GPU Cloud
I-GPU cloud: qasha amaseva we-NVIDIA GPU ngehora
Qasha amaseva we-NVIDIA GPU we-LLM inference, i-fine-tuning kanye nokukhiqiza izithombe. Amanani abukhoma, ukukhokhisa ngehora, izifanekiso ze-PyTorch, vLLM kanye ne-ComfyUI, nokungena nge-SSH.
- Izinsiza ezikhona bukhoma, intengo iqinisekiswa ngaphambi koku-oda
- Izifanekiso ze-PyTorch, vLLM, Ollama kanye ne-ComfyUI
- Amaseva ehora amisiwe akhokhelwa i-disk kuphela
Nge-Cloudj GPU Cloud uqasha iseva ye-GPU ohlwini olukhona bukhoma: khetha imodeli ye-GPU, i-VRAM kanye nediski, qala ngethempulethi efana ne-PyTorch, vLLM, Ollama noma ComfyUI, bese uxhuma nge-SSH. Amaseva ehora akhokhiswa kusuka ebhalansini ye-akhawunti yakho, kanti iseva yamahora emisiwe ikhokhiselwa kuphela idiski yayo.
Amanani okuqasha i-GPU nokutholakala kwazo bukhoma
Uhlu olukhona bukhoma: ukutholakala kanye namanani kuncike emakethe. I-console iqinisekisa intengo yokugcina ngaphambi kokuba u-ode.
| GPU | VRAM nge-GPU ngayinye | Architecture | Ukubala kwe-FP32 | Umkhawulo wamandla | Kusukela | |
|---|---|---|---|---|---|---|
| NVIDIA RTX PRO 6000 Max-Q | 96 GB | Blackwell | ≈ 119 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA RTX PRO 6000 WS | 96 GB | Blackwell | ≈ 119 TFLOPS | 600 W | – | Imininingwane → |
| NVIDIA RTX 5090 | 32 GB | Blackwell | ≈ 107.6 TFLOPS | 575 W | – | Imininingwane → |
| NVIDIA RTX 5090D | 32 GB | Blackwell | ≈ 107.6 TFLOPS | 575 W | – | Imininingwane → |
| NVIDIA RTX PRO 6000 S | 96 GB | Blackwell | ≈ 93.6 TFLOPS | 600 W | – | Imininingwane → |
| NVIDIA RTX 4090 | 24 GB | Ada Lovelace | ≈ 81.4 TFLOPS | 450 W | – | Imininingwane → |
| NVIDIA RTX 6000Ada | 48 GB | Ada Lovelace | ≈ 81.4 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA RTX 6000D | 84 GB | Blackwell | ≈ 77.6 TFLOPS | 600 W | – | Imininingwane → |
| NVIDIA L40S datacenter | 45 GB | Ada Lovelace | ≈ 73.3 TFLOPS | 350 W | – | Imininingwane → |
| NVIDIA RTX PRO 5000 | 48 GB | Blackwell | ≈ 69.6 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA H100 SXM datacenter | 80 GB | Hopper | ≈ 53.5 TFLOPS | 700 W | – | Imininingwane → |
| NVIDIA RTX 5080 | 16 GB | Blackwell | ≈ 53.4 TFLOPS | 360 W | – | Imininingwane → |
| NVIDIA RTX 4080S | 16 GB | Ada Lovelace | ≈ 50.9 TFLOPS | 320 W | – | Imininingwane → |
| NVIDIA RTX 4080 | 16 GB | Ada Lovelace | ≈ 48.6 TFLOPS | 285 W | – | Imininingwane → |
| NVIDIA H100 NVL datacenter | 94 GB | Hopper | ≈ 48.3 TFLOPS | 400 W | – | Imininingwane → |
| NVIDIA H200 NVL datacenter | 140 GB | Hopper | ≈ 48.3 TFLOPS | 600 W | – | Imininingwane → |
| NVIDIA RTX 5070 Ti | 16 GB | Blackwell | ≈ 44.3 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA RTX PRO 4000 | 24 GB | Blackwell | ≈ 44.3 TFLOPS | 145 W | – | Imininingwane → |
| NVIDIA RTX 4070S Ti | 16 GB | Ada Lovelace | ≈ 42.6 TFLOPS | 285 W | – | Imininingwane → |
| NVIDIA H100 PCIE datacenter | 80 GB | Hopper | ≈ 41 TFLOPS | 350 W | – | Imininingwane → |
| NVIDIA RTX 4500Ada | 24 GB | Ada Lovelace | ≈ 38.2 TFLOPS | 210 W | – | Imininingwane → |
| NVIDIA RTX 3090 Ti | 24 GB | Ampere | ≈ 36.4 TFLOPS | 450 W | – | Imininingwane → |
| NVIDIA RTX A6000 | 48 GB | Ampere | ≈ 36.1 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA RTX 4070S | 12 GB | Ada Lovelace | ≈ 35.8 TFLOPS | 220 W | – | Imininingwane → |
