30 December 2011

CUDA at last

CUDA installed, on a recent attempt

- it does work, there's a nice Mandelbrot demo...
I'm interested in Big Integers. Its not clear how much CUDA speeds these up
NVIDIA GPUs have a bunch of floating point multipliers, not a natural fit for integer MUL with carry

Reports vary from "2X" to "order of magnitude" faster on factoring big integers
nvidia
NVIDIA Corporation\NVIDIA GPU Computing SDK 4.0\C\bin\win64\Release\bandwidthTest.exe Starting...
Running on...
Device 0: GeForce GT 525M
Quick Mode
Host to Device Bandwidth, 1 Device(s), Paged memory
Transfer Size (Bytes) Bandwidth(MB/s)
33554432 2330.7
,,,
NVIDIA Corporation\NVIDIA GPU Computing SDK 4.0\C\bin\win64\Release\deviceQuery.exe Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
Found 1 CUDA Capable device(s)
Device 0: "GeForce GT 525M"
CUDA Driver Version / Runtime Version 4.0 / 4.0
CUDA Capability Major/Minor version number: 2.1
Total amount of global memory: 962 MBytes (1008402432 bytes)
( 2) Multiprocessors x (48) CUDA Cores/MP: 96 CUDA Cores
GPU Clock Speed: 1.20 GHz
...
Obviously with just a laptop and a GT525M, I cant be serious about doing big Integer stuff at a competitive level
- but I'm just curious...

Way back, I wrote 386 Assembler code Miller-Rabin Prime tests and a bunch of functions
(Kuttaka.exe, it must be out there )

Occasionally I get a hankering to do big Integers once more

Computers are a million times faster now, they say

GMP is the famous bigInteger library now, buts uses gcc, anti-windows,
a bitch, Even if you go Linux dual boot, you just know that gcc is gonna throw some bunch of errors


a factoring program   YAFU    bbuhrow
- This apparently uses GMP but the exe runs ootb on Windows.
 nice functions like "nextprime"
factored an 80 digit decimal in 4 minutes (Quadratic sieve?)

I want to use my GPU, so I downloaded
msieve    from     gilchrist       this is "Jeff" who people on http://www.mersenneforum.org/ are often thanking

the cuda version
the win32 version worked (factored the 80 digit in 4 minutes)

18/06/2011 02:07 p.m. 870,912 msieve.exe

but the NVIDIA utility reported GPU activity 'none'

tried the big win64 version but got
"vcomp100.dll is missing" - some cuda 64/32 bit mismatch
- not fixed by reloading directX, nor by scattering vcomp100.dll or cudart.dll about

17/06/2011  09:23 p.m.         1,171,968 msieve.exe   CUDA confusion?

I may post a complaint on
mersenneforum
but as of now I dont have any idea if my GPU is active or effective

there is a Number-Field-Sieve implementation CGNFS (from Jeff)
- requires Python 2.6 which is worth having anyway.
I overwrote with the win32 version of msieve and tried the 80 digit using factMsieve.py
- it chugged away for half an hour? then crashed

At the moment I am trying CGNFS on their 100 digit example - hasnt crashed, is onto the sieve
- from the forum it seems its all about tweaking parameters

mpir
MPIR is a fork from GMP
MPIR may be Non-anti-windows and may even be faster
I havnt looked to see if has  a library for VC++

do I still have the level of obsession required ?...

It was oddly nostalgic to look up an article and realise I had tinkered with its algorithms
back in the 90s

MATHEMATICS OF COMPUTATION
VOLUME 53, NUMBER 187
JULY 1989, PAGES 411-414
A New Method for Producing Large Carmichael Numbers
By H. Dubner

08 December 2011

CUDA,OpenCL, NFC

CUDA

Spent Sunday attempting to get CUDA - somehow didnt install
- one page reckoned that CUDA gave only a 2x speedup below 4kbits
Then I tried to get GMP, despite it having no claim to use CUDA
- got tangled in the usual Linux gcc Crapola ... Linux prissy purists insist on handing you the "source" which you have to compile with gcc
but windows gcc is itself fraught with crapola and pils of anxious (outdated?) complaints about it all
..


