Flex your computing power!
suchflex.com
suchflex.com
I understand the 'sharing economy' desire to make use of underutilized resources, but this doesn't seem like an economically feasible way of doing so. The model works for Uber/Lyft: cars are a relatively high upfront cost compared to the cost of gas, but computer hardware is often less expensive upfront than the electricity costs of running it for a year. Additionally, much of the economic value in a service like Uber or Lyft is provided by the driver, not just the use of the car. In this service, the user doesn't provide any value, in fact, they're using up cycles/space that could otherwise be monetized.
For this the suchflex guys earn Gridcoins, which they can sell directly on the market and convert to money. But a user could leave out the middleman alltogether and just mine Gridcoins (or alternatively other cryptocurrency but for that you need ASICS) themselves.
This is harder for the company itself to do, because if they just hire people to go into libraries, universities etc. to install mining bots they might be criminally liable. "Uber for CPU cycles" seems like a less felonious enterprise than installing malware on public-use hardware.
My electricity rate is is 15.7 cents per kw-hour. During typical usage (MS Office, web browser, programming), my Intel 6-core desktop (without LCD monitor on) draws about 150 watts.
For back-of-the-napkin estimates, let's round the kwh cost up to 16 cents and the wattage to 300 watts (to cover scenario of some of cpu cores being 100% pegged). The electricity cost of 24x7 for one year would be ~$413.
What remains would be bandwidth costs -- if any. I like many others have Verizon FiOS and even if others have Comcast or ATT, there's no obvious residential bandwidth costs I can think of to calculate. Maybe... if the homeowner wants to upgrade the speed from 75GB@$99/month to 150@$199/month because the he wants to download the datasets faster. That extra $100 wouldn't have been spent for plain web browsing. So conceivably, that would be $1200 per year. What we don't know is how big the datasets are that must be downloaded. I assume the upload size would be minimal because the compute tasks appear to be variations on "y_output = computecombinationsmontecarlobruteforce(x_input)." The y_output answer would usually be order-of-magnitude(s) smaller than x_input.
Assuming there are no extra bandwidth costs, it would be hard for a company to buy computer hardware for less than a homeowner's $413/year electricity cost.
Perhaps suchflex's particular business model is financially wrong. However in general terms, it does seem possible to find a monetizing sweet spot of computing tasks that takes advantage of the idle and wasted resources of existing home computers. However, if the homeowner has to buy extra hardware that was only dedicated to suchflex, that's probably where the economics won't make as much sense.
That's the issue though, this wouldn't be similar to your typical usage. Instead, if they're using your GPU to train neural networks, it'll be running close to or at full capacity.
I realize that you rounded the costs up, but lets just look at the costs of a GPU often used for machine learning - Nvidia GTX 980 TI. According to Nvidia, it draws 250W under load which according to your figures would result in a yearly cost of roughly $344. That's just for the reference card, a typical card that a consumer would purchase would draw even more. You can buy a 980 TI for a little more than $400. That doesn't even begin to look at hardware actually designed for commercial and research applications.
I think that it's possible to find a way of monetizing computer resources, however, I think it has more to do with arbitraging differences in electricity costs. Suchflex's model certainly wouldn't work where I live (electricity costs in NYC are roughly 20 cents per kw-hour) but parts of the US are under 10 cents. I could see a company attempting to profit from these differences by setting up hardware in a cheap state and negotiating a favorable electricity rate. Heavy computation could then be done on these networks for significantly less than it could in New York or California.
In summary, the value of a consumer's unused computer has more to do with their electricity rate than their hardware.
I don't see how the calculations support that.
For example if we use your worse case scenarios of an entity (such as Suchflex) that had its own datacenter in a 20 cent kwh region and the crowdsourced homecomputers in a 9 cent kwh region, that's a difference of 11 cents.
If we round up the energy usage to 600 watts (pc + GPU), that 11 cents is an annual difference of ~$578. However, for Suchflex to even run computations at all on their own hardware -- whether its 20 or 9 cents -- they have to spend capex of ~$1500 of motherboard+cpu+gpu. That's the $1500 the homeowner already spent for his own purposes. Therefore, Suchflex can redirect $1500 to pay commissions/awards/etc towards pure computations instead of buying their own depreciating hardware (which includes buying/renting the physical datacenters to hold it all).
It seems like the homeowner's hardware is a very significant part of the arbitrage/monetization equation. Yes, there is also potential arbitrage in regional differences of electricity rates. However, the greater arbitrage (at least the first 3 years) is the unused time on residential pc that would have been wasted. That "unused time" arises from computer hardware that was already purchased for other purposes than Suchflex.
My suggestion was not that an entity use crowdsourced home computers, rather that it would be more efficient for a company to setup their own hardware and rent CPU cycles that way. The big difference is that Suchflex is limited to using hardware that consumers regularly purchase, whereas a company could use significantly more energy efficient setups and negotiate a better electricity rate. This is essentially what AWS already offers. Additionally, if you already have to transmit everything remotely, there's no need to stay in the US. Iceland offers rates around 4.3 cents. I chose the 980 TI for my example because it's about as close to perfect as you can find for this scenario while sticking with consumer grade hardware, average setups would be much worse.
My general point is that I don't think Suchflex's model is viable unless, as pliny mentioned, you have access to free electricity through some less-than-legal means (or you live in Iceland).
