geography regression difference

[robe]

I presume geography in 2.0.4 and goegraphy in 2.1 should return the same answer. In benchmarking client's data I found something slightly troubling:

SELECT COUNT(*) FROM face
WHERE ST_DWithin(geog, st_setsrid(st_makepoint(-71.0636, 42.3584), 4326)::geography, 1609.344 );

when I run this on

POSTGIS="2.0.4SVN r11708" GEOS="3.4.0dev-CAPI-1.8.0 r0" PROJ="Rel. 4.8.0, 6 March 2012" GDAL="GDAL 1.10.0, released 2013/04/24" LIBXML="2.7.8" LIBJSON="UNKNOWN" RASTER PostgreSQL 9.2.4, compiled by Visual C++ build 1600, 64-bit

 count
-------
   397
(1 row)


Time: 157.613 ms

On

 POSTGIS="2.1.0rc2 r11726" GEOS="3.4.0dev-CAPI-1.8.0 r0" PROJ="Rel. 4.8.0, 6 Mar
ch 2012" GDAL="GDAL 1.10.0, released 2013/04/24" LIBXML="2.7.8" LIBJSON="UNKNOWN
" RASTER PostgreSQL 9.2.4, compiled by Visual C++ build 1600, 64-bit

 count
-------
   395
(1 row)


Time: 218.903 ms

whammo I got 2 geometries less and lost a good chunk of speed.

I ran a couple of times on each database with similar timings.

I'll try to get permission from client to release this datasource, but in mean time I'll try to narrow down to the offending geometries.