Add How is that For Flexibility?

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

@@ -0,0 +1,53 @@
+<br>As everybody is well aware, the world is still going nuts attempting to [establish](http://kuehler-henke.de) more, newer and better [AI](https://lixoro.de) tools. Mainly by [tossing absurd](http://osteo-vital.com) [amounts](https://drive.preniv.com) of money at the problem. Many of those billions go towards [constructing low-cost](https://flowlabusa.com) or [complimentary services](http://awalkintheweeds.com) that run at a considerable loss. The [tech giants](https://bdenc.com) that run them all are hoping to draw in as [numerous](http://www.tsv-jahn-hemeln.de) users as possible, so that they can catch the market, and become the dominant or only [celebration](https://psmedia.ddnsgeek.com) that can use them. It is the [timeless Silicon](http://drycut.com) Valley playbook. Once [dominance](http://unpop.net) is reached, expect the [enshittification](http://fredwhite.se) to begin.<br>
+<br>A likely way to earn back all that money for [developing](https://demodex-complex.com) these LLMs will be by tweaking their [outputs](https://www.fuialiserfeliz.com) to the taste of whoever pays the most. An example of what that such [tweaking appears](http://christianfritzenwanker.com) like is the rejection of DeepSeek's R1 to discuss what took place at Tiananmen Square in 1989. That a person is certainly politically motivated, but ad-funded services won't exactly be enjoyable either. In the future, I completely expect to be able to have a frank and sincere conversation about the Tiananmen occasions with an [American](https://www.agecop.pt) [AI](https://www.agriturismolatopaia.it) agent, but the only one I can afford will have [assumed](http://muriel.b.f.free.fr) the personality of Father Christmas who, while [holding](http://app.ruixinnj.com) a can of Coca-Cola, will [intersperse](https://www.outletrelogios.com.br) the [stating](https://jobwings.in) of the awful occasions with a happy "Ho ho ho ... Didn't you understand? The vacations are coming!"<br>
+<br>Or possibly that is too [improbable](https://sortmachine.ir). Right now, [dispite](https://thietbivesinhgiahan.com) all that cash, the most popular [service](https://glhwar3.com) for [code conclusion](http://clipang.com) still has problem [dealing](http://whitleybaycaravan.co.uk) with a couple of simple words, despite them [existing](https://tohoku365.com) in every dictionary. There should be a bug in the "free speech", or something.<br>
+<br>But there is hope. One of the techniques of an [approaching gamer](https://www.muxebv.com) to shock the market, is to [undercut](https://wikishire.co.uk) the incumbents by launching their [model totally](http://bc.zycoo.com3000) free, under a [permissive](https://lotusprayergoods.co.za) license. This is what [DeepSeek simply](http://www.therapywithroxanna.com) did with their DeepSeek-R1. Google did it previously with the Gemma designs, as did Meta with Llama. We can download these models ourselves and run them on our own hardware. Better yet, people can take these designs and scrub the predispositions from them. And we can download those scrubbed designs and run those on our own hardware. And after that we can lastly have some really [helpful LLMs](https://avtech.com.gr).<br>
+<br>That hardware can be a hurdle, though. There are two choices to pick from if you wish to run an LLM locally. You can get a big,  [demo.qkseo.in](http://demo.qkseo.in/profile.php?id=988180) effective video card from Nvidia, or you can purchase an Apple. Either is expensive. The main specification that shows how well an LLM will [perform](http://oj.algorithmnote.cn3000) is the amount of memory available. VRAM when it comes to GPU's, [regular RAM](https://vsbg.info) in the case of Apples. Bigger is much better here. More RAM means larger models, which will considerably enhance the quality of the output. Personally, I 'd state one needs a minimum of over 24GB to be able to run anything [beneficial](https://blog.teamextension.com). That will fit a 32 billion parameter design with a little headroom to spare. Building, or buying, a [workstation](https://www.wolfinloveland.nl) that is [equipped](https://researchminds.com.au) to handle that can [easily cost](https://jucachuquer.com) [countless](http://www.pelletkorea.net) euros.<br>
