6 Man Syzygy vs 5 Man Nalimov

[syzygy]

Operating system do not disable its internal filesystem RAM-buffers if drive is of RAMDisk type.

Linux has a ramdisk that is part of the page cache. So files "stored" in the ramdisk are at the same and using the same memory pages cached.

To my knowledge, such ramdisk implementations do not exist on Windows. (Most likely only Microsoft could implement one.)

Even with a Linux-type ramdisk, it would generally be a bad idea to load your favorite 6-piece TBs into the ramdisk, thereby blocking large parts of RAM that the OS can no longer use for other purposes (e.g. other TBs that happen to be more relevant to the current game).

[Nordlandia]

Is Nalimov obsolete by 2016?

[yurikvelo]

Is Nalimov obsolete by 2016?

It is usable to find shortest DTM (solving EG puzzles)

[Nordlandia]

How is USB 3.0 to SATA III in performance.

SSD connected to USB 3.0 with 5-men/6-men syzygy or nalimov.

My idea was to speed up the endgame probing capabilities for my Lenova G70-80 laptop.

[Nordlandia]

The harddrive on my mediocre laptop is HDD.

Can i expect some performance increase in case Syzygy tablebases stored on SSD connected via USB 3.0?

[Nordlandia]

Like that picture.

Need some expertise please.

[yurikvelo]

How is USB 3.0 to SATA III in performance.

SSD connected to USB 3.0 with 5-men/6-men syzygy or nalimov.

My idea was to speed up the endgame probing capabilities for my Lenova G70-80 laptop.

Your laptop has optical drive bay, so it's easy to install SATA 2.5" SSD
I did the same for my laptop, it cost $10 and has no speed/latency penalty.

Also you can swap HDD to optical bay and install new SSD into HDD bay.

Also consider USB uses CPU to transfer data, while ATA/AHCI/SCSI use direct RAM copy

[Dirt]

BTW I am not sure you are going to see much if any Elo gain here. I ran a test a few years back where I cluster tested with and without EGTBs and I found absolutely no gain (or loss) at all.

No visible gain with Nalimov. Everyone seems to see a slight but measurable gain with Syzygy.

[MikeB]

Good position to test 6 MAN TB setup:

