MSI GeForce RTX 3050 LP 6G OC
Exhibit Type: Modern Compact Graphics Accelerator & Silicon Specification Dossier
•
Microarchitecture: NVIDIA Ampere • GA107-325 (8 nm Samsung)
•
Compute Pipeline: 2,304 CUDA Cores • 72 Tensor Cores • 18 RT Cores • 1,492 MHz Boost Clock
•
VRAM & Bus: 6 GB GDDR6 • 96-bit Interface • 168.0 GB/s Bandwidth • PCIe 4.0 x8
•
PassMark G3D Mark: 10,744 (Rank #4 in Museum GPU Collection)
•
Museum Inventory: 1 Physical Unit (MS-V812 • Subsystem ID: 1462:8123)
•
Active Host Pairing: Corsair Carbide Series Air 540 “Cube” Workstation (PCIe Slot 6 • Bus 03:00.0)
•
Microarchitecture: NVIDIA Ampere • GA107-325 (8 nm Samsung)
•
Compute Pipeline: 2,304 CUDA Cores • 72 Tensor Cores • 18 RT Cores • 1,492 MHz Boost Clock
•
VRAM & Bus: 6 GB GDDR6 • 96-bit Interface • 168.0 GB/s Bandwidth • PCIe 4.0 x8
•
PassMark G3D Mark: 10,744 (Rank #4 in Museum GPU Collection)
•
Museum Inventory: 1 Physical Unit (MS-V812 • Subsystem ID: 1462:8123)
•
Active Host Pairing: Corsair Carbide Series Air 540 “Cube” Workstation (PCIe Slot 6 • Bus 03:00.0)
- Official Manufacturer Portal: MSI GeForce RTX 3050 LP 6G OC Product Overview
- Silicon Database Profile: TechPowerUp GPU Database: NVIDIA GeForce RTX 3050 6 GB (GA107)
- Synthetic Performance Authority: PassMark VideoCardBenchmark: GeForce RTX 3050 6GB (Score: 10,744)
- Host Motherboard Exhibit: Supermicro X9SRi-F Server Motherboard (Exhibit #2957)
- Host System Chassis Exhibit: Corsair Carbide Series Air 540 High Airflow ATX Cube Case (Exhibit #2949)
1. Technical Overview & Key Specifications Table
| Category | Property | Specification |
|---|---|---|
| GPU Identity | Commercial Product Name | MSI GeForce RTX 3050 LP 6G OC |
| Board Partner & Vendor | Micro-Star International Co., Ltd. (MSI) / NVIDIA Corporation | |
| MSI Internal Model Code | MS-V812 • Model Code: GeForce RTX 3050 LP 6G OC |
|
| GPU Silicon Code Name | GA107 (Variant: GA107-325-K1-A1 • Revision: A1) |
|
| PCI IDs & Subsystem | PCI Device ID: 10de:2584 • Subsystem ID: 1462:8123 |
|
| Audio Controller ID | PCI Audio ID: 10de:2291 (NVIDIA High Definition Audio) |
|
| Market Segment & Launch Date | Ultra-Efficient Mainstream Desktop / SFF • February 2, 2024 | |
| Physical Silicon | Foundry & Lithography | Samsung Electronics • 8 nm (8N) FinFET Custom Process |
| Transistor Count | 8,700 Million (8.7 Billion) monolithic transistors | |
| Die Surface Area | 200 mm² monolithic die area | |
| Transistor Density | 43.5 Million transistors / mm² | |
| Card Dimensions & Weight | 174 mm × 69 mm × 42 mm • Weight: 311 grams | |
| Form Factor & Bracket | Low-Profile, Dual-Slot (Full-Height ATX bracket mounted; LP bracket included) | |
| Compute Pipeline | Microarchitecture | NVIDIA Ampere (2nd Generation RTX) |
| CUDA Cores / Shaders | 2,304 CUDA Cores (18 Streaming Multiprocessors × 128 Cores/SM) | |
| Tensor Cores | 72 Tensor Cores (3rd Generation with Structural Sparsity) | |
| Ray Tracing (RT) Cores | 18 RT Cores (2nd Generation with Concurrent RT+Graphics) | |
| Texture Mapping Units (TMUs) | 72 TMUs | |
| Render Output Units (ROPs) | 32 ROPs | |
| Base Engine Clock | 1,042 MHz | |
| Boost Engine Clock | 1,492 MHz (MSI Factory OC from 1,470 MHz reference) | |
