Bitcoin client is a resource hog and slow

[RSantana]

If any one is looking for ways to help out or a place to start...

• The 5.2 beta version is taking up 215MB of RAM
• I run the client once a week, and it takes a couple hours to load the 1.5GB block chain and bring it up to date.

[Garr255]

Get more ram, always leave it running. There is little you can do without major modification to improve it. I haven't tried any custom clients, and don't plan to, but some do claim to use less resources.

[Kluge]

Sounds like you need a thin client. Electrum might do what you're looking for.  

[MaxSan]

95mb vs skype at 125mb.

I dont see a problem. I leave mine now running all the time.

[jim618]

With the improvements the bitcoinj team have made to downloading blocks, Multibit downloads around 1000 blocks a minute on a vanilla WiFi.   (It does not save the transactions to disk so it is not doing as much work as the Satoshi client).

I think of it as roughly 1 minute of download for 1 week of blocks.

I haven't really looked at memory size optimisation but think it uses around 100MB RAM when running 4 peers and half a dozen wallets.
Edit: Just had a quick look.   On a Mac with 12 wallets open, 4 peers it uses 133.5MB RAM.

[MaxSan]

multibit is awsome. I use it on my USB for portability. props to you guys.

[Technomage]

Lot of ways to use Bitcoin conveniently, Electrum and Multibit have already been mentioned but web wallets such as the blockchain.info wallet deserve a mention too. Even for storage solutions the Satoshi client is inferior these days because we have Armory that is a heavy client that offers easy ways to safely handle offline wallets.

Bitcoin users really need to know that there are other viable options these days than the regular client. There are a lot of options. I still use the Satoshi client for some of my wallets but if I was actually using Bitcoin for commerce regularly, which I'm still not, I would not be using the Satoshi client. It's too slow unless you're not in a hurry to do anything.

[bitplane]

95mb vs skype at 125mb.

I dont see a problem. I leave mine now running all the time.

I run bitcoind on a server, when bitcoind starts up it sits at around 60MB, this is with less than 10 peer connections and no RPC connections.
Compared with transmission-daemon, which I have seeding over 10GB of files and with over 50 connections, it uses 20MB.

The biggest problem though is that bitcoind leaks memory so it has to be restarted every so often (that's if the kernel's oomkiller doesn't murder it first).

Is there a bug and test case for this? I might have a look at it, sounds like something I could tackle.

[spiccioli]

With the improvements the bitcoinj team have made to downloading blocks, Multibit downloads around 1000 blocks a minute on a vanilla WiFi.   (It does not save the transactions to disk so it is not doing as much work as the Satoshi client).

I think of it as roughly 1 minute of download for 1 week of blocks.

I haven't really looked at memory size optimisation but think it uses around 100MB RAM when running 4 peers and half a dozen wallets.
Edit: Just had a quick look.   On a Mac with 12 wallets open, 4 peers it uses 133.5MB RAM.

jim618

I've just downloaded it, I did not know it existed before your message.

It is great! The only missing feature being encrypted wallets!

spiccioli

[Blind]

web wallets such as the blockchain.info wallet deserve a mention too

Stop advocating web wallets, they're a tragedy in the making. With web wallet you must trust site owner that he won't run away with your money, and even if he is a saint, his competence to secure the service properly. This is exceptionally difficult task as the prize is big.

Thin clients are the way to go and there should be verification mechanism to ensure downloaded copy is free of worms.

[Technomage]

Stop advocating web wallets, they're a tragedy in the making. With web wallet you must trust site owner that he won't run away with your money, and even if he is a saint, his competence to secure the service properly. This is exceptionally difficult task as the prize is big.

Thin clients are the way to go and there should be verification mechanism to ensure downloaded copy is free of worms.

You probably do not know about the current generation of web wallets, what I'm talking about is not like MyBitcoin. The service provider can't run away with your money because they do not have your private key. They will only know an encrypted version of the private key and it will be decrypted entirely in your browser when used. There is even a wallet integrity tool available which checks that the wallet service hasn't been compromised.

