New:Houdini 2 Chess Benchmarks are Acceptable!

[Sedat Canbaz]

Hello Chess Friends,

I have just started a new Houdini 2 BenchMark List

*Note:I switched from Houdini 1.5a to Houdini 2,due to H2.0 supports up to 32 cores

*How to run Houdini 2 Chess Benchmarks (on YouTube):
http://youtu.be/vz9PWWff7z0

For More Information:
http://www.sedatcanbaz.com/chess/houdin ... enchmarks/

Kind Regards,
Sedat Canbaz

[Sedat Canbaz]

New Update:

Code: Select all

kN/s    Mate Processor              Speed  Cores L.P  EXE  Hardware User
23263   57s  Intel Core i7 980X     @4.60GHz  6  ON   x64  Rolando Acosta

For Full BenchMark List:
http://www.sedatcanbaz.com/chess/houdin ... enchmarks/

Best Regards,
Sedat

[Sedat Canbaz]

Last Update:

Code: Select all

kN/s    Mate Processor              Speed  Cores L.P  EXE  Hardware User
16033   30s  AMD Phenom II X6 1090T @4.22GHz  6  ON   x64  Tiago Alves

Kind Regards,
Sedat

[Sedat Canbaz]

Dear Friends,

Last update:

Code: Select all

kN/s    Mate Processor              Speed  Cores L.P  EXE  Hardware User
40259   14s  2x Intel Xeon X5667    @4.46GHz 12  ON   x64  Kim Burcham

For Full Benchmark List:
http://www.sedatcanbaz.com/chess/houdin ... enchmarks/

Note:Kim Burcham's Bench results on Youtube:
http://www.youtube.com/watch?v=lvMfSBtBg9Y

BTW,its just a fantastic bench score, super fast machine !!

Best Regards,
Sedat

[Terry McCracken]

Pretty powerful. Doing almost as many nodes as Deep Blue in 1996.

[h1a8]

Last Update:

Code: Select all

kN/s    Mate Processor              Speed  Cores L.P  EXE  Hardware User
16033   30s  AMD Phenom II X6 1090T @4.22GHz  6  ON   x64  Tiago Alves

Kind Regards,
Sedat

I would love to see some mobile benchmarks (like i7-2670 and i7-2630). My new laptop should be arriving this week so I can supply a Houdini 1.5 i7-2630 benchmark.

[kgburcham]

Pretty powerful. Doing almost as many nodes as Deep Blue in 1996.

Not sure Terry but I think Robert Hyatt said 200,000kns vs my 40,000kns.
So Deep Blue was five times as fast. Of course I have no idea how good the search was, and of course I assume we are not comparing apples to apples. All we know is that Deep Blue was good enough to beat one of the best humans in chess history. We also know that todays programs on modern hardware is----no let me say it this way. Deep Blue was to Kasparov as todays programs are to todays superGMs. In Both examples the machines dominate. I also feel that the superGM should have a more difficult time with the best of today vs years ago.
kgburcham

[Terry McCracken]

Pretty powerful. Doing almost as many nodes as Deep Blue in 1996.

Not sure Terry but I think Robert Hyatt said 200,000kns vs my 40,000kns.
So Deep Blue was five times as fast. Of course I have no idea how good the search was, and of course I assume we are not comparing apples to apples. All we know is that Deep Blue was good enough to beat one of the best humans in chess history. We also know that todays programs on modern hardware is----no let me say it this way. Deep Blue was to Kasparov as todays programs are to todays superGMs. In Both examples the machines dominate. I also feel that the superGM should have a more difficult time with the best of today vs years ago.
kgburcham

That was the improved version of Deep Blue in 1997. In 1996 it averaged 50M nodes per sec.

However, you're right that the nodes per second isn't the same but we don't have Deep Blue to test. Too bad.

[Sedat Canbaz]

Last Update:

Code: Select all

kN/s    Mate Processor              Speed  Cores L.P  EXE  Hardware User
12527   35s  AMD Phenom II X6 1090T  3.20GHz  6  ON   x64  VodkaMaster

Kind Regards,
Sedat

[rbarreira]

Pretty powerful. Doing almost as many nodes as Deep Blue in 1996.

Not sure Terry but I think Robert Hyatt said 200,000kns vs my 40,000kns.
So Deep Blue was five times as fast. Of course I have no idea how good the search was, and of course I assume we are not comparing apples to apples. All we know is that Deep Blue was good enough to beat one of the best humans in chess history. We also know that todays programs on modern hardware is----no let me say it this way. Deep Blue was to Kasparov as todays programs are to todays superGMs. In Both examples the machines dominate. I also feel that the superGM should have a more difficult time with the best of today vs years ago.
kgburcham

There's a paper written by the authors of Deep Blue which states that its parallel efficiency was just 10%. So their 200-1000 Mnodes/s would equal just 20-100 Mnodes/s in single-threaded mode. Thanks to the tightly coupled cores in the x86 architecture, Houdini's parallel efficiency should be much higher so even the 1997 Deep Blue was, for practical purposes, not incredibly faster than today's PCs (at least in terms of effective NPS, I don't know how complex its eval was compared to today's programs).

BTW the X5667 is a 4-core CPU, so why does it say 12 cores with two of those CPUs?