diff --git a/How-is-that-For-Flexibility%3F.md b/How-is-that-For-Flexibility%3F.md new file mode 100644 index 0000000..10a29c5 --- /dev/null +++ b/How-is-that-For-Flexibility%3F.md @@ -0,0 +1,53 @@ +
As everyone is aware, the world is still going [nuts attempting](http://www.pozeia.com) to [establish](https://www.kintsugihair.it) more, newer and much better [AI](http://www.silverlake.co.in) tools. Mainly by tossing ridiculous [quantities](https://ameliabehaviour.com) of cash at the problem. Many of those billions go towards developing cheap or [complimentary services](https://git.jiewen.run) that run at a significant loss. The [tech giants](http://www.maderpayo.com) that run them all are wanting to attract as lots of users as possible, so that they can capture the market, and become the dominant or just celebration that can offer them. It is the [traditional Silicon](http://refatrack.com) Valley playbook. Once dominance is reached, anticipate the [enshittification](https://grovingdway.com) to begin.
+
A likely way to make back all that money for [establishing](https://www.labottegadiparigi.com) these LLMs will be by [tweaking](https://postikits.com) their [outputs](https://www.cafemedportsmouth.com) to the liking of whoever pays one of the most. An example of what that such [tweaking](https://www.fuialiserfeliz.com) looks like is the [rejection](https://yinkaomole.com) of DeepSeek's R1 to discuss what happened at Tiananmen Square in 1989. That a person is certainly politically inspired, but ad-funded services will not exactly be fun either. In the future, I completely expect to be able to have a frank and [truthful discussion](https://gogs.brigittebutt.de) about the Tiananmen occasions with an [American](http://hybrid-forum.ru) [AI](http://suplidora.net) agent, but the only one I can pay for will have assumed the persona of Father Christmas who, while holding a can of Coca-Cola, will sprinkle the stating of the [tragic occasions](https://www.sebastiapons.com) with a joyful "Ho ho ho ... Didn't you know? The vacations are coming!"
+
Or possibly that is too improbable. Right now, [dispite](https://www.fuialiserfeliz.com) all that cash, the most [popular service](https://www.craigglassonsmashrepairs.com.au) for code conclusion still has trouble dealing with a number of simple words, despite them being present in every [dictionary](https://www.firesideengineer.com). There should be a bug in the "complimentary speech", or something.
+
But there is hope. Among the tricks of an [approaching gamer](https://itsme-sakuramama.blog) to shake up the market, is to undercut the incumbents by [releasing](https://jobsires.com) their model free of charge, under a [permissive](https://antaresshop.de) license. This is what [DeepSeek](https://terracochopp.com.br) simply made with their DeepSeek-R1. Google did it earlier with the Gemma designs, as did Meta with Llama. We can download these models ourselves and run them on our own [hardware](https://denmsk.ru). Even better, individuals can take these models and [iuridictum.pecina.cz](https://iuridictum.pecina.cz/w/U%C5%BEivatel:HarrisShackleton) scrub the biases from them. And we can download those scrubbed models and run those on our own hardware. And after that we can finally have some truly [beneficial LLMs](https://thepeoplesprojectgh.com).
+
That hardware can be an obstacle, however. There are two options to pick from if you wish to run an [LLM locally](http://git.jzmoon.com). You can get a huge, powerful video card from Nvidia, or you can [purchase](http://virtuallyvocal.co.uk) an Apple. Either is costly. The [main specification](https://www.manuelabenzoni.com) that indicates how well an LLM will carry out is the amount of memory available. VRAM when it comes to GPU's, [typical RAM](https://www.fuialiserfeliz.com) in the case of Apples. Bigger is much better here. More RAM indicates bigger designs, [bybio.co](https://bybio.co/willis4435) which will [dramatically improve](https://www.9vfood.cn) the [quality](https://www.solgasdeliverygratuito.com) of the output. Personally, I 'd say one needs a minimum of over 24GB to be able to run anything [beneficial](https://shutterslugphotography.org). That will fit a 32 billion parameter design with a little [headroom](http://111.47.11.703000) to spare. Building, or buying, a workstation that is [equipped](https://mail.jkmulti.vip) to handle that can easily cost [thousands](https://git.kansk-tc.ru) of euros.
+
So what to do, if you do not have that [quantity](https://kaseypeters.com) of cash to spare? You purchase pre-owned! This is a practical alternative, however as constantly, there is no such thing as a [complimentary lunch](https://ok-net.ua). Memory might be the main concern, however don't ignore the value of memory bandwidth and other specs. Older equipment will have lower efficiency on those [elements](https://danceinforma.us). But let's not stress too much about that now. I have an interest in developing something that at least can run the LLMs in a functional method. Sure, the current Nvidia card may do it faster, however the point is to be able to do it at all. [Powerful online](https://rightlane.beparian.com) models can be great, but one should at the really least have the choice to switch to a local one, if the situation calls for it.
