non consitent names of engines in ccrl

Questions and comments related to CCRL testing study
Post Reply
Uri Blass
Posts: 39
Joined: Tue Jun 20, 2006 12:44 pm
Sign-up code: 0

non consitent names of engines in ccrl

Post by Uri Blass »

I find that some engines have different names between blitz and long time control and in some cases it is even not clear which engines were tested exactly for comparison between the lists.


For example Scorpio family

blitz has the following:

1 Scorpio 2.0 32-bit (5-men-egbb) 4CPU 2826 +23 −24 28.1% +157.9 28.1% 688
88.4%
Scorpio 2.0 32-bit (5-men-egbb) 2CPU 2804 +25 −25 33.5% +114.6 31.2% 570
94.3%
Scorpio 1.9JA 2CPU 2768 +35 −36 33.5% +112.0 32.7% 275
75.8%
Scorpio 2.0 32-bit (5-men-egbb) 2754 +11 −11 50.4% −1.1 42.8% 2900
56.6%
Scorpio 1.9 (5-men-egbb) 2752 +23 −23 49.1% +6.3 29.8% 640
52.5%
Scorpio 2.0 64-bit (5-men-egbb) 2751 +28 −28 49.4% +3.0 54.4% 351
53.2%
Scorpio 1.81 2749 +13 −13 44.6% +31.0 27.5% 2501
51.5%
Scorpio 1.91JA 2749 +27 −28 28.2% +160.6 25.5% 513
50.2%
Scorpio 1.91 2748 +76 −76 50.0% −3.2 38.5% 52
51.6%
Scorpio 1.9JA (5-men-egbb) 2747 +34 −35 33.4% +112.6 33.9% 286
54.8%
Scorpio 2.0 32-bit (no EGBBs) 2744 +17 −17 48.2% +10.0 56.3% 997
53.6%
Scorpio 1.82 2743 +18 −18 50.1% −12.2 24.4% 1249
68.9%
Scorpio 1.84 2736 +17 −18 28.0% +160.2 26.2% 1264
54.1%
Scorpio 1.91 (5-men-egbb) 2735 +13 −14 47.4% +17.1 32.9% 1954
75.9%
Scorpio 1.9JA 2728 +14 −14 31.3% +136.2 27.4% 1972
59.5%
Scorpio 1.8 2726 +12 −13 38.0% +89.8 26.9% 2523
59.5%
Scorpio 1.81 2CPU 2722 +29 −30 45.8% +30.8 31.9% 382
55.1%
Scorpio 1.84 (5-men-egbb) 2718 +53 −52 56.7% −47.8 35.0% 120
80.2%
Scorpio 1.84 2CPU 2688 +39 −41 28.2% +171.1 25.8% 248
86.9%
Scorpio 1.82 2CPU 2646 +67 −71 28.8% +162.1 32.1% 78
50.9%
Scorpio 1.9BH 32-bit 2642 +265 −586 0.0% +375.0 0.0% 5



long time control has the following unclear engines:

1)Scorpio 2.0 4CPU 2842 +34 −34 35.4% +97.0 37.0% 284
99.5%

Is it identical to
Scorpio 2.0 32-bit (5-men-egbb) 4CPU?

2)Scorpio 1.84 2777 +32 −32 45.4% +21.4 28.7% 355
54.3%

Is it identical to
Scorpio 1.84 or maybe it is Scorpio 1.84 (5-men-egbb)?

3)Scorpio 1.9 32-bit 2CPU 2774 +39 −40 37.1% +85.7 36.4% 217
56.2%
similiar to Scorpio 1.9JA 2CPU
Are they different?

4)Scorpio 1.91 2771 +17 −17 43.1% +44.2 38.0% 1157
81.0%
Is it Scorpio1.91 or Scorpio 1.91 (5-men-egbb)?

5)Scorpio 1.9 2755 +29 −30 41.3% +57.0 33.0% 391
70.4%

Is it Scorpio 1.9 (5-men-egbb) or maybe Scorpio 1.9BH 32-bit?


6)Scorpio 2.0 2731 +34 −34 46.6% +22.3 32.6% 291

Is it
Scorpio 2.0 32-bit (5-men-egbb) or Scorpio 2.0 32-bit (no EGBBs)
or maybe Scorpio 2.0 64-bit (5-men-egbb) ?
User avatar
Graham Banks
Posts: 26879
Joined: Sun Dec 18, 2005 5:47 pm
Sign-up code: 0
Location: Auckland, NZ

Re: non consitent names of engines in ccrl

Post by Graham Banks »

Hi Uri,

I'll point your post out to those who maintain the databases.

