| News |
CMUCL has a faster unicode reader now. Tests show that the CMUCL can process utf8 streams upto two times faster. A serious bug in SSE2 support has been fixed. (The sum or difference of a complex and a float would sometimes produce an incorrect imaginary part.)
CMUCL 20a patch 000 A serious bug in the printer was found in the Unicode builds. The bug caused '|\|| (vertical bar) and '|`| (backquote) to be printed without escapes. This patch can be used to fix this issue. Of course, the next snapshot will have this issue fixed too. See Install for instructions on how to apply a patch.
CMUCL has a faster unicode reader now. Tests show that the CMUCL can process utf8 streams upto two times faster. A serious bug in SSE2 support has been fixed. (The sum or difference of a complex and a float would sometimes produce an incorrect imaginary part.)
CMUCL 20a patch 000 A serious bug in the printer was found in the Unicode builds. The bug caused '|\|| (vertical bar) and '|`| (backquote) to be printed without escapes. This patch can be used to fix this issue. Of course, the next snapshot will have this issue fixed too. See Install for instructions on how to apply a patch.
CMUCL 20a is released. The release notes and binaries for the supported platforms are available from the release/20a directory of the download sites. Due to the 20a release, no snapshots are available for 2009-10.
CMUCL 20a pre-release 1 is available.
The first builds with Unicode support are available. These builds have support for Unicode that is mostly working.
CMUCL 19f has been released. The release notes and binaries for the supported platforms are available from the release/19f directory of the download sites. Due to the 19f release, no snapshots are available for 2009-03.
The Feburary snapshot binaries have been uploaded to the snapshots/2009-02 directory of the download sites.
The January snapshot binaries have been uploaded to the snapshots/2009-01 directory of the download sites.
These snapshots include support for X87 and SSE2. CMUCL will automatically detect if the chip supports SSE2 or not. If so, the SSE2 core will be used. If not, the X87 (legacy) core will be used. For special cases, the user may override this automatic selection by using the -fpu flag. The SSE2 core should work somewhat better than the X87 core on chips that support both. In particular some round-off issues are gone, and complex arithmetic with SSE2 is faster and sometimes much faster than X87.
Also see older news:
