Either those numbers are low, or the bursts at the end of a kernel run are over 1x/Gen1 250MB/s bus capacity. I tested with 2 R9 380s at the same clock speeds (950/1600) and the card in a 16x slot gets ~25H/s vs 21 for the card on a 1x riser.
Thanks for this data. This perf impact of x1 vs x16 is a bit more than what I thought it would be. The transfers are bursty at the end of an equihash run. Basically with a very fast GPU every ~40 msec a 4MB blob is sent over the PCIe link. It wouldn't be difficult to make the code more efficient and have the kernel filter more of the invalid solutions (duplicate inputs) in order to reduce this 4MB down to a few dozen kB. |
|
|
|
|
CPU speed should not affect Sol/s by more than ±10%. The silentarmy kernel continuously streams 10-80 MB/s (depending on your GPU speed) of potential solution data back to the CPU for analysis, but this is easily handled even by weak CPUs.
|
|
|
|
|
I updated the title ("Linux only").
Nvidia support will come in v4, and Windows support probably in v5. I updated the initial post with a roadmap.
|
|
|
|
|
Thanks. I can confirm the openssl developers have never added an iteration count to the "enc" format. I saw a patch floating around, but it never got committed.
|
|
|
|
You are wrong. All other Bitcoin clients I have ever tried show me which address is the intended recipient of the coins. - Electrum: view transaction details → the recipient address is the one not highlighted (the yellow-background one is the change address)
- Bitcoin Wallet for Android: shows the recipient address in the main view
- Bitcoin Core: "bitcoin-cli listtransaction" shows the recipient in the "address" attribute.
Again, Multibit is the only client that fails to show the difference between the change address and the recipient address. (I haven't tried the new 0.4 release, but from the brief changelog it doesn't look like this was fixed.) |
|
|
|
I am using the following steps to create from scratch a wallet on a (secure, offline) Linux machine that can be later imported by Bitcoin Wallet's "restore wallet" feature: - Using bitcoinj's wallet-tool utility from the current github master branch: $ wallet-tool --wallet=wallet create
- I want to know the first address where it can receive coins so: $ wallet-tool --wallet=wallet current-receive-addr
- Encrypt it: $ openssl enc -e -aes-256-cbc -base64 <wallet >wallet.enc && rm wallet
I can think a few details that are important to get right: - Use a recent version of bitcoinj. Older ones produce wallets that are not BIP32-compliant deterministic wallets.
- Make sure to use a strong password for encryption that is equivalent to at least approx. 80 bits of entropy, because "openssl enc" does not use iterated hashing to compute the key & IV. For example a random lowercase alphanumeric password should be at least 16-character long (eg. "9n0y27xhq3k2h7f8" is ~83 bits of entropy).
Are there any pitfalls to think of? Is it expected that all future versions of Bitcoin Wallet should be able to import wallet.enc? By the way, Andreas: there seems to be an assumption in Crypto.java that iterative hashing is used when backing up the wallet, but this is false. The "openssl enc" format does not support iterative hashing. |
|
|
|
|
I know, I said that ("I have to open up the transaction on blockchain.info and match the addresses with the amount of coins I sent to to identify the recipient's address").
My point is the wallet itself should tell me where I sent my coins. Why the heck do I have to check the transaction in a third-party website? This is incredibly lame and absurd.
|
|
|
|
I am a long-time Bitcoin user and decided to try Multibit HD for the first time. I installed the current version (MHD 0.3.0) and used it to sent coins to a recipient. When I inspect the transaction details, the recipient is shown as 2 addresses, however Multibit doesn't even indicate which one is the recipient and which one is my change. I just can't believe that this wallet doesn't answer the simple question "where were my coins sent to?" Seriously? Isn't it one of the most basic use case you could ask a wallet for? "Show me where I sent my coins?" I have to open up the transaction on blockchain.info and match the addresses with the amount of coins I sent to to identify the recipient's address? This issue made me frustrated enough that it enticed me to log on and post on bitcointalk.org (which I hadn't done in 2 years). Honestly these sorts of issues (subpar wallet software) is one of the reasons slowing down Bitcoin's adoption.  |
|
|
|
We just released a rev2 of our chip  It would be nice to have this topic updated with latest info! Sure, what are your Mhash/Joule numbers? PS: sorry I have not been monitoring this thread for a while. |
|
|
|
100 times difficulty ~ 100,000,000 GH/s ~ 1,000,000,000 USD investments in mining gear
Who's gonna pay that?
