|
On Phoenix, I had to go into the config file and set bfi_int = False, otherwise the miner kernel wouldn't run, even though it seemed like it simply skipped patching. After testing dozens of different settings, performance scales similarly to previous post-11.12 / >SDK 2.5 drivers:
Max v11.11 (included SDK 2.5): 225.29Mhash/s (Worksize 256/vectors/memclock=294MHz/GPU=970MHz)
Max this version: 221.03Mhash/s (Worksize 128/vectors4/memclock=800-1000MHz/GPU=970MHz)
Radeon 5770 (stock 850MHz/1200MHz RAM) (similar miner optimizations to all 58xx)
vanitygen is still broken like it is on all newer drivers. I'm also getting desktop glitches when not doing anything at core overclocks that were stable before, I had to reduce the overclock.
|
|
|
|
http://support.amd.com/us/kbarticles/Pages/amdcatalyst132betadriver.aspxCompletely broke all my software: Phoenix 2.0.0/phatk2: [02/26/2013 04:58:47] Welcome to Phoenix v2.0.0
[0 Khash/s] [0 Accepted] [0 Rejected] [DISCONNECTED]C:\Python27\lib\site-package
s\pyopencl\__init__.py:36: CompilerWarning: Non-empty compiler output encountere
d. Set the environment variable PYOPENCL_COMPILER_OUTPUT=1 to see more.
"to see more.", CompilerWarning)
[04:58:49] [GPU 0] Finding inner ELF...
[04:58:49] [GPU 0] Patching inner ELF...
[04:58:49] [GPU 0] Inner ELF does not have 2 .text sections!
[04:58:49] [GPU 0] Sections are: [('', 0, 0), ('.shstrtab', 168, 40), ('.text',
2044, 64368), ('.data', 66412, 4736), ('.symtab', 71148, 16), ('.strtab', 71164,
2)]
[04:58:49] [GPU 0] Fatal error: Failed to apply BFI_INT patch to kernel! Is BFI_
INT supported on this hardware?
[04:58:50] [GPU 0] Finding inner ELF...
[04:58:50] [GPU 0] Patching inner ELF...
[04:58:50] [GPU 0] Inner ELF does not have 2 .text sections!
[04:58:50] [GPU 0] Sections are: [('', 0, 0), ('.shstrtab', 168, 40), ('.text',
2044, 64368), ('.data', 66412, 4736), ('.symtab', 71148, 16), ('.strtab', 71164,
2)]
[04:58:50] [GPU 0] Fatal error: Failed to apply BFI_INT patch to kernel! Is BFI_
INT supported on this hardware?
[04:58:50] Connected to server
[04:58:50] Server gave new work; passing to WorkQueue
[04:58:50] New block (WorkQueue)
[04:58:50] Currently on block: 223214
[04:59:46] Server gave new work; passing to WorkQueue
[0 Khash/s] [0 Accepted] [0 Rejected] [RPC (+LP)] vanitygen: >oclvanitygen.exe -i -k -p 0 -d 0 -o 10plusout.txt 1vanity
Difficulty: 27763956579
Compiling kernel, can take minutes...LLVM ERROR: Cannot select: 0x5be9c20: i8 = setcc 0x5be5628, 0x5bea418, 0x5be4ae8 [I
D=80]
0x5be5628: i32 = AMDILISD::ADD 0x5bea418, 0x5bea0e8 [ID=63]
0x5bea418: i32,ch = CopyFromReg 0x500c8f8, 0x5bea280 [ORD=179] [ID=51]
0x5bea280: i32 = Register %vreg40 [ORD=179] [ID=14]
0x5bea0e8: i32,ch = load 0x500c8f8, 0x5bea390, 0x3d011f8<LD4[getelementptr inbounds ([8 x i32] addrspace(2)* @modulu
s, i32 0, i32 7)]> [ORD=178] [ID=55]
0x5bea390: i32 = Constant<28> [ID=39]
0x3d011f8: i32 = undef [ORD=150] [ID=2]
0x5bea418: i32,ch = CopyFromReg 0x500c8f8, 0x5bea280 [ORD=179] [ID=51]
0x5bea280: i32 = Register %vreg40 [ORD=179] [ID=14]
|
|
|
|
I remade the blank meme image so you don't have to use that fucking shitbox of a site linked all over this thread.  |
|
|
|
|
Thermaltake is just a company name slapped on PSUs manufactured by many different companies, so it is hard to make any generalizations about particular ones. It is better to recommend a company that actually manufactures it's own power supplies, so it's not just a front company that can disappear or rename itself after it has spoiled it's name by selling junk.
