LibreOffice 4.2 Alpha 1 To Bring Many Improvements

[phoronix]

Phoronix: LibreOffice 4.2 Alpha 1 To Bring Many Improvements

LibreOffice was bumped today for version 4.2.0 Alpha 1, the next major update to the popular open-source office suite...

http://www.phoronix.com/vr.php?view=MTUwNDk

[przemoli]

Michael. You will end up with LO as first OpenCL benchmark :P

Anyway. Why no OpenCL benchmarks? *coins* minining is quite popular. You could gain some views/clicks with such tests.
(No automated benchmarks? No corpo interest? Just bblobs would be fine. And getting people to use such sw on Linux would put more preasure on all big 3 to support it better)

[curaga]

Haha, mining as a bench:
- every time he runs it, chance of $$$
- if the PTS profile mines to his wallet, $$$ from users too


* I have no idea if mining requires the password/private key. If it doesn't, ka-ching.

[bridgman]

Michael has already published a number of OpenCL benchmarks, eg :

http://www.phoronix.com/scan.php?page=news_item&px=MTQwMDI

[GreatEmerald]

Haha, mining as a bench:
- every time he runs it, chance of $$$
- if the PTS profile mines to his wallet, $$$ from users too

Chance? The odds of getting a single bitcoin at this point is zero (unless you want to get into ludicrous precision territory).

[curaga]

Maybe he mined some of the copy-coins instead (lite- or whatever).

[DrYak]

* I have no idea if mining requires the password/private key. If it doesn't, ka-ching.

No, mining (or more precisely that part of mining where you actually win coins*) functions exactly like a payment:
You can send money TO ANY public key (= bitcoin address).
BUT you can only spend money only FROM THE WALLET YOU OWN (you need the private key that corresponds to the previous public key, so you can sign order for the money to be transfered to another public key).

Mining is just a special type of payment: at the end of each block of the block chain, a certain amount of bitcoin (initially 50, but now 25) can be transferred "out-of-no-where" (i.e.: without any private key signing them) into any bitcoin address (into any public key).
There's no private key signing those BTCs out of a previous account, only the new BTC are magically transfered into an arbitrary address (traditionally, the guy who managed to find the correct hash to validate the block wrote himself 50 BTC into said block. Nowadays the difficulty is so high that this mining is done by lone individuals anymore, but by pools, so the 25BTC new coins end-up distributed among all people who contributed to the hashing).

In addition to that, nowadays mining and transactions run in separate processes.
A mining process is an OpenCL process (or, specially with bitcoins, a Verilog design running on actual FPGA or ASIC silicon), that only compute hashes.
It recieves only commands like "please compute hash in range A to B, trying to find C", then does those computations and return eventual results. No keys involved at any step.
Then, completely seperately (even more so with FPGA/ASIC where the mining doesn't even run on the same machine) a bitcoin deamon is in charge of listening the whole bitcoin network and keeping the "public ledger" up to date, this deamon is the one manipulating the keys, and adding transaction to the ledger (using the private key) and validating the blocks (by sending hash commands to the miner. Most often this happens over a network socket to another deamon like "bfgminer" which will then actual pilot the hardware) (Nowadays, there's a pool standing in the way between bitcoind and the miner, used to distribute the hashing between several miners and then ordering bitcoind to share the 25BTC among all miners).

*: People tend to forget, but mining is also what keeps the bitcoin system functioning. Each block in the block chain need to be validated and mining (computing hashes) is what does it.
These days, chance of wining the coin are lower, and most of people end up spending more on hardware and energy than earning bitcoins back. But none the less this mining is what keeps the transactions hapening and the bitcoins flowing around.

Chance? The odds of getting a single bitcoin at this point is zero (unless you want to get into ludicrous precision territory).

These days *NOBODY* does mine alone. Instead people bond together and form pools to mine together.
It might be possible for Micheal to set-up an official "Phoronix" P2Pool node which gathers the works of several such Coin OpenCL Benchmarks, and then keeps a certain percentage.

Individual benchmarker still have a very low chance of gaining anything - they might more or less consider mining as a form of "funny lottery".

But Phoronix might get a very low trickle of pennies (distributed over the whole bunch of benchmarks connecting to the P2Pool node either from the node's "tax" as mentioned above, or donations, or people who don't input an actual bitcoin address plain give up any earnings) which could add some minor contribution to the donations (hey Micheal, consider this! I am available for help if you need to setup this ;-) )

Maybe he mined some of the copy-coins instead (lite- or whatever).

This is actually very interesting and important:

A. first from a benchmarking point of view:
the has function tested is completely different.
- traditional bitcoin use SHA-256 ? (two SHA256 hashes one after the other. So almost exclusively only bit operation. Tests only the bitlogic part of the CPU or GPU tested).
- litecoin use Scrypt, which is by design much more resource intensive, specially memory intensive. It puts quite some load on memory bandwidth (thus it also test memory access).
- primecoin is a rather new comer, which does maths instead of cryptographic hash and relies on finding primes (and will exercice the arithmetic part of the device)

B. from money point of vew:
- bitcoin (because it's so simply) is currently best mined on custom hardware (ASIC, FPGA). mining bitcoin efficiently is beyond the reach of GPUs (there it's more a very rare lottery)
- litecoin are currently mined mostly on OpenCL hardware (graphic cards, mostly)
- primecoin are currently CPU only. (but are new on the market, so aren't currently worth much).

Thus considering several test is useful, because:
- it test different scenarios
- it might elevate the chance of getting a few pennies back (specially with litecoins).