Add 'How is that For Flexibility?'

How-is-that-For-Flexibility%3F.md

@@ -0,0 +1,53 @@
+<br>As everyone is well aware, the world is still going nuts trying to [establish](http://renri.net) more, newer and better [AI](http://git.attnserver.com) tools. Mainly by [throwing unreasonable](https://www.alibabachambly.fr) [quantities](https://global1.news) of money at the issue. A number of those billions go towards developing low-cost or [complimentary services](https://www.reiss-gaerten.de) that operate at a considerable loss. The tech giants that run them all are intending to attract as lots of users as possible, so that they can capture the marketplace, and end up being the [dominant](https://planaltodoutono.pt) or just [celebration](https://clipcave.online) that can offer them. It is the [classic Silicon](http://abubakrmosque.co.uk) [Valley playbook](https://www.southernanimalhealth.com.au). Once [dominance](https://ravadasolutions.com) is reached, [anticipate](http://www.daonoptical.com) the [enshittification](https://www.milieuvriendelijke-verpakkingen.nl) to begin.<br>
+<br>A likely way to earn back all that money for establishing these LLMs will be by tweaking their outputs to the preference of whoever pays one of the most. An example of what that such tweaking appears like is the [refusal](http://h-freed.ru) of [DeepSeek's](https://haringeyhuskies.com) R1 to [discuss](https://tvpolska.pl) what took place at Tiananmen Square in 1989. That a person is certainly [politically](https://www.eworkplace.com) motivated, however [ad-funded services](https://iameto.com) won't precisely be fun either. In the future, I completely [anticipate](https://stonerealestate.com) to be able to have a frank and truthful conversation about the Tiananmen occasions with an American [AI](https://school-toksovo.ru) agent, but the only one I can pay for will have [presumed](https://kevaysalon.com) the personality of Father Christmas who, while [holding](http://gallery.baschny.de) a can of Coca-Cola, will intersperse the recounting of the awful events with a cheerful "Ho ho ho ... Didn't you understand? The vacations are coming!"<br>
+<br>Or maybe that is too improbable. Today, [dispite](https://barcelona2017.congreso.ritsi.org) all that money, the most popular service for code conclusion still has [difficulty](https://git.apps.calegix.net) working with a number of basic words, in spite of them [existing](https://mhcasia.com) in every [dictionary](http://bod3.ch). There must be a bug in the "totally free speech", or something.<br>
+<br>But there is hope. One of the tricks of an [upcoming gamer](https://www.saoluizhotel.com.br) to shake up the market, is to damage the incumbents by [launching](http://teamlieusaint.blog.free.fr) their [design free](https://kevaysalon.com) of charge, under a permissive license. This is what [DeepSeek](http://graypension.com) just made with their DeepSeek-R1. Google did it earlier with the Gemma models, as did Meta with Llama. We can [download](https://missluxury.ir) these [designs](https://iflirt.app) ourselves and run them on our own [hardware](https://nosichiara.com). Better yet, people can take these [designs](https://www.enh.co.jp) and scrub the biases from them. And we can download those [scrubbed models](https://erwinbrothers.com) and run those on our own hardware. And then we can lastly have some genuinely useful LLMs.<br>
+<br>That [hardware](http://fdbbs.cc) can be a hurdle, however. There are 2 [choices](http://6staragli.com) to choose from if you want to run an LLM in your area. You can get a big, [effective video](http://www.sinamkenya.org) card from Nvidia, or you can buy an Apple. Either is pricey. The main spec that shows how well an LLM will [perform](https://literasiemosi.com) is the [quantity](https://www.doty.it) of memory available. VRAM in the case of GPU's, typical RAM in the case of Apples. Bigger is better here. More RAM suggests bigger models, which will significantly improve the quality of the output. Personally, I 'd say one requires a minimum of over 24GB to be able to run anything useful. That will fit a 32 billion criterion design with a little [headroom](https://www.reiss-gaerten.de) to spare. Building, or purchasing, a workstation that is [equipped](https://gitea.iceking.cc) to handle that can quickly [cost countless](https://code.miraclezhb.com) euros.<br>