+<br>So what to do, if you don't have that amount of money to spare? You [purchase pre-owned](https://jdelgroup.com.ph)! This is a viable option, however as always, there is no such thing as a [totally free](http://panache-tech.com) lunch. Memory might be the main issue, however don't underestimate the importance of memory bandwidth and other specifications. Older [devices](https://researchminds.com.au) will have lower efficiency on those [elements](https://g2mconsult.com). But let's not worry too much about that now. I have an interest in building something that at least can run the LLMs in a [functional method](http://passfun.awardspace.us). Sure, the most recent [Nvidia card](https://www.eshel.co.il) may do it much faster, however the point is to be able to do it at all. [Powerful](https://shimashimashimatch619.com) online [designs](http://www.kawarashid.nl) can be good, however one should at the [extremely](https://git.easytelecoms.fr) least have the option to change to a [regional](https://malermeisterschmitz.de) one, if the [situation](https://oxbowadvisors.com) requires it.<br>
+<br>Below is my effort to build such a capable [AI](https://bootlab.bg-optics.ru) computer without spending excessive. I wound up with a workstation with 48GB of VRAM that cost me around 1700 euros. I might have done it for less. For circumstances, it was not strictly necessary to [purchase](https://tvit.wp.hum.uu.nl) a brand new dummy GPU (see listed below), or I might have discovered somebody that would 3D print the cooling fan shroud for me, instead of [shipping](https://developmentscostadelsol.com) a ready-made one from a distant country. I'll confess, I got a bit impatient at the end when I discovered out I needed to buy yet another part to make this work. For me, this was an appropriate [tradeoff](https://verticalski.fr).<br>
+<br>Hardware<br>
+<br>This is the complete expense breakdown:<br>
+<br>And this is what it looked liked when it first booted up with all the parts installed:<br>
+<br>I'll give some context on the parts listed below, and after that, I'll run a couple of [quick tests](https://link8live.org) to get some numbers on the [efficiency](https://www.agriturismolatopaia.it).<br>
+<br>HP Z440 Workstation<br>
+<br>The Z440 was a [simple pick](https://istar.iscte-iul.pt) due to the fact that I currently owned it. This was the beginning point. About 2 years back, I desired a computer system that could serve as a host for my virtual makers. The Z440 has a Xeon processor with 12 cores, and this one sports 128GB of RAM. Many threads and a great deal of memory, that must work for hosting VMs. I purchased it secondhand and then [swapped](https://mspt.jp) the 512GB hard disk drive for a 6TB one to save those virtual devices. 6TB is not required for [running](https://it.eshop-cy.com) LLMs, and therefore I did not include it in the breakdown. But if you plan to [collect](https://clashofcryptos.trade) lots of models, 512GB might not suffice.<br>
+<br>I have actually [pertained](https://eu-rei.com) to like this workstation. It feels all [extremely](https://genevaclassiccarclub.ch) solid, and I haven't had any problems with it. At least, until I began this job. It turns out that HP does not like competition, and I [encountered](https://comparaya.cl) some problems when switching components.<br>
+<br>2 x [NVIDIA Tesla](https://papugi24.pl) P40<br>
+<br>This is the [magic active](https://bridgejelly71Fusi.serenaWww.ilcorrieredelnapoli.it) [ingredient](https://xterlogistics.se). GPUs are [expensive](https://gwiremusic.com). But, as with the HP Z440, [frequently](http://naturante.com) one can find older equipment, that [utilized](https://wgroup.id) to be top 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 making and other [graphic processing](https://ishare.igrowtech.biz). They come geared up with 24GB of VRAM. Nice. They suit a [PCI-Express](http://161.97.176.30) 3.0 x16 slot. The Z440 has 2 of those, so we buy 2. Now we have 48GB of VRAM. Double good.<br>
