't get into the open side.

After trying water cooling in a few rigs and I can't recommend it to anybody unless they're doing extreme overclocking and need the ultimate cooling system. Water cooling comes at a price, however. Hoses break, fittings leak, pumps cease and burn out. I probably wouldn't spend any money on a rig as old as yours unless it's for parts that can be salvaged later and put on a new rig eg. a new disk, more RAM, a DVD burner. Coolers aren't a standard fit so they usually won't fit onto your next rig. IMHO putting a new cooler on an old rig is a bad investment unless you intend to keep the rig for at least another year.

Ones approach to upgrades and mods always depend on disposable income, how much time/labor you can invest and how much you can build and fabricate on your own. Building your own gear requires tools which aren't cheap either but they are a good investment because they enable you to build stuff for less than half the price of solutions you see in stores. And you can create solutions that you can't buy anywhere. If you have a measuring tape, tin snips, pliers, a drill and a few screwdrivers you can build very elaborate ducts from 4 inch round heating duct and adapt them to virtually any fan, intake port, exhaust port or existing duct. You can get 90 degree and 45 degree elbows, tees and vaqrious other components that make it easy. For example you could build a duct that connects to the hot exhaust vent(s) on your computer at one end and to the wood burner's chimney at the other. IIRC you are in the UK where central heating is not widely used so 4" round duct might not be as plentiful and cheap as it is here in Canada. The kind of duct material we use for venting clothes dryers to the outdoors is very useful too.

The chimney the wood burner connects to might not be the only exit available to you. If you have a window you can remove all or some of the glass and replace it with a piece of wood. Cut a hole through the wood and attach your duct over the hole. You can put a fan there too to help suck the air from the computer and blow it out the window. To keep rain from entering the duct put an elbow outside the window and turn it so it points down.

Another option is to put a small mains powered box fan on a window sill with the flow pointing to the outdoors. Place your computer on a stand that is tall enough to position the computer so that the hot air gets sucked into the fan in the window before it has a chance to dissipate into the room. A friend of mine does something similar. He has the fan in the window but the computer is down on the floor. With scissors, glue and duct tape he build a simple square duct out of cardboard that fits crudely around the fan at the top and loosely around the back of the computer. The fan creates enough suction at the bottom of the duct that all the hot air leaving the computer gets pulled into the duct and pushed out the window in spite of all the gaps and leaks in the duct. If he wanted to he could seal the gaps and leaks with duct tape but it works well enough the way it is so why bother. It isn't a work of art but it hides nicely behind the window drape. The room stays much cooler in the summer and his computer no longer overheats.

I have the same problem: BOINC cashe is set to 0.8 day, but server sends to me too much WUs (I have got more than 180 WU - ~5 days, and the transfer was condinued, until I have just disable new tasks manually).
But A@H isn't my main project, BOINC periodically is switched to other projects, which takes 100% CPU, and run them several days. Therefore I am forced to manually adjust the number of jobs to exclude deadline.

Now I use BoincTasks, it can disable/enable new tasks by the script (not good, about in the limits of 100 WUs), but I think it isn't right way - the project must regulate max task number in the client cashe.

boroda3,

I think maybe your problem is caused by a bug in the BOINC scheduler that occurs if your host uses a GPU. I am not sure of all the details but I think the solution most volunteers recommend is to use a very small cache. Try 0.1 instead of 0.8.

I am in the UK, but central heating is very common here ;)
We have ctrl heating & a wood burner.

I'd rather not leave the side off (though I have done in the summer before), my land lady's got curious cats ;), + the rig is in my bedroom so I'd rather avoid adding noise, which unfortunately rules out the screen mesh (good idea though).

