[WIP] Hardware comparison with prices

[TheVirus]

I created a site that I'm currently working on, but it's at a point where I'm willing to take in advice on new features or bug fixes (still plenty to fix).

http://www.thevir.us/bitcoin

It currently uses NewEgg for prices and uses its RSS feeds, which I am going to change since it's not 100% perfect. I plan on adding retail sites for different countries, so please feel free to paste your own recommendations and I'll work on adding them in.

How it works:
Once a day it gets the Wiki hardware comparison page and caches it locally. It then parses that and lists it in a table with the prices available (when you select a card series). I cached the content locally so as to not cause any extra hits on the wiki as well as keeping it as fast as possible.

Future changes:
Comments/ratings
More retail sites
Compare multiple cards
$/MHs column

The reason I did it is because it's something the Wiki was lacking and would be difficult to add as prices change almost daily. While the Wiki data is ~24 hours old, the pricing data is live. I'm open to comments/complaints/suggestions.

Thanks.

[Sukrim]

European prices (mostly Germany, Austria, partly Poland + UK): http://geizhals.at/ (english version: http://skinflint.co.uk/)

I suggest you also to add a graph like I did once:

Field for entering price per kWH

X-Axis: Time in days

Y-Axis: (Hashes solved after x days [Hashrate per second*60*60*24*number of days]) / (Money paid after x days [Initial price + Watts/1000*24*price per kWh*number of days])

The card that is on top there at day X will give you the most hashes for your money invested, so if you stop mining at this day in the future, you should get as many of these cards at the price you have up there as possible.

WARNING: This graph would assume a constant difficulty! You can easily adjust it by letting people input an additional "network growth per day" value (currently something around 4%!) and reducing the "Hashes solved after x days" by this percentage. This however might not give true estimeates over very long time, as the network cannot grow arbitrarily large, so both "extremes" (endless expansion and 0 expansion) are not 100% realistic.

I can show you some examples, if you need them, how graphs like this could look like.
Also you'll need fairly correct hash rates + Watt values for this to make sense!

[TheVirus]

European prices (mostly Germany, Austria, partly Poland + UK): http://geizhals.at/ (english version: http://skinflint.co.uk/)

I suggest you also to add a graph like I did once:

Field for entering price per kWH

X-Axis: Time in days

Y-Axis: (Hashes solved after x days [Hashrate per second*60*60*24*number of days]) / (Money paid after x days [Initial price + Watts/1000*24*price per kWh*number of days])

The card that is on top there at day X will give you the most hashes for your money invested, so if you stop mining at this day in the future, you should get as many of these cards at the price you have up there as possible.

WARNING: This graph would assume a constant difficulty! You can easily adjust it by letting people input an additional "network growth per day" value (currently something around 4%!) and reducing the "Hashes solved after x days" by this percentage. This however might not give true estimeates over very long time, as the network cannot grow arbitrarily large, so both "extremes" (endless expansion and 0 expansion) are not 100% realistic.

I can show you some examples, if you need them, how graphs like this could look like.
Also you'll need fairly correct hash rates + Watt values for this to make sense!

Thanks for that, shouldn't be too difficult once I get the rest of these kinks with Newegg worked out. I can grab the current difficulty via: http://blockexplorer.com/q/getdifficulty Adding pricing for kWH will be a bit tricky but should be doable if the info is there. I am going to setup a database to store all of these values locally instead of grabbing the data remotely for each search. I will have it update the prices daily.

[Sukrim]

The beauty is, you won't need difficulty at all:
Either do a more "dirty" version by just assuming difficulty is adjusted daily (-4% of the previous day each day)
or
Do a more "clean" version by letting people input a steady difficulty jump, then reduce 14 days by this amount and apply the percentage accordingly to the "total hashes calculated".

As the absolute values are not that interesting anyways, you don't even need to label your Y-Axis - the curve on the top is the best one at that hashrate, price and wattage.


