Houdini 1.03 is available

[Robert Houdart]

Artiom,

This is a bug which several users have reported, and thanks to Herbert L on CCC forum I've been able to reproduce it on the development platform.
The correction should be fairly straightforward (a single line of code to add); I'll provide a fixed release tomorrow Saturday 17th July.

I'll keep you informed, sorry to all users for the inconvenience.

Robert

[John Blake]

Thanks Robert for another Houdini update.

Cheers!

[sirabc]

I found some games here. http://www.open-chess.org/viewtopic.php ... t=10#p1484

[White "Houdini 1.01 x64 8_CPU"]
[Black "IvanHoe-B 63Mod12c"]
[Result "1/2-1/2"]
[TimeControl "600"]

{AMD Phenom(tm) II X6 1090T Processor 3895 MHz
W=21.8 plies; 9,350kN/s; Perfect 2009.ctg
B=21.6 plies; 10,751kN/s; 4,395,485 TBAs; Perfect 2009.ctg}

H1.01 on an amdx6 works fine! I'm sure he/she(?) would like to try any new compiles, but didn't sound interested in betatesting

While looking for more information on the x5355, I found this. http://stackoverflow.com/questions/2483 ... xeon-x5355

I am writing an application using boost threads and using boost barriers to synchronize the threads. I have two machines to test the application.
Machine 1 is a core2 duo (T8300) cpu machine (windows XP professional - 4GB RAM) where I am getting following performance figures :

Number of threads :1 , TPS :21
Number of threads :2 , TPS :35 (66 % improvement)

further increase in number of threads decreases the TPS but that is understandable as the machine has only two cores.

Machine 2 is a 2 quad core ( Xeon X5355) cpu machine (windows 2003 server with 4GB RAM) and has 8 effective cores.

Number of threads :1 , TPS :21
Number of threads :2 , TPS :27 (28 % improvement)
Number of threads :4 , TPS :25
Number of threads :8 , TPS :24

As you can see, performance is degrading after 2 threads (though it has 8 cores). If the program has some bottle neck , then for 2 thread also it should have degraded.

Any idea? , Explanations ? , Does the OS has some role in performance ? - It seems like the Core2duo (2.4GHz) scales better than Xeon X5355 (2.66GHz) though it has better clock speed.

Thank you -Zoolii

Could this be the problem? Maybe it's isolated to the x5355?

[Robert Houdart]

{AMD Phenom(tm) II X6 1090T Processor 3895 MHz
W=21.8 plies; 9,350kN/s; Perfect 2009.ctg
B=21.6 plies; 10,751kN/s; 4,395,485 TBAs; Perfect 2009.ctg}

H1.01 on an amdx6 works fine!

Well, not really. In this game Houdini is 15% slower than Ivanhoe, whereas in reality it should be 5% to 10% faster.
In other words, Houdini experiences a 20% performance hit on this Phenom X6 compared to high-end Intel processors.

While looking for more information on the x5355, I found this. http://stackoverflow.com/questions/2483 ... xeon-x5355
...
Could this be the problem? Maybe it's isolated to the x5355?

Yes, this could exactly be the problem, the information you've found is very relevant to our case.
If this processor is a double quad core with shared memory, the memory contention of the spin locks could absolutely kill the performance.
For this kind of architecture the "lockless hashing" promoted by Bob Hyatt should be used. I have since long coded the hashless locking in Houdini, but haven't activated it because for the most common hardware the spinlocks have better performance.
If your friend is interested, I could just compile him/her a 8_CPU version with "Hyatt hashing". Just let me know.

Robert

[robbolito]

Q9550-XP64, Blitz:10' 0

Code: Select all

1   Ivanhoe-B 57d_whm04_w64  3190  +6/-1/=31 56.58%   21.5/38
2   Houdini 1.03 x64 4_CPU   3185  +1/-6/=31 43.42%   16.5/38

There were 40 games but Houdini caused 2 exceptions and lost on time so those 2 games were removed from the list

[sirabc]

If this processor is a double quad core with shared memory, the memory contention of the spin locks could absolutely kill the performance.
For this kind of architecture the "lockless hashing" promoted by Bob Hyatt should be used. I have since long coded the hashless locking in Houdini, but haven't activated it because for the most common hardware the spinlocks have better performance.
If your friend is interested, I could just compile him/her a 8_CPU version with "Hyatt hashing". Just let me know.

A shiny new compile as a present? Who wouldn't want it! I'll keep an eye on this thread for this compile. Hopefully this is the problem and we'll see Houdini crunch some nodes.

Thanks!

[LetoAtreides82]

In my private testing of the x64 4CPU version of 1.02 I had it 20 elo below Rybka 4 x64 4CPU after 500 games:
http://www.open-chess.org/viewtopic.php?f=3&t=224

Will 1.03 overtake Rybka 4? Can't wait to try out the fixed version tomorrow!

[robbolito]

Games:Best Trio-Milwaukee 2010
Site/ Country: AMDX6-1090T, United States
Level: Blitz 10/0
Hardware: AMD Phenom(tm) II X6 1090T Processor6x 3895 MHz with 4,096 MB Memory
Operating system: Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)

Code: Select all

                                         1                    2                    3                    
1   Houdini 1.03 x64 8_CPU   3185   +46  10.5 - 9.512.0 - 8.0**    22.5/40
2   Ivanhoe-B 57d_whm04_w64  3190    -5  9.5 - 10.510.5 - 9.5 **   20.0/40
3   Deep Rybka 4 x64         3185   -41  8.0 - 12.09.5 - 10.5  **  17.5/40

Very nice win by Houdini.

[Robert Houdart]

Houdini 1.03a is now available from http://www.cruxis.com/chess/houdini.htm .

This version contains a bug-fix for the multi-PV analysis hanging in some positions. Even if you don't use multi-PV analysis it is recommended that you replace your existing 1.03 installations with the new build.

I apologize for the inconvenience.

Robert

[robbolito]

Q9550-XP64, Blitz:10' 0
Site/ Country: Q9550-XP64, United States
Level: Blitz 10/0
Hardware: Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz4x 3500 MHz with 4,096 MB Memory
Operating system: Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)

Code: Select all

1   Houdini 1.03 x64 4_CPU  3185  +9/-7/=24 52.50%   21.0/40
2   Deep Rybka 4 x64        3190  +7/-9/=24 47.50%   19.0/40