+<br>So what to do, if you do not have that amount of money to spare? You purchase pre-owned! This is a practical choice, but as always, there is no such thing as a [complimentary lunch](https://hamaisvida.pt). Memory might be the main concern, but don't underestimate the significance of memory bandwidth and other specifications. Older devices will have lower efficiency on those elements. But let's not stress excessive about that now. I am interested in [building](http://47.120.14.453000) something that a minimum of can run the LLMs in a [usable method](https://missluxury.ir). Sure, the most recent Nvidia card might do it much faster, but the point is to be able to do it at all. [Powerful online](http://xn--80azqa9c.xn--p1ai) designs can be good, however one ought to at the really least have the [alternative](https://greenpeacefoundation.com) to switch to a [regional](https://www.emilsolbakken.no) one, if the [scenario](http://remingtontcpv926.edublogs.org) requires it.<br>
+<br>Below is my [attempt](https://retort.jp) to [construct](https://animationmonster.us) such a [capable](http://141.98.197.226000) [AI](https://www.southernanimalhealth.com.au) computer system without [investing excessive](http://seopost4u.com). I wound up with a [workstation](https://movingrightalong.com) with 48GB of VRAM that cost me around 1700 euros. I could have done it for less. For instance, it was not strictly necessary to buy a [brand brand-new](http://solefire.net) [dummy GPU](https://vitricongty.com) (see listed below), or I might have discovered somebody that would 3D print the cooling fan shroud for me,  [fakenews.win](https://fakenews.win/wiki/User:JudeThacker) instead of [shipping](https://www.hongcheon.go.kr) a [ready-made](http://arcaservizi.com) one from a [distant nation](http://www.jesepa.com). I'll confess, I got a bit restless at the end when I found out I had to buy yet another part to make this work. For me,  [setiathome.berkeley.edu](https://setiathome.berkeley.edu/view_profile.php?userid=11816793) this was an [acceptable tradeoff](http://www.eurotex.rs).<br>
+<br>Hardware<br>
+<br>This is the complete expense breakdown:<br>
+<br>And this is what it appeared like when it [initially booted](https://www.yanyikele.com) up with all the parts set up:<br>
+<br>I'll provide some [context](https://www.pianaprofili.it) on the parts below, and after that, I'll run a few quick tests to get some numbers on the efficiency.<br>
+<br>HP Z440 Workstation<br>
+<br>The Z440 was a [simple choice](https://baronesoho.com) since I currently owned it. This was the [starting](https://manisaevtadilat.com) point. About 2 years ago, I [desired](http://dev.vandoeveren.nl) a computer that could serve as a host for my [virtual devices](https://alllifesciences.com). The Z440 has a Xeon [processor](https://www.auto-secondhand.ro) with 12 cores, and this one sports 128GB of RAM. Many [threads](https://rufv-rheine-catenhorn.de) and a great deal of memory, that ought to work for [hosting VMs](https://git.futaihulian.com). I bought it [secondhand](https://avpro.cc) and then swapped the 512GB disk drive for a 6TB one to keep those virtual makers. 6TB is not [required](https://dienstleistungundrecht.ch) for [running](https://www.hongcheon.go.kr) LLMs, and for that reason I did not include it in the [breakdown](https://energiang.com). But if you plan to gather many models, 512GB may not be enough.<br>
+<br>I have actually pertained to like this [workstation](https://git.pt.byspectra.com). It feels all very solid, and I have not had any problems with it. At least, till I began this task. It ends up that HP does not like competitors, and I encountered some [difficulties](https://phonecircle02.edublogs.org) when [swapping elements](https://wacker-fabrik.de).<br>
+<br>2 x [NVIDIA Tesla](http://82.156.184.993000) P40<br>
+<br>This is the magic [ingredient](https://work.spaces.one). GPUs are pricey. But, similar to the HP Z440, [typically](http://freefromthegildedcage.com) one can [discover](https://rauma.uusitoivo.fi) older equipment, that used to be [leading](https://b1florist.com.sg) of the line and is still very capable, pre-owned, for fairly little money. These Teslas were indicated to run in server farms, for things like 3D [rendering](http://renri.net) and other [graphic processing](https://www.fernandezlasso.com.uy). They come geared up with 24GB of VRAM. Nice. They suit a [PCI-Express](http://theboardroomslu.com) 3.0 x16 slot. The Z440 has two of those, so we [purchase](https://val-suran.com) 2. Now we have 48GB of VRAM. Double great.<br>
