Author

Topic: [XMR] Monero - A secure, private, untraceable cryptocurrency - page 771. (Read 4671575 times)

full member
Activity: 141
Merit: 100
Hrmmm... in my opinion, the "only keep blockchain and run the latest daemon" is a not how these networks were initially designed / envisioned - i.e.,  a full node is one that mines - it can create its own blocks. IMO, A non-mining node is like an auditing relay - for instance, a non mining node will still play a part in the hardfork vote (if I understand things correctly). And if one transacts on a non-mining node, they're also participating more than just keeping a copy.  Indeed, there's a full blown discussion raging about what a node is and how many are needed somewhere in bitcointalk.

"other services outside the box" not entirely sure what you mean. The pool software definitely needs to run on the same silicon.

Welp, I'm designing such a device. It won't be pretty, as they will be handmade (you can do ANYTHING with a dremel!) . I was asking about the SSD because you can get much larger HDs with a HDD, and I'm imagining these devices will just sit next to your router for years, so ideally a 10 year device will be at least 500 gigs.

The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

Thanks for the clarifications, as I am just an average end-user lacking in-depth knowledge of "how it works". When I was talking about "services" I meant the merchants, whose primary task is handling transactions. If the "boxed" node is cheap, easy-to-setup and has enough power to handle thousands of transactions - I think it might receive popularity. Non-mining node can be also purchased by enthusiasts to support the network health. The key could be in comparing the yearly 24/7 electricity costs of a regular PC (or VPS subscription) VS purchase + running costs of a boxed node.

Now I see - you are trying to build the full node with mining capability. Which is a very complicated task indeed )) Keep up the good work!
full member
Activity: 201
Merit: 100

Agreed.  And with a great final statement:
Quote
One advantage is now Big Brother can't blame me for creating the strongest anonymity :) They can go after you instead :P
legendary
Activity: 1260
Merit: 1008
The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

The best solution for a 10 year low- or no-maintenance device is to go completely fanless. Forget the GPU unless you can find something decent that is fanless and can be mounted in a fanless chassis (which requires very low thermal output) and instead use a high efficiency CPU with good hash rate like the one we talked about on IRC (hash rate is roughly 120 for 15 watts, possibly less) with a brick PSU. It might require maintenance if the HD dies, but that's about it. Switching to SSD would reduce that. Even with with HD you can't do it $300 though. Not quite yet, but maybe next year.

Unfortunately what kills the price here is that the really cheap CPUs (Celeron, Atom, etc.) can't mine XMR well, but many of them can mine AEON. If you accept a much lower hash rate and efficiency you can still do it, and probably come in below $300 for the whole system.


Indeed, fanless would be the way to go, but yeah those devices we chatted about on irc really provide minimal hash for the buck IMO re: xmr. The 750 tis only pull 33 watts for 250 h/s, and the 750 (1 gb version) even less so for 200 h/s.

That's the same or likely better (since you also have to add the rest of the system overhead for the GPU system) efficiency for the CPU without the need for fans and a second device (size, assembly effort, etc). It is half the total hash rate, but that isn't really huge difference in terms of making a contribution to the network with a low-maint, cheap-to-operate device. Personally I'd rather have something that sits there completely silent for 10 years with zero attention from me than twice the hash rate. If I want more hash I can buy two. But I agree that is a value judgement.

Still it is hard to build that for an attractive price point <$300, so kind of irrelevant right now.

Quote
as it stands, i might be able to get away with fanless. I'll have to run tests once its enclosed.

I doubt you can get 33 W GPU, CPU, HD, etc in a box with no fan. I don't even know whether the GPU can run fanless on its own. If I search for fanless 750-ti all I see are water cooled things (EDIT: I found some with heat pipes and massive heat sinks, but you still have to remove the heat from the box). Though maybe the low power usage of cryptonight mining means a custom built low-efficiency cooling solution could work. I still doubt you will get it in a box. The best you can likely achieve is a well designed box with one quality, relatively expensive, fan (which might last several years) that draws air across multiple devices for cooling. Even that would be a nice design accomplishment.

Quote
that and those devices are already built

I think you can get some high efficiency CPUs that are on raw boards or possibly even socketed where you can design and build your own box. When I've looked for that I've generally found them even more expensive though, so I disregarded it.

we shall see. I'm close. I finally got the slew of M3X0.50 screws in to mount everything and some nylon washers. Thank god (or, well, humanity, or specifically whatsisface that runs wikipedia) for the wikipedia computer mounting screws page.

