GeekRadar / The comparison desk
COMPARE.
GPUIntel Data Center GPU Max NEXT
Graphics processor
- Video memory
- 128 GB*
- Memory bus
- 8,192 bit*
- Rated memory bandwidth
- 3,210 GB/s*
Benchmark results
- LLM Inference: llama.cpp Bench: llama-2-7b-Q4_0 (Q4_0) · version unreported · score8.21* tokens/sPlatform unreported Compare results
- LLM Inference: MLPerf: llama2-70b-99 (UINT4) · version unreported · score527.51* tokens/sPlatform unreported Compare results
- LLM Inference: MLPerf: llama2-70b-99.9 (UINT4) · version unreported · score527.51* tokens/sPlatform unreported Compare results
- LLM Inference: MLPerf: llama3.1-8b (UINT4) · version unreported · score1,337.11* tokens/sPlatform unreported Compare results
- LLM Inference: MLPerf: llama3.1-8b-edge (UINT4) · version unreported · score1,652.47* tokens/sPlatform unreported Compare results
Processing & platform
- Architecture
- Generation 12.5*
- Process node
- 7 nm*
- GPU cores
- 20,480*
- Graphics interface
- PCIe 5.0 x16*
- Reported chip die area
- 1,280 mm²*
- Transistors
- 100,000 million*
- GPU L2 cache
- 408 MB*
- Graphics processor
- Rialto*
- Texture mapping units
- 1,280*
- Raster output units
- 0*
- Ray tracing cores
- 160*
- Chip foundry
- Intel*
- Card or module format
- OAM Module*
- Declared GPU TDP
- 800 W*
- Suggested system PSU
- 1,200 W*
- External power connectors
- 8-pin EPS*
- DirectX / feature level
- 12 (12_1)*
- OpenGL
- 4.6*
- OpenCL
- 3.0*
- Shader model
- 6.6*
Memory & storage
- Memory options
- 128*
- Video memory
- 128 GB*
- Video memory type
- HBM2e*
- Memory bus
- 8,192 bit*
- Rated memory bandwidth
- 3,210 GB/s*
Power specifications
- Thermal design power
- 800 W*
System configuration
Detailed specifications
Specifications 1 · 42 fields
- proc transistors
- 100,000 million*
- proc die size
- 1,280* mm²
- cores
- 20,480*
- rt cores
- 160*
- base clock
- 900*
- boost clock
- 1,600*
- memory clock
- 1,565*
- ram
- 128*
- name
- Intel Data Center GPU Max NEXT*
- test name Graphics benchmark result
- LLM Inference: MLPerf: llama3.1-8b (UINT4)*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- test name Graphics benchmark result
- LLM Inference: MLPerf: llama2-70b-99 (UINT4)*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- test name Graphics benchmark result
- LLM Inference: llama.cpp Bench: llama-2-7b-Q4_0 (Q4_0)*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- test name Graphics benchmark result
- LLM Inference: MLPerf: llama2-70b-99.9 (UINT4)*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- test name Graphics benchmark result
- LLM Inference: MLPerf: llama3.1-8b-edge (UINT4)*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- ram type
- HBM2e*
- bus width
- 8,192*
- bus interface
- PCIe 5.0 x16*
- ram bandwidth
- 3.21 TB/s*
- fp16 performance
- 65.54*
- fp32 performance
- 65.54*
- fp64 performance
- 65.54*
- score Graphics benchmark result
- 527.5125*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- score Graphics benchmark result
- 8.21*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- score Graphics benchmark result
- 527.5125*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- score Graphics benchmark result
- 1,652.47*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- score Graphics benchmark result
- 1,337.1125*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- tdp
- 800* W
- suggested psu
- 1,200* W
- slot width
- OAM Module*
- power connectors
- 8-pin EPS*
- is retail board
- No*
- gpi value
- 62.12*
- unit Graphics benchmark result
- tokens/s*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- unit Graphics benchmark result
- tokens/s*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- unit Graphics benchmark result
- tokens/s*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- unit Graphics benchmark result
- tokens/s*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- unit Graphics benchmark result
- tokens/s*
Scope and conditions
- benchmark_context
- suite
- Per-row third-party test name*
- test_configuration_verified
- No*
- comparable
- No*
- vendor
- int*
- manufacturer
- INTEL*
- arch name
- Generation 12.5*
- proc foundry
- Intel*
- proc process size
- 7* nm
Specifications 2 · 36 fields
- DirectX / feature level
- 12 (12_1)*
- OpenCL
- 3.0*
- OpenGL
- 4.6*
- Shader model
- 6.6*
- Architecture
- Generation 12.5*
- Graphics interface
- PCIe 5.0 x16*
- Memory bus
- 8,192* bit
- Reported shader processors
- 20,480*
- Reported theoretical FP16 peak
- 65.54* TFLOPS
Scope and conditions
- Arithmetic density
- Not stated*
- Reported theoretical FP32 peak
- 65.54* TFLOPS
Scope and conditions
- Arithmetic density
- Not stated*
- Reported theoretical FP64 peak
- 65.54* TFLOPS
Scope and conditions
- Arithmetic density
- Not stated*
- Declared L1 cache configuration
- 64 KB (per EU)*
- Declared GPU L2 cache
- 408 MB*
- Reported display outputs
- No outputs*
- Theoretical pixel fill rate
- 0 MPixel/s*
- External power connectors
- 8-pin EPS*
- Reported chip die area
- 1280 mm²*
- Chip foundry
- Intel*
- Process node declaration
- 7 nm*
- Transistors
- 100,000 million*
- Graphics processor
- Rialto*
- Reported video memory
- 128* GB
- Rated memory bandwidth
- 3.21 TB/s*
- Video memory type
- HBM2e*
- Raster output units
- 0*
- Ray tracing cores
- 160*
- Reported shading units
- 20,480*
- Card or module format
- OAM Module*
- Suggested system PSU
- 1200 W*
- 2:4 tensor sparsity
- No*
- Declared GPU TDP
- 800 W*
- Theoretical texture fill rate
- 2,048.0 GTexel/s*
- Texture mapping units
- 1,280*
- GPU cores
- 20,480*
- Video memory
- 128* GB
- Video memory type
- HBM2e*
0 selected Compare 0
* Provisional data