GPU Comparison
RADEON
AMD Radeon Instinct MI250X
CORE STATE
Aldebaran
VRAM
128 GB
CLOCK SPEED
1700 MHz
TDP
500 W
BUS WIDTH
8192 bit
ARCHITECTURE
CDNA 2.0
PROCESS
6 nm
LAUNCH DATE
2021
VS
GPU
Arc A370M
CORE STATE
DG2-128
VRAM
4 GB
CLOCK SPEED
2050 MHz
TDP
35 W
BUS WIDTH
64 bit
ARCHITECTURE
Xe-HPG
PROCESS
6 nm
LAUNCH DATE
2022
PERFORMANCE BENCHMARKS
geekbench_opencl
geekbench_vulkan
DETAILED SPECIFICATIONS
SPECIFICATION
Instinct MI250X
A370M
Core Specs
Shading Units
14,080
1,024
-92.7%
Shaders
14,080
1,024
-92.7%
TMUs
880
64
-92.7%
ROPs
0
32
+∞%
Compute Units
220
—
Execution Units
—
128
Clocks
Base Clock
1000 MHz
1550 MHz
Boost Clock
1700 MHz
2050 MHz
Memory Clock
1600 MHz
3.2 Gbps effective
1750 MHz
14 Gbps effective
Memory
Memory Size
128 GB
4 GB
VRAM (MB)
131,072
4,096
-96.9%
Memory Type
HBM2e
GDDR6
Memory Bus
8192 bit
64 bit
Bandwidth
3.28 TB/s
112.0 GB/s
Cache
L1 Cache
16 KB (per CU)
—
L2 Cache
16 MB
4 MB
Performance
Pixel Rate
0 MPixel/s
65.60 GPixel/s
Texture Rate
1,496.0 GTexel/s
131.2 GTexel/s
FP32 (TFLOPS)
47.87 TFLOPS
4.198 TFLOPS
FP64 (TFLOPS)
47.87 TFLOPS (1:1)
1,049.6 GFLOPS (1:4)
FP16 (TFLOPS)
383.0 TFLOPS (8:1)
8.397 TFLOPS (2:1)
AI/RT
RT Cores
—
8
XMX Cores
—
128
Power
TDP
500 W
35 W
TDP (W)
500
35
-93.0%
Suggested PSU
900 W
—
Power Connectors
2x 8-pin
—
Architecture
Architecture
CDNA 2.0
Xe-HPG
GPU Name
Aldebaran
DG2-128
Generation
Radeon Instinct
(MIx)
Alchemist
(Arc 3 Mobile)
Process Size
6 nm
6 nm
Transistors
58,200 million
7,200 million
Die Size
724 mm²
157 mm²
Foundry
TSMC
TSMC
Density
80.4M / mm²
45.9M / mm²
AMD MCM
MCM
2
—
API Support
DirectX
—
12 Ultimate (12_2)
OpenGL
—
4.6
Vulkan
—
1.4
OpenCL
3.0
3.0
Shader Model
—
6.6
Physical
Slot Width
Dual-slot
IGP
Length
267 mm
10.5 inches
—
Height
111 mm
4.4 inches
—
Outputs
No outputs
Portable Device Dependent
Bus Interface
PCIe 4.0 x16
PCIe 4.0 x8
Other
Production
—
End-of-life
Predecessor
FirePro Data Center
—