I neglected to mention that I have at least 2 750 ti boards where the manufacture fan assembly has been removed, because the configuration of the pci-eX1 next to the pcieX16 makes it impossible to get a riser in between with that bulky thing. With a fan just blowing over the area, its at a comfortable 53C, which is about 8C over a card with the fan still on them.

though im really tempted to just flip the orientation of the GPU card, cut a hole in the case, and cement on one of them copper dell server heatsinks.
legendary
Activity: 3836
Merit: 4969
Doomed to see the future and unable to prevent it

i friggin hate reddit sometimes. Impossible to easily find where the latest is on that friggin thread.

I completely agree! Just following that conversation was a serious PITA.
legendary
Activity: 1260
Merit: 1008

i friggin hate reddit sometimes. Impossible to easily find where the latest is on that friggin thread.
legendary
Activity: 3836
Merit: 4969
Doomed to see the future and unable to prevent it
legendary
Activity: 2968
Merit: 1198
The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

The best solution for a 10 year low- or no-maintenance device is to go completely fanless. Forget the GPU unless you can find something decent that is fanless and can be mounted in a fanless chassis (which requires very low thermal output) and instead use a high efficiency CPU with good hash rate like the one we talked about on IRC (hash rate is roughly 120 for 15 watts, possibly less) with a brick PSU. It might require maintenance if the HD dies, but that's about it. Switching to SSD would reduce that. Even with with HD you can't do it $300 though. Not quite yet, but maybe next year.

Unfortunately what kills the price here is that the really cheap CPUs (Celeron, Atom, etc.) can't mine XMR well, but many of them can mine AEON. If you accept a much lower hash rate and efficiency you can still do it, and probably come in below $300 for the whole system.


Indeed, fanless would be the way to go, but yeah those devices we chatted about on irc really provide minimal hash for the buck IMO re: xmr. The 750 tis only pull 33 watts for 250 h/s, and the 750 (1 gb version) even less so for 200 h/s.

That's the same or likely better (since you also have to add the rest of the system overhead for the GPU system) efficiency for the CPU without the need for fans and a second device (size, assembly effort, etc). It is half the total hash rate, but that isn't really huge difference in terms of making a contribution to the network with a low-maint, cheap-to-operate device. Personally I'd rather have something that sits there completely silent for 10 years with zero attention from me than twice the hash rate. If I want more hash I can buy two. But I agree that is a value judgement.

Still it is hard to build that for an attractive price point <$300, so kind of irrelevant right now.

Quote
as it stands, i might be able to get away with fanless. I'll have to run tests once its enclosed.

I doubt you can get 33 W GPU, CPU, HD, etc in a box with no fan. I don't even know whether the GPU can run fanless on its own. If I search for fanless 750-ti all I see are water cooled things (EDIT: I found some with heat pipes and massive heat sinks, but you still have to remove the heat from the box). Though maybe the low power usage of cryptonight mining means a custom built low-efficiency cooling solution could work. I still doubt you will get it in a box. The best you can likely achieve is a well designed box with one quality, relatively expensive, fan (which might last several years) that draws air across multiple devices for cooling. Even that would be a nice design accomplishment.

Quote
that and those devices are already built

I think you can get some high efficiency CPUs that are on raw boards or possibly even socketed where you can design and build your own box. When I've looked for that I've generally found them even more expensive though, so I disregarded it.
legendary
Activity: 1260
Merit: 1008
The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

The best solution for a 10 year low- or no-maintenance device is to go completely fanless. Forget the GPU unless you can find something decent that is fanless and can be mounted in a fanless chassis (which requires very low thermal output) and instead use a high efficiency CPU with good hash rate like the one we talked about on IRC (hash rate is roughly 120 for 15 watts, possibly less) with a brick PSU. It might require maintenance if the HD dies, but that's about it. Switching to SSD would reduce that. Even with with HD you can't do it $300 though. Not quite yet, but maybe next year.

Unfortunately what kills the price here is that the really cheap CPUs (Celeron, Atom, etc.) can't mine XMR well, but many of them can mine AEON. If you accept a much lower hash rate and efficiency you can still do it, and probably come in below $300 for the whole system.


Indeed, fanless would be the way to go, but yeah those devices we chatted about on irc really provide minimal hash for the buck IMO re: xmr. The 750 tis only pull 33 watts for 250 h/s, and the 750 (1 gb version) even less so for 200 h/s.

as it stands, i might be able to get away with fanless. I'll have to run tests once its enclosed.

that and those devices are already built. whats the fun in that?
hero member
Activity: 1874
Merit: 840
Keep what's important, and know who's your friend
With the proposed addition of Confidential Transactions to the Monero core (hopefully some day)...

