Two Nvidia data center GPUs, one Pascal and one Turing: 24 GB GDDR5 against 16 GB GDDR6 and 8% more memory bandwidth on the P40.
At a glance
Updated 12 hours ago- Memory
- 24 GB
- GDDR5
- Memory bandwidth
- 346 GB/s
- From
- --
- Memory
- 16 GB
- GDDR6
- Memory bandwidth
- 320 GB/s
- From
- $0.42 /GPU/hr
- 1 provider
Price comparison
Cheapest listed price per GPU per hour by billing type, and the median across providers.
| Billing | P40 | T4G | Difference |
|---|---|---|---|
| On-demand | -- | $0.42 | |
| Reserved | No reserved listings | $0.18 | |
| Median | -- | $1.10 |
Median: Median across providers, one vote each, so a large catalog does not outweigh a small one.
Price history
P40 vs T4G price history
Weekly median on-demand price per GPU per hour
Specs comparison
| Spec | P40 | T4G | Difference |
|---|---|---|---|
| Memory | |||
| Capacity | 24 GB GDDR5 | 16 GB GDDR6 | 1.5x more on P40 |
| Bandwidth | 346 GB/s | 320 GB/s | 1.1x higher on P40 |
| Bus width | 384-bit | 256-bit | 1.5x wider on P40 |
| Compute · dense | |||
| FP16 / BF16 | 65 TFLOPS | ||
| INT8 | 47 TOPS | 130 TOPS | 2.8x higher on T4G |
| FP32 | 12 TFLOPS | 8.1 TFLOPS | 1.5x higher on P40 |
| Platform | |||
| Interconnect | PCIe only | PCIe only | |
| Board power | 250 W | 70 W | 3.6x higher on P40 |
| Released | Q3 2016 | Q3 2020 | 4 years newer on T4G |
| Precision | |||
| Supported formats | FP16FP32INT8 | FP16FP32INT4INT8 | |
Compute: Vendor peak figures, dense and per GPU. Not measured throughput.
Source: P40 datasheet and T4G datasheet.
Where to rent them
Providers listing these GPUs, those with both first, so you can switch without moving clouds.
| Provider | P40 /GPU/hr | T4G /GPU/hr |
|---|---|---|
AWS
|
-- | On-Demand $0.42 |
Similar GPUs
Other comparisons in the same class.
Our data for Nvidia P40 was last updated on Sept. 7, 2026, and for Nvidia T4G on Sept. 7, 2026.