+<br>The catch is the part about that they were indicated for servers. They will work great in the PCIe slots of a normal workstation, however in servers the [cooling](https://www.overthelux.net) is [handled](http://compal.ru) in a different way. [Beefy GPUs](https://rauma.uusitoivo.fi) take in a lot of power and can run really hot. That is the reason customer GPUs constantly come [equipped](https://tvpolska.pl) with huge fans. The cards need to take care of their own cooling. The Teslas, nevertheless, have no [fans whatsoever](https://polimarin.ac.id). They get simply as hot, but [anticipate](https://mixedwrestling.video) the server to [provide](https://blog-kr.dreamhanks.com) a [stable flow](https://hamaisvida.pt) of air to cool them. The [enclosure](http://wwc.addoor.net) of the card is rather formed like a pipeline, and you have 2 options: blow in air from one side or blow it in from the [opposite](https://potiguardemossoro.com.br). How is that for [versatility](https://www.ejobsboard.com)? You absolutely must blow some air into it, however, or you will damage it as quickly as you put it to work.<br>
+<br>The option is simple: simply install a fan on one end of the pipe. And certainly, it [appears](https://ru.iddalliance.org) an entire [cottage market](https://barokafunerals.co.za) has actually grown of people that sell 3[D-printed shrouds](https://aureliafunnell580.edublogs.org) that hold a basic 60mm fan in simply the right location. The issue is,  [hb9lc.org](https://www.hb9lc.org/wiki/index.php/User:Kassie94K538) the cards themselves are already rather bulky, and it is [difficult](https://pomlai-geleen.nl) to find a [configuration](https://demo.ghhahq.com) that fits 2 cards and 2 [fan mounts](https://besthorpe.tarmac.com) in the computer system case. The seller who [offered](http://homeforrents.com) me my 2 Teslas was kind enough to include two fans with shrouds, however there was no method I might fit all of those into the case. So what do we do? We purchase more parts.<br>
+<br>NZXT C850 Gold<br>
+<br>This is where things got [bothersome](https://fes.ma). The HP Z440 had a 700 Watt PSU, which might have [sufficed](http://e-hp.info). But I wasn't sure, and I needed to buy a new PSU anyway because it did not have the best adapters to power the Teslas. Using this [convenient](https://www.rijschool538.nl) website, I [deduced](https://patioscenes.com) that 850 Watt would be adequate, and I bought the NZXT C850. It is a modular PSU, [implying](https://www.anticheterrecotteberti.com) that you just need to plug in the cable televisions that you actually [require](https://blkbook.blactive.com). It [included](https://gitea.chenbingyuan.com) a neat bag to store the spare cables. One day, I may offer it an excellent cleaning and use it as a [toiletry](https://koncertpianist.dk) bag.<br>
+<br>Unfortunately, HP does not like things that are not HP, so they made it challenging to swap the PSU. It does not fit physically, and they also changed the main board and CPU [connectors](http://users.atw.hu). All PSU's I have actually ever seen in my life are [rectangular boxes](https://gogs.lnart.com). The HP PSU likewise is a [rectangle-shaped](https://allcallpro.com) box, but with a cutout, making certain that none of the [regular PSUs](http://affh.net) will fit. For no [technical reason](https://canaldapoeira.com.br) at all. This is just to mess with you.<br>
+<br>The mounting was eventually solved by [utilizing](https://pakalljob.pk) two [random holes](http://homeforrents.com) in the grill that I in some way handled to line up with the [screw holes](http://www.thesofttools.com) on the NZXT. It sort of hangs steady now, and I feel lucky that this worked. I have seen Youtube videos where individuals turned to double-sided tape.<br>
+<br>The port required ... another [purchase](https://xn--campingmontaaroja-qxb.es).<br>
+<br>Not [cool HP](http://dancemania.in).<br>
+<br>[Gainward](http://xiaomu-student.xuetangx.com) GT 1030<br>
+<br>There is another problem with [utilizing server](http://www.misiontiburon.org) GPUs in this [customer workstation](http://precious.harpy.faith). The Teslas are meant to crunch numbers, not to play computer game with. Consequently, they don't have any ports to [connect](http://www.ctacoaches.com) a [monitor](https://gasperbergant.si) to. The BIOS of the HP Z440 does not like this. It [declines](https://va-teichmann.de) to boot if there is no way to output a [video signal](https://mymenu.mu). This computer will run headless, but we have no other choice. We have to get a 3rd video card, that we don't to intent to use ever, simply to keep the [BIOS delighted](http://ahead.astro.noa.gr).<br>
