NVIDIA B300 SXM6 AC vs NVIDIA GeForce RTX 5090 D V2 Comparison

NVIDIA
GEFORCE

NVIDIA B300 SXM6 AC

CORE STATE GB110
VRAM 288 GB
CLOCK SPEED 2032 MHz
TDP 1100 W
BUS WIDTH 8192 bit
ARCHITECTURE Blackwell Ultra
nm
PROCESS 5 nm
LAUNCH DATE 2025
VS
NVIDIA
GEFORCE

GeForce RTX 5090 D V2

CORE STATE GB202
VRAM 24 GB
CLOCK SPEED 2407 MHz
TDP 575 W
BUS WIDTH 384 bit
ARCHITECTURE Blackwell 2.0
nm
PROCESS 5 nm
LAUNCH DATE 2025

PERFORMANCE BENCHMARKS

geekbench_opencl
369,831
N/A
3dmark_3dmark_steel_nomad_dx12
N/A
16,504

Analysis: NVIDIA B300 SXM6 AC vs NVIDIA GeForce RTX 5090 D V2

FAQ

Q: What is the primary benchmark result recorded for the NVIDIA B300 SXM6 AC?

A: The database records a Geekbench OpenCL score of 369,831 for the B300 SXM6 AC, placing it in the 100th percentile of all GPUs.

Q: What is the primary benchmark result recorded for the NVIDIA GeForce RTX 5090 D V2?

A: The database records a 3DMark Steel Nomad DX12 score of 16,504 for the RTX 5090 D V2, placing it in the 59th percentile of all GPUs.

Q: How does the B300 SXM6 AC compare to its nearest rival, the NVIDIA B200?

A: The B300 SXM6 AC scores 7% higher than the NVIDIA B200 in the recorded Geekbench OpenCL test, with scores of 369,831 versus 345,482.

Q: How does the RTX 5090 D V2 compare to its closest rival, the NVIDIA T400?

A: The RTX 5090 D V2 scores 16,504, which is effectively identical to the NVIDIA T400's 16,508, a delta of 0%.

Q: What are the memory capacities of the two GPUs?

A: The B300 SXM6 AC features 288 GB of HBM3e memory, while the RTX 5090 D V2 features 24 GB of GDDR7 memory.

Q: What are the thermal design power (TDP) ratings for each card?

A: The B300 SXM6 AC has a TDP of 1100 W with a suggested PSU of 1500 W, while the RTX 5090 D V2 has a TDP of 575 W with a suggested PSU of 950 W.

Architecture Differences

The B300 SXM6 AC and the RTX 5090 D V2 represent two distinct branches of NVIDIA's Blackwell generation. The B300 SXM6 AC uses the GB110 chip and is classified under the Blackwell Ultra architecture, part of the Server Blackwell (Bxx) generation. In contrast, the RTX 5090 D V2 uses the GB202 chip and is classified under Blackwell 2.0 architecture, part of the GeForce 50 series.

The process node for both is 5 nm at TSMC, but the die characteristics differ substantially. The B300 SXM6 AC integrates 208,000 million transistors on a 1628 mm² die, resulting in a transistor density of 127.8M per mm². The RTX 5090 D V2 integrates 92,200 million transistors on a 750 mm² die, yielding a density of 122.9M per mm². The B300 SXM6 AC's die is more than twice the physical size and carries over twice the transistor count.

The B300 SXM6 AC is built for server workloads with an SXM module slot width, no display outputs, and no DirectX, OpenGL, or Vulkan API support. The RTX 5090 D V2 is a dual-slot consumer card with 1x HDMI 2.1b and 3x DisplayPort 2.1b outputs, supporting DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The B300 SXM6 AC uses PCIe 6.0 x16, while the RTX 5090 D V2 uses PCIe 5.0 x16. The server card has no listed RT cores, while the GeForce card includes 170 RT cores. Both have tensor cores, but the B300 SXM6 AC has 592 tensor cores, while the RTX 5090 D V2 has 680.

The B300 SXM6 AC's memory subsystem uses HBM3e with an 8192-bit bus and 8.19 TB/s bandwidth. The RTX 5090 D V2 uses GDDR7 with a 384-bit bus and 1.34 TB/s bandwidth. The server card is designed for massive data throughput, while the consumer card prioritizes lower latency and higher clock speeds.

The Verdict

The recorded data shows two GPUs with fundamentally different purposes. The B300 SXM6 AC sits at the absolute top of the database with a 100th percentile ranking and a Geekbench OpenCL score of 369,831. It leads its nearest rival, the NVIDIA B200, by 7%, the H200 NVL by 10.4%, the AMD Instinct MI300X by 16.3%, and the L40S by 25%. This is a compute-oriented server accelerator for workloads that demand massive memory capacity, 288 GB, and extreme bandwidth, 8.19 TB/s.

