GPUSPECINDEX

NVIDIA · Kepler (derived from codename prefix) · 2012

GeForce GT 640 OEM

GeForce GT 640 OEM: 1 GB GDDR5, 50 W, 384 CUDA cores. Launched April 24, 2012.

Memory
1 GB
Power
50 W
Bus width
128 bit
Memory1 GB34 of 94 in GeForce GT
Power50 W28 of 91 by higher draw in GeForce GT
CUDA cores3841 of 104 in GeForce GT

Affiliate links — we may earn a commission.

Memory

Capacity
1 GB
Memory type
GDDR5
Bus width
128 bit
Memory clock
891 MHz
Bandwidth
57.02 GB/s

Compute

CUDA cores
384
SM count
2
TMUs
32
ROPs
16
FP32
612.096 GFLOPS
FP64
25.504 GFLOPS
Pixel fill rate
12.75 GP/s
Texture fill rate
25.5 GT/s

Clocks

Boost clock
797 MHz

Board

TDP
50 W
Bus interface
PCIe 3.0 x16
External power required
No
Form factor
desktop

Physical

Architecture
Kepler (derived from codename prefix)
Codename
GK107-320-A2
Process
28 nm
Transistors
1.27 billion
Die size
118 mm²
Series
GeForce GT
Launch date
April 24, 2012

Not recorded by the source: generation, base clock and launch msrp.

Silicon, made visible

Die area is drawn against the largest recorded die in the catalog. Shader dots use a labeled, card-specific ratio so both early and modern GPUs remain legible.

118 mm² die · 384 shadersDie area drawn to scale against the catalog maximum of 1,540 square millimeters. Each dot = 4 shaders.118 mm² die · 384 shadersEach dot = 4 shadersDashed frame = largest catalog die (1,540 mm²)

Within its series, the GeForce GT 640 OEM ranks 14th of 54 for FP32 throughput and 34th of 94 for memory capacity. Its 50 W TDP is within the 75 W PCIe slot-power ceiling.

Source: Wikipedia, revision 1369488585 · catalog loaded 2026-08-16