So please, next time make sure you know what you're talking about before you start spreading FUD.

[Blind]

Stop advocating web wallets, they're a tragedy in the making. With web wallet you must trust site owner that he won't run away with your money, and even if he is a saint, his competence to secure the service properly. This is exceptionally difficult task as the prize is big.

Thin clients are the way to go and there should be verification mechanism to ensure downloaded copy is free of worms.

You probably do not know about the current generation of web wallets, what I'm talking about is not like MyBitcoin. The service provider can't run away with your money because they do not have your private key. They will only know an encrypted version of the private key and it will be decrypted entirely in your browser when used. There is even a wallet integrity tool available which checks that the wallet service hasn't been compromised.

So please, next time make sure you know what you're talking about before you start spreading FUD.

Sounds like advanced techno science-fiction stuff, but you still rely on javascript delivered from possibly compromised service to decrypt your data. What stops rogue site owner or hackers from adding literally one line of code to have your keys shipped to their door, once you authenticate and have keys in the open? Even wallet integrity check won't help you as you can get served malicious javascript per user basis. Back to square one, putting faith in site owner and his ability to protect the service.

[zer0]

bitcoin-qt is using ~71megs of memory and only 2% CPU on my debian laptop right now
though took me almost 2 days to fully xfer the blockchain when I first installed

[memvola]

Sounds like advanced techno science-fiction stuff, but you still rely on javascript delivered from possibly compromised service to decrypt your data. What stops rogue site owner or hackers from adding literally one line of code to have your keys shipped to their door, once you authenticate and have keys in the open? Even wallet integrity check won't help you as you can get served malicious javascript per user basis. Back to square one, putting faith in site owner and his ability to protect the service.

I myself don't use web wallets, but it's not exactly back to square one either. Wallet verifier won't fail unless github is compromised, so it's probably safe enough for pocket money. Multisig as an authentication option will fix the rest of these issues.

[Technomage]

Sounds like advanced techno science-fiction stuff, but you still rely on javascript delivered from possibly compromised service to decrypt your data. What stops rogue site owner or hackers from adding literally one line of code to have your keys shipped to their door, once you authenticate and have keys in the open? Even wallet integrity check won't help you as you can get served malicious javascript per user basis. Back to square one, putting faith in site owner and his ability to protect the service.

It's not back to square one. The wallet verification tool is run by the user on the user's computer. The point of it is to verify that the service is not sending malicious code. Also as memvola pointed out, P2SH will increase security further. In fact I believe that blockchain.info will probably be one of the first, if not the first wallet that actually has P2SH functionality in it. The site owner is already working on P2SH features.

Currently even if person x is not using the verification tool and the wallet got compromised, the vulnerability would be there only as long as people are still logging in. The site would be taken down and warnings posted all over the place if this happened so the window of opportunity is small in any case, so we have come far from MyBitcoin. With verifications and especially P2SH, it becomes basically worry-free entirely.

[Blind]

Sounds like advanced techno science-fiction stuff, but you still rely on javascript delivered from possibly compromised service to decrypt your data. What stops rogue site owner or hackers from adding literally one line of code to have your keys shipped to their door, once you authenticate and have keys in the open? Even wallet integrity check won't help you as you can get served malicious javascript per user basis. Back to square one, putting faith in site owner and his ability to protect the service.

I myself don't use web wallets, but it's not exactly back to square one either. Wallet verifier won't fail unless github is compromised, so it's probably safe enough for pocket money. Multisig as an authentication option will fix the rest of these issues.

My bad, I assumed integrity check comes in form of a web service that pulls entire wallet's frontend and compares checksums. Stand alone tool is better, but as a matter of fact, github got compromised few days ago by bored russian teenager, so this method is not entirely bullet proof. P2SH might save the day, but if you're going through such hops, why not just use trusted open source thin client? Download once, verify (ideally against hash stored in the blockchain so cannot be manipulated), and forget. Another upside is, in case of DoS attack/domain takedown/unpaid hosting bill event, you're not cut off from your resources, that is whole idea of Bitcoin, be your own bank and don't rely on 3rd party entity.