The RTX 5090 D V2 occupies a very different position. Its 3DMark Steel Nomad DX12 score of 16,504 places it in the 59th percentile, with nearest rivals within 0.9% or less. The NVIDIA T400, AMD Radeon PRO W7500, NVIDIA RTX PRO 6000 Blackwell, and AMD Radeon RX 5700 XT all score within a narrow band of 16,361 to 16,508. This suggests the 5090 D V2 is a well-balanced consumer graphics card, not an outlier in either direction.

The choice between these two depends entirely on the workload class. The B300 SXM6 AC is for server deployments where compute density, memory capacity, and bandwidth are paramount. The RTX 5090 D V2 is for desktop systems where display output, API support, and moderate power draw are required. The data does not indicate that one is "better" overall; it shows two specialized tools.

Specification Differences

The two GPUs differ across nearly every specification field. The B300 SXM6 AC uses the GB110 chip with Blackwell Ultra architecture, while the RTX 5090 D V2 uses the GB202 chip with Blackwell 2.0 architecture. The server card is from the Server Blackwell (Bxx) generation; the GeForce card is from the GeForce 50 generation.

Transistor count: 208,000 million versus 92,200 million. Die size: 1628 mm² versus 750 mm². Transistor density: 127.8M per mm² versus 122.9M per mm².

Base clock: 1665 MHz versus 2017 MHz. Boost clock: 2032 MHz versus 2407 MHz. Memory clock: 2000 MHz (8 Gbps effective) versus 1750 MHz (28 Gbps effective).

Memory size: 288 GB versus 24 GB. Memory type: HBM3e versus GDDR7. Bus width: 8192 bit versus 384 bit. Bandwidth: 8.19 TB/s versus 1.34 TB/s.

Shading units: 18,944 versus 21,760. TMUs: 592 versus 680. ROPs: 24 versus 176. RT cores: not listed versus 170. Tensor cores: 592 versus 680.

Pixel rate: 48.77 GPixel/s versus 423.6 GPixel/s. Texture rate: 1,202.9 GTexel/s versus 1,636.8 GTexel/s. FP32: 76.99 TFLOPS versus 104.8 TFLOPS. FP16: 76.99 TFLOPS (1:1) versus 104.8 TFLOPS (1:1).

TDP: 1100 W versus 575 W. Slot width: SXM Module versus dual-slot. Power connectors: none listed versus 1x 16-pin. Suggested PSU: 1500 W versus 950 W. Bus interface: PCIe 6.0 x16 versus PCIe 5.0 x16. Display outputs: none versus 1x HDMI 2.1b and 3x DisplayPort 2.1b.

APIs: the B300 SXM6 AC has N/A for DirectX, OpenGL, and Vulkan; the RTX 5090 D V2 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Dimensions are only listed for the RTX 5090 D V2: 304 mm length, 137 mm height, 48 mm width.

Release dates differ: the B300 SXM6 AC launched on September 10, 2025, while the RTX 5090 D V2 launched on August 14, 2025. Predecessors also differ: Server Hopper for the B300 SXM6 AC, GeForce 40 for the RTX 5090 D V2. Successors are Server Rubin and GeForce 60, respectively.

Head-to-Head Benchmarks

The database contains no direct head-to-head benchmark results between the B300 SXM6 AC and the RTX 5090 D V2. Each card has a single recorded benchmark from a different test suite, so a direct numerical comparison is not possible. The B300 SXM6 AC was tested with Geekbench OpenCL, and the RTX 5090 D V2 was tested with 3DMark Steel Nomad DX12.

What the data does show is each card's standing within its own benchmark context. The B300 SXM6 AC's Geekbench OpenCL score of 369,831 is 7% higher than the NVIDIA B200's 345,482. It is 10.4% higher than the NVIDIA H200 NVL's 334,891, 16.3% higher than the AMD Instinct MI300X's 317,994, and 25% higher than the NVIDIA L40S's 295,763. These are substantial margins, indicating the B300 SXM6 AC is a clear performance leader in its compute-oriented test.

The RTX 5090 D V2's 3DMark Steel Nomad DX12 score of 16,504 is within 0% of the NVIDIA T400's 16,508, within 0.5% of the AMD Radeon PRO W7500's 16,415, within 0.6% of the NVIDIA RTX PRO 6000 Blackwell's 16,408, and within 0.9% of the AMD Radeon RX 5700 XT's 16,361. The margins here are negligible, showing the RTX 5090 D V2 is clustered tightly with its peers in this particular DX12 test.

The percentile rankings reinforce the separation. The B300 SXM6 AC is in the 100th percentile of all GPUs, meaning it outperforms virtually every other recorded GPU in its benchmark. The RTX 5090 D V2 is in the 59th percentile, meaning it sits slightly above the median but well within the middle of the distribution.

Where Each One Wins

The B300 SXM6 AC wins decisively in compute-scale metrics. Its 288 GB of HBM3e memory with 8.19 TB/s bandwidth is unmatched by the RTX 5090 D V2's 24 GB of GDDR7 with 1.34 TB/s. The 8192-bit bus width provides a 21x advantage in memory bus width, and the bandwidth advantage is over 6x. For workloads that fit within the memory capacity, the B300 SXM6 AC's FP32 throughput of 76.99 TFLOPS is substantial, though it trails the RTX 5090 D V2's 104.8 TFLOPS in raw compute.

