Building a Supercomputer for a Senior Project

[TheSeether]

First off, this is going to be a personal supercomputer, so it wont be rows of towers like you think when you hear supercomputer, think of it more a rendering build. 

Well a little bit about me, I live in a little town called Shoshone Idaho, I live and work on a cattle/farming ranch (roughly 1000 acres of land). I do everything from running a tractor to branding and roping calves, and feeding 150 cows on a daily basis. Contradictory to my lifestyle, I love technology. When i was 13 I saved up my money, bought a broken dirt bike, fixed it, and sold it a year later making 1000 dollars after expenses. I saved up a little more working money and a couple months later I bought an ASUS G75 laptop for my 14th birthday. My dad said it was a the biggest waste of money and mistake in my life. I however I have used it everyday to do homework and everything else you can do with a good laptop for the past 3 years, and its the best investment I have ever made. ( I'm 17 As of last August).

 

Now on the subject of my senior project. As you can tell, my father was born and raised on a ranch just like I was, so he is very non-tech savvy and refuses to be when I try to teach him anything. He probably types 2 letters a minute and has to call me when he wants to use the keyboard. So as you would guess he was not very excited like i was when i told him i wanted to build a supercomputer for my senior project. He will back me up in everything I do, but to him im speaking a different language when i start talking about computers. I have a reference build for computer, and the cost it was close to 10 Grand. I nor do my family have 10 grand to spend on a computer. Sure if it were a new tractor or piece of farming equipment my dad would be happy to put 10 grand towards that because it would pay itself back. However he doesn't see any return in building a supercomputer, so i could get 700 dollars MAX if anything from my father, so there is the reason behind the sob story.

 

So what I came here for Is a Sponsorship(s) and a mentor to help me through my project. I have tried to reach out to multiple YouTube channels because this would be a cool story for a YouTube video and repay them for there help/sponsorship through a video on there channel, but i have came up empty handed on everyone that I reached out to. I would Ideally like to get each of the company's that I would request PC parts from (Intel,Asus,Nvidia,EK,Corsair,etc) explain there relationship to me and my senior project, therefore everyone involved gets a guess you could say credit, if you would, to helping me with this, as to lets say one company or person. My Mentor would just be someone that will say that they will help and guide me through the process of my senior project (building supercomputer).

 

Why I am doing this is because of where I came from and Idaho as a whole. I doubt anyone in this world thinks of technology when they think about Idaho. I'm not trying to change that but as some of you may know Idaho is not top notch in its education. In fact Idaho is 50th, that's right, the worst state in the nation for how many high school students go on to college. This is mainly because Idaho is a very agricultural state, and i know many people who just don't because they have to take up what there family did agriculturally before them. I want this story to go big. Not just for me but for education in Idaho. Maybe this project will bring new light and importance to a subject that frankly, people just want to sweep under the rug.

 

So please, anyone interested in helping me with sponsorship's anyone who is really tech savvy that would like to help me as my Mentor. Anyone who could help me or offer a sponsorship in getting my supercomputer built I would love to tell you more and would love your help. If I cant get any help, I have a plan B, which is downgrading from top of the line everything and the cheapest I can go under 700$ Budget. So please, anyone who could offer me a sponsorship, or someone who would be kind enough to be my mentor, please respond. I need all the help and exposure I can get to make this actually happen.

[Sauron]

I'd be glad to help you pick the parts, bu for the actual money you're in the wrong place I'm afraid. Also, don't expect Linus to personally reply here - he has a business mail that you can use to contact him for sponsorships and such. But be warned, I don't think he's even done anything like this and he will most likely refuse. If it's just for demonstration purposes, you could build a cluster with 20ish raspberry pis

[Decon]

what you don't realize is that you CAN build a supercomputer for $700 with used server hardware... I haven't spent more than 1k on hardware and I have ( IMO ) a monstrous setup by buying used server gear

 

don't see how building a PC is gonna be news to anyone though, if you seek notoriety you're much better off programming something or founding a robotics club...

