With the beta flac.exe linked in Maurits's post, I get the following on an AMD Phenom II X4 970 @3.5GHz:
Set 1: a few albums (wav 44.1kHz 16bit): 2:33:09, 1 630 978 738 bytes
flac 1.2.1 lv8 0:51 (180x) 809 645 691 (49,64%) [+/- 0MB]
flac 1.3.1 lv8 0:24 (383x) 807 720 851 (49,52%) [- 1,8MB]
flac 1.3.1 lv8e 1:27 (106x) 805 230 363 (49,37%) [- 4,2MB]
tak 2.3.0 p4m 1:28 (104x) 755 009 174 (46,29%) [-52,1MB]
Set 2: hires album (wav 48kHz 24bit): 1:13:58, 1 280 307 724 bytes
flac 1.2.1 lv8 1:04 ( 69x) 918 175 228 (71,71%) [+/- 0MB]
flac 1.3.1 lv8 0:26 (171x) 916 311 947 (71,56%) [- 1,8MB]
flac 1.3.1 lv8e 2:02 ( 36x) 915 515 373 (71,50%) [- 2,5MB]
tak 2.3.0 p4m 0:54 ( 82x) 893 934 332 (69,82%) [-23,1MB]
Set 3: hires album (wav 96kHz 24bit): 0:53:48, 1 863 301 258 bytes
flac 1.2.1 lv8 1:29 (36x) 1 222 429 935 (65,60%) [+/- 0MB]
flac 1.3.1 lv8 0:33 (98x) 1 216 111 317 (65,26%) [- 6,0MB]
flac 1.3.1 lv8e 2:50 (19x) 1 204 972 348 (64,66%) [-16,6MB]
tak 2.3.0 p4m 1:06 (49x) 1 168 977 920 (62,73%) [-51,0MB]
This is using foobar with 4x encoding threads from and to a RAMdisk. TAK simply for (personal) reference.
Great job on that encoding speed boost! It doesn't seem to matter whether you're encoding a "normal" or a hires album with flac anymore.
8e is not really worth it compared to TAK though - too slow and too little gain. Leaving the -e parameter out of the lv 8 preference was a solid decision.