How do you find outputs to mix with on the block chain if you don't know the amounts associated with those outputs?

If you are mixing with other outputs on the blockchain you don't necessarily have the "Amount key" to determine if the amounts match yours.

Hmm, after thinking about it a bit if you don't know what the amount associated with an output is on the blockchain, you could essentially mix with all outputs instead of like-amount outputs...hmm

Does that sound right?

I haven't read the latest paper by shen but this is one thing that I would like to hear an answer to.



I would imagine, all the coins will pool together in a block and then mix all of it.  Afterwards, send the coins to the corresponding parties each sender wants to send too. It wouldn't necessarily matter how many coins are being sent, and wouldn't need to know how much coins you are mixing with...  I think that's the just of it, if I read into it right...
legendary
Activity: 2492
Merit: 1491
LEALANA Bitcoin Grim Reaper
With the proposed addition of Confidential Transactions to the Monero core (hopefully some day)...

How do you find outputs to mix with on the block chain if you don't know the amounts associated with those outputs?

If you are mixing with other outputs on the blockchain you don't necessarily have the "Amount key" to determine if the amounts match yours.

Hmm, after thinking about it a bit if you don't know what the amount associated with an output is on the blockchain, you could essentially mix with all outputs instead of like-amount outputs...hmm

Does that sound right?

I haven't read the latest paper by shen but this is one thing that I would like to hear an answer to.

hero member
Activity: 1874
Merit: 840
Keep what's important, and know who's your friend
The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

The best solution for a 10 year low- or no-maintenance device is to go completely fanless. Forget the GPU unless you can find something decent that is fanless and can be mounted in a fanless chassis (which requires very low thermal output) and instead use a high efficiency CPU with good hash rate like the one we talked about on IRC (hash rate is roughly 120 for 15 watts, possibly less) with a brick PSU. It might require maintenance if the HD dies, but that's about it. Switching to SSD would reduce that. Even with with HD you can't do it $300 though. Not quite yet, but maybe next year.

Unfortunately what kills the price here is that the really cheap CPUs (Celeron, Atom, etc.) can't mine XMR well, but many of them can mine AEON. If you accept a much lower hash rate and efficiency you can still do it, and probably come in below $300 for the whole system.




Just curious.. is an AMD Athlon a "really cheap" CPU that won't be as efficient?  I'm currently in the process of learning Ubuntu and the command line systems, and was wondering if it would be worth it to put my hashing power towards XMR, or just go for AEON?
legendary
Activity: 2968
Merit: 1198
The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

The best solution for a 10 year low- or no-maintenance device is to go completely fanless. Forget the GPU unless you can find something decent that is fanless and can be mounted in a fanless chassis (which requires very low thermal output) and instead use a high efficiency CPU with good hash rate like the one we talked about on IRC (hash rate is roughly 120 for 15 watts, possibly less) with a brick PSU. It might require maintenance if the HD dies, but that's about it. Switching to SSD would reduce that. Even with with HD you can't do it $300 though. Not quite yet, but maybe next year.

Unfortunately what kills the price here is that the really cheap CPUs (Celeron, Atom, etc.) can't mine XMR well, but many of them can mine AEON. If you accept a much lower hash rate and efficiency you can still do it, and probably come in below $300 for the whole system.


legendary
Activity: 2380
Merit: 1085
Money often costs too much.
The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

No fan lasts that long. I'm woundering if a http://uk.wikipedia.org/wiki/%D0%95%D0%BB%D0%B5%D0%BC%D0%B5%D0%BD%D1%82_%D0%9F%D0%B5%D0%BB%D1%8C%D1%82%D1%8C%D1%94 could do the trick of cooling down the hot spot whilest recovering some electrical energy. Recuperation. Tesla cars do more of that than my car has denoted total capacity.

Don't know why Whikipedia doesn't feature english translations on http://en.wikipedia.org/wiki/Jean_Charles_Athanase_Peltier work.
hero member
Activity: 785
Merit: 500
BURST got Smart Contracts (AT)
With "sudo apt-get install libdb++-dev"  the error stopped.


monero -> fatal error  db_cxx.h missing

It's optional in master, but I'm not sure what that build script expects, so you may want to install libdb++-dev (or whatever is the package name for the Berkeley DB C++ layer on your distro), for that header. The binaries seem to have been built though, so everything's still OK, you'll be using LMDB (which is the preferred DB).