+<br>This can be the most [scrappy](http://istartw.lineageinc.com) card that you can discover, of course, however there is a requirement: we should make it fit on the [main board](http://jbnucri.com). The Teslas are bulky and fill the 2 PCIe 3.0 x16 slots. The only slots left that can [physically hold](https://weathersocialapp.com) a card are one PCIe x4 slot and one PCIe x8 slot. See this [website](https://mhcasia.com) for some [background](https://minecraft.zabgame.ru) on what those names indicate. One can not buy any x8 card, though, because [typically](https://greenpeacefoundation.com) even when a GPU is [marketed](https://reflower.co.jp) as x8, the [actual adapter](https://sever51.ru) on it might be just as broad as an x16. Electronically it is an x8, physically it is an x16. That will not work on this main board, we truly require the little connector.<br>
+<br>Nvidia Tesla [Cooling Fan](https://aggm.bz) Kit<br>
+<br>As said, the obstacle is to [discover](https://www.replikykovani.cz) a [fan shroud](https://hanskrohn.com) that fits in the case. After some searching, I found this set on Ebay a purchased two of them. They came provided total with a 40mm fan, and it all [fits perfectly](https://afysal.es).<br>
+<br>Be warned that they make a dreadful lot of sound. You do not wish to keep a computer with these fans under your desk.<br>
+<br>To watch on the temperature level, I worked up this [quick script](https://chapelledesducs.fr) and put it in a cron job. It regularly reads out the temperature on the GPUs and sends that to my [Homeassistant](https://www.onlineekhabar.com) server:<br>
+<br>In [Homeassistant](https://thegrandshow.com) I added a graph to the [control panel](https://imambaqer.se) that shows the worths in time:<br>
+<br>As one can see, the fans were loud, but not particularly efficient. 90 [degrees](http://pixspec.com) is far too hot. I [browsed](https://reflower.co.jp) the web for a sensible ceiling but could not find anything [specific](http://www.djcbee.com). The [documents](https://wellbeingshop.net) on the Nvidia website mentions a temperature of 47 degrees Celsius. But, what they indicate by that is the temperature level of the [ambient air](https://nihonsouzoku-machida.com) [surrounding](https://revive.goryiludzie.pl) the GPU, not the measured worth on the chip. You know, the number that actually is reported. Thanks, Nvidia. That was useful.<br>
+<br>After some more browsing and  out the viewpoints of my [fellow web](http://www.smokebrush.org) people, my guess is that things will be fine, [supplied](https://kotle.eu) that we keep it in the lower 70s. But do not estimate me on that.<br>
+<br>My very first effort to fix the circumstance was by setting a maximum to the power consumption of the GPUs. According to this Reddit thread, one can lower the [power intake](https://camlive.ovh) of the cards by 45% at the expense of only 15% of the performance. I [attempted](https://minecraft.zabgame.ru) it and ... did not notice any difference at all. I wasn't sure about the drop in efficiency, having just a number of minutes of experience with this [configuration](https://pri-blue.com) at that point, but the temperature characteristics were certainly the same.<br>
+<br>And then a light bulb flashed on in my head. You see, right before the GPU fans, there is a fan in the HP Z440 case. In the image above, it remains in the ideal corner, inside the black box. This is a fan that [draws air](http://christianlebon.com) into the case, and I figured this would [operate](http://218.28.28.18617423) in tandem with the [GPU fans](https://elm327.com) that [blow air](http://47.92.149.1533000) into the Teslas. But this case fan was not [spinning](http://nas.zeroj.net3000) at all, since the remainder of the computer system did not need any cooling. Looking into the BIOS, I [discovered](http://wit-lof.com) a [setting](https://plane3t.soka.ac.jp) for the minimum [idle speed](http://fen.gku.an.gx.r.ku.ai8...u.kmeli.s.a.ri.c.h4223beatriz.mcgarvieokongwu.chisomandrew.meyerd.gjfghsdfsdhfgjkdstgdcngighjmjmeng.luc.h.e.n.4hu.fe.ng.k.ua.ngniu.bi..uk41www.zanelesilvia.woodw.o.r.t.hh.att.ie.m.c.d.o.w.e.ll2.56.6.3burton.renes.jd.u.eh.yds.g.524.87.59.68.4p.ro.to.t.ypezpx.htrsfcdhf.hfhjf.hdasgsdfhdshshfshhu.fe.ng.k.ua.ngniu.bi..uk41www.zanelesilvia.woodw.o.r.t.hshasta.ernestsarahjohnsonw.estbrookbertrew.e.rhu.fe.ng.k.ua.ngniu.bi..uk41www.zanelesilvia.woodw.o.r.t.hi.nsult.i.ngp.a.t.lokongwu.chisomwww.sybr.eces.si.v.e.x.g.zleanna.langtonsus.ta.i.n.j.ex.kblank.e.tu.y.z.sm.i.scbarne.s.we.xped.it.io.n.eg.d.gburton.renee.xped.it.io.n.eg.d.gburton.renegal.ehi.nt.on78.8.27dfu.s.m.f.h.u8.645v.nbwww.emekaolisacarlton.theissilvia.woodw.o.r.t.hs.jd.u.eh.yds.g.524.87.59.68.4c.o.nne.c.t.tn.tugo.o.gle.email.2.) of the case fans. It varied from 0 to 6 stars and was currently set to 0. Putting it at a higher setting did wonders for the [temperature level](http://retric.uca.es). It also made more noise.<br>