If you want to take apart and diagnose what failed on the power supplies (likely cheap Chinese capacitors) you won't be able to return them for warranty service. Typically if they have a one year warranty, the things will start exploding about 1.5 years in.
|
|
|
|
Holy cowwwww. I had noticed no one had hit the less than 1 yet and SDice was up about 25% on it already so she was due for a hit. Yes, clearly that's the way universe.math.random(onlyInYourHead) works... Are there funds available to exploit this for marketing Erik? Virtually all marketing for high prize games/lottos are about what the winners do with... or even more directly - what would you do with...!
Blow it all gambling! I think that the winner may prefer not to have the winnings associated with a likeness of himself, lest he discover why Uncle Sam inc. really doesn't like online gambling. |
|
|
|
I agree with Mike Hearn that this is likely the cause. The PC I had this problem with is a 5 year old laptop which just might be breaking down slowly but surely. It's not a problematic PC otherwise but I guess LevelDB is intensive enough to be problematic if there is even a small flaw with the HD. Maybe I need a new laptop.  Or take it apart and clean out five years worth of lint and fuzz and put new thermal compound on the CPU cooler:  |
|
|
|
"Most transactions require a fee for quick processing. What priority do you want to give your transaction?"
Button: "Normal Priority" (fills in the edit box with the median fee)
Button: "High Priority" (fills in the edit box with double the median fee)
Button: "Low Priority" (fills in the edit box with half the median fee)
Button: "No Fee (may take a long time)" (fills in the edit box with zero)
I'm sure the language could be cleaned up and I'm sure there are a zillion alternate ways to scale the median for "low" and "high" but you get the idea. The user need only press a button to indicate the relative priority of the transaction.
To Bitcoin, your transaction already has something else that is is called priority - the age of the coin inputs being sent multiplied by the value of the transaction in BTC. When this priority is high enough, a transaction qualifies to be free if it has no spammy payment outputs in it, and it qualifies to be placed in the area of a block reserved for free transactions (although this might not be big enough for all the free high priority transactions). If a transaction is not calculated high priority, than sending a transaction with less than the minimum per-kb fee will guarantee a transaction that takes hours or days or is just ignored by miners and not relayed by other nodes. A no-fee option is no option unless you want to guarantee a forum with dozens more people a week that need to manually edit transactions out of their wallet that will never confirm. That makes the fee the minimum (required and included) + the optional fee. Transactions go through relatively quickly now if they include just the minimum fee (when required), usually the next block. In a future when many fee-paying transactions are waiting for block inclusion, any recommendation for a "good" optional fee amount might not be what an experienced user would have typed in themselves. You are in the right frame of mind, how to recommend more fee in the user interface without a constant flood of questions of how to use it - "what's the minimum optional fee that will work for me?"... "I paid more but it still took 20 minutes"..."Why does it recommend such a big fee when it's not required", etc. |
|
|
|
More: To force width it appears that class="post" is what needs a style. It looks like to fix old posts below a reply or "view all posts by" you need to add to "windowbg", "windowbg2", "windowbg3": <tr class="catbg">
<td colspan="2" align="left" class="smalltext">
<div style="float: right;">Posted on: February 21, 2013, 02:03:01 PM</div>
Posted by: greyhawk
</td>
</tr><tr class="windowbg2">
<td colspan="2" class="smalltext" id="msg1547342" width="100%">
<div align="right" class="smalltext"><a href="#top" onclick="return insertQuoteFast(1547342);">Insert Quote</a></div>
<div class="post">I concur.</div>
</td> |
|
|
|
Fixed.
