IBlazingSQL ikhiphe ikhodi yayo yomthombo yokusetshenziswa kwe-GPUS ukusheshisa ukucutshungulwa kwedatha

Iphrojekthi entsha yomthombo ovulekile ifuna ukuthatha ama-analytics ayise ezingeni elilandelayo futhi ngukuthi abantu abasemuva IBlazingSQL isanda kumemezela ukuthi ikhiphe ikhodi yomthombo yenjini yayo ye-SQL, esetshenziswa kuma-GPU ukusheshisa ukucutshungulwa kwedatha. IBlazingSQL akuyona i-DBMS ephelele, kepha ibekwe njengenjini yokuhlaziya nokucubungula amasethi wedatha amakhulu, afaniswa nemisebenzi yawo ne-Apache Spark.

Kulabo abangajwayele iBlazingSQL kufanele bakwazi lokho le yinjini ye-SQL esheshiswe nge-GPU eyakhelwe kuhlelo lwe-RAPIDS okuyiqoqo lemitapo yolwazi yomthombo ovulekile wokusebenzisa ama-analytics wokuphela kokuphela namapayipi wesayensi yedatha kuma-GPU.

Ngokusho kweqembu, IBlazingSQL yadalelwa ukubhekana nezindleko, ubunzima nokuhamba kancane abasebenzisi ababhekana nakho lapho besebenza emihlanganweni emikhulu kwedatha. IBlazingSQL ilungele ukwenza imibuzo yomuntu ngamunye yokuhlaziya kumasethi wedatha amakhulu (amashumi ama-gigabytes) agcinwe kumafomethi we-tabular (isib. Izingodo, izibalo zeNetFlow, njll.).

Ukusebenza ne-GPU, kusetshenziswa iqoqo lemitapo yolwazi ye-RAPIDS abOkunye kuthuthukiswe ngokubandakanyeka kwe-NVIDIA, okukuvumela ukuthi wakhe izinhlelo zokucubungula nokuhlaziya ezisebenza ngokuphelele ohlangothini lwe-GPU (kunikezwa isikhombimsebenzisi sePython ukusebenzisa ama-primitives we-CUDA asezingeni eliphansi nokubala okufanayo).

IBlazingSQL inikeza amandla wokusebenzisa i-SQL esikhundleni se-API Ukucutshungulwa kwedatha ye-cuUDF (kususelwa ku-Apache Arrow) okusetshenziswa yi-RAPIDS. IBlazingSQL ungqimba olungeziwe olusebenza ngaphezulu kwe-cuDF futhi lisebenzisa umtapo wezincwadi we-cuIO ukufunda idatha kusuka kudiski.

Imibuzo ye-SQL yile ukuhumusha kumakholi womsebenzi we-cuUDF, avumela idatha ukuthi ilayishwe kwi-GPU futhi wenze imisebenzi yokuhlanganisa, yokuhlanganisa nokuhlunga kuyo. Isekela ukwenziwa kokulungiselelwa okusatshalaliswa okuhlanganisa izinkulungwane zama-GPU.

Ukusetshenziswa kwe I-SQL ivumela i-RAPIDS ukuthi ihlanganiswe nezinhlelo ezikhona zokuhlaziya ngaphandle kokubhala ama-processor ngaphandle kokusebenzisa ukulayishwa okuphakathi kwedatha kwi-DBMS eyengeziwe, ngenkathi kugcinwa ukuhambisana okugcwele nazo zonke izingxenye ze-RAPIDS, kuhunyushwa ukusebenza okukhona ku-SQL nokuqinisekisa ukusebenza kwezinga le-cuDF. Kufaka ukusekelwa kokuhlanganiswa nemitapo yolwazi ye-XGBoost kanye ne-cuML ukuxazulula imisebenzi yokuhlaziya neyokufunda ngomshini.

I-BlazingSQL ingaqhuba imibuzo emafayilini amafulethi ngamafomethi we-CSV ne-Apache Parquet etholakala ezinhlelweni zenethiwekhi namafu ezifana ne-HDSF ne-AWS S3, idlulisela ngqo umphumela kwimemori ye-GPU.

Ngenxa yokusebenza kokufanisa kwi-GPU nokusetshenziswa kwememori yevidiyo esheshayo, ukwenziwa kombuzo eBlazingSQL kushesha okuphindwe izikhathi ezingama-20 kune-Apache Spark.

IBlazingSQL yenza kube lula ukusebenza nedatha - esikhundleni samakhulu wezingcingo zomsebenzi we-cuDF, ungakwenza ngombuzo owodwa we-SQL.

"IBlazingSQL ibhekelela lokhu kukhathazeka kwamakhasimende hhayi ngenjini ye-SQL GPU esheshayo, kodwa futhi nokugxila ekushiseni," kusho uRodrigo Aramburu, oyi-CEO yeBlazingSQL, kubhulogi elandelayo. "Ngemigqa embalwa nje yekhodi, iBlazingSQL ingabuza idatha yakho eluhlaza, noma ngabe ihlala kuphi, futhi isebenzisane ne-RAPIDS yakho kanye nesitaki sezibalo."

I-BlazingSQL inika amandla abasebenzisi ukuthi babuze idatha yedatha yedatha yamabhizinisi isetha ngqo kwimemori ye-GPU njenge-GPU DataFrame (GDF). I-GDF yiphrojekthi enikezela ukusekelwa kokusebenzisana phakathi kwezicelo ze-GPU. Iphinde ichaze ungqimba lwedatha yememori ye-GPU ejwayelekile.

"Ngokusebenzisa umcibisholo we-Apache kuma-GPU nokuhlanganisa ne-Dask, iBlazingSQL izothuthukisa ukusebenza komthombo ovulekile futhi idlulise igagasi elilandelayo lokusebenzisana kuhlelo olusheshayo lwe-data science ecosystem."

Kulabo abanentshisekelo kufanele bakwazi lokho ikhodi ibhalwe ku-C ++ nge-python interface yabasebenzisi futhi umthombo ovulekile ungaphansi kwelayisense le-Apache 2.0.

Isixhumanisi yilokhu.


Yiba ngowokuqala ukuphawula

Shiya umbono wakho

Ikheli lakho le ngeke ishicilelwe. Ezidingekayo ibhalwe nge *

*

*

  1. Ubhekele imininingwane: Miguel Ángel Gatón
  2. Inhloso yedatha: Lawula Ugaxekile, ukuphathwa kwamazwana.
  3. Ukusemthethweni: Imvume yakho
  4. Ukuxhumana kwemininingwane: Imininingwane ngeke idluliselwe kubantu besithathu ngaphandle kwesibopho esisemthethweni.
  5. Isitoreji sedatha: Idatabase ebanjwe yi-Occentus Networks (EU)
  6. Amalungelo: Nganoma yisiphi isikhathi ungakhawulela, uthole futhi ususe imininingwane yakho.