You assumed prices of $10 per Ghash/s. It is well-known current ASIC vendors have insane margins on the raw silicon. As a matter of fact, Cointerra is already selling pre-orders $3 per Ghash/s. Watch the price drop to $1 or less per Ghash/s over the next 12 months. Therefore a $100 million investment in mining hardware is all you need for the network to grow to 100 Phash/s. This will happen by September 2014. Watch. One year later... It is now September 2014, and the network has grown from 1.0 Phash/s to 250 Phash/s: I was right. In fact we crossed 100 Phash/s in June 2014, so my estimation of when we would grow by 100x was correct with an error margin of ±90 days. (Man, it has been a while since I logged on bitcointalk.) |
|
|
|
|
Updated the post to say the TP-LINK is included.
|
|
|
|
Remember our Redhash units? Well we, Thin Air Ventures, are liquidating a stock of 30 unpopulated machines (ie. without hash units). Each one comes exactly with: - custom 4U red rackable case - 3 fans - TP-LINK WR403N mini computer - power distribution unit board - control unit board - 4 backpanels - all internal power and data cables - 32 heatsinks (for hash units) - no ATX power supply - no hash units Condition: - AS IS, WITH ALL FAULTS - about 1 in 5 units contains faults such as non-working control unit board or TP-LINK, broken pin headers, etc You might be interested in buying these unpopulated Redhash units to service existing functional units, or to assemble your own Avalon chips on hash unit boards and quickly get them up and running, etc. Price and shipping: - We are asking merely $250 per unit (excluding shipping) - Minimum order quantity: 5 units - Shipping: worldwide, or local pickup at Lake Forest, CA Contact us if interested: sales@thinairventures.com |
|
|
|
we should remember that Bitcoin and Bitgold are not similar at all, despite their names suggesting otherwise.
You are kidding right? Bit gold is incredibly similar to Bitcoin, it has: - decentralization - chains of proof-of-work - timestamps (ie. ordering in the block chain) - digital signatures to represent transactions |
|
|
|
I agree with this theory, that Nick Szabo is Satoshi.
To me, the biggest piece of evidence that very few people discuss, is the fact that Nick Szabo attempted to post-date his "bit gold" post from 2005 to 2009, after the release of Bitcoin.
That's funny. So the witch hunt continues. Anyway Nick Szabo created the "bitgold" after getting inspiration from Wei Dai (who in 1998 created the digital currency b-money). And Dai was inspired by David Chaum, who first proposed cryptographic system for untraceable payments in 1982. The list will go on and on....... But why would Nick Szabo suddenly attempt to post-date his "bit gold" post, literraly 2 months after the Bitcoin whitepaper was published? Also, the design of bit gold is a lot closer to bitcoin, than b-money or what Chaum worked on. But why would Satoshi, who was so thorough in hiding his trails, settle with the name 'Bitcoin', despite its apparent similarity to ' bit gold'? If he was familiar with Szabo's work he should not have overlooked this, especially he should not on one hand coin such a name, and on the other hand pretend to have no knowledge of Szabo's work. Simple: because Satoshi/Szabo isn't perfect. He committed small mistakes. Szabo loved the system he invented so much that he spent non-negligible time thinking about a name for it (bit gold)! So, then he implements it, and it was hard for him to not take inspiration from the name bit gold that he loved, so he comes up with bitcoin. Perhaps he thought about people commenting on the name similarity, but he may have thought "what the heck, I love the name, it's just a name, I am fine with it being similar". |
|
|
|
I agree with this theory, that Nick Szabo is Satoshi.
To me, the biggest piece of evidence that very few people discuss, is the fact that Nick Szabo attempted to post-date his "bit gold" post from 2005 to 2009, after the release of Bitcoin.
That's funny. So the witch hunt continues. Anyway Nick Szabo created the "bitgold" after getting inspiration from Wei Dai (who in 1998 created the digital currency b-money). And Dai was inspired by David Chaum, who first proposed cryptographic system for untraceable payments in 1982. The list will go on and on....... But why would Nick Szabo suddenly attempt to post-date his "bit gold" post, literraly 2 months after the Bitcoin whitepaper was published? Also, the design of bit gold is a lot closer to bitcoin, than b-money or what Chaum worked on. |
|
|
|
|
I agree with this theory, that Nick Szabo is Satoshi.
To me, the biggest piece of evidence that very few people discuss, is the fact that Nick Szabo attempted to post-date his "bit gold" post from 2005 to 2009, after the release of Bitcoin.
|
|
|
|
But the important thing to remember is, since they specialize in FPU(therefore probably being horribly weak at INT math)
It is absolutely false. Supercomputers are built from CPUs and GPUs that can do both integer and floating point operations just as fast as each other. As a matter of fact: a modern Intel or AMD CPU core can execute 8 single precision floating point operations per cycle, or 8 32-bit integer operations per cycle. That's the exact same capacity. |
|
|
|
As I said on HackerNews: This article is posted on DealBook (strictly speaking an entity independent from the NYTimes). DealBook is a financial-news website founded by Andrew Ross Sorkin [1]. This person has always had an anti-Bitcoin stance, probably because he clearly does not understand Bitcoin at all (see him interview the Winklevoss twins about Bitcoin: http://www.youtube.com/watch?v=1oGuKEazS5o ) and people tend to be dubious or scared by something they don't understand. [1] http://en.wikipedia.org/wiki/Andrew_Ross_Sorkin |
|
|
|
|
Show Posts