| Compute Throughput (FP32) | 6.875 TFLOPS Boost (4.799 TFLOPS Base) | |
| Memory Subsystem | Dedicated VRAM Size | 6,144 MB (6 GB) GDDR6 |
| Memory Bus Width | 96-bit (3 × 32-bit discrete memory channels) | |
| Memory Clock Frequency | 1,750 MHz (14.0 Gbps effective data rate) | |
| Peak Theoretical Bandwidth | 168.0 GB/s (168,000 MB/s) | |
| Level 2 (L2) Cache | 2,048 KB (2 MB) High-Speed Shared L2 Cache | |
| ECC Memory Support | Non-ECC GDDR6 | |
| Display & Media I/O | Physical Video Outputs | 1 × DisplayPort 1.4a + 2 × HDMI 2.1 |
| Maximum Digital Resolution | 7680 × 4320 @ 60 Hz (8K UHD) • 4K @ 120 Hz HDR | |
| Maximum Active Displays | Up to 3 concurrent independent displays | |
| Hardware Video Acceleration | 7th Gen NVENC (H.264, HEVC up to 8K) • 5th Gen NVDEC (AV1, VP9, H.264, HEVC) | |
| Content Protection | HDCP 2.3 compliant | |
| Thermal & Electrical | Thermal Design Power (TDP) | 70 Watts Maximum Rated Power Consumption |
| PCIe Slot Power Limit | 75.000 Watts (Draws 100% power through PCIe slot contacts) | |
| Auxiliary Power Connector | None (Zero 6-pin or 8-pin PCIe cables required) | |
| Cooling Solution | Dual active mini cooling fans over extruded aluminum fin heatsink | |
| Recommended System PSU | 300 Watts | |
| Bus & Interconnect | Bus Interface Capability | PCI Express 4.0 ×8 (Physical ×16 form factor, electrical ×8 lanes) |
| Negotiated Host Bus | PCI Express 3.0 ×8 (8.0 GT/s) on Supermicro X9SRi-F Slot 6 | |
| Root Port Routing | Connected to CPU Root Port 2a (00:02.0) direct to Xeon E5-2680 |
|
| Resizable BAR (ReBAR) | Hardware support up to 8 GB (BAR 0: 16 MB, BAR 1: 256 MB active) | |
| API & Platform Support | DirectX Support | DirectX 12 Ultimate (Feature Level 12_2) • Shader Model 6.8 |
| OpenGL / Vulkan | OpenGL 4.6 • Vulkan 1.3 | |
| Compute Frameworks | CUDA Compute Capability 8.6 • OpenCL 3.0 • DirectCompute | |
| Operating System Drivers | Linux nouveau DRM driver (DRI active) • NVIDIA 550.xx+ Display Drivers |
2. GA107 Silicon Topology & Ampere Architecture
The GA107 silicon die represents the high-efficiency foundation of NVIDIA’s Ampere family. Fabricated on Samsung’s 8 nm (8N) FinFET node, GA107 integrates 8.7 billion transistors into a compact 200 mm² die. In the 6 GB RTX 3050 configuration, the die is configured with 18 active Streaming Multiprocessors (SMs), each delivering double the FP32 execution throughput of previous generations alongside 3rd-generation Tensor Cores and 2nd-generation RT Cores:
+======================================================================================================================+
| NVIDIA GA107-325 [RTX 3050 LP 6G] SILICON DIE & SM TOPOLOGY |
+======================================================================================================================+
HOST INTERFACE: PCI EXPRESS 4.0 / 3.0 x8 BUS (ROOT PORT 00:02.0)
|
v
+------------------------------------+
| GIGA-THREAD ENGINE & HOST DISPATCH |
+------------------------------------+
|
+------------------------------------------+------------------------------------------+
| |
v v
+------------------------------------+ +------------------------------------+