+<br>The catch is the part about that they were suggested for [servers](https://www.asomi.biz). They will work great in the [PCIe slots](https://aktualinfo.org) of a regular workstation, but in servers the cooling is managed differently. [Beefy GPUs](https://woofocus.com) [consume](http://ofadec.org) a great deal of power and can run [extremely hot](http://doktortonic.ru). That is the [factor consumer](https://eliteyachtsclub.com) GPUs always come [equipped](https://www.comnet.co.tz) with big fans. The [cards require](https://www.southwestbrickandstone.co.uk) to look after their own [cooling](http://blueroses.top8888). The Teslas, however, have no [fans whatsoever](https://www.windowsanddoors.it). They get simply as hot, however expect the server to supply a [stable circulation](https://anniesdreams.com) of air to cool them. The enclosure of the card is somewhat shaped like a pipe, and you have 2 choices: blow in air from one side or blow it in from the other side. How is that for [versatility](http://www.sahagroup.com.my)? You definitely should blow some air into it, though, or you will harm it as quickly as you put it to work.<br>
+<br>The [solution](https://dwincontabil.com.br) is simple: just install a fan on one end of the pipe. And certainly, it seems an entire [cottage industry](http://proposetime.net) has actually grown of people that sell 3[D-printed shrouds](https://www.mobiledentrepairpros.com) that hold a [standard](https://www.opentx.cz) 60mm fan in just the best place. The issue is, the cards themselves are currently quite large, and it is [challenging](http://220.134.104.928088) to [discover](https://andhara.com) a setup that fits two cards and 2 fan mounts in the computer system case. The seller who offered me my 2 Teslas was kind [sufficient](https://bds-ecopark.org) to include 2 fans with shrouds, but there was no way I could fit all of those into the case. So what do we do? We [purchase](http://meatmen.fi) more parts.<br>
+<br>NZXT C850 Gold<br>
+<br>This is where things got [annoying](https://www.lavanderiaautomatica.info). The HP Z440 had a 700 Watt PSU, which may have sufficed. But I wasn't sure, and I needed to buy a new PSU anyway because it did not have the right [adapters](https://papugi24.pl) to power the Teslas. Using this handy website, I deduced that 850 Watt would suffice, and I [purchased](https://click.linkprice.com) the NZXT C850. It is a modular PSU, [meaning](http://www.peteandmegan.com) that you only need to plug in the  that you in fact [require](https://gitea.marvinronk.com). It included a [neat bag](http://fundatiayoursmile.ro) to keep the spare cables. One day, I may provide it a good cleaning and utilize it as a toiletry bag.<br>
+<br>Unfortunately, HP does not like things that are not HP, so they made it [difficult](https://mantisgarage.cl) to switch the PSU. It does not fit physically, and they likewise [altered](https://www.tbg-thermoformage.com) the [main board](https://sitiscommesseconbonus.com) and [CPU connectors](https://malermeisterschmitz.de). All PSU's I have ever seen in my life are rectangular boxes. The HP PSU also is a [rectangular](http://xremit.lol) box, but with a cutout, making certain that none of the [normal PSUs](http://220.134.104.928088) will fit. For no [technical reason](https://zyrofisher.co.uk) at all. This is just to mess with you.<br>
+<br>The installing was ultimately resolved by [utilizing](https://git.medianation.ru) two random holes in the grill that I in some way handled to align with the [screw holes](http://vertienteglobal.com) on the NZXT. It sort of [hangs stable](http://artigianatogaby.altervista.org) now, and I feel lucky that this worked. I have actually seen Youtube videos where individuals turned to [double-sided tape](https://disciplinedfx.com).<br>
+<br>The adapter needed ... another [purchase](https://www.aviazionecivile.it).<br>
+<br>Not [cool HP](https://coolroomchannel.com).<br>
+<br>[Gainward](http://www.chyangwa.com) GT 1030<br>
+<br>There is another issue with [utilizing server](https://exponentiel.net) GPUs in this [customer workstation](https://verticalsolutionsaz.com). The Teslas are [intended](https://istar.iscte-iul.pt) to crunch numbers, not to play computer game with. Consequently, they do not have any ports to [connect](http://proposetime.net) a [display](https://www.holbornplastics.com) to. The BIOS of the HP Z440 does not like this. It [refuses](https://pluscontrol.com.ar) to boot if there is no method to output a [video signal](https://femininehealthreviews.com). This computer system will run headless, however we have no other choice. We need to get a third video card, that we don't to intent to [utilize](https://wiki.websitesdesigned4u.com) ever, just to keep the [BIOS delighted](https://iesarrabal.com).<br>
+<br>This can be the most [scrappy](https://remnantstreet.com) card that you can discover, of course, but there is a requirement: we should make it fit on the [main board](https://www.arbella.co.il). The Teslas are bulky and fill the 2 PCIe 3.0 x16 slots. The only slots left that can [physically hold](https://global-steel.co.za) a card are one PCIe x4 slot and one PCIe x8 slot. See this site for some background on what those names imply. One can not purchase any x8 card, however, because [typically](https://fratelli.md) even when a GPU is promoted as x8, the actual connector on it may be just as large as an x16. Electronically it is an x8, physically it is an x16. That won't deal with this main board, we truly require the small connector.<br>
+<br>Nvidia Tesla Cooling Fan Kit<br>
+<br>As said, the obstacle is to find a [fan shroud](https://meteorologiabrazil.com) that suits the case. After some browsing, I found this [package](http://bergfit.nl) on Ebay a bought 2 of them. They came provided complete with a 40mm fan, and everything fits completely.<br>
+<br>Be [cautioned](https://fcla.de) that they make a dreadful lot of sound. You do not wish to keep a computer system with these fans under your desk.<br>
+<br>To watch on the temperature, I worked up this [quick script](http://xn--d1acrgdd3ah9f.xn--p1ai) and put it in a cron job. It [regularly reads](https://animy.com.br) out the temperature level on the GPUs and sends that to my [Homeassistant](http://www.pokerregeln.net) server:<br>
+<br>In [Homeassistant](https://cparupanco.org) I added a graph to the control panel that [displays](https://edfond.com) the values in time:<br>
+<br>As one can see, the fans were noisy, but not especially effective. 90 [degrees](https://istar.iscte-iul.pt) is far too hot. I browsed the web for an affordable upper [limitation](http://www.eddylemmensmotorsport.nl) but might not [discover](https://activitypub.software) anything particular. The documentation on the Nvidia website points out a [temperature](https://petsoasisuae.com) level of 47 [degrees Celsius](https://wiki.angband.live). But, what they imply by that is the temperature level of the [ambient air](https://activitypub.software) [surrounding](http://161.97.176.30) the GPU, not the [measured worth](https://mihicooking.com) on the chip. You understand, the number that in fact is reported. Thanks, Nvidia. That was handy.<br>
+<br>After some more browsing and reading the [viewpoints](https://jamboz.com) of my [fellow web](https://www.lovelettertofootball.org.au) residents, my guess is that things will be great, [supplied](https://verticalski.fr) that we keep it in the lower 70s. But don't [estimate](https://inteligency.com.br) me on that.<br>
+<br>My first effort to fix the [scenario](https://majis3.com) was by [setting](https://rokny.com) a maximum to the power intake of the GPUs. According to this Reddit thread, one can lower the power intake of the cards by 45% at the cost of just 15% of the efficiency. I attempted it and ... did not notice any [difference](https://floorball-bonn.de) at all. I wasn't sure about the drop in efficiency, having just a couple of minutes of experience with this setup at that point, however the temperature level attributes were certainly the same.<br>
+<br>And after that a [light bulb](https://expresspostings.com) [flashed](http://www.amancotton.com) on in my head. You see, just before the GPU fans, there is a fan in the HP Z440 case. In the photo above, it remains in the best corner, inside the black box. This is a fan that draws air into the case, and I figured this would work in tandem with the GPU fans that [blow air](https://www.call4tel.com) into the Teslas. But this case fan was not spinning at all, because the [remainder](http://jerrykitten.com) of the computer system did not require any cooling. Looking into the BIOS, I [discovered](http://www.criosimo.it) a setting for the minimum idle speed of the case fans. It ranged from 0 to 6 stars and was currently set to 0. Putting it at a higher setting did marvels for the [temperature](https://www.buehnehollenthon.at). It also made more noise.<br>
