Page 1 of 3

Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 6:40 am
by geots
Komodo 1.3 and 2.03 32bit in 40/3 repeating both average somewhere between 270kn and 350kn- nodes per second. Still in 40/3 repeating, Komodo 3 averages around 170 to 220 kns But with Komodo 4, again 32bit and 40/3 repeating, the node count drops to between 70kns to 85 kns. I have yet to see Komodo 4 go over 85kns.

In summary, it seems the newer the version, the smaller the node count displayed. Is this intentional, is the node count being partially hidden, or what?


Best,

george

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 6:48 am
by Graham Banks
geots wrote:Komodo 1.3 and 2.03 32bit in 40/3 repeating both average somewhere between 270kn and 350kn- nodes per second. Still in 40/3 repeating, Komodo 3 averages around 170 to 220 kns But with Komodo 4, again 32bit and 40/3 repeating, the node count drops to between 70kns to 85 kns. I have yet to see Komodo 4 go over 85kns.

In summary, it seems the newer the version, the smaller the node count displayed. Is this intentional, is the node count being partially hidden, or what?


Best,

george
Komodo 4 32-bit reaches around 130 kns here.

Image

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 7:42 am
by lkaufman
geots wrote:Komodo 1.3 and 2.03 32bit in 40/3 repeating both average somewhere between 270kn and 350kn- nodes per second. Still in 40/3 repeating, Komodo 3 averages around 170 to 220 kns But with Komodo 4, again 32bit and 40/3 repeating, the node count drops to between 70kns to 85 kns. I have yet to see Komodo 4 go over 85kns.

In summary, it seems the newer the version, the smaller the node count displayed. Is this intentional, is the node count being partially hidden, or what?


Best,

george
In Komodo 4 we changed the method of node counting to a way that we considered more appropriate, not to hide anything. But I think it only reduced stated node counts by 10% or so, nothing like what you are reporting. I'd like to know if others are showing similar huge drops in NPS from Komodo 3 to Komodo 4, especially on 32 bit. It's possible but unlikely that something is wrong with the 32 bit version, as we hardly checked it out at all. We'll have to investigate. I'm pretty certain at least that nothing is wrong with the 64 bit version; early results are in line with expectations.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 7:47 am
by lkaufman
Graham Banks wrote:[Komodo 4 32-bit reaches around 130 kns here.
]

Thanks, but could you tell us what Komodo 3 does on the same hardware? It should be in the 140-150 range unless something is wrong with Komodo 4 32 bit.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:00 am
by Graham Banks
lkaufman wrote:Thanks, but could you tell us what Komodo 3 does on the same hardware? It should be in the 140-150 range unless something is wrong with Komodo 4 32 bit.
Will take a look shortly and let you know.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:03 am
by geots
lkaufman wrote:
geots wrote:Komodo 1.3 and 2.03 32bit in 40/3 repeating both average somewhere between 270kn and 350kn- nodes per second. Still in 40/3 repeating, Komodo 3 averages around 170 to 220 kns But with Komodo 4, again 32bit and 40/3 repeating, the node count drops to between 70kns to 85 kns. I have yet to see Komodo 4 go over 85kns.

In summary, it seems the newer the version, the smaller the node count displayed. Is this intentional, is the node count being partially hidden, or what?


Best,

george
In Komodo 4 we changed the method of node counting to a way that we considered more appropriate, not to hide anything. But I think it only reduced stated node counts by 10% or so, nothing like what you are reporting. I'd like to know if others are showing similar huge drops in NPS from Komodo 3 to Komodo 4, especially on 32 bit. It's possible but unlikely that something is wrong with the 32 bit version, as we hardly checked it out at all. We'll have to investigate. I'm pretty certain at least that nothing is wrong with the 64 bit version; early results are in line with expectations.


Thn Larry, I think it would be wise for me to hold off on any 32 bit testing at the moment. Because I can tell you that on the same machine, Komodo 3's node count is over twice the node count of Version 4.


Regards,

george

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:13 am
by Jimmy Huggins
Larry I have around 90kns on my 32 bit machine. But my is around 2.14 Ghz so I don't have the fastest PC.

Komodo 3 is around 300kns

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:15 am
by Graham Banks
Graham Banks wrote:
lkaufman wrote:Thanks, but could you tell us what Komodo 3 does on the same hardware? It should be in the 140-150 range unless something is wrong with Komodo 4 32 bit.
Will take a look shortly and let you know.
I can add that in an endgame position later in the same game, Komodo 4 32-bit is consistently up around 220 kns.
This is on a Q6600 quad using 1CPU.

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:15 am
by geots
lkaufman wrote:
geots wrote:Komodo 1.3 and 2.03 32bit in 40/3 repeating both average somewhere between 270kn and 350kn- nodes per second. Still in 40/3 repeating, Komodo 3 averages around 170 to 220 kns But with Komodo 4, again 32bit and 40/3 repeating, the node count drops to between 70kns to 85 kns. I have yet to see Komodo 4 go over 85kns.

In summary, it seems the newer the version, the smaller the node count displayed. Is this intentional, is the node count being partially hidden, or what?


Best,

george
In Komodo 4 we changed the method of node counting to a way that we considered more appropriate, not to hide anything. But I think it only reduced stated node counts by 10% or so, nothing like what you are reporting. I'd like to know if others are showing similar huge drops in NPS from Komodo 3 to Komodo 4, especially on 32 bit. It's possible but unlikely that something is wrong with the 32 bit version, as we hardly checked it out at all. We'll have to investigate. I'm pretty certain at least that nothing is wrong with the 64 bit version; early results are in line with expectations.


After a double-check to be sure, on my machine, Komodo3 averages a node count of 185,and Komodo averages a node count of 85. The count between 2 different machines would not help. This comparison is with them playing each other. It bothers me enought to where I will hold off testing till I hear further from you.


Again- Best,

george

Re: Don- Don't understand Komodo 4?? A bug??

Posted: Mon Dec 26, 2011 8:17 am
by geots
Jimmy Huggins wrote:Larry I have around 90kns on my 32 bit machine. But my is around 2.14 Ghz so I don't have the fastest PC.

Mine is around 2995 and my node count is no better than yours.


george


PS: I will have to hear from Larry and Don to test it "as is" before I put any cpu time into testing the 32 bit.