[TheSeether]

I'd be glad to help you pick the parts, bu for the actual money you're in the wrong place I'm afraid. Also, don't expect Linus to personally reply here - he has a business mail that you can use to contact him for sponsorships and such. But be warned, I don't think he's even done anything like this and he will most likely refuse. If it's just for demonstration purposes, you could build a cluster with 20ish raspberry pis

No I wasn't looking for anything like Linus or like that. Just hypothetically speaking someone who works at a company, who would be interested in helping sponsor/fund my senior project. That would be great but I wasn't expecting anything of that nature to happen.

[TheSeether]

what you don't realize is that you CAN build a supercomputer for $700 with used server hardware... I haven't spent more than 1k on hardware and I have ( IMO ) a monstrous setup by buying used server gear

 

don't see how building a PC is gonna be news to anyone though, if you seek notoriety you're much better off programming something or founding a robotics club...

Yes I fully understand and realize that, thats why its my plan B in the first place. But would it be even close to a machine that cost 10x more? Well common sense it wouldnt. Not seeking notoriety but with how a sponsorship commonly works is that when a company or person sponsors something they pay or help for some portion of a project and they have there name written all over it so they get business exposure, that's why im trying to reach out to businesses or someone willing to fund my project, because frankly that's the only way I see this working.

[Decon]

Yes I fully understand and realize that, thats why its my plan B in the first place. But would it be even close to a machine that cost 10x more? Well common sense it wouldnt. Not seeking notoriety but with how a sponsorship commonly works is that when a company or person sponsors something they pay or help for some portion of a project and they have there name written all over it so they get business exposure, that's why im trying to reach out to businesses or someone willing to fund my project, because frankly that's the only way I see this working.

 

If you're serious about this and you would like advice, add me on steam @ Decon

 

 

as mentioned below I re-read your post; why do you even need this?

 

[Kalle Hänninen]

So you practically want really good computer? Cuz I don't understand what would you need it for.

 

And you can build cluster just by using BOINC, free of charge, but I doubt that you have the necessary skills to do so.

[TheSeether]

I would like to disprove your idea that you can't build a supercomputer from used server hardware... How much do you think a server grade 12 core AMD opteron CPU would cost?  I think you will find the results quite surprising...  

 

If you're serious about this and you would like advice, add me on steam @ Decon\

 

 

as mentioned below I re-read your post; why do you even need this?

 

 

 

So for my senior project I need a physical product of some sort. Some kind of product that I would make, that is related to a career field that I am interested in. Originally i was just going to build a computer and im done. School board said that there isint anything super amazing about that, why not a build a really good performance computer, and instead of having your Cardboard Tri fold, Get 4 or 6 monitors set up and use them as if they were your tri-fold. Another member of the school board suggested so people understand the performance gap, when i present my product to the school board I do a live benchmark of the computer im building vs one of the lab computers here at my school. For it to pass there expectations this computer should be able to have as much power as 10 or more of the computers at my school. The computers here at my school all have a really old Sapphire X1550 GPU, A 500 gig Western digital hard drive, gigabyte GA-945GCM MB, and an intel E5300 processor. So I was thinking of just running a geekbench for the processor, passmark for graphics, and just tell them or represent how long it would take to move an identical 300 gig file on both of the machines. So that's my requirements from my school board if you would. They said if i could make it capable of 10 of the computers here at my school that would pass, but 15 is where i should shoot for. So basically im shooting for 1000% to 1500% better scores on all the benchmarks and they would be happy.