I found that the ~/.bitmonero directory was empty.
With: wget -O blockchain.bin http://monero.cc/downloads/blockchain/linux/blockchain.bin
I could get the blockchain.bin (4530181601).

That bin file is now obsolete (it's meant to work with the obsolete binaries available on the site). bitmonerod will resync from peers. Or if you have a slow network connection, you can convert that bin file into a LMDB database (blockchain_export, then blockchain_import).

When I try to start ./bitmonerod I get the following error:
~/bitmonero/build/release/bin$ ./bitmonerod
Creating the logger system

What does it mean, how to fix?

It means it's starting. It's doing just this, and doesn't print more ?

~/bitmonero/build/release/bin$ ./simplewallet
Creating the logger system
terminate called after throwing an instance of 'std::runtime_error'
  what():  locale::facet::_S_create_c_locale name not valid
Aborted (core dumped)

I'd forgot about that. Looks like the C++ locale stuff barfs on some locales, not sure why, needs looking at.
Try running with empty LANG and LC_ALL environment variables for now.


I added to /etc/environment the line: LC_ALL="en_US.UTF-8" and restarted the system.

Now "cd bitmonero/build/release/bin/ && ./bitmonerod" the pan shows a lots more ;-)

simplewallet also works.

Thanks for your help.
legendary
Activity: 1105
Merit: 1000

and how much would you be willing to pay?

how much would you be willing to pay if it could also mine (~200 or 250 or 500 h/s)?

how much would you be willing to pay if it could mine and also ran the pool software (so, you'd have your own private pool)?

how much would u pay if it could run additional decentralized software?

and why ssd?

I don't think the price should exceed 200$, and the "dedicated" means that box should only keep blockchain and run the latest daemon. All other services should run outside the box, as they are able to communicate with daemon over TCP/IP network, aren't they? Perhaps there are security or performance issues in TCP/IP versus shared memory, I don't know. I assume, when running daemon and "service" on the same LAN subnet, it should work OK.

Well yeah, using SSD might be an exaggeration. My own testing showed that starting / exiting daemon on SSD is way faster that on regular drive. So when you need to do some daemon maintenance, SSD should reduce the downtime. In regular operations (in my case, one wallet communicating with daemon  Cheesy), indeed, there is no difference between drive types.

Hrmmm... in my opinion, the "only keep blockchain and run the latest daemon" is a not how these networks were initially designed / envisioned - i.e.,  a full node is one that mines - it can create its own blocks. IMO, A non-mining node is like an auditing relay - for instance, a non mining node will still play a part in the hardfork vote (if I understand things correctly). And if one transacts on a non-mining node, they're also participating more than just keeping a copy.  Indeed, there's a full blown discussion raging about what a node is and how many are needed somewhere in bitcointalk.

"other services outside the box" not entirely sure what you mean. The pool software definitely needs to run on the same silicon.

Welp, I'm designing such a device. It won't be pretty, as they will be handmade (you can do ANYTHING with a dremel!) . I was asking about the SSD because you can get much larger HDs with a HDD, and I'm imagining these devices will just sit next to your router for years, so ideally a 10 year device will be at least 500 gigs.

The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.

If Monero goes "anywhere", 1 TB will likely not be enough space in 10 years.
legendary
Activity: 1260
Merit: 1008

and how much would you be willing to pay?

how much would you be willing to pay if it could also mine (~200 or 250 or 500 h/s)?

how much would you be willing to pay if it could mine and also ran the pool software (so, you'd have your own private pool)?

how much would u pay if it could run additional decentralized software?

and why ssd?

I don't think the price should exceed 200$, and the "dedicated" means that box should only keep blockchain and run the latest daemon. All other services should run outside the box, as they are able to communicate with daemon over TCP/IP network, aren't they? Perhaps there are security or performance issues in TCP/IP versus shared memory, I don't know. I assume, when running daemon and "service" on the same LAN subnet, it should work OK.

Well yeah, using SSD might be an exaggeration. My own testing showed that starting / exiting daemon on SSD is way faster that on regular drive. So when you need to do some daemon maintenance, SSD should reduce the downtime. In regular operations (in my case, one wallet communicating with daemon  Cheesy), indeed, there is no difference between drive types.

