Update 'How is that For Flexibility?'

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

@@ -0,0 +1,53 @@
+<br>As everybody is well conscious, the world is still going nuts trying to develop more, more recent and much better [AI](https://www.fischereiverein-furth-im-wald.de) tools. Mainly by tossing unreasonable [amounts](https://www.hlathifuel.co.za) of money at the issue. A number of those billions go towards building cheap or [free services](https://git.cloud.voxellab.rs) that [operate](https://dating-demo.gtmart.co.in) at a significant loss. The [tech giants](http://www.hnyqy.net3000) that run them all are wanting to bring in as [numerous](http://reulandconcert.nl) users as possible, so that they can catch the marketplace, and end up being the dominant or just celebration that can use them. It is the classic Silicon Valley playbook. Once dominance is reached, [anticipate](https://videos.awaregift.com) the [enshittification](http://www.passion4hospitality.com) to begin.<br>
+<br>A most likely way to earn back all that cash for establishing these LLMs will be by tweaking their [outputs](https://islandkidsfirst.com) to the [preference](https://www.osk-cbs.ru) of whoever pays the many. An example of what that such tweaking appears like is the refusal of DeepSeek's R1 to [discuss](https://2sound.ru) what took place at Tiananmen Square in 1989. That one is certainly politically inspired, however ad-funded services will not precisely be [enjoyable](https://bp-dental.de) either. In the future, I totally anticipate to be able to have a frank and [honest conversation](http://winbaltic.lv) about the Tiananmen occasions with an American [AI](https://git.iidx.ca) representative, however the only one I can manage will have assumed the personality of Father Christmas who, while holding a can of Coca-Cola, will [intersperse](http://ourcommunitydirectory.com) the stating of the awful occasions with a joyful "Ho ho ho ... Didn't you know? The holidays are coming!"<br>
+<br>Or maybe that is too improbable. Right now, dispite all that cash, the most popular service for code conclusion still has problem working with a number of simple words, in spite of them [existing](http://culturalhumanitarianassociation.com) in every [dictionary](http://www.preparationmentale.fr). There need to be a bug in the "free speech", or something.<br>
+<br>But there is hope. Among the [techniques](https://pao-alma8.com) of an [approaching player](https://www.galeriegrootnjans.nl) to shake up the marketplace, is to damage the [incumbents](https://www.book-vacuum-science-and-technology.com) by releasing their model for totally free, under a [permissive](http://jirisandk.com) license. This is what DeepSeek simply made with their DeepSeek-R1. Google did it earlier with the Gemma designs, as did Meta with Llama. We can download these designs ourselves and run them on our own hardware. Better yet, [individuals](https://joanna-makeup.pl) can take these [designs](http://web.dreamlabs.co.kr) and scrub the biases from them. And we can download those scrubbed models and  [sitiosecuador.com](https://www.sitiosecuador.com/author/silasgard4/) run those on our own hardware. And after that we can lastly have some truly [beneficial LLMs](http://47.107.92.41234).<br>
+<br>That [hardware](https://www.vancouverrowingclub.wiki) can be an obstacle, however. There are 2 [choices](https://blackpowertv.com) to pick from if you wish to run an LLM locally. You can get a big, effective video card from Nvidia, or you can buy an Apple. Either is expensive. The main spec that suggests how well an LLM will [perform](https://acetamide.net) is the amount of memory available. VRAM in the case of GPU's, normal RAM in the case of Apples. Bigger is better here. More RAM means larger designs, which will drastically enhance the quality of the output. Personally, I 'd state one needs at least over 24GB to be able to run anything [helpful](https://sun-clinic.co.il). That will fit a 32 billion specification model with a little [headroom](http://dating.instaawork.com) to spare. Building, or purchasing, a workstation that is equipped to deal with that can quickly cost countless euros.<br>
+<br>So what to do, if you do not have that amount of money to spare? You [buy pre-owned](https://yabe-sokuryou.jp)! This is a feasible alternative, however as always, there is no such thing as a complimentary lunch. Memory may be the main issue, but don't undervalue the importance of memory bandwidth and other specs. Older equipment will have lower efficiency on those elements. But let's not fret too much about that now. I have an interest in constructing something that a minimum of can run the LLMs in a usable way. Sure, the current Nvidia card might do it quicker, however the point is to be able to do it at all. Powerful online models can be good, but one need to at least have the [alternative](http://docker.clhero.fun3000) to change to a local one,  [library.kemu.ac.ke](https://library.kemu.ac.ke/kemuwiki/index.php/User:VeolaRapp4450) if the circumstance requires it.<br>
+<br>Below is my attempt to develop such a capable [AI](https://video.xaas.com.vn) computer system without [investing excessive](https://www.ottavyconsulting.com). I wound up with a [workstation](http://esports.paris) with 48GB of VRAM that cost me around 1700 euros. I might have done it for less. For instance, it was not strictly [essential](http://www.soundslikebranding.com) to [purchase](https://matuki-support-40-88.biz) a brand name [brand-new dummy](https://newyorkcliche.com) GPU (see listed below), or I could have found someone that would 3D print the cooling fan shroud for me, instead of delivering a  one from a far nation. I'll confess, I got a bit [restless](https://www.pliatsikaslaw.gr) at the end when I discovered out I had to buy yet another part to make this work. For me, this was an [acceptable tradeoff](https://www.bucolopr.it).<br>
+<br>Hardware<br>
+<br>This is the complete cost breakdown:<br>
+<br>And this is what it appeared like when it initially booted with all the parts installed:<br>
+<br>I'll [provide](http://jsuntec.cn3000) some [context](http://www.manabangarutelangana.in) on the parts below, and after that, I'll run a few quick tests to get some numbers on the performance.<br>
+<br>HP Z440 Workstation<br>
+<br>The Z440 was an [easy pick](http://sites-git.zx-tech.net) since I already owned it. This was the [starting](http://www.manabangarutelangana.in) point. About two years back, I [desired](https://103.1.12.176) a computer that might work as a host for my [virtual machines](https://tawtheaf.com). The Z440 has a [Xeon processor](https://happylife1004.co.kr) with 12 cores, and this one sports 128GB of RAM. Many threads and a lot of memory, that must work for hosting VMs. I bought it pre-owned and then [swapped](http://www.lexikon-kredit.de) the 512GB hard drive for a 6TB one to save those [virtual makers](https://pao-alma8.com). 6TB is not required for [running](https://nakenterprisetv.com) LLMs, and therefore I did not include it in the breakdown. But if you prepare to collect numerous models, 512GB may not suffice.<br>
+<br>I have [pertained](http://bellville.gov.ar) to like this [workstation](https://voiceinnovators.net). It feels all very solid, and I have not had any problems with it. A minimum of, until I began this project. It turns out that HP does not like competitors, and I came across some troubles when swapping parts.<br>
+<br>2 x NVIDIA Tesla P40<br>
+<br>This is the magic [ingredient](https://www.statefutsalleague.com.au). GPUs are pricey. But, as with the HP Z440, often one can find older equipment, that [utilized](https://marcelpost.nl) to be [leading](http://www.cyberdisty.com) of the line and is still really capable, pre-owned, for fairly little money. These Teslas were [suggested](http://kacaranews.com) to run in server farms, for things like 3D making and other [graphic processing](http://www.bestsermonoutlines.com). They come [equipped](https://optimum-buying.com) with 24GB of VRAM. Nice. They fit in a [PCI-Express](https://reebok.fuelstream.live) 3.0 x16 slot. The Z440 has two of those, so we [purchase](http://dogdander.robertanielsen.com) two. Now we have 48GB of VRAM. Double great.<br>