[Decon]

So for my senior project I need a physical product of some sort. Some kind of product that I would make, that is related to a career field that I am interested in. Originally i was just going to build a computer and im done. School board said that there isint anything super amazing about that, why not a build a really good performance computer, and instead of having your Cardboard Tri fold, Get 4 or 6 monitors set up and use them as if they were your tri-fold. Another member of the school board suggested so people understand the performance gap, when i present my product to the school board I do a live benchmark of the computer im building vs one of the lab computers here at my school. For it to pass there expectations this computer should be able to have as much power as 10 or more of the computers at my school. The computers here at my school all have a really old Sapphire X1550 GPU, A 500 gig Western digital hard drive, gigabyte GA-945GCM MB, and an intel E5300 processor. So I was thinking of just running a geekbench for the processor, passmark for graphics, and just tell them or represent how long it would take to move an identical 300 gig file on both of the machines. So that's my requirements from my school board if you would. They said if i could make it capable of 10 of the computers here at my school that would pass, but 15 is where i should shoot for. So basically im shooting for 1000% to 1500% better scores on all the benchmarks and they would be happy.

so you want to build a PC to beat school PC's? Cause with those specs you listed just about anything you buy now will beat their PC's by so much its insane (especially for GPU it would be much beyond 1500% even on a $700 budget).  Still don't get how using off-the-shelf components to build a PC is a senior project though...  As I said earlier you'd be better off with programming something or building a robot if you want to do well on your senior project.

[unijab]

If you have the programming skills you could build a cluster of these

 

http://www.newegg.com/Product/Product.aspx?Item=N82E16813190005

[HippY]

What do you want to use the "super computer"?

Because the benching it once is extremely wasteful. I think if you would have a software to run that would be actually useful, than it would be a good experience.

 

I would turn it around. show that for $700/1000 you can build a PC that is several times as good as the PC-s there, so like a value/performance build.

Because lets be honest, building a supercomputer out of $10K that is 15x better than 1 really old PC is not a big thing.

It is like I want to prove that there are quicker cars out there by buying a BMW M3 and race it with Linus old lambo. 

Yeah, it will win... but I have not really proved any point am I not?

[TheSeether]

What do you want to use the "super computer"?

Because the benching it once is extremely wasteful. I think if you would have a software to run that would be actually useful, than it would be a good experience.

 

I would turn it around. show that for $700/1000 you can build a PC that is several times as good as the PC-s there, so like a value/performance build.

Because lets be honest, building a supercomputer out of $10K that is 15x better than 1 really old PC is not a big thing.

It is like I want to prove that there are quicker cars out there by buying a BMW M3 and race it with Linus old lambo. 

Yeah, it will win... but I have not really proved any point am I not?

Well selling once I'm finished with my senior project, most likely sell it, because il be going into basic training a few months after i graduate. Yeah it really isint, its just taking parts from a shelf and putting them together, im aware of that. My school board however already approved this, so I cant really change, and they are completely satisfied with what I pitched to them. I wasn't going for my project being the thing to bring Idahos education problems to light if that's what you were thinking, that was going to be possible really if a larger-ish company did a sponsorship for my project, probably learn about the Dont Fail Idaho project, and started pulling strings from there.

 

 

If you have the programming skills you could build a cluster of these

 

http://www.newegg.com/Product/Product.aspx?Item=N82E16813190005

How many do you think I would need?

[TheSeether]

so you want to build a PC to beat school PC's? Cause with those specs you listed just about anything you buy now will beat their PC's by so much its insane (especially for GPU it would be much beyond 1500% even on a $700 budget).  Still don't get how using off-the-shelf components to build a PC is a senior project though...  As I said earlier you'd be better off with programming something or building a robot if you want to do well on your senior project.

Sorry I thought I responded to you that day but obviously I didn't. Well our whole entire senior project revolves around a career field, and a product, to represent our time and effort invested in learning about that career field. Now my school board preety much said that same thing, would making a custom high performance computer actually a job? And if so is there a particular business? So I just told them about Xidax. They looked up Xidax, found out its literally just a company who builds high performance computers that are marked up 200%. They said your good to go. But because they already approved this project, I cant really change without going through the whole submit and get it approved process, and like I said they are 100% completely satisfied with what I pitched them, so I don't really see a motive behind changing my subject now.

