This question wouldn't happen to be related to wanting to obtain the key used to encrypt the passwords in the Adobe password leak would it?
If so, based on what I've read, the notion that the passwords are encrypted with 3DES is an assumption based on the block size. The cipher used may not be 3DES, it may be some other 8 byte block sized cipher. Single DES for example. This post on the Sophos Naked Security blog talks about this. Unless you have better information than I do that might be something to keep in mind.
A while back on Reddit there was a thread with an OpenCL bitslice single DES cracker here. Since 3DES is basically just DES done three times, that code should be able to be modified to do what you want.
I had a very quick play at the time and IIRC on my 8 x AMD 7970 GPU system I was getting an estimated time of 128 days to brute force single DES. Its likely that proper tuning and optimisation could bring this time down, but it does mean that brute forcing 3DES would still take a rather long time. Just how much longer will depend on how the 3DES is done - there are various modes by which 3DES can be implemented, including EDE (Encrypt Decrypt Encrypt) mode where the same exact 56 bit key can be used three times. This is a backwards compatibility mode which is functionally equivalent to single DES, so it could be brute forced like single DES. I wouldn't put it past someone who uses ECB block chaining to make a dumb mistake like this, but of course its not something you can count on.
If so, based on what I've read, the notion that the passwords are encrypted with 3DES is an assumption based on the block size. The cipher used may not be 3DES, it may be some other 8 byte block sized cipher. Single DES for example. This post on the Sophos Naked Security blog talks about this. Unless you have better information than I do that might be something to keep in mind.
A while back on Reddit there was a thread with an OpenCL bitslice single DES cracker here. Since 3DES is basically just DES done three times, that code should be able to be modified to do what you want.
I had a very quick play at the time and IIRC on my 8 x AMD 7970 GPU system I was getting an estimated time of 128 days to brute force single DES. Its likely that proper tuning and optimisation could bring this time down, but it does mean that brute forcing 3DES would still take a rather long time. Just how much longer will depend on how the 3DES is done - there are various modes by which 3DES can be implemented, including EDE (Encrypt Decrypt Encrypt) mode where the same exact 56 bit key can be used three times. This is a backwards compatibility mode which is functionally equivalent to single DES, so it could be brute forced like single DES. I wouldn't put it past someone who uses ECB block chaining to make a dumb mistake like this, but of course its not something you can count on.