So this was changed? https://bitcointalk.org/Themes/custom1/style.css?fin11: #bodyarea .windowbg .td_headerandpost img {max-width: 100%}The preview post feature doesn't show the auto-shrunk version of images though.. Actually the whole thread below where you post a response still has full-size images So I put something like this:  And now no in-post scroll to make it useful...? (edit: my post has a scroll bar...  ) Another place to fix width is in the "show last posts" per user. One wide post (code or image) and the whole page goes wide: https://bitcointalk.org/index.php?action=profile;u=63589;sa=showPostsThe mobile phone user would be served by a SMF extension that followed the link and made sure the linked file isn't huge. I could kill a page and blow up your 4G by linking a never-ending php as an image. |
|
|
|
You are getting more than 8 connections because Bitcoin now supports uPnP, communicating with your router and automatically configuring port forwarding, so you are able to make a full network connection, also allowing inbound traffic. Just as you had to download 5GB of blockchain from other peer-to-peer nodes in order to use Bitcoin, Bitcoin is now sharing your copy of the blockchain and relaying transactions to others. Bitcoin has no native rate limiting, although here is a thread discussing third-party solutions: https://bitcointalk.org/index.php?topic=100779.0 |
|
|
|
I just noticed, this is not yet a block-size thread. To change that, would you be so kind and make a graph of MB/block? It's not blockchain.info's job to make a graph that you or I can make ourselves:  |
|
|
|
|
Those are long-term Bitcoin holders releasing their caches of Bitcoin into the marketplace, redistributing the Bitcoins for the benefit of new adopters. When others see this altruistic gift to the marketplace, sometimes they are moved enough to join in too.
|
|
|
|
|
You must not just move blk* files, but the ENTIRE data directory.
In bitcoin pre 0.8.0, you can move the blockchain after closing bitcoin with the detach option, but the files to move are blk000x.dat AND blkindex.dat, AND you must remove any files in the database subdirectory of the destination Bitcoin data directory.
In Bitcoin 0.8.0, the above files are not used, all subdirectories must be copied to the new server. It has also been reported that the new leveldb database format is not as portable, i.e. a datadir created on Linux might not work on Windows.
The reason for detaching the pre-0.8.0 Berkeley database before migration is similar, part of it's data structure and journaling is stored in the log files in the database subdirectory, and these are platform-specific, as the internal data will depend on the Bitcoin and library versions and platform you are running. Even copying the entire data directory may not work without "detach" if the destination is a different build environment.
|
|
|
|
|
The correct approach here is to PUT BACK YOUR ORIGINAL WALLET. We can't know if your backup is good or not, especially if you have been gambling like crazy since the backup.
Then we will need to correct the balance in your wallet and regain control of any payments that you have sent out that have never been confirmed.
To do this I would advise the use of pywallet. In your Bitcoin client, note the transaction number of every payment both sent to you and that you have sent that has zero confirmations. Then shut down Bitcoin and run pywallet with the web interface, and remove the transactions from your wallet that have never confirmed. Go to the pywallet page to learn it's requirements (python) and how to run it (pywallet --web).
When you restart Bitcoin, do so with the -rescan option. This will rescan the blockchain looking for payments to your addresses and restore the actual balance of addresses you control. There should be no payments shown without the green "more than 6 confirmations" icon. Then you can send your balance to an address in a new wallet.
|
|
|
|
This will likely get more exposure and response if you were to post it in the main blockchain.info thread. blockchain.info and mtgox are just the largest of many web wallets and exchanges, there is nothing "official" about them. blockchain.info has to connect to mtgox to retrieve the last trading price for your convenience. The first question to answer for support would be if the price is reported incorrectly, because Bitcoin has gone up and down. http://bitcoincharts.com/charts/mtgoxUSD#rg10ztgSzm1g10zm2g25 |
|
|
|
|
Show Posts