Re cases, yea I hear you on that, when I got this case ~6yrs ago I cut open the fan holes, the rear one I was able to fully open up but the front one had the bloody power switch mounted across the front of it!! So I had to leave a strip of metal for the switch to mount on, that goes right across the front middle of the fan! Also the fan mounting screw points are set back on stamped out metal plates(presumably to give clearance for the power switch), which means I can't cut the fan hole right back there either! Grrr, I never would of bought this case if I'd known that :/. I might look again at that front fan hole & see if I can redesign at least the fan mountings. Btw I have plenty of tools as I'm a car mechanic, would a gas axe be over kill? ;) heheh.

Re water cooling, yea reliability is the main reason I don't want to go there.
Whilst looking at the Thermalright 120 cooler I found a recent revision which includes support for S1155, 1156 & 1366 so I could re-use it for a latter rig (sandybridge would be nice :), unless that's been replaced now!).

Wouldn't connecting a duct to the chimney allow smoke back through? Or do you think the air output from the fan will be enough to keep it out? (assuming no power cuts happen to kill the rig! lol). Anyhow the landlady might not allow that kind of mod anyway, she certainly wouldn't allow the window mods, which would be impractical here as I only have 2 large windows.
I'm pretty much snookered here, though I hope to move out within a year so that will resolve that problem anyway,....... err hopefully! lol

So for now I've upgraded the front fan from a 1200 RPM fan to a 2200 RPM fan, which atm I run at 1400 RPM (I can actually make use of the fan controller now which was previously just left at 100% all the time). I can at least crank the fans up when I'm up/out in the hotter months. I'm going to upgrade the rear 1 too.
Strangely with just the frt fan upgraded & cranked up to 100%, whilst the CPU temps do drop a little (~2C) the system & the uppermost HDD temps go up!

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

Btw I have plenty of tools as I'm a car mechanic, would a gas axe be over kill? ;) heheh.

Great! A guy with tools who knows how to use them without poking out his eye. I'm a welder, we'll get along fine :) You can use a gas axe on material that thin but it has to be a small one that is adjusted just so. Even then you must be able to cut very quickly. That takes practise your average mechanic usually doesn't get on the job. I've done it in a pinch but torches that small are very rare even in well equipped welding shops. A plasma torch is the preferred tool for hot cutting very thin material as they are much easier to adjust to a very low heat output. The problem with them is that they must be aligned rather precisely to the material which means you cannot have more than about 8 mm gap between the torch and the material else the machine either won't cut or the cutting will be erratic. That might be a problem inside a case due to corners, stuff sticking out, etc. Or it might not be a problem in your situation. You can find plasma cutters at auto body shops as well as welding shops. You know how it works... buy the man a pint and he'll take care of you.

Re water cooling, yea reliability is the main reason I don't want to go there.

Ahh but you're a mechanic. Unlike most people you know the art of preventive maintenance. You know hoses shrink and clamped connections need to be retightened once in a while and you know to not tighten it until you pinch the hose off. You've seen rotted radiators so you know that if the coolant gets low you probably should not top up the reservoir with water. On the other hand you also know "the fewer moving parts the better".

Whilst looking at the Thermalright 120 cooler I found a recent revision which includes support for S1155, 1156 & 1366 so I could re-use it for a latter rig (sandybridge would be nice :), unless that's been replaced now!).

Sandybridge has been replaced by Ivybridge which was replaced by Haswell. I don't know details regarding socket revisions, you may very well find that cooler will be compatible with revisions for some time. BTW, we're talking revisions within the Core 2 family not major technology shift. Haswell isn't that much more powerful than Sandybridge so don't ignore Sandybridge just because it's 2 revisions behind. Prices on Sandybridge have dropped to help usher in Haswell, stocks are depleting, you might get a better deal if you wait or you might find they are all gone.

Wouldn't connecting a duct to the chimney allow smoke back through? Or do you think the air output from the fan will be enough to keep it out?

The dominant "force" moving exhaust gases in that system would be convection (hot gases are less dense and tend to rise) which will always carry any exhaust upward. Therefore if the duct from the computer comes up from below, the exhaust from the woodburner will never push back down into computer. If the duct slopes upward from the point it connects to existing duct/chimney over to the computer then yes convection would tend to push smoke into the computer. So put the computer well below the point where its duct will tie into existing duct/chimney and keep the computer's duct always sloping upward.

