|
gptkbp:instance_of
|
gptkb:Graphics_Processing_Unit
|
|
gptkbp:ai
|
gptkb:Yes
Multiple frameworks supported.
|
|
gptkbp:aienhancements
|
gptkb:Yes
|
|
gptkbp:aiinference_support
|
gptkb:Yes
|
|
gptkbp:aitraining_support
|
gptkb:Yes
|
|
gptkbp:application
|
Machine Learning, Data Analytics
|
|
gptkbp:architecture
|
GCN 5th Gen
|
|
gptkbp:community_support
|
Strong community support.
|
|
gptkbp:compatibility
|
Compatible with various software tools.
|
|
gptkbp:compute_power
|
High compute power for AI.
|
|
gptkbp:cooling_system
|
Active cooling
|
|
gptkbp:direct_xsupport
|
gptkb:Direct_X_12
|
|
gptkbp:ecosystem
|
Part of AMD's ROCm ecosystem.
|
|
gptkbp:focus
|
Data science and analytics.
|
|
gptkbp:form_factor
|
gptkb:PCIe
Full-height, Full-length
|
|
gptkbp:fp64_performance_ratio
|
1:16
|
|
gptkbp:fuel_economy
|
Designed for energy efficiency.
High efficiency for data centers.
|
|
gptkbp:gpu
|
gptkb:Yes
|
|
gptkbp:gpuarchitecture
|
gptkb:Vega
|
|
gptkbp:has_ability
|
5.0
|
|
gptkbp:has_units
|
gptkb:64
|
|
gptkbp:high_performance_computing_support
|
gptkb:Yes
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
AMD Radeon Instinct MI25
|
|
gptkbp:key_feature
|
High memory bandwidth, AI acceleration.
|
|
gptkbp:manufacturer
|
gptkb:AMD
|
|
gptkbp:manufacturing_process
|
14 nm
|
|
gptkbp:market_position
|
High-end GPU for AI and HPC.
|
|
gptkbp:market_segment
|
AI and Deep Learning
|
|
gptkbp:memory_type
|
gptkb:HBM2
2048-bit
|
|
gptkbp:number_of_cores
|
4096
|
|
gptkbp:open_glsupport
|
gptkb:Open_GL_4.5
|
|
gptkbp:partnerships
|
Collaborations with AI companies.
|
|
gptkbp:peak_fp16_performance
|
24.6 TFLOPS
|
|
gptkbp:peak_fp32_performance
|
12.3 TFLOPS
|
|
gptkbp:performance
|
gptkb:High
1/16 FP32
12.3 TFLOPS
24.6 TFLOPS
AI and machine learning tasks.
Optimized for AI workloads.
|
|
gptkbp:power_connector
|
8-pin and 6-pin
|
|
gptkbp:predecessor
|
AMD Radeon Instinct MI24
|
|
gptkbp:product_line
|
Radeon Instinct series.
|
|
gptkbp:project
|
Continued advancements in AI technology.
|
|
gptkbp:ram
|
gptkb:memory
32 GB
512 GB/s
32 GB HBM2
|
|
gptkbp:release_date
|
gptkb:2017
|
|
gptkbp:release_notes
|
Designed for machine learning and high-performance computing.
|
|
gptkbp:release_target
|
Enterprise and research institutions.
|
|
gptkbp:released
|
gptkb:Radeon_Instinct
|
|
gptkbp:slisupport
|
gptkb:Yes
|
|
gptkbp:stall_speed
|
1300 MHz
|
|
gptkbp:successor
|
gptkb:AMD_Radeon_Instinct_MI50
|
|
gptkbp:supports
|
gptkb:Tensor_Flow
gptkb:ROCm
gptkb:Py_Torch
gptkb:Open_CL
|
|
gptkbp:target_market
|
gptkb:research_center
|
|
gptkbp:tdp
|
300 W
|
|
gptkbp:thermal_design_power
|
300 W
|
|
gptkbp:trends
|
Growing demand for AI solutions.
|
|
gptkbp:user_base
|
Data scientists, researchers.
|
|
gptkbp:virtualization_support
|
gptkb:Yes
|
|
gptkbp:bfsParent
|
gptkb:NVIDIA_V100
|
|
gptkbp:bfsLayer
|
5
|