NVIDIA GPUs whakatapua me te whakawhitinga-metal. Ka whakaterena ngā tauira, ka whakahaere i te whakawāteatanga, te whakawāteatanga ataata, me te whakatere i te rorohiko pūtaiao i runga i te tono.
+-------------------------+
| NVIDIA-SMI 550.54.15 |
| Driver: 550.54.15 |
| CUDA Version: 12.4 |
+-------------------------+
| GPU Name | Mem |
| 0 Tesla P40 | 24GB |
+-------------------------+
| GPU Util: 0% |
| Memory: 0MiB / 24576MiB|
+-------------------------+
$ python train.py --model llama
Loading model... _
NVIDIA GPUs whakatapua me te PCIe katoa hei mahi nui rawa
E whakawhānui ana mātou i a tātau GPU. Ka wātea mai ngā tauira NVIDIA GPU tāpiri me te VRAM tiketike ake me ngā hanganga hōu ake.
E hiahiatia ana he GPUs ā-kaupapa-wāhanga (A100, H100), he whakaritenga GPU-maha rānei? Tātai ki a mātou hei kōrerorero i ōna hiahia.
Mai i ngā whatunga ā-ira whakaakona ki te whakarārangi i ngā tirohanga 3D, ka whakahaeretia e te GPU VPS ērā katoa
Mā te whakaako tauira ako ā-roto me PyTorch, TensorFlow, JAX rānei. Mā te whakamātautau i ngā tauira reo nui pēnei i te LLaMA, Mistral, me te Stable Diffusion i runga i ngā pūrere GPU motuhake.
Ka tīmata →Whakataunga ngā tauira AI i roto i te whakanaotanga me te whakawāteatanga GPU iti. Mā te LLMs, te whakawhanaketanga whakaahua, te mōhiotanga reo, me ngā API tirohanga rorohiko i te tauine.
Ka tīmata →Ka tere ake te whakarerekētanga o te ataata me NVENC, te whakarārangi i ngā tirohanga 3D me Blender, Maya rānei, me te tukanga i ngā pūwhitinga pāpāho ātaahua me te whakarerekētanga GPU-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā
Ka tīmata →Ka whakahaeretia ngā hihiri rāpoi ngota, ngā whakamātautau āhuarangi, ngā hihiri waikawa rorohiko, me ētahi atu utu mahi HPC e whakateretia ana e ngā GPUs CUDA-ka taea.
Ka tīmata →Ka tere ake ngā pūrere raraunga me RAPIDS, cuDF, me cuML. Ka tuwhera i ngā huinga raraunga nui rawa, ka whakahaere tātaritanga GPU-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-ā-
Ka tīmata →Ka hanga, ka whakawātea i ngā taupānga AI me ngā pukapuka Jupyter, te waewae o te waehere VS, me ngā taiao whanaketanga CUDA katoa. He tere ake te whakawhanaketanga me ngā rauemi GPU whakatapua.
Ka tīmata →Ko te tikanga o te PCIe tika e kore ana te whakawāteatanga o te whakawāteatanga. E kōrero hāngai ana tōtou waehere ki ngā pūrere GPU.
Kāore he whakatūnga matatini, he whakapāpāho rānei. Hiko i tōna GPU, kōwhiri i tētahi OS, me te whakawātea i roto i ngā minu.
Ko te utu marama kāore he kānataraki. Whakapae ki te wā e hiahiatia ana e koe ngā GPUs, whakaiti i te wā kāore.
Ka whakatū i tētahi anga, putanga kaiwhakaari, CUDA toolkit rānei. Ko tōtou pūnaha, ko tōtou ture. Ka uru ki a SSH mai i te rā tuatahi.
Ko ngā tohutohu kua whakaritea i mua mō ngā anga AI/ML rongonui, ngā utauta hanga, me ngā taupānga whakatere GPU
Ko ngā mea katoa e hiahiatia ana e koe kia mōhio ki te GPU VPS whakanohotanga
Ko te GPU VPS he pūnaha pūnaha ā-pūnaha e whai ana i ngā pūrere NVIDIA GPU. E whakarato ana i te āhei ōrite ki tētahi VPS paerewa engari me ngā āheinga tukanga whakarara kaha mō te whakaakoranga AI / ML, te whakawāteatanga, te whakawāteatanga, me ngā taumahi rorohiko pūtaiao.