[Decon]

Sorry I thought I responded to you that day but obviously I didn't. Well our whole entire senior project revolves around a career field, and a product, to represent our time and effort invested in learning about that career field. Now my school board preety much said that same thing, would making a custom high performance computer actually a job? And if so is there a particular business? So I just told them about Xidax. They looked up Xidax, found out its literally just a company who builds high performance computers that are marked up 200%. They said your good to go. But because they already approved this project, I cant really change without going through the whole submit and get it approved process, and like I said they are 100% completely satisfied with what I pitched them, so I don't really see a motive behind changing my subject now.

I think you made a mistake with your project choice then to be honest, cause AFAIK there's basically no way to get a sponsorship unless you already have some recognition (popular modder, record-holding OCer, big youtube channel etc...)  Companies aren't just gonna drop 10k on someone to "build a fast PC for a science project"  It still makes no sense what you are trying to achieve here; why not build a 700$ system and call it a day (that would certainly fit the "build a PC" career path)

[dalekphalm]

Sorry I thought I responded to you that day but obviously I didn't. Well our whole entire senior project revolves around a career field, and a product, to represent our time and effort invested in learning about that career field. Now my school board preety much said that same thing, would making a custom high performance computer actually a job? And if so is there a particular business? So I just told them about Xidax. They looked up Xidax, found out its literally just a company who builds high performance computers that are marked up 200%. They said your good to go. But because they already approved this project, I cant really change without going through the whole submit and get it approved process, and like I said they are 100% completely satisfied with what I pitched them, so I don't really see a motive behind changing my subject now.

I agree with @Decon, doesn't seem like you're going to get the sponsorship you'd like.

 

Have you tried reaching out to local Idaho Computer Repair/Builder businesses? I'm sure there are tons of Mom and Pop repair shops. You might be able to get them to donate a single part here and there - or one of the bigger chains in a bigger city might help sponsor the whole build.

 

That's infinitely more probable then getting YouTube Channel sponsorship from Linus or other Tech Channels.

 

Have you actually tried reaching out to Linus? Sending him/Luke/Nick a PM, tagging them in a post, sending their business email a message, etc.

 

I think you made a mistake with your project choice then to be honest, cause AFAIK there's basically no way to get a sponsorship unless you already have some recognition (popular modder, record-holding OCer, big youtube channel etc...)  Companies aren't just gonna drop 10k on someone to "build a fast PC for a science project"  It still makes no sense what you are trying to achieve here; why not build a 700$ system and call it a day (that would certainly fit the "build a PC" career path)

I agree. I'm just really confused as to the purpose of the "Supercomputer". Because, let's be honest. It's not a supercomputer. Unless you drop $100,000+, it's still not a supercomputer. A "Fast PC", or a "Fast Server" also aren't supercomputers.

 

What is the purpose of the Server? What will it do - if anything - besides running a benchmark suite? Will you demonstrate some usage scenarios of the Server? Eg: Rendering some video or 3D animations? Crunching simulations? etc.

 

PC Building definitely is a career path - but it's not a particularly lucrative one for most people in the business. For every Xidax, there are 1000 "Joe's Computer Repair", where they make razor thin margins and can barely keep the lights on.

 

If you are passionate about PC building - especially "Boutique" PC's (That is - high end, flashy, custom watercooloed, nice cable management, making them look "tricked out", etc), then you can certainly work towards this. But I would only do so if it's a huge passion - AND as long as you keep in mind that this field - much like being an artist or a musician - is a very tough and very competitive industry.

[Guest]

what you don't realize is that you CAN build a supercomputer for $700 with used server hardware... I haven't spent more than 1k on hardware and I have ( IMO ) a monstrous setup by buying used server gear

 

don't see how building a PC is gonna be news to anyone though, if you seek notoriety you're much better off programming something or founding a robotics club...

 

Are you willing to share some info on that build? I'm just curious as hell now.

 

 

- snip -

 

You basically screwed yourself, buddy. Building a PC that outperforms an E5300 is not a problem. Building a cluster is not a problem either. But on a budget of $700? It's gonna be goddamn impossible. 