I can think of only one situation where it might not work as I've described. If the woodburner is of the newer type that burns the wood slowly then I think there will be less convection compared to old fireplaces where you have a good flame going. With less convection it might not work properly but I emphasize "might not" so consult an expert before deciding.

I have seen hundreds of natural gas fired water heaters with exhaust ducts that tie directly into the exhaust duct from the furnace and it meets building code. I've asked licensed furnace installers and they say all you have to do is keep everything sloping up and convection will do the rest. Mind you they're talking about a furnace and a hot water heater. A woodburner is not exactly the same as a gas fired furnace and we're talking about dangerous carbon monoxide so get the opinion of a licensed heating expert before you do it. The people who make, sell or install the woodburner would be a good place to start and there are plenty of websites that deal with DIY home improvements.

(assuming no power cuts happen to kill the rig! lol).

With sufficient convection the extra fan is redundant. I would put it in to ensure good airflow when the woodburner is not fired up and you have no convection in the system.

Anyhow the landlady might not allow that kind of mod anyway, she certainly wouldn't allow the window mods, which would be impractical here as I only have 2 large windows.

I didn't give details but the way I would mod the windows would be easily reversible and leave no damage. Just keep the parts you remove and put it all back to original condition when you leave.

I'm pretty much snookered here, though I hope to move out within a year so that will resolve that problem anyway,....... err hopefully! lol

If not then maybe you can use some of my ideas in your new place.

So for now I've upgraded the front fan from a 1200 RPM fan to a 2200 RPM fan, which atm I run at 1400 RPM (I can actually make use of the fan controller now which was previously just left at 100% all the time). I can at least crank the fans up when I'm up/out in the hotter months. I'm going to upgrade the rear 1 too.
Strangely with just the frt fan upgraded & cranked up to 100%, whilst the CPU temps do drop a little (~2C) the system & the uppermost HDD temps go up!

That's a common problem with your approach. You've changed the airflow inside the case in a way that is causing turbulence. Turbulence can cause the air to simply go around in circles inside the case rather than entering and leaving smoothly. Apparently the turbulence has reduced airflow around the upper HDD as well as overall airflow. Apparently only airflow around the CPU has been enhanced. As a mechanic you know more horsepower doesn't always translate into greater speed or acceleration. If you boost the horsepower you might need to change tires and suspension before that additional horsepower actually reaches the ground. This is a similar example. Higher fan RPM (and I know RPM isn't analogous to HP) can actually yield lower CFM, depends on fan blade geometry and other factors. And even if you do install higher CFM it won't do much good if it causes turbulence inside the case. Not saying higher CFM is guaranteed to increase turbulence, it depends on where the fan is located, the geometry of obstructions and free paths inside the case, etc.

Now you see why I maintain that if removing the side of the case doesn't fix the problem then just leave the stuff in the case alone and try to decrease the temperature of the air going into the case. For me the easiest way to do that is to vent the computer's exhaust outdoors and/or pull air from the outdoors into the case (with ducts) if the outdoor temp is cooler than indoor temp. That kind of solution seems daunting at first (at least it was for me) but if you think about it for a while you can find a way to make it work. That's me... take it slow, think, work smart not hard, choose a solution that scales easily.

You might find that upgrading the other fan resolves what seems to be a turbulence problem but don't be surprised if it doesn't. One suggestion (which you are probably already aware of) is to have half of the fan power pulling air in and the other half pushing air out.

... the exhaust from the woodburner will never push back down into computer.

What if something (bird) blocks the chimney at the top?
Will you have then 'smoked' computer ;)

---

- ALF - "Find out what you don't do well ..... then don't do it!" :)

Smoked computer, Lol! Good pun and a good point!

I will never learn to not say never.

Actually, if the chimney is blocked then the firebox will fill with smoke and the fire will extinguish itself. I would have to test to be sure but I think adequate slope on the duct would prevent a huge amount of smoke from reaching the computer.

