[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [computer-go] Pattern Matcher
No PGO. Even the dead slow Cygwin i used.
MP2600 (2.127Ghz) 256KB cache. I might have crippled something modifying
hashtable size and putting an AND there and doing x++, but the speedup is
clearly showing.
I see now it can get speeded up a lot more by allocating 1 table instead of 2.
But well there is another 100 optimizations and they call cost time.
It's 5.3 million nps already now.
Using 1048583 transposition table entries with 8 probes.
Solving 9-ply position after 444333377 . . .
score = -2 (-) work = 27
334588191 pos / 62001 msec = 5396.5 Kpos/sec
store rate = 0.004
- 0.280 < 0.198 = 0.052 > 0.196 + 0.274
909721 23904 15676 15340 16963 16064 13475 11108
7950 6128 4126 2933 1882 1196 846 473
345 164 126 68 30 30 14 10
2 6 0 3 0 0 0 0
At 11:18 9-11-2004 -0500, yonik wrote:
>I tried this on a Opteron in 64 bit mode (RH Linux EL 3), but I get a
>segfault.
>It looks like the C code isn't 64 bit clean...
>
>I also tried it on a 2.8GHz Xeon running RedHat EL 3.0:
>gcc3.2.3 -O2: 1257.7 Kpos/sec
>Java 5.0 -server: 1132.4 Kpos/sec
>
>So in this quick test, the java version is 10% slower.
>
>-Yonik
>
>----- Original Message -----
From: "John Tromp" <John.Tromp@xxxxxxxxxxxxxxxxx>
>To: "computer-go" <computer-go@xxxxxxxxxxxxxxxxx>
>Sent: Tuesday, November 09, 2004 8:38 AM
>Subject: Re: [computer-go] Pattern Matcher
>
>
>> PS: it does seem C compilers have improved lately:
>> gcc -O gets
>> 95994066 pos / 46619 msec = 2059.1 Kpos/sec
>> while IBMJava2-142 gets
>> 95994066 pos / 72105 msec = 1331.3 Kpos/sec
>> on solving the position 4443333 on my AMD Athlon(tm) XP 2700+
>machine.
>> So, C is almost 55% faster. More than the 25% gap I noticed a few
>years
>> back, but nothing like the "more than 200% gap" Vincent would have
>you
>> believe.
>
>_______________________________________________
>computer-go mailing list
>computer-go@xxxxxxxxxxxxxxxxx
>http://www.computer-go.org/mailman/listinfo/computer-go/
>
>
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/