If you have some nice, pretty and meaningful graphs like this set up, I bet people will start to run for your website!

[TheVirus]

Added $/MH (cost per megahash/second).
Refined Newegg search.
Added foundation for graphs.
Setup the database to hold card/price data.

Moving search data to individual files. The search script (in the long run) will run once a day to update the prices. All prices will be stored locally in the DB so that it doesn't need to load external pages each time.

[TheVirus]

Added refresh button.
Added 'All' to the comboboxes.
Added average prices and cost per MH to main table.

[BombaUcigasa]

I like this site better: http://www.bitminer.info/

It's practical, it has automated computing, and relevant and clear data on a nice and clean layout. Yours has nice sort-tables that seem to fail on $/MH column and maybe some cool charts I can't see yet.

[TheVirus]

I like this site better: http://www.bitminer.info/

It's practical, it has automated computing, and relevant and clear data on a nice and clean layout. Yours has nice sort-tables that seem to fail on $/MH column and maybe some cool charts I can't see yet.

My site isn't even close to finished. The reason the $/MH fails is because either the price isn't listed or the MH/s isn't listed. I plan on adding more sites to gather pricing data from so people from outside the US can get an idea of what to do.

I plan on gathering data for each card, even if I have to do it manually (things like wattage are unlikely to change). The graphing is almost done, I got my listbox up where you can choose the cards to compare and all I need to do is have it harvest the data from the database and graph it.

I have a lot of ideas in store for the site and hope that a lot of people find use in it.

[TheVirus]

Got graphs to a stable point. Select a few cards and click Refresh and it'll calculate the cost of running the card over a period of 7 days. I have a lot more planned for it, including other graphs, and will be working on implementing them in the next few days.

[BombaUcigasa]

I suspect you are an American because you use reverse logic measurements for Mhash/$ (like you do on your car consumption measurements), the charts work nicely now, except for the power per dollar awkwardness.

The graphing part I couldn't get it to work, nothing happens.

[TheVirus]

I suspect you are an American because you use reverse logic measurements for Mhash/$ (like you do on your car consumption measurements), the charts work nicely now, except for the power per dollar awkwardness.

The graphing part I couldn't get it to work, nothing happens.

I'm American, so I am terrible at math The graph shows how much MHash/day per $ you should expect. The price is determined by the cost of the card, plus the cost of power (watt consumption). It uses a static value of $0.10 per kWh. I will allow the user to enter in their own as well as change the currency type, I just need to work on mining data from other sites.

The charts on the first two tabs are $/MH, which is backwards, and I'll fix that. I'm not sure why the graph doesn't work for you. Select a few cards from the selection box (click the +) and then click Refresh and it should draw the graph.

[cothoms]

Suggestion:  Also include provision for multi-GPU setups, given that wattage doesn't linearly increase with additional GPUs on a mobo.

[TheVirus]

Suggestion:  Also include provision for multi-GPU setups, given that wattage doesn't linearly increase with additional GPUs on a mobo.

Wattage is determined from the wiki. If there is more than 1 entry for the model, then I take the average (ignoring null values).

Edit: Seems graphs are broken for some reason. Data returning looks correct. Will work on it now.

[TheVirus]

Think I fixed graphs. I switched from jqPlot to flot since I've used flot in the past and seems to be able to do what I want easily. Going to work on adding in more graphs to show different types of data.

If you select a card with no known price then it doesn't get drawn. I'm going to write a quick check tomorrow and have it show up in the legend with no data so the user doesn't think it's broken.

[SlipperySlope]

I run chrome and found the site unreadable - sorry.

[TheVirus]

I run chrome and found the site unreadable - sorry.

I've only tested in Firefox, since that's the browser I currently use. I'll see what I can do for Chrome.

Edit: Yep, 100% broken in Chrome. Not sure why.

Edit2: Appears to be flot that breaks it.

Edit3: It was an invalid JS line. It works now.