[MonetDB-users] Benchmark on MonetDB/X100
I want to use MonetDB for benchmarking on the TPC-H workload. According to the paper "MonetDB/X100 - A DBMS In The CPU Cache", the X100 module achieves significant acceleration against MonetDB/MIL on Query 1 of TPC-H. Still, when I search your site, there is no reference about how it is possible to activate X100 inside MonetDB. I have only found X100 in your repository for mserver5 in Sourceforge, but it is not currently active. Therefore, I have the following questions: a) Has X100 been integrated to the Mserver5 version of MonetDB, or included as an extension module? b) Does Mserver5 use vector processing as proposed in the paper "MonetDB/X100: Hyper-Pipelining Query Execution" (CIDR 2005) ? c) How is it possible to use MonetDB/X100 for benchmarking purposes? Are there any instructions online? Thank you.
I want to use MonetDB for benchmarking on the TPC-H workload. According to the paper "MonetDB/X100 - A DBMS In The CPU Cache", the X100 module achieves significant acceleration against MonetDB/MIL on Query 1 of TPC-H. Still, when I search your site, there is no reference about how it is possible to activate X100 inside MonetDB. I have only found X100 in your repository for mserver5 in Sourceforge, but it is not currently active. Therefore, I have the following questions: That code is not used at all, and was not the X100code itself, but a mimick of the code to see the differences and overhead of MAL interpretation a) Has X100 been integrated to the Mserver5 version of MonetDB, or included as an extension module? No, the X100 engine is a separately developed track, which is only used for research purposes only. The X100 developers so far have not been tempted to integrated the engine as a back-end for
Konstantinos Krikellas wrote: Dear Konstantinos, thank you for your interest in the MonetDB product family. the M5 software stack. The code base is far from being deployed out-of-the-box by others.
b) Does Mserver5 use vector processing as proposed in the paper "MonetDB/X100: Hyper-Pipelining Query Execution" (CIDR 2005) ? M5 shares the kernel with the M4 line, but has been significantly improved in terms of language interpretation overhead and overall performance due to much better optimization infrastructure.
The code for the Database Cracking is visible in the code being distributed, but has not been turned on, pending further integration and quality assurance.
c) How is it possible to use MonetDB/X100 for benchmarking purposes? Are there any instructions online? For this you have to contact Peter Boncz, the code owner. regards, Martin Kersten Thank you. ------------------------------------------------------------------------
------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ ------------------------------------------------------------------------
_______________________________________________ MonetDB-users mailing list MonetDB-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-users
participants (2)
-
Konstantinos Krikellas
-
Martin Kersten