Radeon Instinct MI455X vs Tesla T4

Primary details

GPU architecture, market segment, value for money and other general parameters compared.

Place in the ranking	265	not rated
Place by popularity	not in top-100	not in top-100
Power efficiency	26.29	no data
Architecture	Turing (2018−2022)	CDNA 5.0 (2026)
GPU code name	TU104	MI450 256CU
Market segment	Workstation	Workstation
Release date	13 September 2018 (7 years ago)	23 July 2026 (recently)

Detailed specifications

General parameters such as number of shaders, GPU core base clock and boost clock speeds, manufacturing process, texturing and calculation speed. Real power consumption of some graphics cards can well exceed their nominal TDP, especially if overclocked.

Pipelines / CUDA cores	2560	32768
Core clock speed	585 MHz	1000 MHz
Boost clock speed	1590 MHz	2400 MHz
Number of transistors	13,600 million	320,000 million
Manufacturing process technology	12 nm	2 nm
Power consumption (TDP)	70 Watt	1400 Watt
Texture fill rate	254.4	2,457.6
Floating-point processing power	8.141 TFLOPS	157.3 TFLOPS
ROPs	64	no data
TMUs	160	1024
Tensor Cores	320	no data
Ray Tracing Cores	40	no data
L1 Cache	2.5 MB	8 MB
L2 Cache	4 MB	192 MB

Form factor & compatibility

Information on compatibility with other computer components. Useful when choosing a future computer configuration or upgrading an existing one. For desktop graphics cards it's interface and bus (motherboard compatibility), additional power connectors (power supply compatibility).

Interface	PCIe 3.0 x16	PCIe 5.0 x16
Length	168 mm	no data
Width	1-slot	EAM Module
Supplementary power connectors	None	None

VRAM capacity and type

Parameters of VRAM installed: its type, size, bus, clock and resulting bandwidth. Integrated GPUs have no dedicated video RAM and use a shared part of system RAM.

Memory type	GDDR6	HBM4
Maximum RAM amount	16 GB	432 GB
Memory bus width	256 Bit	24576 Bit
Memory clock speed	1250 MHz	1900 MHz
Memory bandwidth	320.0 GB/s	23.3 TB/s

Connectivity and outputs

This section shows the types and number of video connectors on each GPU. The data applies specifically to desktop reference models (for example, NVIDIA’s Founders Edition). OEM partners often modify both the number and types of ports. On notebook GPUs, video‐output options are determined by the laptop’s design rather than the graphics chip itself.

Display Connectors

No outputs

No outputs

API and SDK support

List of supported 3D and general-purpose computing APIs, including their specific versions.

DirectX	12 Ultimate (12_1)	N/A
Shader Model	6.5	N/A
OpenGL	4.6	N/A
OpenCL	1.2	3.0
Vulkan	1.2.131	N/A
CUDA	7.5	-
DLSS	+	-

Pros & cons summary

---

Recency	13 September 2018	23 July 2026
Maximum RAM amount	16 GB	432 GB
Chip lithography	12 nm	2 nm
Power consumption (TDP)	70 Watt	1400 Watt