Nothing before Bitcoin offered such great incentives for hacking/scamming, as system grows/matures, methods will only get better.

[memvola]

github got compromised few days ago by bored russian teenager, so this method is not entirely bullet proof

Yeah, but it makes it a long shot. I wouldn't choose to depend on a web wallet for my entire fortune, but could still use it for daily small transactions and maybe backup purposes (depending on the scheme).

why not just use trusted open source thin client?

Makes a lot of sense of course. One reason I can think of is a lack of multiplatform thin client that works on all your devices and you are not too enthusiastic about shuffling money through wallets. For instance, I probably could run Electrum on my N900 (not sure) if I spent some time on it, whereas blockchain.info/wallet would work instantly.

A javascript web-wallet with automated backups and a local verifier is analogous to a thin client that has forced updates. Forced updates aren't good though. Also, even with multisig, I'd probably prefer to maintain my secondary keys myself. However, for most people, this might not prove to be more reliable.

I think the structure will evolve to the point where wallets would be portable between clients and the standard two-factor authentication would work through the network itself. Then you can use whatever method is more convenient for you and fall back to the other method in case of emergencies. Even if you insist on relying on online services, you could keep a copy of your wallets on multiple sites. Of course, no combination of these sites shall have clear access to all the keys necessary to complete a transaction (which is tricky if you know nothing about how Bitcoin works).

To make the concept easier to grasp for laymen, banks/wallets and clearing houses/escrows could be totally separated, all of which provide a thin-client service as the standard transaction method (so that you would only need to access remote encrypted keys in case of emergencies). Don't see that happening even in the far future though.

[Deafboy]

I probably could run Electrum on my N900 (not sure) if I spent some time on it

Yes, you can. But you will need easydeb chroot. Python used on maemo5 is prehistoric.

[memvola]

I probably could run Electrum on my N900 (not sure) if I spent some time on it

Yes, you can. But you will need easydeb chroot. Python used on maemo5 is prehistoric.

Nice...   I'll try to run spesmilo while I'm at it too. Thanks for the tip.

Especially my wife will be happy when she sees this on her N900. She just started actively using Bitcoin, and constantly asking what her preferred client should be.

[Blind]

Makes a lot of sense of course. One reason I can think of is a lack of multiplatform thin client that works on all your devices and you are not too enthusiastic about shuffling money through wallets. For instance, I probably could run Electrum on my N900 (not sure) if I spent some time on it, whereas blockchain.info/wallet would work instantly.

A javascript web-wallet with automated backups and a local verifier is analogous to a thin client that has forced updates. Forced updates aren't good though. Also, even with multisig, I'd probably prefer to maintain my secondary keys myself. However, for most people, this might not prove to be more reliable.

I think the structure will evolve to the point where wallets would be portable between clients and the standard two-factor authentication would work through the network itself. Then you can use whatever method is more convenient for you and fall back to the other method in case of emergencies. Even if you insist on relying on online services, you could keep a copy of your wallets on multiple sites. Of course, no combination of these sites shall have clear access to all the keys necessary to complete a transaction (which is tricky if you know nothing about how Bitcoin works).

To make the concept easier to grasp for laymen, banks/wallets and clearing houses/escrows could be totally separated, all of which provide a thin-client service as the standard transaction method (so that you would only need to access remote encrypted keys in case of emergencies). Don't see that happening even in the far future though.

Forced updates are out of question, this creates a nasty attack vector. Ideally I would be running few versions behind, if protocol permits.

I don't have much against web wallets per se, one can find them useful for keeping spare change, I just don't like when they're recommended to newcomers as a viable method to store your bitcoins. We are all lazy and trying to remove friction from any part of our lives, and as such web wallets are appealing, but there are risks involved and every individual choosing such service should be well aware of them. Often newcomers don't have this knowledge, so ideally this recommendation should be avoided. I'd hate to read in few months that one of them was hacked and people lost their entire bitcoin wallet because they thought they kept their coins in Fort Knox.