The Week in Chess

Friday, April 5, 2013

Stockfish 2.3.1 x64 Lost to Moron 1.0 With Unbelievable Score!

Not exactly.

A few days ago, while performing my routine tournaments for the Owl Computer Chess Engines Rating List, I set-up 64-bit matches for Stockfish against Houdini 3, Robodini 1.1 and Critter 1.6a.  These are ran in an AMD quad core, Linux/Windows OS together with another unfinished tournament which made up 4 total simultaneous tournaments in one computer.  This was the same configuration where Stockfish lost to Fire and Firenzina terribly in test matches some days ago.  I wanted to be absolutely sure whether Stockfish sucks when there are many processes competing with CPU resources.  After 50 rounds each, I stopped the fight, declaring TKO against the badly cooked Stockfish.

Here is the result:
 
Rank Engine ELO Games Score% Draws% Points Win Loss Draw
1 Houdini 3 Pro x64 207 50 99.00% 2.00% 49.5 49 0 1
2 Robodini 1.1 x64 204 50 98.00% 0.00% 49.0 49 1 0
3 Critter 1.6a x64 -32 50 92.00% 16.00% 46.0 42 0 8
4 Stockfish 2.3.1 x64 -378 150 3.67% 6.00% 5.5 1 140 9

With this disgusting result, I was forced to let Stockfish have a tournament with other engines in a one to one combat with no other tournaments running in an AMD dual core computer.  Previously, I don't have problems with Stockfish in simultaneous tournaments because my computers were configured to have separate virtual machines for each, and only 1 core was used.  The purpose was to have a level playing field where SMP capable chess engines cannot take advantage of their extra power which would give the idea of the relative strength of the algorithms behind the chess engines.  The current configuration enables the utilization of available CPU cores and it is running in only one instance of Windows XP x64 under Linux Virtualbox.

This is a very uncomfortable situation where an engine is given special attention to compete favorably. Also, when Stockfish is involved, my computers will perform at less than optimal because only a single tournament could be run which would delay the completion of the tournament schedule considerably.

To the Stockfish Team, please look at this problem, whether it is related with the SMP/threads algorithm.


Download the tournament results here:


I have also included the CPU technical information below for "debugging" purposes.

processor    : 0
vendor_id    : AuthenticAMD
cpu family    : 16
model        : 4
model name    : AMD Phenom(tm) II X4 945 Processor
stepping    : 3
microcode    : 0x10000af
cpu MHz        : 3000.000
cache size    : 512 KB
physical id    : 0
siblings    : 4
core id        : 0
cpu cores    : 4
apicid        : 0
initial apicid    : 0
fpu        : yes
fpu_exception    : yes
cpuid level    : 5
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt npt lbrv svm_lock nrip_save
bogomips    : 5999.56
TLB size    : 1024 4K pages
clflush size    : 64
cache_alignment    : 64
address sizes    : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

processor    : 1
vendor_id    : AuthenticAMD
cpu family    : 16
model        : 4
model name    : AMD Phenom(tm) II X4 945 Processor
stepping    : 3
microcode    : 0x10000af
cpu MHz        : 3000.000
cache size    : 512 KB
physical id    : 0
siblings    : 4
core id        : 1
cpu cores    : 4
apicid        : 1
initial apicid    : 1
fpu        : yes
fpu_exception    : yes
cpuid level    : 5
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt npt lbrv svm_lock nrip_save
bogomips    : 6000.38
TLB size    : 1024 4K pages
clflush size    : 64
cache_alignment    : 64
address sizes    : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

processor    : 2
vendor_id    : AuthenticAMD
cpu family    : 16
model        : 4
model name    : AMD Phenom(tm) II X4 945 Processor
stepping    : 3
microcode    : 0x10000af
cpu MHz        : 3000.000
cache size    : 512 KB
physical id    : 0
siblings    : 4
core id        : 2
cpu cores    : 4
apicid        : 2
initial apicid    : 2
fpu        : yes
fpu_exception    : yes
cpuid level    : 5
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt npt lbrv svm_lock nrip_save
bogomips    : 6000.39
TLB size    : 1024 4K pages
clflush size    : 64
cache_alignment    : 64
address sizes    : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

processor    : 3
vendor_id    : AuthenticAMD
cpu family    : 16
model        : 4
model name    : AMD Phenom(tm) II X4 945 Processor
stepping    : 3
microcode    : 0x10000af
cpu MHz        : 3000.000
cache size    : 512 KB
physical id    : 0
siblings    : 4
core id        : 3
cpu cores    : 4
apicid        : 3
initial apicid    : 3
fpu        : yes
fpu_exception    : yes
cpuid level    : 5
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt npt lbrv svm_lock nrip_save
bogomips    : 6000.38
TLB size    : 1024 4K pages
clflush size    : 64
cache_alignment    : 64
address sizes    : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

No comments:

Post a Comment

Chessdom News