Summary: | [new ebuild] sci-biology/phred-071220 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Martin Mokrejš <mmokrejs> |
Component: | New packages | Assignee: | Andrey Kislyuk (RETIRED) <weaver> |
Status: | RESOLVED FIXED | ||
Severity: | enhancement | CC: | sci-biology |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
URL: | http://phrap.org/phredphrapconsed.html | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
phred-071220.ebuild
files/phred Manifest mathtest.cpp |
Description
Martin Mokrejš
2009-01-01 16:09:20 UTC
Created attachment 177014 [details]
phred-071220.ebuild
Similar ebuild should be made available for phred-020425-distrib.tar.Z (note the .Z extension instead of .gz in newer beta distfile). Created attachment 177106 [details]
files/phred
installed into /etc/env.d/21phred
Created attachment 177107 [details]
Manifest
Now available in the science overlay Compiles&installs fine, thanks. <quote> From: David Gordon Subject: Re: phred, phrap, consed, autofinish, swat, cross_match I cannot provide assistance in building consed from source due to the potentially huge amount of time involved. CONSED_HOME is used by many programs, including scripts, so that executables can be called directly by full path. It is also used by consed for particular files it needs. You are right that phd2fasta is packaged twice--this is not necessary in your case. There are some runtime checks for versions of some programs. On Thu, 1 Jan 2009, Martin MOKREJŠ wrote: > BTW, while reading README.TXT I wonder what CONSED_HOME is needed for. > Is not $PATH sufficient? Or do you want to say the binaries have to be > in a same directory? I don't understand why. Want to place into /usr/bin/ > as usual. > > I am puzzled with phd2fasta. I received it from Brent directly. > Why is is packaged under consed_sources again? Where can I find in the > source tree which version is it? > > Are there any runtime checks for min. version of the required programs > or are the version numbers only listed in the README.TXT file? > Thanks, > Martin <quote> In gentoo-x86. On quad-core Intel machine running 64bit gentoo linux: # g++ -mfpmath=387 mathtest.cpp -o mathtest # ./mathtest 0 - -0.010493125305267737152 1 - -0.0026089216791607503648 2 - -0.00065133848759705019645 # g++ -mfpmath=387 -m128bit-long-double mathtest.cpp -o mathtest # ./mathtest 0 - -0.010493125305267737152 1 - -0.0026089216791607503648 2 - -0.00065133848759705019645 # g++ -mfpmath=387 -m96bit-long-double mathtest.cpp -o mathtest # ./mathtest 0 - -0.010493125305267737152 1 - -0.0026089216791607503648 2 - -0.00065133848759705019645 # g++ -mfpmath=sse -m96bit-long-double mathtest.cpp -o mathtest # ./mathtest 0 - -0.010493125305267803071 1 - -0.0026089216791607139356 2 - -0.00065133848759701962194 # g++ -mfpmath=sse -m128bit-long-double mathtest.cpp -o mathtest # ./mathtest 0 - -0.010493125305267803071 1 - -0.0026089216791607139356 2 - -0.00065133848759701962194 # On old P4-M machine running 32-bit gentoo linux: $ g++ -mfpmath=387 mathtest.cpp -o mathtest $ ./mathtest 0 - -0.010493125305267768377 1 - -0.0026089216791607742173 2 - -0.00065133848759708293935 $ g++ -mfpmath=387 -m128bit-long-double mathtest.cpp -o mathtest $ ./mathtest 0 - -0.010493125305267768377 1 - -0.0026089216791607742173 2 - -0.00065133848759708293935 $ g++ -mfpmath=387 -m96bit-long-double mathtest.cpp -o mathtest $ ./mathtest 0 - -0.010493125305267768377 1 - -0.0026089216791607742173 2 - -0.00065133848759708293935 $ g++ -mfpmath=sse -m96bit-long-double mathtest.cpp -o mathtest mathtest.cpp:1: warning: SSE instruction set disabled, using 387 arithmetics $ ./mathtest 0 - -0.010493125305267768377 1 - -0.0026089216791607742173 2 - -0.00065133848759708293935 $ g++ -mfpmath=sse -m128bit-long-double mathtest.cpp -o mathtest mathtest.cpp:1: warning: SSE instruction set disabled, using 387 arithmetics $ ./mathtest 0 - -0.010493125305267768377 1 - -0.0026089216791607742173 2 - -0.00065133848759708293935 $ Created attachment 185120 [details]
mathtest.cpp
Simple c++ code sensitive to floating point precision issues with 387 and sse math.
|