 

Let's say, you can get lots of cheap LGA775 motherboards, there's a lot of them for pennies, and build a cluster. Pack them with Q6600s, they're also cheap. And here you run into a giant wall. What about powering them up? Cooling? Noise? 

 

If it was my task, I'd go with an old 2-4 CPU Supermicro motherboard (in case of 2 CPUs I'd probably be able to fit it in a standard ATX case) and pack it with LGA1366 Xeons. But that's a) overkill b) I wouldn't really be building anything, just putting off-the-shelf components together. And selling it would probably prove to be goddamn impossible. Also, I'd basically have the PC equivalent of an old Mac Pro. 

 

Unless by "personal supercomputer" you mean the likes of 18-core Intel Xeon paired with 128 gigs of DDR4 RAM and quad Titans... To this I've got to say - good luck. 

 

You'd be better off if you were to build a custom case, a nice watercooled computer or whatever else. I'd be willing to chip in and donate some parts or whatever. But this is - at least in my opinion - a lost cause. 

 

Also, as far as personal feedback is concerned - you aren't going to win anyone over with a sob story. If someone is going to sponsor you, they need a reason to do so. "We helped a kid complete his school assignment" isn't going to have any marketing value, because a lot of people is going to knock at their doors and come with lots of stories like "I'm a high-schooler too, can I have some money/parts?" or "My family is poor, can I have a PC? It's good PR for you!" - and companies exist mostly to sell things. Neither this is a story that will go big, mostly because people are doing this every single day. You aren't doing anything new nor innovative. It's been done bajilion of times, and the fact you are doing this isn't really changing anything. Now, with a nice custom PC case you could probably get some publicity. But with a powerful computer? Unless it is something that's never been done before - no way. 

 

If I were you, I'd change my subject. If you're going to stick to it - best of luck. You're going to need it. 

[Decon]

Are you willing to share some info on that build? I'm just curious as hell now.

$400 USD gets you a matched pair of used 8 core LGA 2011 Xeons: http://www.ebay.com/itm/Pair-Intel-Xeon-E5-2670-2-6Ghz-20MB-Cache-8-GT-S-TDP-8-Core-Processor-CPU-SR0KX-/381465228319?hash=item58d119741f:g:fYEAAOSwo6lWKDYZ

 

Last month I got a pair of 6 core / 12 thread Xeons for $140 shipped

 

Pretty crazy IMO

 

Warning Ebay link above solely for reference purposes

[Guest]

$400 USD gets you a matched pair of used 8 core LGA 2011 Xeons: http://www.ebay.com/itm/Pair-Intel-Xeon-E5-2670-2-6Ghz-20MB-Cache-8-GT-S-TDP-8-Core-Processor-CPU-SR0KX-/381465228319?hash=item58d119741f:g:fYEAAOSwo6lWKDYZ

 

Last month I got a pair of 6 core / 12 thread Xeons for $140 shipped

 

Pretty crazy IMO

 

Warning Ebay link above solely for reference purposes

 

I'm also assuming one of dual-CPU Supermicro motherboards and lots of RAM? 

 

On one hand I want to build this right now... On the other hand - I'm gonna think about that.

 

Also, have you tried gaming on that monster?

 

Thanks!

[Decon]

I'm also assuming one of dual-CPU Supermicro motherboards and lots of RAM? 

 

On one hand I want to build this right now... On the other hand - I'm gonna think about that.

 

Also, have you tried gaming on that monster?

 

Thanks!

no I only use it for VM's

 

I'm running both processors on an X8DAH+-F with 48gb ECC DDR3 (1/3 slots filled)

[Guest]

no I only use it for VM's

 

I'm running both processors on an X8DAH+-F with 48gb ECC DDR3 (1/3 slots filled)

 

That's what I thought. I'm actually curious how it would perform as a gaming machine, although I'd build mine for video editing/some rendering work/vm duties/general purpose server anyway. 

 

Thanks a lot!