8
7
6
5
4
3
2
1
	a	b	c	d	e	f	g	h

4k3/8/P1r5/8/1N6/8/6B1/4K3 b - -

Most engines without 6 MAN TB will not score vs engine with 6 Man TB. All engines with 6 MAN TB setup properly will score the point even against with 6 MAN EGTB.
[Event "Computer Chess Game"] [Site "Mac-Pro.local"] [Date "2016.04.24"] [Round "-"] [White "Crafty-25.1aT01.syz"] [Black "Komodo 9.42"] [Result "1-0"] [TimeControl "10+0"] [FEN "4k3/8/P1r5/8/1N6/8/6B1/4K3 b - - 0 1"] [SetUp "1"] 1... Rc1+ {-1010.00/5} 2. Kd2 Ra1 {-1010.00/5 0.1} 3. Kc3 Kd7 {-1010.00/5 0.1} 4. Kc4 Ra5 {-1010.00/5 0.1} 5. Nc6 Ra2 {-1010.00/5 0.1} 6. Bf3 Ra3 {-1010.00/5 0.1} 7. Kb5 Rb3+ {-1010.00/5 0.1} 8. Ka5 Ra3+ {-1010.00/5 0.1} 9. Kb6 Rb3+ {-1010.00/5 0.1} 10. Ka7 Kd6 {-1010.00/5 0.1} 11. Bd1 Rc3 {-1010.00/5 0.1} 12. Na5 Kc7 {-1010.00/5 0.1} 13. Nb3 Re3 {-1010.00/5 0.1} 14. Nc5 Re5 {-1010.00/5 0.1} 15. Nd3 Rd5 {-1010.00/5 0.1} 16. Be2 Rd6 {-1010.00/5 0.1} 17. Bf1 Rd4 {-1010.00/5 0.1} 18. Ne5 Re4 {-1010.00/5 0.1} 19. Nc6 Rg4 {-1010.00/5 0.1} 20. Be2 Rf4 {-1010.00/5 0.1} 21. Na5 Rf5 {-1010.00/5 0.1} 22. Nc4 Kc6 {-1010.00/5 0.1} 23. Bg4 Rf4 {-1010.00/5 0.1} 24. Bf3+ Kc5 {-1010.00/5 0.1} 25. Nd2 Kb5 {-1010.00/5 0.1} 26. Bh5 Rh4 {-1010.00/5 0.1} 27. Bg6 Rh6 {-1010.00/5 0.1} 28. Be8+ Kc5 {-1010.00/5 0.1} 29. Ne4+ Kb4 {-1010.00/5 0.1} 30. Bd7 Rh7 {-1010.00/5 0.1} 31. Nf6 Rf7 {-1010.00/5 0.1} 32. Nd5+ Ka5 {-1010.00/5 0.1} 33. Nb6 Rf6 {-1010.00/5 0.1} 34. Nc4+ Kb4 {-1010.00/5 0.1} 35. Ne5 Kc5 {-1010.00/5 0.1} 36. Nc6 Rd6 {-1010.00/5 0.1} 37. Nb8 Rb6 {-1010.00/5 0.1} 38. Bc6 Rb3 {-1010.00/5 0.1} 39. Bb7 Kd6 {-1010.00/5 0.1} 40. Nc6 Kc7 {-1010.00/5 0.1} 41. Ne7 Rh3 {-1010.00/20 0.4} 42. Nd5+ Kd7 {-1010.00/5 0.1} 43. Kb6 Rb3+ {-1010.00/5 0.1} 44. Kc5 Ra3 {-1010.00/5 0.1} 45. Bc6+ Kc8 {-1010.00/28 0.2} 46. Kb6 Rb3+ {-1010.00/5 0.1} 47. Bb5 Rb2 {-1010.00/5 0.1} 48. Nc7 Ra2 {-1010.00/5 0.1} 49. Bc4 Ra4 {-1010.00/27 0.2} 50. Nb5 Kd7 {-1010.00/27 0.2} 51. a7 Ke7 {-1010.00/26 0.1} 52. Kb7 Kf6 {-1010.00/25 0.4} 53. a8=Q Rxa8 {-1010.00/27 0.3} 54. Kxa8 Ke5 {-1010.00/28 0.3} 55. Ka7 Kf6 {-1010.00/25 0.3} 56. Kb6 Ke5 {-1010.00/25 0.3} 57. Kc5 Ke4 {-1010.00/24 0.3} 58. Nd6+ Ke5 {-1010.00/24 0.1} 59. Bd5 Kf4 {-1010.00/25 0.3} 60. Kd4 Kg5 {-1010.00/5 0.1} 61. Ke5 Kg6 {-1010.00/5 0.1} 62. Nf5 Kh7 {-1010.00/27 0.2} 63. Kf6 Kh8 {-1010.00/5 0.1} 64. Nh4 Kh7 {-1010.00/5 0.1} 65. Ng6 Kh6 {-1010.00/5 0.1} 66. Bg8 Kh5 {-1010.00/5 0.1} 67. Ne5 Kh4 {-1000.20/22 0.1} 68. Kf5 Kg3 {-1000.19/23 0.1} 69. Ng4 Kg2 {-1000.18/23 0.1} 70. Bc4 Kf3 {-1000.17/23 0.1} 71. Bf1 Kg3 {-1010.00/5 0.1} 72. Be2 Kh4 {-1000.16/23 0.1} 73. Nf2 Kg3 {-1010.00/5 0.1} 74. Nd1 Kh4 {-1000.12/23 0.1} 75. Kf4 Kh3 {-1000.11/5 0.1} 76. Bf3 Kh2 {-1000.07/22 0.1} 77. Nf2 Kg1 {-1000.06/5 0.1} 78. Kg3 Kf1 {-1000.05/5 0.1} 79. Nd3 Kg1 {-1000.04/5 0.1} 80. Be2 Kh1 {-1000.03/5 0.1} 81. Nf2+ Kg1 {-1000.02/5 0.1} 82. Nh3+ Kh1 {-1000.01/5 0.1} 83. Bf3# {White mates} 1-0

[yurikvelo]

Good position to test 6 MAN TB setup:

Komodo 9.3, no Syzygy see correct lines up to move 10 instantly (1 sec from clean hash).

Code: Select all

D=24, 9 MN
+1,58	1. ... Rc1+ 2.Ke2 Ra1 3.Kd3 Kd7 4.Kc4 Ra5 5.Nc6 Ra2 6.Bf3 Ra3 7.Kb5 Rb3+ 8.Ka5 Ra3+ 9.Kb6 Rb3+ 10.Ka7 Kd6 11.Nd4 Rd3
+4,46	1. ... Rc7 2.Bb7 Rc1+ 3.Kd2 Ra1 4.Nc6 Ra2+ 5.Kd3 Ra3+ 6.Kc4 Kf7 7.a7 Ra4+ 8.Kd5 Rxa7 9.Nxa7 Kf6 10.Ke4 Ke6 11.Bd5+ Kf6

11.Nd4 is draw, the only winning move is 11. Bd1

And here Komodo is in trouble, it always prefer (a bit) 11. Nd4 vs 11.Bd1

Against opponent with Syzygy, Komodo probably would draw this winning position

SF Dev prefer 11.Nd4 up to D=28/50 (40.5 MN on clean hash), but switch to 11. Bd1 at D=29+ (64MN+)
11.Bd1 and 11.Nd4 has the same eval, so it's not easy move and SF time management should allow some time to reach 60MN+