The B300 SXM6 AC also wins on the server-side infrastructure front. It uses PCIe 6.0 x16, a generation ahead of the RTX 5090 D V2's PCIe 5.0 x16. It has no display outputs, which is appropriate for a compute accelerator, and its API support is listed as N/A, indicating it is not designed for graphics rendering workloads.

The RTX 5090 D V2 wins on clock speeds and graphics-oriented specifications. Its base clock of 2017 MHz and boost clock of 2407 MHz are significantly higher than the B300 SXM6 AC's 1665 MHz and 2032 MHz. Its pixel rate of 423.6 GPixel/s is nearly 9x higher than the B300 SXM6 AC's 48.77 GPixel/s, and its texture rate of 1,636.8 GTexel/s is 36% higher. Its FP32 and FP16 throughput of 104.8 TFLOPS exceeds the B300 SXM6 AC's 76.99 TFLOPS.

The RTX 5090 D V2 wins for any workload requiring display output. It has 1x HDMI 2.1b and 3x DisplayPort 2.1b, supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, and includes 170 RT cores for ray tracing. Its 176 ROPs versus 24 ROPs gives it a massive advantage in rasterization tasks. Its 680 TMUs and 21,760 shading units also exceed the B300 SXM6 AC's 592 TMUs and 18,944 shading units.

The RTX 5090 D V2 wins on power efficiency in practical terms. Its TDP of 575 W with a suggested PSU of 950 W is far below the B300 SXM6 AC's 1100 W TDP and 1500 W suggested PSU. It is a dual-slot card with a 16-pin connector, making it suitable for desktop installation, while the B300 SXM6 AC is an SXM module with no listed power connector, indicating a server chassis requirement.

The data indicates the B300 SXM6 AC wins for memory-bound server compute, while the RTX 5090 D V2 wins for graphics rendering and desktop use. The benchmark scores are not directly comparable due to different test suites, but each card's position relative to its own peers is clear. The B300 SXM6 AC is a top-tier compute accelerator; the RTX 5090 D V2 is a mid-pack graphics card with strong clock speeds and display features.

DETAILED SPECIFICATIONS

SPECIFICATION
B300 SXM6 AC
RTX 5090 D V2
Core Specs
Shading Units
18,944
21,760 +14.9%
Shaders
18,944
21,760 +14.9%
TMUs
592
680 +14.9%
ROPs
24
176 +633.3%
SM Count
148
170 +14.9%
Clocks
Base Clock
1665 MHz
2017 MHz
Boost Clock
2032 MHz
2407 MHz
Memory Clock
2000 MHz 8 Gbps effective
1750 MHz 28 Gbps effective
Memory
Memory Size
288 GB
24 GB
VRAM (MB)
294,912
24,576 -91.7%
Memory Type
HBM3e
GDDR7
Memory Bus
8192 bit
384 bit
Bandwidth
8.19 TB/s
1.34 TB/s
Cache
L1 Cache
256 KB (per SM)
128 KB (per SM)
L2 Cache
126 MB
96 MB
Performance
Pixel Rate
48.77 GPixel/s
423.6 GPixel/s
Texture Rate
1,202.9 GTexel/s
1,636.8 GTexel/s
FP32 (TFLOPS)
76.99 TFLOPS
104.8 TFLOPS
FP64 (TFLOPS)
1,202.9 GFLOPS (1:64)
1.637 TFLOPS (1:64)
FP16 (TFLOPS)
76.99 TFLOPS (1:1)
104.8 TFLOPS (1:1)
AI/RT
RT Cores
—
170
Tensor Cores
592
680 +14.9%
Power
TDP
1100 W
575 W
TDP (W)
1,100
575 -47.7%
Suggested PSU
1500 W
950 W
Power Connectors
—
1x 16-pin
Architecture
Architecture
Blackwell Ultra
Blackwell 2.0
GPU Name
GB110
GB202
Generation
Server Blackwell (Bxx)
GeForce 50
Process Size
5 nm
5 nm
Transistors
208,000 million
92,200 million
Die Size
1628 mm²
750 mm²
Foundry
TSMC
TSMC
Density
127.8M / mm²
122.9M / mm²
API Support
DirectX
—
12 Ultimate (12_2)
OpenGL
—
4.6
Vulkan
—
1.4
OpenCL
3.0
3.0
CUDA
10.3
12.0
Shader Model
—
6.9
Physical
Slot Width
SXM Module
Dual-slot
Length
—
304 mm 12 inches
Height
—
137 mm 5.4 inches
Outputs
No outputs
1x HDMI 2.1b3x DisplayPort 2.1b
Bus Interface
PCIe 6.0 x16
PCIe 5.0 x16
Other
Launch Price
—
2,299 USD
Production
Active
Active
Predecessor
Server Hopper
GeForce 40
Successor
Server Rubin
GeForce 60
View B300 SXM6 AC Details View GeForce RTX 5090 D V2 Details