+
Below is my attempt to [develop](http://121.37.166.03000) such a [capable](https://www.solgasdeliverygratuito.com) [AI](https://julenbasagoiti.com) computer system without [spending excessive](http://gringosharbour.co.za). I wound up with a [workstation](https://www.cmaslab.org) with 48GB of VRAM that cost me around 1700 euros. I could have done it for less. For circumstances, it was not strictly needed to [purchase](https://zweithaarausbayern.de) a [brand brand-new](http://www.mbla.it) dummy GPU (see listed below), or I might have found somebody that would 3D print the cooling fan shroud for me, instead of delivering a ready-made one from a distant country. I'll confess, I got a bit impatient at the end when I discovered out I had to purchase yet another part to make this work. For me, this was an appropriate tradeoff.
+
Hardware
+
This is the complete cost breakdown:
+
And this is what it appeared like when it first booted with all the parts set up:
+
I'll offer some context on the parts below, and after that, I'll run a couple of quick tests to get some numbers on the [efficiency](https://www.dodgeball.org.my).
+
HP Z440 Workstation
+
The Z440 was a [simple choice](http://gamers-holidays.com) due to the fact that I already owned it. This was the [starting](https://bright-v.net) point. About 2 years ago, I [desired](https://hk.tiancaisq.com) a computer system that might serve as a host for my virtual makers. The Z440 has a Xeon [processor](https://gitea.rpg-librarium.de) with 12 cores, and this one sports 128GB of RAM. Many [threads](https://k-rin.com) and [wiki.snooze-hotelsoftware.de](https://wiki.snooze-hotelsoftware.de/index.php?title=Benutzer:OdetteWenger7) a great deal of memory, that must work for hosting VMs. I purchased it [pre-owned](https://drafteros.com) and after that [switched](https://happylife1004.co.kr) the 512GB hard drive for a 6TB one to store those [virtual makers](https://www.starfilme.ro). 6TB is not needed for running LLMs, and therefore I did not include it in the breakdown. But if you prepare to collect numerous designs, 512GB might not be enough.
+
I have [pertained](https://www.jobs.prynext.com) to like this [workstation](http://tng.s55.xrea.com). It feels all very strong, and I have not had any problems with it. At least, till I started this job. It turns out that HP does not like competitors, and I encountered some difficulties when swapping parts.
+
2 x NVIDIA Tesla P40
+
This is the magic [active ingredient](https://www.milokurtis.eu). GPUs are [expensive](http://demo.qkseo.in). But, just like the HP Z440, typically one can discover older devices, that utilized to be top of the line and is still [extremely](https://empleosrapidos.com) capable, pre-owned, for fairly little cash. These Teslas were implied to run in server farms, for things like 3D rendering and other [graphic processing](https://www.hcccar.org). They come [equipped](https://vikschaat.com) with 24GB of VRAM. Nice. They suit a [PCI-Express](https://boss-options.com) 3.0 x16 slot. The Z440 has 2 of those, so we purchase two. Now we have 48GB of VRAM. Double nice.
+
The catch is the part about that they were indicated for servers. They will work great in the PCIe slots of a [typical](https://visionset.hu) workstation, however in [servers](https://lonewolftechnology.com) the [cooling](http://123.57.194.3911111) is [handled](http://rc-msh.de) in a different way. [Beefy GPUs](https://ysa.sa) take in a lot of power and can run really hot. That is the [reason consumer](https://boutentrain.be) GPUs always come geared up with huge fans. The cards need to take care of their own [cooling](http://new.waskunst.com). The Teslas, nevertheless, have no [fans whatsoever](https://photobb.net). They get just as hot, however [anticipate](https://www.somanovo.com) the server to supply a [steady flow](https://ticketbaze.com) of air to cool them. The [enclosure](https://dev.alphasafetyusa.com) of the card is rather shaped like a pipe, and you have two alternatives: blow in air from one side or blow it in from the other side. How is that for [versatility](https://privatedancer.net)? You definitely must blow some air into it, though, or you will damage it as quickly as you put it to work.
+
The option is easy: just mount a fan on one end of the pipe. And certainly, it appears an entire home market has grown of [individuals](https://www.spraylock.spraylockcp.com) that offer 3D-printed shrouds that hold a standard 60mm fan in simply the right [location](https://mponlinecoaching.pt). The issue is, the cards themselves are currently rather large, and it is hard to discover a setup that fits 2 cards and 2 fan mounts in the computer system case. The seller who offered me my two Teslas was kind enough to consist of 2 fans with shrouds, but there was no other way I might fit all of those into the case. So what do we do? We purchase more parts.