I was kidding about the gas axe at my (old) work, it's a large beast ;) (I've just changed job). But interesting to know it could be done with much smaller ones. I might be able to find someone with a plasma cutter but I think it would just be easier if I stuck with an air saw or dye grinder.

Re water cooling, yea the fewer moving parts the better ;)

I remember reading about Ivybridge now you mention it, I also seem to recall their was some major limitation on their overclocking abilities (was it down to their process size?) & that Sandybridge was a much better overclocker.
Not heard of Haswell, time to read up again then! ;)

The woodburner is only a few years old, don't know if that makes it a slow burner by default?

With sufficient convection the extra fan is redundant. I would put it in to ensure good airflow when the woodburner is not fired up and you have no convection in the system.

You misunderstand me, I was referring to if their was a power cut & thus my PC shut off their would be no outward airflow, & if the burner happened to be on at the time the smoke then might enter my PC.

Re windows, like I said I only have 2 large windows, it would not be practical to remove even just 1 as I would loose a huge amount of light into the room. Not to mention you can't really dismantle double glazing (& the landlady wouldn't allow it, like I said ;) ).

Re case fans, yea I already have a push/pull setup, the front fan draws air in, the rear exhausts it.
And upgrading the rear fan did indeed lower case & upper HDD temps, though it only provided a ~1C drop in CPU temps! This is doing my head in! Maybe it's because the inlet fan was at 2200 RPM & the exh @ 1800 RPM? Imbalanced air flow? (their both 120 mm fans btw). Although the PSU is exhausting too.

I've just speed them both upto 1800 RPM (default is 1400 RPM), see if that helps.

And understood you on turbulence, I was thinking a similar thing.
Maybe I should take my PC to my old work, slap a transparent panel on its side & borrow their smoke machine (used for finding air leaks)to see what the flow is like!
Do you think vaporised oil smoke would leave a residue inside my PC??

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7

I remember reading about Ivybridge now you mention it, I also seem to recall their was some major limitation on their overclocking abilities (was it down to their process size?) & that Sandybridge was a much better overclocker.
Not heard of Haswell, time to read up again then! ;)

I also recall reading something about Ivybridge not overclocking well but I also have this fuzzy memory about another article that said people who were complaining that they don't overclock well were attempting to overclock the locked models. As with Sandybridge the K models (eg. i7-2600K) are the unlocked models. I don't overclock so I don't know much more than that.

I'm looking at ordering parts for a new system soon and I am convinced that if you want Intel rather than AMD and it's going to be a cruncher rig then the Xeon line is the best choice. They don't have built in graphics processors like the i models therefore they use less power, run cooler and cost less. For example compare this Haswell based Xeon with this Ivybridge based i5 . Both are 4 core but the Xeon has HT so it performs like an i7. The Xeon has bigger level 3 cache than the i5 and also has PCI-E 3.0 whereas the i5 is only PCI-E 2.0. All the new video cards are PCI-E 3.0 and while they may not be hampered much by PCI-E 2.0 the next generation probably will be. Also, the i5 is only USB 2.0 whereas the Xeon is USB 3.0. Note the Xeon is only $10 more than the i5 so you're getting i7 Haswell performance (minus the integrated video) for very close to Ivybridge i5 price. The integrated GPU is useless for crunching so IMHO the Xeon is the way to go. Don't know if it overclocks well but I don't need that anyway.

The woodburner is only a few years old, don't know if that makes it a slow burner by default?

I'm not sure. Google the make and model.

With sufficient convection the extra fan is redundant. I would put it in to ensure good airflow when the woodburner is not fired up and you have no convection in the system.

You misunderstand me, I was referring to if their was a power cut & thus my PC shut off their would be no outward airflow, & if the burner happened to be on at the time the smoke then might enter my PC.

I understood that. I wante4d to say "no, smoke won't enter the computer if there is sufficient convection" though I might have worded it in such a confusing way that it appears I missed your point.