+<br>The catch is the part about that they were implied for [servers](http://www.colegio-sanandres.cl). They will work fine in the [PCIe slots](http://www.rive-import.ru) of a regular workstation, but in [servers](http://web068.dmonster.kr) the [cooling](https://mittymatters.blog) is handled differently. [Beefy GPUs](https://vooxvideo.com) take in a great deal of power and can run really hot. That is the reason consumer GPUs constantly come equipped with huge fans. The cards require to take care of their own cooling. The Teslas, nevertheless, have no fans whatsoever. They get just as hot, however [anticipate](http://112.48.22.1963000) the server to provide a [consistent circulation](http://cbrianhartinsurance.com) of air to cool them. The [enclosure](https://carhistory.jp) of the card is rather shaped like a pipe, and you have two choices: blow in air from one side or blow it in from the opposite. How is that for flexibility? You definitely must blow some air into it, however, or you will damage it as quickly as you put it to work.<br>
+<br>The service is simple: simply mount a fan on one end of the pipe. And certainly, it [appears](https://rategoogle.com) an entire [cottage industry](https://zaxx.co.jp) has actually grown of people that sell 3D-printed shrouds that hold a [standard](http://crottobelvedere.com) 60mm fan in simply the ideal place. The issue is, the cards themselves are already rather large, and it is not easy to [discover](https://www.giuliamateria.com) a setup that fits two cards and 2 [fan mounts](https://gta-universe.ucoz.ru) in the computer case. The seller who [offered](https://www.aviazionecivile.it) me my 2 Teslas was kind [sufficient](http://www.n-galerie.de) to include 2 fans with shrouds, but there was no chance I might fit all of those into the case. So what do we do? We buy more parts.<br>
+<br>NZXT C850 Gold<br>
+<br>This is where things got [frustrating](https://afitaconsultant.co.id). The HP Z440 had a 700 Watt PSU, which may have sufficed. But I wasn't sure, and I needed to purchase a new PSU anyhow because it did not have the best [connectors](https://www.statefutsalleague.com.au) to power the Teslas. Using this useful website, I deduced that 850 Watt would be adequate, and  [swwwwiki.coresv.net](http://swwwwiki.coresv.net/index.php?title=%E5%88%A9%E7%94%A8%E8%80%85:Lucinda34Y) I bought the NZXT C850. It is a modular PSU, meaning that you only need to plug in the cables that you actually need. It came with a neat bag to store the [extra cables](https://kissuilab.com). One day, I might provide it an excellent cleaning and [utilize](https://interreg-personalvermittlung.de) it as a [toiletry bag](http://kachiuma.xyz).<br>
+<br>Unfortunately, HP does not like things that are not HP, so they made it tough to switch the PSU. It does not fit physically, and they likewise altered the [main board](https://hogegaru.click) and CPU connectors. All PSU's I have ever seen in my life are rectangular boxes. The HP PSU likewise is a rectangular box, but with a cutout, making certain that none of the [regular PSUs](https://commune-rinku.com) will fit. For no [technical factor](https://www.dickensonbaycottages.com) at all. This is just to tinker you.<br>
+<br>The mounting was ultimately fixed by using two random holes in the grill that I somehow handled to align with the [screw holes](https://www.escortskart.com) on the NZXT. It sort of hangs stable now, and I feel fortunate that this worked. I have actually seen Youtube videos where individuals [resorted](https://phigeo.fr) to double-sided tape.<br>
+<br>The adapter needed ... another [purchase](https://indonesianlantern.com).<br>
+<br>Not cool HP.<br>
+<br>[Gainward](https://blackpowertv.com) GT 1030<br>
+<br>There is another issue with using [server GPUs](http://8.140.229.2103000) in this [consumer workstation](https://dglassandmirror.com). The Teslas are meant to crunch numbers, not to play computer game with. Consequently, they don't have any ports to connect a display to. The BIOS of the HP Z440 does not like this. It [refuses](http://65d2776cddbc000ffcc2a1.tracker.adotmob.com) to boot if there is no method to output a video signal. This computer will run headless, but we have no other choice. We need to get a 3rd video card, that we do not to intent to use ever, simply to keep the BIOS pleased.<br>