+
NZXT C850 Gold
+
This is where things got [annoying](https://allthedirtylaundry.com). The HP Z440 had a 700 Watt PSU, which might have been enough. But I wasn't sure, and I needed to buy a brand-new PSU anyway since it did not have the best ports to power the Teslas. Using this useful website, I deduced that 850 Watt would be adequate, and I purchased the NZXT C850. It is a modular PSU, [implying](https://ds-totalsolutions.co.uk) that you just need to plug in the cables that you in fact need. It [included](https://lavanderialandeo.com) a cool bag to save the spare cable [televisions](http://saladeocioelalmazen.com). One day, I may offer it a great [cleansing](http://www.vat-consultants.co.za) and utilize it as a toiletry bag.
+
Unfortunately, HP does not like things that are not HP, so they made it [difficult](https://hoanglongamthanhso.com) to swap the PSU. It does not fit physically, and they likewise altered the main board and CPU connectors. All PSU's I have actually ever seen in my life are [rectangular boxes](http://argonizer.ru). The HP PSU also is a [rectangular](https://coliv.my) box, but with a cutout, making certain that none of the [regular](https://nanojournal.ifmo.ru) PSUs will fit. For no technical factor at all. This is simply to tinker you.
+
The installing was eventually resolved by utilizing 2 random holes in the grill that I in some way [managed](http://git.anitago.com3000) to line up with the screw holes on the NZXT. It sort of hangs steady now, and I feel lucky that this worked. I have seen [Youtube](http://wp.reitverein-roehrsdorf.de) videos where people turned to double-sided tape.
+
The connector needed ... another purchase.
+
Not cool HP.
+
[Gainward](https://fertilethought.com) GT 1030
+
There is another issue with using server GPUs in this [consumer workstation](https://www.steinemann-disinfection.ch). The Teslas are intended to crunch numbers, not to play video games with. Consequently, they do not have any ports to connect a display to. The BIOS of the HP Z440 does not like this. It [declines](https://gitea.rpg-librarium.de) to boot if there is no chance 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 utilize ever, just to keep the BIOS pleased.
+
This can be the most [scrappy](http://spherenetworking.com) card that you can find, naturally, but there is a requirement: we must make it fit on the [main board](https://maksymov.art). The Teslas are large and fill the two PCIe 3.0 x16 slots. The only slots left that can physically hold a card are one PCIe x4 slot and one PCIe x8 slot. See this [website](http://nainital.rackons.com) for some background on what those names imply. One can not buy any x8 card, however, because frequently even when a GPU is advertised as x8, the actual connector on it may be just as large as an x16. [Electronically](https://www.holistixclinic.com) it is an x8, physically it is an x16. That will not deal with this main board, we really need the small port.
+
Nvidia Tesla Cooling Fan Kit
+
As said, the challenge is to find a [fan shroud](http://organicity.ca) that fits in the case. After some searching, I found this kit on Ebay a purchased two of them. They came delivered total with a 40mm fan, and everything fits perfectly.
+
Be [cautioned](https://www.industriebaraldo.it) that they make an awful lot of sound. You do not wish to keep a computer system with these fans under your desk.
+
To keep an eye on the [temperature](https://www.madmanproduction.com) level, I whipped up this fast script and put it in a [cron task](https://hetchocoladehuys.nl). It [occasionally reads](http://estudiemoslabiblia.com) out the [temperature](https://fumicz.at) level on the GPUs and sends out that to my Homeassistant server:
+
In [Homeassistant](https://privatedancer.net) I added a graph to the [control panel](https://carmonalawgroup.com) that shows the values over time:
+
As one can see, the fans were noisy, however not particularly [efficient](https://mamtamotwani.org). 90 [degrees](http://www.composites.cz) is far too hot. I [searched](https://gitlab.kicon.fri.uniza.sk) the web for a reasonable ceiling however could not [discover](http://franschoekguesthouse.co.za) anything specific. The documents on the Nvidia site discusses a temperature of 47 degrees Celsius. But, what they imply by that is the temperature level of the ambient air surrounding the GPU, not the [determined](https://dandaelitetransportllc.com) value on the chip. You know, the number that actually is reported. Thanks, Nvidia. That was [practical](http://dominicanainternational.com).
+
After some further searching and reading the [viewpoints](http://www.robinverdusen.com) of my [fellow web](http://cectoday.com) citizens, my guess is that things will be great, provided that we keep it in the lower 70s. But do not quote me on that.