| GRAPHICS PROCESSING CLUSTER 0 | | GRAPHICS PROCESSING CLUSTER 1 |
| | | |
| +-------------------------------+ | | +-------------------------------+ |
| | STREAMING MULTIPROCESSORS | | | | STREAMING MULTIPROCESSORS | |
| | (9 SM Blocks Active) | | | | (9 SM Blocks Active) | |
| | - 1,152 Concurrent FP32 Cores | | | | - 1,152 Concurrent FP32 Cores | |
| | - 36 Tensor Cores (Gen 3) | | | | - 36 Tensor Cores (Gen 3) | |
| | - 9 Ray Tracing Cores (Gen 2) | | | | - 9 Ray Tracing Cores (Gen 2) | |
| | - 36 Texture Units (TMUs) | | | | - 36 Texture Units (TMUs) | |
| | - 1,152 KB L1 Data / Shared | | | | - 1,152 KB L1 Data / Shared | |
| +-------------------------------+ | +-------------------------------+ |
+------------------------------------+ +------------------------------------+
| |
+------------------------------------------+------------------------------------------+
|
v
+------------------------------------+
| 2,048 KB (2 MB) HIGH-SPEED L2 POOL |
+------------------------------------+
|
+------------------------------------------+------------------------------------------+
| | |
v v v
+-----------------------+ +-----------------------+ +-----------------------+
| MEMORY CONTROLLER 0 | | MEMORY CONTROLLER 1 | | MEMORY CONTROLLER 2 |
| - 32-bit GDDR6 Bus | | - 32-bit GDDR6 Bus | | - 32-bit GDDR6 Bus |
| - 10 ROP Units | | - 11 ROP Units | | - 11 ROP Units |
+-----------------------+ +-----------------------+ +-----------------------+
| | |
v v v
2,048 MB GDDR6 2,048 MB GDDR6 2,048 MB GDDR6
(1750 MHz Clock) (1750 MHz Clock) (1750 MHz Clock)
+-------------------------------------------------------------------------------------------------------------+
| AGGREGATE MEMORY SUBSYSTEM: 6,144 MB (6 GB) GDDR6 • 96-BIT BUS • 168.0 GB/s THEORETICAL BANDWIDTH |
+=============================================================================================================+
3. PassMark Synthetic Benchmarks & Generational Stack Performance
In the standardized PassMark PerformanceTest 3D Graphics benchmark suite, the MSI GeForce RTX 3050 LP 6G OC achieves a composite G3D Mark of 10,744, earning Rank #4 in Mateo’s Museum GPU Collection. Most remarkably, it achieves this high performance while consuming only 70 Watts, establishing the highest power efficiency (153.5 G3D points per Watt) of any discrete desktop graphics card in the museum:
PassMark G3D Mark
10,744
Rank #4 in Museum GPU Collection
PassMark G2D Mark
682
Fluent 2D Desktop & Multi-Window GUI
Performance Per Watt
153.5 pts/W
Museum Efficiency Leader (10,744 / 70W)
Speed vs Quadro K4000
+295.3%
Nearly 4× Faster Than T7610 GPU
| Hardware Model | Architecture & Silicon | VRAM Buffer | TDP (Watts) | PassMark G3D | Relative Variance |
|---|---|---|---|---|---|
| MSI RTX 2060 Ventus 6G OC | Turing (TU106, 12 nm, 1920 cores) | 6 GB GDDR6 (192-bit) | 160W | 14,111 | +31.3% (Museum Rank #1) |
| AMD Radeon Instinct MI50 | Vega 20 GL (7 nm, 3840 cores) | 32 GB HBM2 (4096-bit) | 300W | 13,571 | +26.3% (Museum Rank #2) |
| NVIDIA RTX 3060 Laptop GPU | Ampere (GA106, 8 nm, 3840 cores) | 6 GB GDDR6 (192-bit) | 130W | 13,102 | +21.9% (Museum Rank #3) |
