Radeon HD 4670 AGP vs B100
We compared two Desktop platform GPUs: 1024MB VRAM Radeon HD 4670 AGP and 96GB VRAM B100 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Radeon HD 4670 AGP 's Advantages
Lower TDP (59W vs 1000W)
B100 's Advantages
Released 15 years and 4 months late
Boost Clock1837MHz
More VRAM (96GB vs 1024GB)
Larger VRAM bandwidth (4.10TB/s vs 25.60GB/s)
16576 additional rendering cores
Score
Benchmark
FP32 (float)
Radeon HD 4670 AGP
480
B100
+12833%
62080
Graphics Card
Jul 2009
Release Date
Nov 2024
Radeon R700
Generation
Server Blackwell
Desktop
Type
Desktop
AGP 8x
Bus Interface
PCIe 5.0 x16
Clock Speeds
-
Base Clock
1665 MHz
-
Boost Clock
1837 MHz
800 MHz
Memory Clock
2000 MHz
Memory
1024MB
Memory Size
96GB
GDDR3
Memory Type
HBM3e
128bit
Memory Bus
4096bit
25.60GB/s
Bandwidth
4.10TB/s
Render Config
4
Compute Units
-
-
SM Count
132
320
Shading Units
16896
32
TMUs
528
8
ROPs
24
-
Tensor Cores
528
16 KB (per CU)
L1 Cache
256 KB (per SM)
128 KB
L2 Cache
50 MB
Theoretical Performance
6.000 GPixel/s
Pixel Rate
44.09 GPixel/s
24.00 GTexel/s
Texture Rate
969.9 GTexel/s
-
FP16 (half)
248.3 TFLOPS
480.0 GFLOPS
FP32 (float)
62.08 TFLOPS
-
FP64 (double)
31.04 TFLOPS
Board Design
59W
TDP
1000W
250 W
Suggested PSU
1400 W
1x DVI
1x HDMI 1.3a
1x VGA
Outputs
No outputs
Graphics Processor
RV730
GPU Name
GB102
RV730 XT
(215-0719047)
GPU Variant
-
TeraScale
Architecture
Blackwell
TSMC
Foundry
TSMC
55 nm
Process Size
5 nm
0.514 billion
Transistors
104 billion
146 mm²
Die Size
Unknown
Graphics Features
10.1 (10_1)
DirectX
N/A
3.3
OpenGL
N/A
1.1
OpenCL
3.0
N/A
Vulkan
N/A
-
CUDA
10.1
4.1
Shader Model
N/A