I tēnei wā e tuku ana mātou i ngā GPU NVIDIA Tesla P40 me te 24GB o VRAM ia. Ko ēnei GPU e tino pai ana i te AI / ML, te tauira, te whakarerekētanga o te tāurunga, me te rorohiko pūtaiao. Ka tāpirihia ētahi atu tauira GPU i a mātou e whakawhānui ana i a tātau hanganga.
Ka hoatu e te GPU te āheitanga ki tōna VPS, te āheitanga mārō ki ngā pūrere GPU ōkiko. Kāore i te rite ki ngā GPU tiritiri, whakawātea rānei, ka whiwhi i te mahi katoa o te GPU me te kore utu. Ka whakawhitiwhiti ā-GPU ō rātou taupānga pērā ki te mahi i runga i tētahi mīhini whakatapua.
Ko ngā tauira GPU VPS e tae ana ki Ubuntu me te kōwhiringa kia whakatū i ngā kaiārahi NVIDIA me te CUDA toolkit. Ka taea hoki e koe te whakatū i tōna putanga kaiārahi e manakohia ana. Ka whakarato tātuhi mātou mō te whakatū i ngā anga ML rongonui pēnei i a PyTorch me te TensorFlow.
Heoi, ka taea e koe te hono i ngā GPU maha ki tētahi VPS kotahi mō ngā utu mahi e whai hua ana i te whakaakoranga GPU maha, i te tukanga whakarara rānei. Tātai ki tātau tīma hoko mō ngā whakaritenga GPU maha me te utu.
He tino pai te GPU VPS mō te whakaakoranga tauira AI/ML me te whakawāteatanga, te whakamāramatanga mōhiotanga o te tauira reo nui, te kitenga rorohiko, te whakarerekētanga me te whakawāteatanga ataata, ngā whakamātautau pūtaiao, me ētahi atu taumahi e whai hua ana i te hiko whakatinana whakahuatahi nui.
Ko te GPU VPS e whakawhiwhia ana ki te pūtake o te marama, pēnei i a tātau kaupapa VPS paerewa. E utu ana anake koe mo te wā e whakaritea ana tō tātau pūnaha. Tātai mai ki a mātou mō te utu o nāianei i te mutunga o tātau whakawhiwhinga GPU.
Ka haere ngā tauira GPU VPS i runga i ngā pūrere GPU-kaupapa, nō reira ko te whakawhānuitanga mai i tētahi VPS paerewa e hiahiatia ana kia whakaritea he pūnaha hōu. Heoi anō, ka taea e koe te neke atu i tōtou raraunga me ngā whakaritenga. Ka taea e tātau rōpū tautoko te āwhina ki te tukanga neke.
Heoi, ko ngā kaupapa VPS katoa tae atu ki te GPU VPS e whai ana i te parenga DDoS taumata-wāhi, te āheitanga pūtake katoa, te hono IPv4 me te IPv6, me ngā āhuatanga haumarutanga ōrite ki a tātau VPS Cloud paerewa e whakarato ana.
Ka taea te GPU VPS me Ubuntu 22.04 LTS me Ubuntu 24.04 LTS, he tautoko pai ake a NVIDIA. E tautoko ana hoki a Debian 12. E whakarongo ana tātau i a Ubuntu 24.04 LTS mō te tauritetanga o te utauta CUDA hōu.
Kāore, kāore he whakawhiwhinga iti rawa. Ka whakawhiwhia te GPU VPS i ia marama, ā, ka taea e koe te whakakore i tētahi wā. E whakapono ana tātau ki te whakawātea, ki te whakawātea i te pūkoro mārō.
He ngāwari noa iho te waihanga i tētahi kāwanatanga, te kōwhiri i tētahi whakaritenga GPU VPS, te kōwhiri i tōna pūnaha whakahaere, me te whakawhānui. Ka noho tōna pūnaha GPU-powered i ngā minu me te āheitanga pūtake katoa me tōna GPU e wātea ana mā nvidia-smi.
Ka whiwhi hiko NVIDIA GPU whakatapua i roto i ngā minu. Kāore he whakawhiwhinga, kāore he whakawhiwhinga.