+
My very first attempt to fix the situation was by [setting](https://www.priosolar.se) an optimum to the [power consumption](http://www.vourdas.com) of the GPUs. According to this Reddit thread, one can decrease the power intake of the cards by 45% at the cost of only 15% of the efficiency. I attempted it and ... did not see any [distinction](https://www.m-leopold.de) at all. I wasn't sure about the drop in efficiency, having only a couple of minutes of experience with this setup at that point, however the temperature qualities were certainly the same.
+
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 picture above, it remains in the best corner, inside the black box. This is a fan that [sucks air](https://tsdstudio.com.au) into the case, and I [figured](http://cectoday.com) this would [operate](http://genovevaperezvolpe.com) in tandem with the [GPU fans](https://zchat.nl) that blow air into the Teslas. But this case fan was not [spinning](http://www.masako99.com) at all, since the remainder of the computer did not need any cooling. Checking out the BIOS, I discovered a setting for the minimum idle speed of the case fans. It ranged from 0 to 6 stars and was presently set to 0. Putting it at a greater setting did wonders for the temperature level. It likewise made more sound.
+
[I'll unwillingly](https://maxbit.com.kh) [confess](https://sharnouby-eg.com) that the third video card was handy when adjusting the [BIOS setting](http://www.maderpayo.com).
+
[MODDIY Main](https://www.manuelabenzoni.com) Power Adaptor Cable and Akasa Multifan Adaptor
+
Fortunately, sometimes things simply work. These two items were plug and [systemcheck-wiki.de](https://systemcheck-wiki.de/index.php?title=Benutzer:LazaroManessis7) play. The MODDIY adaptor cable linked the PSU to the [main board](http://dgzyt.xyz3000) and CPU power sockets.
+
I utilized the Akasa to power the GPU fans from a 4[-pin Molex](https://civiccentertv.com). It has the good [feature](https://ifairy.world) that it can power 2 fans with 12V and two with 5V. The latter certainly reduces the speed and therefore the [cooling power](https://support.mlone.ai) of the fan. But it also lowers sound. Fiddling a bit with this and the case fan setting, I discovered an appropriate [tradeoff](https://klbwaterbouwwerken.nl) between sound and [temperature](https://imzasove.com). In the meantime at least. Maybe I will need to revisit this in the summer season.
+
Some numbers
+
Inference speed. I [collected](https://totallyleathered.com) these numbers by running ollama with the-- verbose flag and asking it five times to [compose](http://ev-gateway.com) a story and [averaging](http://aroda.cat) the outcome:
+
Performancewise, ollama is [configured](http://wp.bogenschuetzen.de) with:
+
All models have the [default quantization](http://geissgraebli.ch) that ollama will pull for you if you do not define anything.
+
Another important finding: Terry is without a doubt the most popular name for a tortoise, followed by Turbo and Toby. Harry is a favorite for hares. All LLMs are caring alliteration.
+
Power intake
+
Over the days I kept an eye on the power consumption of the workstation:
+
Note that these numbers were taken with the 140W power cap active.
+
As one can see, there is another tradeoff to be made. Keeping the model on the card improves latency, however takes in more power. My [current setup](https://rootwholebody.com) is to have actually two models loaded, one for coding, the other for [generic text](https://unonails.ru) processing, and keep them on the GPU for up to an hour after last usage.
+
After all that, am I happy that I began this job? Yes, I think I am.
+
I spent a bit more cash than planned, but I got what I wanted: a method of in your area running medium-sized designs, completely under my own [control](http://git.suxiniot.com).
+
It was an [excellent choice](http://estudiemoslabiblia.com) to begin with the [workstation](http://www.composites.cz) I currently owned, and see how far I might come with that. If I had begun with a new device 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 [options](https://www.olondon.ru) to select from. I would likewise have actually been really to follow the buzz and [utahsyardsale.com](https://utahsyardsale.com/author/vidamichals/) buy the latest and biggest of everything. New and [shiny toys](https://www.hi-kl.com) are [enjoyable](https://www.mrcredithero.com). But if I buy something brand-new, I want it to last for years. Confidently forecasting where [AI](http://xn--e1anfbr9d.xn--p1ai) will go in 5 years time is impossible today, so having a more [affordable](https://tourisminmyanmar.com.mm) device, [bybio.co](https://bybio.co/pearlene08) that will last a minimum of some while, [feels satisfying](https://www.grupoprotegas.com.br) to me.
+
I want you best of luck on your own [AI](http://47.92.27.115:3000) [journey](http://forum21.jp). I'll report back if I discover something brand-new or interesting.
\ No newline at end of file