Alenka
Data
database engine
SHOW HN2012
364points
60comments
#24of 1,677 in 2012
Top 2%of 2012 by points
How HN took it
MixedNo clear lean
6 positive, 10 neutral or asking, 1 mixed, 4 critical, out of 21 comments about the project. Thread score +0.13.
Idea: 6 for, 0 againstDocs: 0 for, 1 againstPerformance: 0 for, 1 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Impressive GPU database idea and implementation
From 4 comments
Very, very cool. I'm always happy to see more ways to utilize the tremendous performance/price of GPUs. So, what are the pros and cons of this approach? For a given hardware and…
corysama on HNI love this kind of novel thinking. I'd definitely like to read more detailed articles on how this works.
jrmg on HNIf you look up columnar databases, you can find a whole host of research on how these things store data. Storing values column by column to do vector processing is pretty common,…
jdf on HNgithub would be better for browsing. But the source is only 117k - I'm surprised how much of a barrier a few clicks and a few seconds was for me. (Partly, I expected it to be…
6ren on HN
Objections
Column stores are poor for writes and transactional work
From 1 comment
Column store, compare it with MonetDB (open source) and similar. This design is suitable for read-mostly databases. It's a trade-off, faster reads and much better compression.…
alecco on HN
GPU transfer overhead and disk streaming limit performance on large data
From 2 comments
Congratulations, well done! Our team has built an in-memory DB (hybrid, both column and row store available, also using advanced compression). We have investigated GPUs for…
mrich on HNLooks like a very interesting approach to me given the recent developments. Still I have a few questions regarding your numbers. Can you specify how much data you actually stream…
grundprinzip on HN
SourceForge hosting is hard to browse
From 3 comments
Excellent work! I have a very small (marketing-related) nitpick: Would you consider hosting your project on github instead of SourceForge? I don't say this to be trendy, github…
breckinloggins on HNAnother practical consideration: I clicked on the "manual.txt" file and SF redirected me to a download page. I would have preferred to read the file in-place (like on github).
davvid on HNI second this. I'd rather be at the dentist than on a SourceForge page. I swear it's the "myspace.com" of open source hosting.
astrodust on HN
Requests
Move the project to GitHub
From 2 comments
Excellent work! I have a very small (marketing-related) nitpick: Would you consider hosting your project on github instead of SourceForge? I don't say this to be trendy, github…
breckinloggins on HNAnother practical consideration: I clicked on the "manual.txt" file and SF redirected me to a download page. I would have preferred to read the file in-place (like on github).
davvid on HN
More detailed documentation and overviews
From 2 comments
Interesting. I wonder how a GPU could best be put to use in a database engine. Joins, I suppose? Would be cool to read more about this implementation, however the readme is a bit…
ericflo on HNgithub would be better for browsing. But the source is only 117k - I'm surprised how much of a barrier a few clicks and a few seconds was for me. (Partly, I expected it to be…
6ren on HN
Questions
Does it work with ATI/AMD GPUs or only NVIDIA?
From 1 comment
nvidia gpus only, right? does it work with ati as well? CUDA is probably more advanced, so it's no surprise.
2AM on HN
Why CUDA rather than OpenCL?
From 2 comments
antonmks, did you consider using OpenCL instead? If yes - why CUDA outweighed it?
pavelkaroukin on HNWhether it's still relevant or not CUDA has a lot of mindshare in the academic community. Partially this is due to marketing and having their product out first, but also OpenCL…
mappu on HN
Whether the database is stored on disk or in memory
From 2 comments
That looks absolutely brilliant. So is the database all stored in the GPU memory? Or in main memory? Or does your engine handle storing to disk as well? Pretty novel use of a GPU…
chubs on HNThank you ! The data are stored on a disk and only the columns necessary for transformations are brought to GPU. In case the data does not fit into a limited GPU memory, it is…
antonmks on HN
Alternatives named
MonetDB
From 2 comments
Column store, compare it with MonetDB (open source) and similar. This design is suitable for read-mostly databases. It's a trade-off, faster reads and much better compression.…
alecco on HNIf you look up columnar databases, you can find a whole host of research on how these things store data. Storing values column by column to do vector processing is pretty common,…
jdf on HN
SAP HANA
From 1 comment
You can find info here (unfortunately not just an easy download) http://www.sap.com/hana/index.epx
mrich on HN
PGStrom
From 1 comment
This is awesome. :) Someone did something similar for PostgreSQL here: http://wiki.postgresql.org/wiki/PGStrom
foobarbazetc on HN
kdb/Q and Kona
From 1 comment
Super cool, and thanks for sharing! Are you familiar with APL/K/Q/J ? These implement column store databases which a programming language significantly more expressive and…
beagle3 on HN