Rybka Rental program is a nonsense.

[Quicksort]

Any wealthy chess enthusiast would rather invest $30000 in a Cray Cx1
(Windows HPC Server 2008) with Houdini, rather than renting R4 cluster
for an equivalent amount. By the way, CX1 is only the first step in Cray' s
plan to take advantage from Windows hegemony.

What will the R4 cluster be worth when Cray 256-cores Windows power computers will be available ? Can amateur clustering compare with supercomputing experience ? Harsh times ahead for Rybka connection.

Qapla, Robert !

[Jason Konopka]

It sounds like an interesting idea for the future, but how capable and effective is Houdini currently of parallelization past 8 cores? While I don't like the fact that much time was spent on the Rybka Cluster that should probably have been spent on bugfixes like 3+ and 4.1, it cannot be denied that the Rybka Cluster has amazing capabilities and its parallel search capabilities are well-known and proven in action.

[notyetagm]

It sounds like an interesting idea for the future, but how capable and effective is Houdini currently of parallelization past 8 cores? While I don't like the fact that much time was spent on the Rybka Cluster that should probably have been spent on bugfixes like 3+ and 4.1, it cannot be denied that the Rybka Cluster has amazing capabilities and its parallel search capabilities are well-known and proven in action.

+1

[Quicksort]

Robert has already clearly stated he was about to propose to "selected" Houdini users an evaluation version supporting more than 8 threads.

My point is: how can Lukas Cimiotti, in the long term and whatever his skills are, compete with experienced Cray engineers ?

Do you think a man like Robert, given his scientific background and interest in chess programming will be stopped by an x-threads issue ?

Believe me, for all these reasons Houdini stock is clearly a "strong buy" !

Cheers,

Alain

[Dann Corbit]

The CX1 is a cluster machine.
SMP programs won't utilize all the cores.

Of course, Robert could easily take a model like that of Scorpio to get an MPICH2 interface to his program.

I wonder how many people really need a $25K machine (for the ultra low end model) to play chess.

I guess it is fewer than the number of people who will buy cluster time. So (if that is the case) then the model Vas has chosen may be a good one.

[M ANSARI]

I don't think anyone on any of the computer chess forums is a candidate customer for the Rybka Cluster. Most likely it would be a professional chess player who needs powerful hardware + software combination that he can access remotely. Imagine you are Anand and playing in a tourney in Spain, your next game is against Shirov and he just beat Carlsen in your favorite pet line. So you have to prepare in case he runs the same line against you, and you only have your laptop as an analysis assistant. Obviously having access to the cluster would be worth it then.

Another possible user would be a second to the top players, where he has a long train ride or has a long layover at an airport. He could use this spare time and still have access to hardware that is stronger than anything out there.

I guess you could say you can have your own hardware connected remotely as well, but most professional chess players are not really into cutting edge hardware, and the setup required with available tools is quite difficult and not robust. So letting someone else do it for you for a fee seems reasonable. Again I am pretty sure nobody here fits this profile, but there might be a market for that.

[Gian-Carlo Pascutto]

it cannot be denied that the Rybka Cluster has amazing capabilities and its parallel search capabilities are well-known

Really?

What's the speedup of Rybka cluster on 40 cores (+-5 cluster nodes) and on 100 cores (+- 12 cluster nodes)?

[Mike S.]

SedatChess gives a node rate of 18205 kN/s for 200 CPUs:

http://sedatchess.110mb.com/index.php?p=1_72

Does that mean that the speedup per doubling the number of CPUs remains constant for Cluster-R4, even with that many CPUs? I was trying to make a rough estimation in comparison to the fastest i7 in the list, and the factor would be ~1.78 if I got the math right. Which is not sure.

[Gian-Carlo Pascutto]

The speedup of a parallel program is unfortunately is not the same as the NPS. Unless that's accounted for in the NPS number, in which case you still have to take Vasik's word for it. And then you *still* only have a raw number for 200 CPUs, which doesn't tell you much about how much cluster nodes that actually is. (25 nodes x 8 CPU/node?)

[Sylwy]

I don't think anyone on any of the computer chess forums is a candidate customer for the Rybka Cluster. ..............[words].................. Again I am pretty sure nobody here fits this profile, but there might be a market for that.

Waiting for a bit of access time at Vas & Lukas ProfessionalChessCluster GmbH :




S