As already indicated in other threads **, am underway in doing a comparison on a dual boot, Ubuntu 14.04 LTS with 4.2.1 kernel and Windows 10-Pro, both 64 bit, CPU Q6600 locked to stock speed of 2.4Ghz.

The Linux part is done, having completed 3 T400 and 1 T300. This has established that within T400 there can be a variance of nearly an hour, depending on Batch number.

7.14 HST1_000115_000056_AT0019_T300_F00056_S00001_1 20:22:07 (20:11:14) 4/12/2016 10:54:59 AM 4/12/2016 10:58:59 AM 99,11 211.38 MB 240.96 MB
7.14 HST1_000628_000003_AT0008_T400_F00028_S00001_1 18:14:01 (18:03:34) 4/12/2016 9:48:55 AM 4/12/2016 9:49:57 AM 99,04 183.17 MB 246.35 MB
7.14 HST1_000601_000071_AC0021_T400_F00096_S00001_1 19:23:23 (19:06:15) 4/12/2016 4:05:14 AM 4/12/2016 4:08:15 AM 99,12 181.28 MB 250.80 MB
7.14 HST1_000601_000066_AC0021_T400_F00091_S00001_1 19:12:42 (19:02:43) 4/11/2016 2:59:48 PM 4/11/2016 3:00:48 PM 99,13 196.00 MB 260.15 MB

With an efficiency of > 99%, at least we know that I/O on this platform is not an issue [There's discussion of systems having many many > 32 threads. Not a topic here... a problem of the few blessed with such horsepower!]

Upon booting into W10-64 Pro, managed to immediately fetch 4. The OS on same machine has never given > 98% efficiency, in fact on the 4 that ran previously at 1.6Ghz of T325 type, never got above 97.6%, after 37 hours of computing, but still a small deviation.

7.14 hst1 HST1_000650_000005_KC0007_T400_F00030_S00001_1 00:21:05 (00:16:46) 79,56 1,082 01d,08:07:38 09d,23 4/12/2016 12 PM [1] 00:05:21 69.49 MB 106.11 MB
7.14 hst1 HST1_000650_000091_KC0008_T400_F00016_S00001_1 00:21:05 (00:16:46) 79,55 1,071 01d,08:27:26 09d,23 4/12/2016 12 PM [1] 00:05:27 69.18 MB 105.76 MB
7.14 hst1 HST1_000650_000090_KC0008_T400_F00015_S00001_0 00:21:02 (00:16:43) 79,49 1,084 01d,08:00:00 09d,23 4/12/2016 12 PM [1] 00:05:21 70.10 MB 106.77 MB
7.14 hst1 HST1_000650_000014_KC0007_T400_F00039_S00001_1 00:20:58 (00:16:42) 79,66 1,072 01d,08:15:35 09d,23 4/12/2016 12 PM [1] 00:05:17 68.60 MB 105.08 MB

Using the <fraction_done_exact/> tag in app_config has previously proven to provide quite accurate prediction on the completion time [for started tasks], so these have kicked off with 1 day, 8 hours, but initial efficiency is sodden, just 80%. This is on my side a known issue and main reason of not using W10 other than for testing. Cold booting into W10 instead of taking it out of hibernation kicks off all sorts of update and indexing processes in W10 that lasts for hours and hours [given up of stopping them]. Still waiting on the fast-ring update build 10586 as a subscribers, this is still build 10240. At any rate, it's the CPU time comparison moi is after.

TTYL [Earliest tomorrow afternoon at this rate of progress].

** Someone, I think in the Beta forum, laid claim on there being 20% difference in performance between Linux and Windows, without any documenting that I can recollect, so this is a first attempt of backing up the numbers, whatever the outcome will be.