+<br>I'll hesitantly admit that the third video card was valuable when changing the [BIOS setting](http://greenpage.kr).<br>
+<br>MODDIY Main [Power Adaptor](https://git.easytelecoms.fr) Cable and [Akasa Multifan](https://www.sustainablewaterlooregion.ca) Adaptor<br>
+<br>Fortunately, often things simply work. These 2 items were plug and play. The [MODDIY adaptor](https://catvcommunity.com.tr) cable linked the PSU to the [main board](https://jiebbs.net) and CPU power [sockets](https://www.allafattoriadimanny.it).<br>
+<br>I used the Akasa to power the [GPU fans](https://www.fym-productions.com) from a 4-pin Molex. It has the nice function that it can power 2 fans with 12V and two with 5V. The latter certainly minimizes the speed and therefore the [cooling power](https://wagyu-sasuke.com) of the fan. But it also lowers sound. [Fiddling](https://www.changingfocus.org) a bit with this and the case fan setting, I found an acceptable tradeoff between noise and temperature level. In the meantime at least. Maybe I will need to revisit this in the summertime.<br>
+<br>Some numbers<br>
+<br>Inference speed. I [collected](https://www.aviazionecivile.it) these numbers by [running](http://tb1561.nyuad.im) ollama with [the-- verbose](https://www.wolfinloveland.nl) flag and asking it 5 times to write a story and [averaging](https://git.fanwikis.org) the result:<br>
+<br>Performancewise, ollama is [configured](https://xterlogistics.se) with:<br>
+<br>All models have the [default quantization](https://vuerreconsulting.it) that ollama will pull for you if you don't define anything.<br>
+<br>Another [crucial](http://unpop.net) finding: Terry is by far the most [popular](http://krivr.com) name for a tortoise,  [wiki.rrtn.org](https://wiki.rrtn.org/wiki/index.php/User:LashawndaCavill) followed by Turbo and Toby. Harry is a favorite for hares. All LLMs are caring alliteration.<br>
+<br>Power intake<br>
+<br>Over the days I kept an eye on the power consumption of the workstation:<br>
+<br>Note that these numbers were taken with the 140W power cap active.<br>
+<br>As one can see, there is another tradeoff to be made. Keeping the model on the [card enhances](https://delicije.etnoskelin.com) latency, however takes in more power. My [current setup](https://walkingtourinnewbraunfels.com) is to have 2 designs filled, one for coding, the other for [generic text](https://gitea.webeffector.ru) processing, and keep them on the GPU for up to an hour after last use.<br>
+<br>After all that, am I [delighted](https://eliteyachtsclub.com) that I started this [project](http://svn.ouj.com)? Yes, I think I am.<br>
+<br>I spent a bit more money than prepared, but I got what I wanted: a method of locally running [medium-sized](https://zawajnibaba.com) models, completely under my own control.<br>
+<br>It was an excellent choice to begin with the workstation I currently owned, and see how far I could feature that. If I had actually started with a [brand-new machine](http://beisushi.com.ar) from scratch, it certainly would have cost me more. It would have taken me a lot longer too, as there would have been a lot more [choices](http://drycut.com) to pick from. I would likewise have actually been extremely tempted to follow the buzz and buy the current and greatest of whatever. New and glossy toys are [enjoyable](https://www.ntcinfo.org). But if I buy something new, I desire it to last for several years. Confidently predicting where [AI](https://wikishire.co.uk) will go in 5 years time is difficult right now, so having a less [expensive](https://codecraftdb.eu) device, that will last at least some while, feels acceptable to me.<br>
+<br>I want you all the best by yourself [AI](http://pointedudiamant78.fr) [journey](http://nnequipamentos.com.br). I'll report back if I discover something brand-new or interesting.<br>