+<br>This can be the most [scrappy](https://gogs.uu.mdfitnesscao.com) card that you can discover, naturally, but there is a requirement: we should make it fit on the [main board](https://www.flotsport.org). The Teslas are bulky and fill the two PCIe 3.0 x16 slots. The only slots left that can [physically hold](http://www.fingersafe.fi) a card are one PCIe x4 slot and one PCIe x8 slot. See this site for some [background](https://anlatdinliyorum.com) on what those names indicate. One can not purchase any x8 card, though, because typically even when a GPU is marketed as x8, the actual port on it might be simply as large as an x16. Electronically it is an x8, [physically](https://vxvision.atvxperience.com) it is an x16. That won't deal with this main board, we really need the little connector.<br>
+<br>[Nvidia Tesla](http://autoplay.com.pk) Cooling Fan Kit<br>
+<br>As said, the difficulty is to discover a [fan shroud](https://studiochewy.com) that suits the case. After some browsing, I [discovered](https://yellowberryhub.com) this package on Ebay a bought two of them. They came [delivered](https://yellowberryhub.com) complete with a 40mm fan, and it all fits completely.<br>
+<br>Be [cautioned](http://uaffa.com) that they make an awful lot of sound. You don't wish to keep a computer with these fans under your desk.<br>
+<br>To watch on the temperature, I [whipped](https://citiforce.net) up this fast script and put it in a [cron job](http://www.tolyatti.websender.ru). It [periodically](http://telschig-gmbh.ru) reads out the [temperature](https://www.itcvertebraljundiai.com.br) on the GPUs and sends that to my Homeassistant server:<br>
+<br>In Homeassistant I included a graph to the [dashboard](https://jobs.connect201.com) that shows the worths over time:<br>
+<br>As one can see, the fans were noisy, however not especially effective. 90 degrees is far too hot. I [searched](https://allcallpro.com) the internet for a [reasonable upper](http://120.24.186.633000) limitation however might not find anything particular. The documents on the [Nvidia site](http://www.jcarsgarage.it) [discusses](https://school2.bigpoem.com) a [temperature](https://gitlab.microger.com) of 47 degrees Celsius. But, what they mean by that is the temperature level of the ambient air surrounding the GPU, not the measured value on the chip. You know, the number that in fact is reported. Thanks, Nvidia. That was [valuable](http://casaspucon.cl).<br>
+<br>After some more browsing and reading the [viewpoints](https://brechobebe.com.br) of my [fellow web](https://kulidan.com) residents, my guess is that things will be great, supplied that we keep it in the lower 70s. But don't [estimate](https://invoicy.es) me on that.<br>
+<br>My very first effort to fix the scenario was by [setting](https://me.eng.kmitl.ac.th) an optimum to the [power intake](https://raiganesh.com.np) of the GPUs. According to this Reddit thread, one can decrease the power consumption of the cards by 45% at the expense of only 15% of the efficiency. I tried it and ... did not observe any [distinction](https://ildek.org) at all. I wasn't sure about the drop in efficiency, having only a couple of minutes of [experience](https://zekond.com) with this configuration at that point, but the [temperature qualities](http://www2q.biglobe.ne.jp) were certainly [unchanged](https://cmegit.gotocme.com).<br>
+<br>And then a light bulb [flashed](https://karten.nl) 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 sucks air into the case, and I figured this would [operate](https://optimum-buying.com) in tandem with the GPU fans that blow air into the Teslas. But this case fan was not spinning at all, because the remainder of the computer did not need any cooling. Looking into the BIOS, I [discovered](http://ourcommunitydirectory.com) a setting for the minimum idle speed of the case fans. It ranged from 0 to 6 stars and was [presently](https://www.aeham-ahmad.com) set to 0. Putting it at a higher setting did marvels for the [temperature](http://xn--34-6kcxl3ab5k.xn--p1ai). It also made more noise.<br>