| NVIDIA RTX 3090 | 24 GB | Ampere | ≈ 35.4 TFLOPS | 350 W | – | Imininingwane → |
| NVIDIA RTX 3080 Ti | 12 GB | Ampere | ≈ 35.1 TFLOPS | 350 W | – | Imininingwane → |
| NVIDIA A40 datacenter | 45 GB | Ampere | ≈ 29.9 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA RTX 3080 | 10 GB | Ampere | ≈ 29.5 TFLOPS | 380 W | – | Imininingwane → |
| NVIDIA RTX 4070 | 12 GB | Ada Lovelace | ≈ 29.3 TFLOPS | 200 W | – | Imininingwane → |
| NVIDIA L4 datacenter | 22 GB | Ada Lovelace | ≈ 24.2 TFLOPS | 72 W | – | Imininingwane → |
| NVIDIA RTX 5060 Ti | 16 GB | Blackwell | ≈ 22.8 TFLOPS | 180 W | – | Imininingwane → |
| NVIDIA RTX 4060 Ti | 16 GB | Ada Lovelace | ≈ 21.6 TFLOPS | 165 W | – | Imininingwane → |
| NVIDIA RTX A4000 | 16 GB | Ampere | ≈ 20.6 TFLOPS | 140 W | – | Imininingwane → |
| NVIDIA RTX 3070 | 8 GB | Ampere | ≈ 19.8 TFLOPS | 220 W | – | Imininingwane → |
| NVIDIA RTX 3060 Ti | 8 GB | Ampere | ≈ 15.7 TFLOPS | 200 W | – | Imininingwane → |
| NVIDIA A100 PCIE datacenter | 40 GB | Ampere | ≈ 15.6 TFLOPS | 250 W | – | Imininingwane → |
| NVIDIA A100 SXM4 datacenter | 40 GB | Ampere | ≈ 15.6 TFLOPS | 400 W | – | Imininingwane → |
| NVIDIA A800 PCIE datacenter | 80 GB | Ampere | ≈ 15.6 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA RTX 2080 Ti | 11 GB | Turing | ≈ 14.6 TFLOPS | 250 W | – | Imininingwane → |
| NVIDIA RTX 3060 laptop | 12 GB | Ampere | ≈ 12.9 TFLOPS | 80 W | – | Imininingwane → |
| NVIDIA Tesla V100 datacenter | 16 GB | Volta | ≈ 12.5 TFLOPS | 300 W | – | Imininingwane → |
| NVIDIA CMP 170HX datacenter | 64 GB | Ampere | ≈ 12.1 TFLOPS | 250 W | – | Imininingwane → |
| NVIDIA RTX 3060 | 12 GB | Ampere | ≈ 12 TFLOPS | 170 W | – | Imininingwane → |
| NVIDIA Titan Xp | 12 GB | Pascal | ≈ 11.7 TFLOPS | 160 W | – | Imininingwane → |
| NVIDIA RTX A2000 | 6 GB | Ampere | ≈ 11.2 TFLOPS | 70 W | – | Imininingwane → |
| NVIDIA GTX 1080 Ti | 11 GB | Pascal | ≈ 11 TFLOPS | 250 W | – | Imininingwane → |
| NVIDIA GTX 1080 | 8 GB | Pascal | ≈ 7.8 TFLOPS | 130 W | – | Imininingwane → |
| NVIDIA GTX 1070 Ti | 8 GB | Pascal | ≈ 7.4 TFLOPS | 180 W | – | Imininingwane → |
| NVIDIA GTX 1650 | 4 GB | ≈ undefined TFLOPS | W | – | Imininingwane → | |
| NVIDIA Q RTX 8000 | 45 GB | Turing | ≈ 11.9 TFLOPS | 250 W | – | Imininingwane → |
| NVIDIA Tesla P40 datacenter | 24 GB | Pascal | ≈ 9.4 TFLOPS | 250 W | – | Imininingwane → |
| NVIDIA RTX 2070S | 8 GB | Turing | ≈ 8.8 TFLOPS | 210 W | – | Imininingwane → |
| NVIDIA Q RTX 4000 | 8 GB | Turing | ≈ 7.7 TFLOPS | 125 W | – | Imininingwane → |
| NVIDIA RTX 2070 | 8 GB | Turing | ≈ 7.7 TFLOPS | 150 W | – | Imininingwane → |
| NVIDIA RTX 2060S | 8 GB | Turing | ≈ 7.5 TFLOPS | 175 W | – | Imininingwane → |
| NVIDIA GTX TITAN X | 12 GB | Maxwell | ≈ 6.8 TFLOPS | 150 W | – | Imininingwane → |
| NVIDIA RTX 2060 | 12 GB | Turing | ≈ 6.5 TFLOPS | 130 W | – | Imininingwane → |
| NVIDIA GTX 1660 Ti | 6 GB | Turing | ≈ 5.3 TFLOPS | 120 W | – | Imininingwane → |
| NVIDIA GTX 1660 S | 6 GB | Turing | ≈ 3.6 TFLOPS | 125 W | – | Imininingwane → |
Izifanekiso ze-GPU zemisebenzi ye-AI, LLM kanye ne-rendering
I-LLM inference
vLLM, SGLang noma Ollama eno-Open WebUI yokusabalalisa nokuhlola amamodeli olimi.
I-Fine-tuning
Unsloth Studio yokwenza i-fine-tuning yamamodeli olimi kudatha yakho siqu.
Ukwakhiwa kwezithombe namavidiyo
ComfyUI ye-Flux, SDXL kanye nokugeleza komsebenzi wamavidiyo.
Ukuthuthukiswa kwe-PyTorch kanye ne-CUDA
Izithombe ze-PyTorch ne-Ubuntu 24.04 ezine-NVIDIA CUDA, ezitholakala nge-SSH.
I-desktop noma i-VM ephelele
I-desktop ye-Ubuntu kusiphequluli, noma umshini obonakalayo we-KVM ogcwele uma uwudinga.
Indlela yokuqasha iseva ye-GPU ngezinyathelo ezi-4
- 01
Khetha ithempulethi nosayizi wediski
- 02
Khetha i-GPU ehambisanayo ohlwini olukhona bukhoma
- 03
Qinisekisa intengo bese udala iseva kusuka ebhalansini ye-akhawunti yakho
- 04
Xhuma nge-SSH noma nge-terminal yesiphequluli
Amanani kanye nokukhokhisa kwe-GPU cloud
- Ngehora: i-GPU ikhokhiswa ngehora ngenkathi iseva isasebenza. Iseva yamahora emisiwe ikhokhiselwa kuphela idiski yayo.
- Imijikelezo yokukhokha: ngenyanga, njalo ngekota, kabili ngonyaka noma ngonyaka, kanti nanoma isiphi isaphulelo siboniswa ngaphambi kokuba ukhokhe.
- Isitoreji sediski sikhokhiswa nge-GB ngayinye kanyekanye nentengo ye-GPU.
- Amanani asekelwe ku-USD; ungawabonisa nge-VND ngenani lamanje lokushintshisana kwemali.
Imibuzo Evame Ukubuzwa nge-GPU cloud
Imisebenzi elingana kumemori ye-GPU: i-LLM inference kanye ne-fine-tuning, ukukhiqiza izithombe namavidiyo, i-rendering nokuthuthukiswa kwe-CUDA. Izifanekiso ze-PyTorch, vLLM, SGLang, Ollama, Unsloth kanye ne-ComfyUI zikufakela isoftware ngokuzenzakalelayo.
Ithebula elingenhla libala ukulungiselelwa okutholakala njengamanje, okuhlanganisa ne-NVIDIA RTX 5090 ene-32 GB ye-VRAM uma ikhona. Ukutholakala kuyashintsha usuku lonke.
Amanani avela ohlwini lwempahla ye-GPU ebukhoma futhi angashintsha. I-console ikhombisa intengo eqinisekisiwe ngaphambi koku-oda, kanti iseva yakho ikhokhiswa ngaleyo ntengo eqinisekisiwe.
Yebo. Ngaphandle kokukhokha ngehora ungakhetha umjikelezo wanyanga zonke, wekota, wezinyanga eziyisithupha noma wonyaka, kanti nanoma isiphi isaphulelo sikhonjiswa ngaphambi kokuba ukhokhe. Isitoreji se-disk sikhokhiswa nge-GB kanye nentengo ye-GPU.
Ngokukhokhisa ngehora, iseva emisiwe ikhokhiselwa kuphela i-disk yayo. Susa iseva lapho ungasayidingi idatha yayo.
Nge-SSH usebenzisa ukhiye wakho, noma ngetheminali esipheqululini sakho kusuka ku-console. Izifanekiso ze-Desktop ne-VM nazo zinikeza ukungena kokubonakalayo.