| MSI GeForce RTX 3050 LP 6G OC | Ampere (GA107, 8 nm, 2304 cores) | 6 GB GDDR6 (96-bit) | 70W | 10,744 | Current Museum Exhibit (Rank #4) |
| Sapphire RX 580 2048SP OC | Polaris 20 (14 nm, 2048 cores) | 8 GB GDDR5 (256-bit) | 150W | 7,609 | -29.2% (Museum Rank #5) |
| MSI GTX 680 Lightning | Kepler (GK104, 28 nm, 1536 cores) | 2 GB GDDR5 (256-bit) | 195W | 5,610 | -47.8% (Museum Rank #6) |
| HP NVIDIA Quadro K4000 | Kepler (GK106GL, 28 nm, 768 cores) | 3 GB GDDR5 (192-bit) | 80W | 2,718 | -74.7% (Museum Rank #7) |
| Gigabyte GeForce GTS 250 | Tesla (G92b, 55 nm, 128 cores) | 512 MB GDDR3 (256-bit) | 150W | 563 | -94.8% (Museum Rank #9) |
4. Power Efficiency, 75W Slot-Power & Low-Profile Engineering
100% PCIe Slot Power (No Auxiliary Cables)
The defining engineering breakthrough of the 6 GB RTX 3050 revision is its 70 Watt TDP. By fine-tuning voltage-frequency curves and memory bus width to 96-bit, MSI completely eliminated the need for 6-pin or 8-pin auxiliary power cables.
All operational current is drawn directly through the motherboard’s PCIe slot contacts (rated for 75.000W), making it ideal for systems with proprietary power supplies, minimal cable clutter, and enterprise server motherboards without dedicated PCIe GPU power leads.
Low-Profile Dual-Slot Form Factor
Measuring just 174 mm in length, 69 mm in height, and 42 mm in width, the card occupies two expansion slots while fitting into ultra-compact small form factor (SFF) chassis, 2U rackmount enclosures, or standard ATX towers.
MSI packages the card with both a pre-installed low-profile bracket and a full-height ATX expansion bracket. In the museum’s workstation build, the card is mounted using the full-height bracket to seamlessly interface with standard ATX chassis rear I/O slots.
5. Platform Synergy & Host Workstation Pairing
The MSI GeForce RTX 3050 LP 6G OC is actively deployed as the discrete graphics engine in Marius’ custom high-performance workstation build:
Host Machine: Corsair Carbide Series Air 540 “Cube” Workstation (Exhibit #2949)
-
Physical Slot Topology: Populated in PCIe 3.0 Slot 6 (Bus
0000:03:00.0) of the Supermicro X9SRi-F (Exhibit #2957) server motherboard. Slot 6 connects directly to CPU Root Port 2a (00:02.0) of the Intel Xeon E5-2680 (Exhibit #2963). - PCIe Bus Negotiation: The card natively supports PCIe 4.0 x8. On the LGA 2011 Romley platform, the link gracefully negotiates down to PCIe 3.0 x8 (8.0 GT/s), delivering 7.88 GB/s of bidirectional bandwidth which easily satisfies the throughput requirements of the GA107 graphics pipeline with zero bus saturation.
- Direct Airflow Path Thermals: The Corsair Air 540 cube enclosure houses the graphics card directly in the primary airflow path generated by dual front 140 mm intake fans. Cooler ambient air is delivered straight across the card’s dual mini cooling fans and extruded aluminum heatsink, exhausting out the rear expansion grid.
- Auxiliary Cable Elimination: Because the card requires zero external power leads, the main motherboard chamber of the Corsair Air 540 remains entirely unencumbered by bulky 6-pin/8-pin PCIe power cables, maximizing unobstructed airflow across the 160 GB DDR3 memory bank and CPU heatsink.
