# GP104

> 2016 graphics processing unit manufactured by Nvidia

**Wikidata**: [Q135392906](https://www.wikidata.org/wiki/Q135392906)  
**Source**: https://4ort.xyz/entity/gp104

## Summary
GP104 is a graphics processing unit (GPU) model manufactured by Nvidia and published on 2016-05-27. It is based on Nvidia’s Pascal microarchitecture and is associated with the working title “NV134.”

## Key Facts
- GP104 is an instance of a **graphics processing unit model**.
- GP104 is **manufactured by Nvidia**.
- GP104 has a **publication date of 2016-05-27**.
- GP104 uses the **Pascal** microarchitecture.
- GP104 is fabricated using a **16 nm lithography process**.
- GP104 has a **maximum thermal design power (TDP) of 200 W**.
- GP104 has **2560 shading units**, **160 texture mapping units**, and **64 render output units**.
- GP104 cache capacities include **48 KB (L1 cache)** and **2048 MB (L2 cache)**.
- GP104 supports/uses multiple APIs and compute platforms with listed versions: **DirectX 12 (12_1)**, **OpenGL 4.6**, **OpenCL 3.0**, **Vulkan 1.3**, **CUDA 6.1**, and **Shader Model 6.8**.
- GP104’s country of origin is the **United States**.

## FAQs
### Q: What is GP104?
A: GP104 is an Nvidia-manufactured GPU model published in 2016 (2016-05-27). It is based on the Pascal microarchitecture and was associated with the working title NV134.

### Q: What graphics/compute APIs does GP104 support?
A: GP104 is listed as using DirectX 12 (12_1), OpenGL 4.6, OpenCL 3.0, Vulkan 1.3, CUDA 6.1, and Shader Model 6.8. These indicate the graphics and compute interfaces it is designed to work with.

### Q: What are GP104’s core hardware unit counts?
A: GP104 is specified with 2560 shading units, 160 texture mapping units, and 64 render output units. These are key architectural counts used to describe GPU capability.

### Q: What process node and power level is GP104 associated with?
A: GP104 is fabricated on a 16 nm lithography process. Its thermal design power is listed as 200 W (maximum).

## Why It Matters
GP104 matters as a defined Nvidia GPU model from 2016 that represents an implementation of the Pascal microarchitecture on a 16 nm lithography process. In practical terms, it is a concrete hardware target characterized by specific architectural unit counts (shading units, texture mapping units, and render output units), cache sizes, and a maximum TDP value, all of which are commonly used to compare GPU designs and understand performance/power envelopes at a high level.

It is also relevant because it is explicitly associated with a set of graphics and compute APIs and versions—DirectX, OpenGL, Vulkan, OpenCL, CUDA, and Shader Model—indicating the software interfaces it is intended to work with. That makes GP104 useful in contexts such as hardware capability cataloging, driver/API compatibility discussions, and technical documentation that needs to map a GPU model to supported programming and rendering interfaces. Finally, its identification in a GPU specifications database (TechPowerUp GPU specs database ID g793) provides a stable reference point for cross-linking and lookup.

## Notable For
- A **Pascal**-microarchitecture Nvidia GPU model published on **2016-05-27**.
- Fabrication on a **16 nm lithography process**.
- A listed **maximum TDP of 200 W**.
- Explicitly listed usage/support for **DirectX 12 (12_1)**, **Vulkan 1.3**, **OpenGL 4.6**, **OpenCL 3.0**, **CUDA 6.1**, and **Shader Model 6.8**.
- Specific unit counts: **2560 shading units**, **160 TMUs**, and **64 ROPs**.

## Body
### Identification and Classification
- **Name:** GP104
- **Instance of:** graphics processing unit model
- **Manufacturer:** Nvidia
- **Country of origin:** United States
- **Working title:** NV134
- **TechPowerUp GPU specs database ID:** g793

### Release/Publication
- **Publication date:** 2016-05-27
- **Wikidata description (provided):** “2016 graphics processing unit manufactured by Nvidia”

### Architecture and Fabrication
- **Microarchitecture:** Pascal
- **Fabrication method:** 16 nm lithography process

### Power
- **Thermal design power (TDP):** 200 W (**maximum**)

### Core Hardware Counts
- **Number of shading units:** 2560
- **Number of texture mapping units (TMUs):** 160
- **Number of render output units (ROPs):** 64

### Cache / On-Chip Storage
- **L1 cache:** 48 KB
- **L2 cache:** 2048 MB

### Software Interfaces (APIs / Platforms) and Versions
GP104 is listed as using the following, with versions:
- **DirectX:** 12 (12_1)
- **OpenGL:** 4.6
- **OpenCL:** 3.0
- **Vulkan:** 1.3
- **CUDA:** 6.1
- **Shader Model:** 6.8

### Related Entities (Context)
- **Nvidia:** the manufacturer (American multinational technology company).
- **OpenCL:** an open standard for programming heterogeneous computing systems (such as CPUs or GPUs).

## References

1. GPU Specs Database