I think it's theoretically possible for electricity costs to overwhelm hardware costs but so far, I haven't seen any numbers that make this disparity obvious. Some example AWS costs[1]:
g2.2xlarge is $0.65/hour
g2.8xlarge is $2.68/hour
More analysis would be required to see if particular computation tasks can done 15x faster on AWS optimized instances than the unoptimized residential computers ($0.65/$0.043==15x).
Without concrete spreadsheet of tasks, performance runtimes, and cloud costs, I still don't see obvious evidence that AWS (or Google Cloud) will be more cost efficient than unused home computers.
Spot instance prices are typically far less - for g2.2xlarge they average around $0.1/hr - https://ec2price.com/?product=Linux/UNIX&type=g2.2xlarge®...
And note that customers can't run arbitrary or secure workloads with this proposal - they just want to mine crypto on your hardware and give a small percentage of returns to you as rent.
When theDAO was launched, I toyed with the idea of an etherium GAPP similar to Flex but for anything - but it would be a hell of a lot of effort to build, and I'm not sure there is demand - people won't bother for $10/week, especially if it takes up a lot of their HD space, bandwidth and makes their GPU take off (noise).
$480/year for a $400 video card? $240/year for 3TB of disk space?
No discussion of bandwidth.
No discussion of security.
No discussion of liability.
Other concern is that price for storage isn't linear and multiple small packages are more worth than single big. So I could use my 1tb to get 2x 500GB and get 10usd instead of 8usd.
joining it with above server I should be able to get 10usd per TB, so 5 x 6 x 10 = 300USD per month - 155 USD for server = 145USD profit monthly?
PS: site is hosted on heroku
ps2: you can find extra page in google cache: http://webcache.googleusercontent.com/search?q=cache:PpTl0nt...
I also wonder how they calculate the money you get just for having certain hardware. It says a 6 core 3.2Ghz CPU pays out $28/month but what if I have an average of 70% CPU usage the whole time (unrealistic, but still what if)? Do they pay me still $28 even though they only get 30% of what they could get or do they reduce the payment or even worse does the software just take the whole CPU for it self and make the rest of the computer unusable?
I've seen some similar projects (don't remember the names) but none of those actually answered only one of these questions.
The second part is based on Gridcoin, a cryptocurrency that rewards BOINC computations in a decentralized way. So you do all kind of BOINC projects like SETI@home, Rosetta@home (folding proteins),... and the newly generated coins by inflation get distributed according to research done by each user. Again, the important part is that there is no central authority giving out the coins, but each client queries the Boinc projects individually to find out how much research each user has done and the clients then come to a consensus about this, which is stored in the blockchain and is then the basis for the reward each client gets when staking the next time.
Suchflex is basically just a platform built on top of the Gridcoin platform for these "earn money by doing research" projects and simple cryptocurrency mining on the other hand.
Pure PoS cryptocurrencies cannot be mined by suchflex.
Then I was wrong, does suchflex also mine ETH then?
It's pretty easy to set up ETH mining yourself and pocket all the money, but that's probably not the audience this software is intended for (Windows users with access to "free" electricity).
Here's my !stats from the #gridcoin IRC channel:
fediverse> your next staked block will get you: 17 GRC (research owed) <fediverse> per day, average: 5.7 GRC <fediverse> (BTC (per day): 0.00005062 / $ (per day): $0.03 / $ (per 30 days): $0.88 / RAC (estimated): 136287.5 / magnitude (estimated): 28.60)
There's NO way that you're going to make $40/month with a 980 crunching BOINC projects, at least not with gridcoin. Gridcoin is $0.50 - $0.75, depending on where you look, and even the best rigs won't make more than 100 a day at best.
If so, I'm pretty certain that it won't be economical due to the power usage. Great for teenagers with gaming machines that their parents are paying the electricity bill on, and that's about it.
I don't get how they set such a big price for that.
In any case I do agree their numbers look really high (they need to factor in redundancy and their own margin) and could very well turn out to be either a VC cash grab or an outright scam (malware)
I'm curious how the technique presented in "Reusable Garbled Circuits and Succinct Functional Encryption" https://eprint.iacr.org/2012/733.pdf could potentially be utilised in such a system. I'm assuming it currently isn't possible due to computation overhead?
Many of these workunits are optimization problems without a clear solution and rely on algorithms using randomness, stuff like protein folding and bitcoin mining are examples of this. The nice thing about those is that you can send them out to as many clients as you want and just pick the best answer.
On a side note: I found the website on my phone really annoying to read since lots of paragraphs are animated and it keeps slowly sliding from left to right/right to left every time you scroll over it again.
EDIT: Talking about this http://i.imgur.com/JVtcTKg.gif
Just to confirm, regarding your second point: you found the subtle animation unpleasant? Thank you very much for your feedback.
I'm used to sites having transitions or animations as you scroll, but usually that's only happening once.
"Maximize your earnings with a SuchFlex Pro Account"
Couldn't sound like a sketchier pay money to get money scheme if they wanted it to.
Then again, this could also be done without the additional comission by just running Gridcoin.
Was curious to see what operating systems are supported.
Currently, Windows OS is supported. Mac and Linux support will be released early this Fall. Thanks