Bitcoin Forum
May 10, 2024, 04:34:38 AM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: CAL++ GPU Library?  (Read 2951 times)
freeqaz (OP)
Newbie
*
Offline Offline

Activity: 10
Merit: 0


View Profile
March 24, 2011, 12:03:23 PM
 #1

Hey guys, I've been using Pyrit for a while (A GPU program) and for ATI cards there's a library called CAL++ that uses C++ kernel extensions to speed up the speed at which an ATI card can computer. This is up to 80% for the Radeon 5xx0 series. Was interested if you guys had heard anything about it applying to Bitcoin.

I don't know enough about OpenCL at this point to say much, but I know the syntax is similar between the two. Debating learning and porting it over to poclbm, since Pyrit is also Python.

Aloha,
-Free

P.S. If one BTC is 83cents, then the 5 hours it takes me to crunch data makes me very little?
"Bitcoin: mining our own business since 2009" -- Pieter Wuille
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
1715315678
Hero Member
*
Offline Offline

Posts: 1715315678

View Profile Personal Message (Offline)

Ignore
1715315678
Reply with quote  #2

1715315678
Report to moderator
1715315678
Hero Member
*
Offline Offline

Posts: 1715315678

View Profile Personal Message (Offline)

Ignore
1715315678
Reply with quote  #2

1715315678
Report to moderator
1715315678
Hero Member
*
Offline Offline

Posts: 1715315678

View Profile Personal Message (Offline)

Ignore
1715315678
Reply with quote  #2

1715315678
Report to moderator
rikur
Full Member
***
Offline Offline

Activity: 216
Merit: 100


View Profile
March 24, 2011, 04:18:05 PM
 #2

I accidentally stumbled across CAL++ as well. Sounds like a sweet deal. Unfortunately, I couldn't easily build pyrit under windows so I have to give in with the idea for now. And I'm not familiar with C at all.

80% improvement sounds awesome and it just might work. Even a 10-20% improvement to hashing would be major. I think you could get quite a few donations for porting poclbm to use CAL++ instead of the OpenCL wrapper.

I hope someone will look this up !
niooron
Full Member
***
Offline Offline

Activity: 193
Merit: 100


View Profile
March 24, 2011, 06:25:34 PM
 #3

AMD dropped CAL because it was a mess, this won't get anywhere.
freeqaz (OP)
Newbie
*
Offline Offline

Activity: 10
Merit: 0


View Profile
March 24, 2011, 09:47:56 PM
 #4

I accidentally stumbled across CAL++ as well. Sounds like a sweet deal. Unfortunately, I couldn't easily build pyrit under windows so I have to give in with the idea for now. And I'm not familiar with C at all.

80% improvement sounds awesome and it just might work. Even a 10-20% improvement to hashing would be major. I think you could get quite a few donations for porting poclbm to use CAL++ instead of the OpenCL wrapper.

I hope someone will look this up !

I'll look more into this today, since you've actually heard something about it. Unfortunately, there is no support for CAL++ in Windows. People trying to get it to work in cygwin can get CPU stuff to work, but I haven't read anything about successful GPU. So this would be a Linux only thing.

As for AMD dropping CAL support, I couldn't find anything about that online. I searched and somebody mentioned it back in 2008, but they failed to produce any official support to their statement. If you could grab that information, that'd be awesome.

The good thing about CAL is that it supports Radeon 3xx0 series cards. That would be a big improvement for some people, because I have a spare laptop with a Mobility Radeon 3650. OpenCL doesn't work on these cards.
Beremat
Full Member
***
Offline Offline

Activity: 263
Merit: 100


YGOLD is a Defi platform


View Profile
March 24, 2011, 10:21:27 PM
 #5

Milkyway@home used to run a CAL++ application. It seems that it's switched over to OpenCL since then.

Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!