Author: Derek Paquette
Date: 17:56:51 09/15/04
Go up one level in this thread
On September 15, 2004 at 02:00:55, Kurt Utzinger wrote:
>On September 14, 2004 at 20:45:26, Derek Paquette wrote:
>
>>On most programs, the time it takes to reach the next ply, is usually double
>>what it took to reach the previous, so it looks something like this,
>>
>>atleast with shredder and fritz
>>
>>10ply = 40 sec
>>11 ply = 1:20
>>12 ply = 2:40
>>
>>etc.
>>
>>Of course this isn't the exact number but it is generally double, however with
>>Hiarcs 9, it seems to be almost a 5x increase per depth, almost 6 in some cases.
>>
>>So my question is, WHY would a faster cpu make it much stronger, unless of
>>course the cpu was 5x faster, (hiarcs to remind you is not a multiprocessor
>>program)
>>
>>So if I'm searching to depth 12 in 3 minutes,
>>but I wouldn't reach depth 13 for atleast 14 minutes, why would i need a faster
>>cpu? its still going to be stuck on the first calculated move for along time,
>>
>>it doesnt make sense to me, perhaps I am not understanding something critical,
>>
>>Also, what word have we the community received about Hiarcs 10 on PC.
>>
>>Gixxer(the user name) on playserver was VERY active about 4 months ago, I have
>>not seen him in along time.
>>
>>Any word on this new program?
>>He estimated that Hiarcs10 would be 50 elo stronger than hiarcs9, that would put
>>it 21 elo LOWER than shredder 8 currently on the SSDF,
>>
>>So I wonder, is Hiarcs10 being delayed until they can push it well beyond 50
>>elo, so they aren't releasing something that will be even weaker than Junior8?
>>or DF8? a program that would be a generation BEHIND?
>>
>>These questions I would like answered, as I am a big Hiarcs fan, although my
>>fanboyism is dwindling as they have not kept us up to date, and that the
>>performance of the engine might start to turn into a 'second rate' engine as it
>>cannot keep up with the top dogs, although this is all speculation, and I hope
>>non of it is true.
>>
>>Derek Paquette (check my website in two days, for the new opening book, with
>>fixed 1.d4 lines)
>>www.geocities.com/theupandup22
>
>
> Is the following not a counter-example of your theory?
> Analysis on P4 1.8/32 MB hash.
> Kurt
>
>[Event "Angriff"]
>[Site "?"]
>[Date "2000.??.??"]
>[Round "?"]
>[White "Opfer"]
>[Black "1.Sxg6"]
>[Result "1-0"]
>[Annotator "FCP Forum: TL"]
>[SetUp "1"]
>[FEN "1r1r1b1k/1bpq2pp/p4p2/P2np2P/1pBP3N/2P5/1P1B1PP1/R2QR1K1 w - - 0 1"]
>[PlyCount "11"]
>[EventDate "2000.??.??"]
>
>1. Ng6+ hxg6 2. hxg6 Qf5 3. Bd3 Nxc3 4. bxc3 Bf3 5. Bxf5 Bxd1 6. Raxd1 {
>and White wins a piece} 1-0
>
>Opfer - 1.Sxg6
>[D]1r1r1b1k/1bpq2pp/p4p2/P2np2P/1pBP3N/2P5/1P1B1PP1/R2QR1K1 w - - 0 1
>
>Analysis by Hiarcs 9:
>
>1.dxe5 fxe5 2.Rxe5
> +- (1.43) Depth: 1/8 00:00:00
>1.dxe5 fxe5 2.Rxe5
> ± (1.10) Depth: 2/11 00:00:00
>1.dxe5 Bc5
> ± (1.10) Depth: 2/11 00:00:00
>1.dxe5 Bc5
> ± (1.35) Depth: 3/12 00:00:00
>1.dxe5 fxe5 2.cxb4 Be7 3.Nf3
> ± (1.35) Depth: 3/12 00:00:00
>1.dxe5 fxe5 2.Qc2 Qe8 3.Be2
> ± (1.23) Depth: 4/16 00:00:00
>
>1.dxe5 bxc3 2.Bxc3 fxe5 3.Rxe5 Nxc3 4.Qxd7 Rxd7
> ± (1.13) Depth: 5/16 00:00:00 26kN
>
>1.dxe5 Nxc3 2.Bxc3 bxc3 3.Qxd7 Rxd7 4.bxc3 fxe5 5.Rxe5
> ± (1.13) Depth: 6/18 00:00:01 83kN
>
>1.dxe5 Nxc3 2.Bxc3 bxc3 3.Qxd7 Rxd7 4.bxc3 fxe5 5.Rxe5
> ± (0.88) Depth: 7/22 00:00:01 143kN
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 Rxd1 4.Rxd1 bxc3 5.bxc3 fxe5
> ² (0.54) Depth: 7/22 00:00:01 170kN
>
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 Rxd1 4.Rxd1 bxc3 5.bxc3 fxe5 6.Rd7
> ± (0.72) Depth: 8/23 00:00:06 793kN
>
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 Rxd1 4.Rxd1 bxc3 5.bxc3 fxe5 6.Rd7
> ² (0.47) Depth: 9/28 00:00:12 1586kN
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 Rxd1 4.Rxd1 bxc3 5.bxc3 fxe5 6.Rd7 Bd6
> ² (0.46) Depth: 9/28 00:00:13 1696kN
>
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 Rxd1 4.Rxd1 bxc3 5.bxc3 fxe5 6.Rd7 Bd6
> ± (0.71) Depth: 10/29 00:00:51 6511kN
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 bxc3
> ± (0.75) Depth: 10/29 00:00:58 7567kN
>
>1.dxe5 Nxc3 2.Bxc3 Qxd1 3.Raxd1 bxc3 4.Rxd8 Rxd8 5.bxc3 fxe5 6.Nf3 Bxf3 7.gxf3
> ² (0.60) Depth: 11/32 00:02:20 18404kN
>1.Ng6+ hxg6
> ² (0.61) Depth: 11/38 00:05:09 40367kN
>1.Ng6+ hxg6
> ± (1.00) Depth: 11/38 00:05:36 43550kN
>1.Ng6+ hxg6 2.hxg6 Qf5 3.Bd3 Nxc3 4.bxc3 e4 5.Rxe4 Bxe4 6.Bxe4 Qb5 7.Qg4
> ± (1.08) Depth: 11/38 00:06:08 47626kN
>
>1.Ng6+ hxg6 2.hxg6 Qf5 3.Bd3 Nxc3 4.bxc3 e4 5.Rxe4 Bxe4 6.Bxe4 Qb5 7.Qg4
> ± (1.33) Depth: 12/39 00:08:17 62580kN
>1.Ng6+ hxg6 2.hxg6 Qf5 3.Bd3 Nxc3 4.bxc3 e4 5.Rxe4 Bxe4 6.Bxe4
> +- (1.46) Depth: 12/40 00:09:46 73099kN
>
>1.Ng6+ hxg6 2.hxg6 Nf4 3.Bxf4 Qf5 4.Re3 Bd5 5.Rh3+ Kg8 6.Bxd5+ Rxd5 7.Qh5 Qxh5
>8.Rxh5 Bd6 9.dxe5 Bxe5 10.Bxe5
> +- (1.53) Depth: 13/41 00:21:34 163632kN
>
>(Utzinger, MyTown 15.09.2004)
Hrrrmmm...so ...it would almost seem that when playing at tournament time
controls, a faster cpu won't give it more ply, but just allow it to search
through more secondary moves, 2/44 3/44 etc etc.
So I was half right then.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.