Hrmmm... in my opinion, the "only keep blockchain and run the latest daemon" is a not how these networks were initially designed / envisioned - i.e.,  a full node is one that mines - it can create its own blocks. IMO, A non-mining node is like an auditing relay - for instance, a non mining node will still play a part in the hardfork vote (if I understand things correctly). And if one transacts on a non-mining node, they're also participating more than just keeping a copy.  Indeed, there's a full blown discussion raging about what a node is and how many are needed somewhere in bitcointalk.

"other services outside the box" not entirely sure what you mean. The pool software definitely needs to run on the same silicon.

Welp, I'm designing such a device. It won't be pretty, as they will be handmade (you can do ANYTHING with a dremel!) . I was asking about the SSD because you can get much larger HDs with a HDD, and I'm imagining these devices will just sit next to your router for years, so ideally a 10 year device will be at least 500 gigs.

The 1 TB version i built comes to ~$300 with a 750 ti, so you could get 250 h/s. And these are all retail component costs. Perhaps someday I'd be able to get scale prices somehow. Proper cooling has proven to be a pain - its difficult to find high quality small form factor fans - fans that could last 5 years of constant runtime. And then the question of how loud the fans can be.
newbie
Activity: 9
Merit: 0
Looks like make exited with the error you described above. Try installing the Berkeley database libraries for C++ and run the script again.
Code:
sudo apt-get install libdb++-dev
legendary
Activity: 1276
Merit: 1001
monero -> fatal error  db_cxx.h missing

It's optional in master, but I'm not sure what that build script expects, so you may want to install libdb++-dev (or whatever is the package name for the Berkeley DB C++ layer on your distro), for that header. The binaries seem to have been built though, so everything's still OK, you'll be using LMDB (which is the preferred DB).

I found that the ~/.bitmonero directory was empty.
With: wget -O blockchain.bin http://monero.cc/downloads/blockchain/linux/blockchain.bin
I could get the blockchain.bin (4530181601).

That bin file is now obsolete (it's meant to work with the obsolete binaries available on the site). bitmonerod will resync from peers. Or if you have a slow network connection, you can convert that bin file into a LMDB database (blockchain_export, then blockchain_import).

When I try to start ./bitmonerod I get the following error:
~/bitmonero/build/release/bin$ ./bitmonerod
Creating the logger system

What does it mean, how to fix?

It means it's starting. It's doing just this, and doesn't print more ?

~/bitmonero/build/release/bin$ ./simplewallet
Creating the logger system
terminate called after throwing an instance of 'std::runtime_error'
  what():  locale::facet::_S_create_c_locale name not valid
Aborted (core dumped)

I'd forgot about that. Looks like the C++ locale stuff barfs on some locales, not sure why, needs looking at.
Try running with empty LANG and LC_ALL environment variables for now.

hero member
Activity: 785
Merit: 500
BURST got Smart Contracts (AT)
I set up a new computer (Ubuntu server 14.04) and started the command:

Code:
cd ~ && rm -f install_monero.sh && wget https://raw.githubusercontent.com/Quanttek/install_monero/master/install_monero.sh && bash install_monero.sh

It worked for a long time till I saw an error:
monero -> fatal error  db_cxx.h missing

In the directory bitmonero/build release/bin/ I find these files:

Quote
-rwxrwxr-x 1 ronald ronald 24626041 Oct 19 11:38 bitmonerod
-rwxrwxr-x 1 ronald ronald  3157953 Oct 19 11:39 blockchain_converter
-rwxrwxr-x 1 ronald ronald  2879491 Oct 19 11:40 blockchain_export
-rwxrwxr-x 1 ronald ronald  3253779 Oct 19 11:41 blockchain_import
-rwxrwxr-x 1 ronald ronald  1136820 Oct 19 11:32 connectivity_tool
-rwxrwxr-x 1 ronald ronald   743573 Oct 19 11:33 simpleminer
-rwxrwxr-x 1 ronald ronald  5087977 Oct 19 11:34 simplewallet

I found that the ~/.bitmonero directory was empty.
With: wget -O blockchain.bin http://monero.cc/downloads/blockchain/linux/blockchain.bin
I could get the blockchain.bin (4530181601).

When I try to start ./bitmonerod I get the following error:
~/bitmonero/build/release/bin$ ./bitmonerod
Creating the logger system

What does it mean, how to fix?

I get:
~/bitmonero/build/release/bin$ ./simplewallet
Creating the logger system
terminate called after throwing an instance of 'std::runtime_error'
  what():  locale::facet::_S_create_c_locale name not valid
Aborted (core dumped)

- I guess that is a followup error from not running bitmonerod.


Jump to: