1× A4048 GB VRAM
Germany · 12 vCPU · 56 GB RAM · Storage billed separately
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
gpu3.small
Full price
€1.05
per hour, whole instance, 1 GPU
NVIDIA · Ampere
Rent A40
NVIDIA A40, 48 GB.RenderingVirtual desktopsgeneral ml
Full price from
—
in stock, per GPU-hour
Spot from
—
in stock, per GPU-hour
Offers in stock
0 of 16
full price or spot
Providers in stock
0 of 3
full price or spot
Offers (17)
View offer
4× A4048 GB each, 192 GB total
Germany · 48 vCPU · 224 GB RAM · Storage billed separately
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
This GPU supports NVLink, but the provider doesn't say whether this server has it connected. Ask before relying on it.
gpu3.large
Full price
€4.18
per hour, whole instance, 4 GPUs
€1.05/h per GPU
8× A4048 GB each, 384 GB total
Germany · 96 vCPU · 448 GB RAM · Storage billed separately
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
This GPU supports NVLink, but the provider doesn't say whether this server has it connected. Ask before relying on it.
gpu3.huge
Full price
€8.36
per hour, whole instance, 8 GPUs
€1.05/h per GPU
2× A4048 GB each, 96 GB total
Germany · 24 vCPU · 120 GB RAM · Storage billed separately
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
This GPU supports NVLink, but the provider doesn't say whether this server has it connected. Ask before relying on it.
gpu3.medium
Full price
€2.09
per hour, whole instance, 2 GPUs
€1.05/h per GPU
1× A4048 GB VRAM
Netherlands · 48 vCPU · 384 GB RAM · 35 Gbit/s
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
g7.12xlarge.8
Full price
€3.58
per hour, whole instance, 1 GPU
1× A4048 GB VRAM
Netherlands · 48 vCPU · 384 GB RAM · 35 Gbit/s
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
g7.12xlarge.8
Full price
€3.58
per hour, whole instance, 1 GPU
2× A4048 GB each, 96 GB total
Netherlands · 96 vCPU · 768 GB RAM · 40 Gbit/s
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
This GPU supports NVLink, but the provider doesn't say whether this server has it connected. Ask before relying on it.
g7.24xlarge.8
Full price
€7.16
per hour, whole instance, 2 GPUs
€3.58/h per GPU
2× A4048 GB each, 96 GB total
Netherlands · 96 vCPU · 768 GB RAM · 40 Gbit/s
Details
Stock and price checked 14 min ago.
You start it yourself once there is stock.
This GPU supports NVLink, but the provider doesn't say whether this server has it connected. Ask before relying on it.
g7.24xlarge.8
Full price
€7.16
per hour, whole instance, 2 GPUs
€3.58/h per GPU
Full price
from €3.58
per GPU, per hour
Price and availability history
Price per GPU-hour
Excluding VAT.
Providers with stock
Each provider counts once. Full price and spot are separate.
Share of offers in stock
Full price. Offers we can't say are really sold there are left out until they've been in stock once.
Units free
Only providers that report a number. Each line is one provider and one server size, because their counts cover different things and can't be added up. Providers that only say yes or no are under Stock by provider.
Stock by provider
In stock, yes or no, for every provider selling this GPU. A filled band means in stock: green at full price, orange as spot.
NosanaSold out
Reports a number
ExoscaleNo stock data
No stock data shown
T Cloud PublicNo stock data
No stock data shown
Using several together
Can be linked in pairs with an NVLink bridge at up to 112.5 GB/s, if the provider has fitted one. Most providers don't say, so ask before relying on it. Software still has to split the work, so the memory doesn't act as one big pool.
Specifications
3x DisplayPort 1.4 present but disabled by default (configured for virtualization); NVLink pairs two cards to 96 GB. fp64Tflops is derived: NVIDIA states GA102 FP64 is 1/64th of FP32 (37.4 / 64). interconnectPerLinkGBs is bidirectional (14.0625 GB/s each direction per link).
Memory
- VRAM
- 48 GB
- Memory type
- GDDR6
- Memory bandwidthHow fast the GPU reaches its memory. Often limits LLM speed
- 696 GB/s
- Bus width
- 384 bit
- ECC memoryDetects and corrects memory errors
- Yes
Compute
- FP64Speed at 64-bit precision, used for scientific simulation
- 0.58 TFLOPS
- FP32Speed at 32-bit precision
- 37.4 TFLOPS
- FP16Speed at 16-bit precision
- 299.4 TFLOPS
- BF1616-bit AI format. Standard for training and full-quality inference
- 299.4 TFLOPS
- INT88-bit integer speed, used by quantized models
- 598.6 TOPS
- FP8 support
- No
- FP4 support
- No
- Transformer EngineHardware that speeds up FP8 and FP4
- No
- Compute cores
- 10,752
Multi-GPU and connectivity
- Interconnect
- NVLink
- Interconnect bandwidthHow fast linked GPUs exchange data
- 112.5 GB/s
- Links
- 4
- Per link
- 28.125 GB/s
- Max GPUs in fabric
- 2
- PCIe generation
- 4
- PCIe lanes
- 16
- PCIe bandwidth
- 64 GB/s
Sharing and security
- MIG partitioningSplits one GPU into several isolated smaller GPUs
- No
- Max MIG instances
- 0
- vGPU
- Yes
- Confidential computing
- No
Hardware
- Architecture
- Ampere
- Launch year
- 2020
- Process node
- Samsung 8N
- Transistors
- 28.3 billion
- Die size
- 628.4 mm²
- Power (TDP)Maximum power draw
- 300 W
- Cooling
- passive
Video
- NVENC engines
- 1
- NVDEC engines
- 2
- AV1 encode
- No
- AV1 decode
- Yes
Sources (46)
- https://images.nvidia.com/content/Solutions/data-center/a40/nvidia-a40-datasheet.pdf
- https://nvidianews.nvidia.com/_gallery/download_pdf/5fb5906ced6ae50b88b2c862/
- https://www.nvidia.com/en-us/data-center/solutions/confidential-computing/
- https://docs.nvidia.com/590trd1-trusted-computing-solutions-release-notes.pdf
- https://developer.nvidia.com/blog/nvidia-hopper-architecture-in-depth/
- https://developer.nvidia.com/blog/inside-nvidia-blackwell-ultra-the-chip-powering-the-ai-factory-era/
- https://images.nvidia.com/aem-dam/Solutions/geforce/ada/nvidia-ada-gpu-architecture.pdf
- https://developer.nvidia.com/video-encode-decode-support-matrix
- https://docs.nvidia.com/vgpu/gpus-supported-by-vgpu.html
- https://www.nvidia.com/content/PDF/nvidia-ampere-ga-102-gpu-architecture-whitepaper-v2.pdf
- https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Center/a40/NVIDIA%20A40%20Product%20Brief.pdf