6. Verified Empirical Telemetry & Linux Diagnostics (Host: cube • 192.168.111.71)
Hardware telemetry captured during the live system audit of the active Corsair Air 540 “Cube” workstation confirms full hardware detection, PCIe link negotiation, and Linux Direct Rendering Infrastructure operation:
| Diagnostic Parameter | Empirical Live Value | Field Validation Notes |
|---|---|---|
| Operating System & Kernel | Ubuntu 22.04.5 LTS (x86_64) | Linux Kernel 5.15.0-194-generic |
| PCI Device Location | 0000:03:00.0 (VGA compatible controller) |
Physical Slot 6 connected to CPU Root Port 2a (00:02.0) |
| Hardware Identification | 10de:2584 (rev a1) • Subsystem: 1462:8123 |
NVIDIA GA107 Ampere Silicon • MSI Subsystem ID verified |
| Audio Function Location | 0000:03:00.1 (Audio device) |
NVIDIA Audio Controller 10de:2291 (rev a1) • Driver: snd_hda_intel |
| PCIe Interface Capability | PCIe 4.0 x8 (16.0 GT/s, max link width x16) | Native GA107 8-lane interface capability reported by lspci |
| Negotiated PCIe Link | PCIe 3.0 x8 (8.0 GT/s) | Operating at maximum platform speed and full 8-lane electrical width |
| Slot Power Limit | 75.000 Watts | Zero auxiliary 6/8-pin power cables required; 100% slot power delivery |
| Resizable BAR (ReBAR) | BAR 0: 16 MB • BAR 1: 256 MB (hardware support up to 8 GB) | Full hardware ReBAR silicon support validated in kernel memory space |
| Active Kernel Driver | nouveau DRM driver |
Kernel modules: nvidiafb, nouveau • Direct Rendering active |
| VRAM Buffer & Shaders | 6 GB GDDR6 (96-bit, 168.0 GB/s) | 2,304 CUDA Cores • 72 Tensor Cores • 18 RT Cores |
7. Curator’s Assessment & Museum Provenance
-
Museum Hardware Provenance: Mateo’s Museum collection preserves 1 authentic physical unit of the MSI GeForce RTX 3050 LP 6G OC graphics card (MSI Model
MS-V812, Subsystem ID1462:8123). The card is actively deployed as the dedicated visual compute accelerator in Marius’ custom workstation build housed inside the Corsair Carbide Series Air 540 cube chassis. - Unprecedented Power Efficiency: Delivering a PassMark G3D score of 10,744 at a mere 70 Watt TDP (153.5 points per Watt), this card represents the pinnacle of slot-powered graphics engineering. It outperforms legacy 150W-195W high-end GPUs like the MSI GTX 680 Lightning (5,610) and Sapphire RX 580 (7,609) while consuming less than half their electrical power.
- Enterprise & SFF Workstation Utility: Because the card relies entirely on 75W PCIe slot power without auxiliary power connectors, it can be deployed in enterprise server motherboards (such as the Supermicro X9SRi-F) that lack auxiliary 6-pin/8-pin PCIe power drops, bringing modern DirectX 12 Ultimate, Vulkan 1.3, and CUDA compute capabilities to server platforms.
- Modern Multi-Monitor 4K/8K Display Hub: Equipped with dual HDMI 2.1 ports and a DisplayPort 1.4a connection, the card effortlessly powers high-refresh 4K UHD and 8K displays with full HDR and 10-bit color depth, transforming the Xeon E5-2680 server motherboard into an extraordinarily versatile, silent daily-driver workstation.
-
Empirical Telemetry Validation: In live system diagnostics on the active host
cube(192.168.111.71) under Ubuntu 22.04.5 LTS, the card verified flawless PCIe 3.0 x8 negotiation at 8.0 GT/s, complete DRI driver stability under the open-sourcenouveaustack, and full hardware Resizable BAR addressability up to 8 GB.