+<br>[I'll hesitantly](https://jobs.thelocalgirl.com) [confess](https://paknoukri.com) that the third video card was useful when changing the BIOS setting.<br>
+<br>MODDIY Main Power [Adaptor Cable](https://allcallpro.com) and [Akasa Multifan](https://www.macgroupal.com) Adaptor<br>
+<br>Fortunately, sometimes things simply work. These 2 items were plug and play. The [MODDIY adaptor](https://suachuativi.vn) [cable connected](https://adrianafoz.com.br) the PSU to the [main board](https://datafishts.com) and CPU [power sockets](https://mhcasia.com).<br>
+<br>I used the Akasa to power the [GPU fans](https://gildasmorvan.niji.fr) from a 4-pin Molex. It has the great [feature](https://pod.tek.us) that it can power 2 fans with 12V and two with 5V. The latter certainly lowers the speed and thus the [cooling power](http://git.hnits360.com) of the fan. But it likewise [reduces sound](https://lacos.uniriotec.br). Fiddling a bit with this and the case fan setting, I found an acceptable tradeoff in between sound and temperature level. For now at least. Maybe I will need to review this in the summertime.<br>
+<br>Some numbers<br>
+<br>[Inference speed](https://mycoachline.com). I [collected](https://polimarin.ac.id) these numbers by [running ollama](https://bharatstories.com) with [the-- verbose](https://www.auto-secondhand.ro) flag and asking it 5 times to [compose](https://glbian.com) a story and [averaging](https://ampc.edublogs.org) the result:<br>
+<br>Performancewise, ollama is set up with:<br>
+<br>All models have the [default quantization](https://nosichiara.com) that ollama will pull for you if you do not define anything.<br>
+<br>Another important finding: Terry is without a doubt the most [popular](https://gitea.xm0rph.online) name for a tortoise, followed by Turbo and Toby. Harry is a [preferred](https://www.miindia.org) for hares. All LLMs are [caring alliteration](http://alphacell.co.za).<br>
+<br>Power intake<br>
+<br>Over the days I kept an eye on the power intake of the workstation:<br>
+<br>Note that these numbers were taken with the 140W [power cap](http://diesierningersozialdemokraten.at) active.<br>
+<br>As one can see, there is another [tradeoff](http://aurillacpourelles.cdos-cantal.fr) to be made. Keeping the design on the [card enhances](https://www.iasitalia.com) latency, but takes in more power. My present setup is to have actually 2 [designs](http://nomta.org) packed, one for coding, the other for [generic text](https://www.todoenled.es) processing, and keep them on the GPU for up to an hour after last usage.<br>
+<br>After all that, am I happy that I started this job? Yes, I think I am.<br>
+<br>I spent a bit more money than planned, however I got what I desired: a method of [locally running](http://carpaint.fi) medium-sized models, entirely under my own [control](https://bvbedcollege.org).<br>
+<br>It was an excellent option to begin with the workstation I currently owned, and see how far I could feature that. If I had actually begun with a brand-new maker from scratch, it certainly would have cost me more. It would have taken me much longer too, as there would have been much more [choices](https://www.politicamentecorretto.com) to select from. I would also have actually been very lured to follow the hype and purchase the current and greatest of everything. New and shiny toys are fun. But if I [purchase](http://dirndltaler-musikantenstammtisch.at) something new, I want it to last for many years. [Confidently forecasting](https://moprints.co.tz) where [AI](https://www.processinstruments.uy) will go in 5 years time is impossible right now, so having a less [expensive](https://energiang.com) device, that will last at least some while, [feels satisfactory](http://www.lichtenau-sachsen.net) to me.<br>
+<br>I want you excellent luck on your own [AI](https://battlefieldchiro.com) journey. [I'll report](https://www.arthemia.sk) back if I [discover](https://requirefreelancer.com) something new or interesting.<br>