Two things that might help:

1. An engine is only 64-bit if it states so in the name. Otherwises it's 32-bit.
2. All Scorpio versions in 40/40 have been tested using egbbs.

Regards, Graham.
Uri Blass
Posts: 39
Joined: Tue Jun 20, 2006 12:44 pm
Sign-up code: 0

Re: non consitent names of engines in ccrl

Post by Uri Blass »

Thanks
Some more questions:
1)Danasah 2.85 from 40/40 is identical or not identical to DanaSah 2.85k from blitz?
2)Toga II 1.3.4 from 40/40 is identical to Toga II 1.3.4 from 40/4 that has only 14 games or maybe it is identical to
Toga II 1.3x4 that was seriously tested in blitz?
User avatar
Graham Banks
Posts: 26879
Joined: Sun Dec 18, 2005 5:47 pm
Sign-up code: 0
Location: Auckland, NZ

Re: non consitent names of engines in ccrl

Post by Graham Banks »

Uri Blass wrote:Thanks
Some more questions:
1)Danasah 2.85 from 40/40 is identical or not identical to DanaSah 2.85k from blitz?
2)Toga II 1.3.4 from 40/40 is identical to Toga II 1.3.4 from 40/4 that has only 14 games or maybe it is identical to
Toga II 1.3x4 that was seriously tested in blitz?
I'll get Shaun to answer these ones.
I suspect the DanaSah's are the same 2.85 version, but I'm not sure about the Togas.
Uri Blass
Posts: 39
Joined: Tue Jun 20, 2006 12:44 pm
Sign-up code: 0

Re: non consitent names of engines in ccrl

Post by Uri Blass »

Here is some other non consistent name

Twisted Logic 0.080b26 in the long time control may be the same as
Twisted Logic 0.80beta26 in the blitz section.


Petir 4.9999 in the 40/40 when Petir 4.999999 in the 40/4
Are they the same or not the same?


I try to make comparison between programs at different lists(in this case Twisted is not included because it played less than 200 games and I include in the comparison only engines that played at least 200 games in both lists)

So far it seems that Arasan10 earns the most from long time control but basically programs with low rating earn from long time control in most cases so I think that better comparison will be if we have ranking list of the same engines at blitz and long time control.

Edit:Note about arasan
The name of arasan are also non consitent and we have arasan 10 32 bits in the blitz list when only arasan10 in the 40/40 list.
I assume that the 32 bit is tested because nothing is mentioned in the long rating list but I see that in the blitz rating list there are 2 arasan10 when one is 64 bits so I wonder if part of the explanation for the good result of arasan10 is not testing the 64 bits version.

Uri
User avatar
Shaun
Posts: 6889
Joined: Sat May 13, 2006 3:24 pm
Sign-up code: 10159
Location: Brighton. UK

Re: non consitent names of engines in ccrl

Post by Shaun »

Uri Blass wrote:Thanks
Some more questions:
1)Danasah 2.85 from 40/40 is identical or not identical to DanaSah 2.85k from blitz?
2)Toga II 1.3.4 from 40/40 is identical to Toga II 1.3.4 from 40/4 that has only 14 games or maybe it is identical to
Toga II 1.3x4 that was seriously tested in blitz?
Uri

Toga II 1.3x4 is a different engine to 1.3.4...

Shaun
User avatar
Shaun
Posts: 6889
Joined: Sat May 13, 2006 3:24 pm
Sign-up code: 10159
Location: Brighton. UK

Re: non consitent names of engines in ccrl

Post by Shaun »

Hi Uri,

We need to look at consistent naming between 40/4 and 40/40 where possible...

Often at 40/4 we deliberately split out testing without say egbbs to show any difference...

All games should be played with egbbs and TBs where these are supported - we might changing the namings so we show 'no egbbs' rather than adding 'egbb' where we have tested an engine both with and without...