Re windows, like I said I only have 2 large windows, it would not be practical to remove even just 1 as I would loose a huge amount of light into the room. Not to mention you can't really dismantle double glazing (& the landlady wouldn't allow it, like I said ;) ).

Understood about dismantling or modding the window. Can you at least open a window? If you can then you can still push the hot exhaust out. It's not as convenient as modding the window because you might be forced to close the window if it rains. That works here because we get so little rain. I hear you get quite a bit more in many parts of the UK.

Re case fans, yea I already have a push/pull setup, the front fan draws air in, the rear exhausts it.
And upgrading the rear fan did indeed lower case & upper HDD temps, though it only provided a ~1C drop in CPU temps! This is doing my head in! Maybe it's because the inlet fan was at 2200 RPM & the exh @ 1800 RPM? Imbalanced air flow? (their both 120 mm fans btw). Although the PSU is exhausting too.

It's a familiar story, more fans don't always work as well as you think. IIRC you got a better temp drop just by removing the side. You said that won't work because of your landlady's cats and I can see how cats might chew through cheap synthetic screen material or claw it off but if you use aluminum screen and secure it well they won't get through it. My friend says cats can't stand the smell of marigolds so maybe a potted marigold beside your rig?

And understood you on turbulence, I was thinking a similar thing.
Maybe I should take my PC to my old work, slap a transparent panel on its side & borrow their smoke machine (used for finding air leaks)to see what the flow is like!
Do you think vaporised oil smoke would leave a residue inside my PC??

A smoke test would either verify or deny the turbulence theory. I think it would leave a residue that I definitely would not want in a DVD player/burner because it might coat the laser's lens and screw up the beam. If the smoke is not conductive then it should not harm the motherboard. HDDs are sealed pretty good but they are vented in a way that blocks dust. Smoke is finer than dust and might get through the vent. If the smoke is conductive then it might cause shorts between components on the motherboard. If it coated the heat sink it could decrease thermal transfer but I suspect it would wash off easily enough. The conductance is the biggest concern. If it's merely atomised oil then it would be a good insulator and would be OK, however if it's smoke from burned oil then it will have a high carbon content which would make it a very good conductor.

I've seen videos of extremely overclocked rigs submerged in an oil bath for extreme cooling and they run fine. Mind they use mineral oil. IIRC, some high-perf motor oils contain molybdenum which I believe is a conductor so you might be wary of smoke (even if it's just vaporised oil) made from such oil.

I wonder what they use for making smoke for stage shows, rock concerts for example, and how hard it would be to get some of that?

I'm pretty sure they use dry ice aka CO2.

The oil is vaporised not burnt, don't know what kind of oil it is but I might be able to find out.
I hadn't even thought of conductance! I was thinking more about it leaving a sticky residue which the dust would then collect on.

The Xeon sounds like a good choice :), though for me overclocking performance is essential. Here those CPUs are £204 for the Xeon & £175 for the i5, both out of my price range!
Re power cut & smoke, ah ok got ya this time :).

Re window, I can only leave it open when someone's in the house as it's large enough for someone to climb in through & it's just above a low roof. Plus I couldn't get a large bore pipe to it from where the PC is, nor can I move the PC to anywhere else, hence snookered :(.
And yea we get quite a lot of rain ;).

Re leaving the side off, as I mentioned I have done that before, when I go out I close the door (on a wedge) so the cats can't get in there then anyway ;), which just leaves the slightly higher noise & light levels, although it is tolerable.

---

Team AnandTech - SETI@H, Muon1 DPAD, Folding@H, MilkyWay@H, Asteroids@H, LHC@H, POGS, Rosetta@H, Einstein@H,DHPE & CPDN

Main rig - Ryzen 3600, 32GB DDR4 3200, RX 580 8GB, Win10
2nd rig - i7 4930k @4.1 GHz, 16GB DDR3 1866, HD 7870 XT 3GB(DS), Win7