Skip to content

Versus Engine

NVIDIA DGX Spark 64GB vs Radeon RX 9070

Recorded specs, strengths and limitations for NVIDIA DGX Spark 64GB and Radeon RX 9070, side by side. Check the evidence and your requirements before deciding.

Pricing

Not verified here

Current prices and availability are not verified in this summary. Check regional offers and required accessories before buying.

Recorded design scores

NVIDIA DGX Spark 64GB: Not recorded · Radeon RX 9070: Not recorded

Editorial design scores are not performance benchmarks or proof of suitability. Review the full records and your requirements.

Feature notes on record

  • NVIDIA DGX Spark 64GB: NVIDIA Grace Blackwell compute
  • Radeon RX 9070: Achievable frequency: Up to 2520 MHz

NVIDIA DGX Spark 64GB

NVIDIA DGX Spark with 64GB configuration and NVIDIA Grace Blackwell compute.

Where it wins, where it doesn't

Pros

  • Supports up to 100-billion-parameter models
  • Up to 1.7x performance compared to a single system

Cons

  • Limited to 64GB configuration, constraining memory availability for larger models

Radeon RX 9070

Radeon RX 9070 graphics card with up to 2520 MHz achievable frequency and 16 GB GDDR6 memory

Where it wins, where it doesn't

Pros

  • Achievable frequency of up to 2520 MHz
  • 16 GB of GDDR6 memory
  • Up to 640 GB/s memory bandwidth

Cons

  • Requires a minimum PSU of 650W
  • High typical board power of 220W

Specifications

Recorded specifications for NVIDIA DGX Spark 64GB and Radeon RX 9070
SpecificationNVIDIA DGX Spark 64GBRadeon RX 9070
achievable frequency—Up to 2520 MHz
ai accelerators—112
Architecture—RDNA 4
Board Design / Outputs—1x HDMI 2.1b, 3x DisplayPort 2.1a
Board Design / Power Connectors—2x 8-pin
Board Design / Suggested PSU—550 W
Board Design / TDP—220 W
Bus Interface—PCIe 5.0 x16
bus type—PCIe® 5.0 x16
Chip Package—HFCBGA-2069
Clock Speeds / Base Clock—1330 MHz
Clock Speeds / Boost Clock—2520 MHz
Clock Speeds / Game Clock—2070 MHz
Clock Speeds / Memory Clock—2518 MHz
cluster_model_supportup to 200 billion parameters—
computeNVIDIA Grace Blackwell—
configuration64GB—
Density—151.0M / mm²
Die Size—357 mm²
displayport™—2.1a
features—AMD Software Adrenalin Edition, AMD Fluid Motion Frames, AMD Noise Suppression, AMD Radeon™ Super Resolution, AMD Smart Access Memory, AMD Smart Access Video, AMD Privacy View, AMD Radeon™ Boost, AMD Radeon™ Anti-Lag, AMD Radeon™ Image Sharpening, AMD Enhanced Sync Technology, AMD FreeSync™ Technology, AMD Radeon™ Chill, AMD Virtual Super Resolution, AMD FSR™ 'Redstone'
Foundry—TSMC
game frequency—2070 MHz
Generation—Navi IV
gpu memory—AMD Infinity Cache Technology
GPU Name—Navi 48
GPU Variant—Navi 48 XT
Graphics Card—RX 9000
Graphics Features / OpenCL—2.0
Graphics Features / OpenGL—4.6
Graphics Features / Vulkan—1.4
hdmi™—2.1b
Matrix Performance / BF16—144.5 TFLOPS
Matrix Performance / INT4—578.0 TOPS
Matrix Performance / INT8—289.0 TOPS
Matrix Performance / Sparse Matrix—2x Throughput
max memory size—16 GB
Memory / Bandwidth—644.6 GB/s
Memory / Memory Bus—256 bit
Memory / Memory Size—16 GB
memory bandwidth—Up to 640 GB/s
memory interface—256-bit
memory speed—Up to 20 Gbps
memory type—GDDR6
model_supportup to 100-billion-parameter models—
Navi 48 GPU Notes—Super Resolution: FSR 4.1.2, XeSS 3, Frame Generation: FSR 4.0.2, XeSS 3, Ray Regeneration: FSR 1.2.1, Radiance Caching: FSR 0.9.0 (Technical Preview), Compute Units: 4th Gen, Ray Tracing: 3rd Gen, Infinity Cache: 3rd Gen, AI Accelerators: 2nd Gen, Radiance Display Engine: 2nd Gen, Shader ISA: GFX12.0 (gfx1201), Graphics/Compute: 12.0.1, Display Core Next: 4.0.1, Video Core Next: 5.0.0, JPEG Engine: 5.0.0
networkingNVIDIA ConnectX-7—
Numeric Format Support / Matrix—INT1, INT2, INT4, INT8, MXFP4, NVFP4, MXFP6, FP8, MXFP8, FP16, BF16, FP32, TF32, FP64
Numeric Format Support / Vector—INT4, INT8, INT16, INT32, FP16, BF16, FP32, FP64
Package Size—45 mm × 35 mm
peak 4-bit precision (int4 matrix) performance—578 TOPs
peak 8-bit precision (fp8 matrix) performance (e5m2, e4m3)—289 TFLOPs
peak 8-bit precision (int8 matrix) performance—289 TOPs
peak half precision (fp16 matrix) performance—145 TFLOPs
peak half precision (fp16 vector) performance—36.1 TFLOPs
peak pixel fill-rate—Up to 322.6 GP/s
peak single precision (fp32 vector) performance—36.1 TFLOPs
performance_comparisonup to 1.7x performance compared with a single system—
performance_testNVIDIA’s Qwen 3.8 27B test—
Predecessor—Navi III
Process Size—4 nm
Process Type—N4P FinFET
ray accelerators—56
Render Config / L0 Cache—32 KB per CU
Render Config / L2 Cache—8 MB
Render Config / L3 Cache—64 MB
Render Config / Matrix Cores—112
Render Config / RT Cores—56
Render Config / Shading Units—3584
Render Config / TMUs—224
rops—128
software_stackNVIDIA AI software stack—
stream processors—3584
Successor—Alpha Trion
supported rendering format—HDMI™ 4K Support, 4K H264 Decode, 4K H264 Encode, H265/HEVC Decode, H265/HEVC Encode, AV1 Decode, AV1 Encode
Theoretical Performance / FP16—36.13 TFLOPS (1:1)
Theoretical Performance / FP32—36.13 TFLOPS
Theoretical Performance / FP64—1.129 TFLOPS (1:32)
Theoretical Performance / Pixel Rate—322.6 GPixel/s
Theoretical Performance / Texture Rate—564.5 GTexel/s
transistor count—53.9 B
Transistors—53.9 billion
typical board power (desktop)—220W

Questions about NVIDIA DGX Spark 64GB vs Radeon RX 9070

What is the difference between NVIDIA DGX Spark 64GB and Radeon RX 9070?
On record, NVIDIA DGX Spark 64GB's main strength is: Supports up to 100-billion-parameter models. Radeon RX 9070's is: Achievable frequency of up to 2520 MHz.
What are the downsides of NVIDIA DGX Spark 64GB and Radeon RX 9070?
NVIDIA DGX Spark 64GB: Limited to 64GB configuration, constraining memory availability for larger models. Radeon RX 9070: Requires a minimum PSU of 650W.