Over the next few weeks expect to see some changes

Shaun
Uri Blass
Posts: 39
Joined: Tue Jun 20, 2006 12:44 pm
Sign-up code: 0

Re: non consitent names of engines in ccrl

Post by Uri Blass »

Here is a summery of top non consistent names or cases when I am not 100% sure which program played

1)Rybka 2.1 64-bit 2CPU 3075 +26 ?25 73.4% ?166.8 34.3% 571 ? (is it identical to rybka2.1o in the blitz)
2)Deep Shredder 11 64-bit 4CPU 3021 +18 ?17 55.7% ?39.2 43.5% 1031 (? with (5-men-FSB) from blitz or without)
3)Rybka 2.1 32-bit 2CPU 3003 +26 ?26 70.8% ?141.9 41.1% 508?(?Rybka 2.1c 32-bit 2CPU from the blitz)
4)Rybka 1.2 32-bit 2976 +18 ?18 70.1% ?144.8 35.1% 1125 (?=rybka1.2f or rybka1.2n or rybka1.2i)
5)Rybka 2.1 32-bit 2970 +23 ?23 69.5% ?130.8 40.3% 645 (?Rybka 2.1c 32-bit or Rybka 2.1o 32-bit)
6)Rybka 2.1 32-bit (C+1 OSO) 2967 +31 ?30 72.3% ?159.8 28.5% 410 (?=Rybka 2.1c 32-bit (C+1 OSO))
7)Toga II 1.4 beta4 4CPU(can I assume not tested in blitz because only beta5 or no egbb beta4b was tested at blitz)
8)Rybka 1.0 64-bit 2921 +16 ?16 68.7% ?127.2 35.5% 1415 (Rybka 1.0 Beta 64-bit in the blitz list)
9)Rybka 1.0 32-bit 2884 +16 ?16 61.0% ?73.8 40.3% 1337 v (again written as Rybka 1.0 Beta 32-bit in the blitz list)
10)Fruit 051103 2870 +24 ?24 48.2% +10.6 43.0% 554 v (written as Fruit 05/11/03 in the blitz list)
11)Strelka 1.8 2870 +19 ?19 46.1% +26.1 41.3% 927 v (written as Strelka 1.8 32-bit in the blitz list)
12)Toga II 1.3.4 2868 +24 ?24 48.0% +13.7 42.3% 551(?only with HT80 or HT75 or x4 appeared. can I assume safely that was not tested in blitz)
13)Strelka 1.0b 2863 +31 ?31 56.0% ?37.7 39.3% 341 v(32 bit is written in the blitz)
14)Toga II 1.2.1a 2862 +11 ?11 48.3% +9.0 39.4% 2883 v(32 bit is written in the blitz)
15)Glaurung 2 epsilon/5 64-bit 2CPU 2859 +23 ?23 39.3% +73.6 37.9% 618(in the blitz Glaurung 2-epsilon/5 64-bit 2CPU that is more correct)
16)Toga II 1.1a 2824 +18 ?18 55.2% ?33.7 37.0% 1048(32 bit is written in the blitz)
17)Shredder 9 2823 +17 ?17 55.7% ?38.4 37.3% 1162 (only shredder9.1 appears in the blitz,are they the same?)
18)Glaurung 2 epsilon/5 32-bit 2805 +27 ?27 49.0% +3.9 35.3% 462 (in the blitz Glaurung 2-epsilon/5 32-bit)
19)Fritz 8 Bilbao (in the blitz only Fritz 8 that is probably different but I am not sure)
Uri Blass
Posts: 39
Joined: Tue Jun 20, 2006 12:44 pm
Sign-up code: 0

Re: non consitent names of engines in ccrl

Post by Uri Blass »

I continue names when I am unsure abut them or I am sure that they are non consistent

20)Scorpio 1.9 32-bit 2CPU 2774 +39 ?40 37.1% +85.5 36.4% 217 (?Scorpio 1.9JA 2CPU probably different)
21)Glaurung 1.2.1 Avalanche 32-bit 2773 +32 ?32 52.3% ?17.4 35.0% 331(identical to Glaurung 1.2.1 (Avalanche) 32-bit)
22)Chessmaster 11 2771 +18 ?18 41.1% +59.1 37.3% 978 (identical to Chessmaster 11 Default)
23)Scorpio 1.91 2769 +16 ?16 41.1% +58.9 36.9% 1278 (? identical to Scorpio 1.91 (5-men-egbb) or to Scorpio 1.91JA probably to the first)
24)Glaurung 1.2.1 Crusader 32-bit 2764 +32 ?32 52.0% ?17.9 28.5% 347(identical to Glaurung 1.2.1 (Crusader) 32-bit)
25)Movei 00.8.438 (10 10 10) 2761 +37 ?37 49.4% +2.3 34.3% 242(identical to Movei 00.8.438(10 10 10) when the difference is a single space after 438)
26)Scorpio 1.9 2755 +29 ?30 41.3% +56.8 33.0% 391(probably identical to Scorpio 1.9 (5-men-egbb))
27)Scorpio 2.0 2730 +33 ?34 46.6% +21.8 33.6% 298 (identical to Scorpio 2.0 32-bit (5-men-egbb))
Post Reply