So the free arena is Crapola, I need to buy mathematica? ...MatLab has nifty pages on CUDA .. but you have to register to see the price !? (Thousands"??

a Quote:
. I do not think SciLab and Octave are viable alternatives to MATLAB for me. I think Python is a viable alternative for me. There is not much that I can do in MATLAB that I cannot do in Python. There is no question that if I was just starting out I would use Python instead of MATLAB. MATLAB, however, is cheap for me compared to the cost of porting my existing code to Python.
http://www.scientific-computing.com/review1.html

Maybe the best way to get bigint/cuda might be to sign up with one of the global shared computation efforts, one assumes that they will download simply...



Meanwhile: notes on OpenCL (I cant determine if OponCL is a rival of CUDA, or an older obselescant device)

It just so happens that Radeon cards are quick with integer calculations so they were the perfect tool for the job increasing the hash rate by magnitudes of 100+. M
Radeon 5770 is not the fastest card around but packs way more punch than a traditional CPU. It also looks like Radeon cards are faster than their Nvidia counterparts especially when it comes to password brute forcing and the latest Radeon 6990 graphic card should be several times faster than my Radeon 5770 and should make ANY 8 character password obsolete.

he University of Georgia was experimenting with this last year.
gtri
They were able to gang GPU’s to achieve staggering results, even 12 character passwords are crackable in a sort amount of time.


Radeon HD 5970 can execute 3200 32-bit instructions per clock (using its 3200 ALUs or shaders


$350 for the HD5970
AMD Radeon HD 6990: 3072 ALUs x 830 MHz = 2550 billion 32-bit instruction per second


________________________________________________________________________________
OpenCL
compute4cash is using an open-source miner called Poclbm (
bitcoin

August 8, 2011, AMD released the OpenCL-driven AMD Accelerated Parallel Processing (APP) Software Development Kit (SDK) v2.5,


wikipedia


Dell Inspiron i% gas ATI Mobility Radeon HD 5470, Sadly I dont have this???
Mine has NVIDIA GeForce GT 525M - 1GB ???




intel opencl

intel



_______________________
NFC
...if you’re the pioneering (sic) type who likes to swap smartphone batteries in the field: The battery that comes with the stock Galaxy Nexus includes an NFC antenna inside of it (..like Android Beam, for instance). ..So unless you’re swapping that battery cell out for similar one, don’t expect to be waving your Google Wallet around anytime soon.
Wired
_____________________________________________
Google ... integrate Google Wallet into mobile web payments.
.....probably would not make sense yet..... with Fandango.com and MovieTickets.com to offer Google Wallet as a payment option for their mobile web checkout.
...users won't have to try to type their 16-digit credit card number plus the expiration date and security code on their phone.. Typing on mobile phones is a notorious hassle, and mobile Web users often abandon their online shopping carts when they reach the point of entering payment information.
[OK EMV NFC, we have the technology... we can read the CCNumber,ExpDate (but not the 3digit
...Card Security Code (CSC) ... Card Verification Value (CVV) or Card Verfication Code (CVC)]

.. also offer their own apps for iPhone, Android, ...
It's unclear whether Google Wallet will be offered as a payment option for in-app purchases -- but ..."if Google decided to offer this ability, it might "pull the rug out from underneath Apple's 30% tax on in-app purchases."
cnn





______________________________________________________
Auckland Transport's announcement calls the new card an "integrated smartcard," which is, as I've observed before, rather gilding the lily. The service is hardly all that smart, nor very integrated. It's just an electronic cash substitute that can deduct a fare from your card.
...A fully integrated system would include a zonal charging system where passengers were levied for the distance travelled regardless of changes of transport mode on the way.
Even when the new "integrated card" comes in, swapping from bus to train, for example, will incur a new fare.
In a letter to transport committee chairman Mike Lee in August, Mark Lambert, Auckland Transport's manager, public transport operations, outlined how the organisation planned to implement the "integrated public transport network" model over the next three or four years. He painted a four-tiered network of routes, the backbone being a rapid transit network of electrified rail and dedicated busways, interlinked with a secondary grid of bus services.
But absent in this three to four-year plan is any mention of a simplified fare system. One that doesn't punish the commuter for changing modes mid-journey. The new $87 million "integrated ticketing system" is capable [sic] of doing these calculations. When will it be asked to is the question.


nzherald
Auckland's much-hyped smartcard system hardly worthy of the name By Brian Rudman 5:30 AM Friday Nov 11, 2011