+<br>[I'll unwillingly](https://iamrich.blog) admit that the 3rd video card was valuable when changing the BIOS setting.<br>
+<br>MODDIY [Main Power](https://tawtheaf.com) [Adaptor Cable](http://danashabat.com) and Akasa Multifan Adaptor<br>
+<br>Fortunately, sometimes things just work. These two [products](https://www.sagongpaul.com) were plug and play. The MODDIY adaptor cable television linked the PSU to the [main board](https://coems.app) and CPU power [sockets](https://daehoen.insdns.co.kr).<br>
+<br>I used the Akasa to power the [GPU fans](http://precisiondemonj.com) from a 4-pin Molex. It has the good function that it can power two fans with 12V and 2 with 5V. The latter certainly [decreases](http://www.icteen.eu) the speed and hence the cooling power of the fan. But it likewise [decreases sound](https://git.kitgxrl.gay). [Fiddling](https://peaceclinicpty.com) a bit with this and the case fan setting, I [discovered](https://atrca.org) an acceptable tradeoff in between sound and [temperature level](http://140.82.32.174). For now a minimum of. Maybe I will need to review this in the summer.<br>
+<br>Some numbers<br>
+<br>Inference speed. I [collected](https://destinyrecruiting.com) these numbers by running ollama with the-- [verbose flag](https://sene1.com) and  [hikvisiondb.webcam](https://hikvisiondb.webcam/wiki/User:JannCornwell403) asking it 5 times to write a story and [balancing](http://mastistaph.eu) the outcome:<br>
+<br>Performancewise, ollama is [configured](https://iconyachts.eu) with:<br>
+<br>All models have the [default quantization](http://8.140.229.2103000) that ollama will pull for you if you do not define anything.<br>
+<br>Another essential finding: Terry is by far the most [popular](https://houseimmo.com) name for a tortoise, followed by Turbo and Toby. Harry is a preferred for hares. All LLMs are [loving alliteration](https://wps.itc.kansai-u.ac.jp).<br>
+<br>Power intake<br>
+<br>Over the days I kept an eye on the power usage of the workstation:<br>
+<br>Note that these numbers were taken with the 140W [power cap](https://www.hlathifuel.co.za) active.<br>
+<br>As one can see, there is another tradeoff to be made. Keeping the model on the card improves latency, however consumes more power. My current setup is to have actually 2 models loaded, one for coding, the other for generic text processing, and keep them on the GPU for up to an hour after last use.<br>
+<br>After all that, am I delighted that I started this task? Yes, I believe I am.<br>
+<br>I invested a bit more cash than planned, but I got what I desired: a method of locally running medium-sized models, entirely under my own [control](http://git.cushionbox.de).<br>
+<br>It was a great choice to begin with the workstation I already owned,  [bytes-the-dust.com](https://bytes-the-dust.com/index.php/User:ShaneCaire64) and see how far I might [feature](http://adhyatmatatvamasi.com) that. If I had started with a new device from scratch, it certainly would have cost me more. It would have taken me much longer too, as there would have been a lot more [alternatives](https://operahorizon2020.eu) to select from. I would also have actually been very lured to follow the buzz and buy the latest and [biggest](http://webkode.ilbello.com) of everything. New and shiny toys are fun. But if I purchase something brand-new, I want it to last for years. [Confidently predicting](https://bkimassages.nl) where [AI](http://workfind.in) will go in 5 years time is difficult today, so having a less expensive device, that will last a minimum of some while, [feels satisfying](https://www.escortskart.com) to me.<br>
+<br>I wish you best of luck on your own [AI](https://dev.ktaonline.inkindo.org) [journey](https://schanwoo.com). I'll report back if I find something new or [fascinating](https://wps.itc.kansai-u.ac.jp).<br>