NVIDIA GeForce RTX 3050 6 GB Review - The Fastest Slot-Powered GPU 106

NVIDIA GeForce RTX 3050 6 GB Review - The Fastest Slot-Powered GPU

(106 Comments) »

Introduction

NVIDIA Logo

A few weeks ago, NVIDIA released the new GeForce RTX 3050 6 GB as its new entry level discrete GPU—here's our review. This new GPU slipped into NVIDIA's product stack with much less fanfare than the RTX 40 SUPER series, and replaces the likes of the GTX 1650 and GTX 1660 as sub-$200 options. In case you haven't noticed, the latest crop of discrete graphics cards are expensive—three-figure prices are normal, gone are the days of the $70 entry-level, because processor integrated graphics solutions have grown powerful over the years. Besides an entry level price, the RTX 3050 6 GB is particularly energy efficient, and is designed to operate without additional power connectors, needing just the PCIe slot power to run.

The GeForce RTX 3050 6 GB that we review today is based on the older Ampere graphics architecture, which meets the full DirectX 12 Ultimate feature set, including real time ray tracing; as well as support for DLSS. Is the card fast enough for ray tracing? We'll answer this question later in the review. If you recall, NVIDIA's argument against ray tracing for entry-level products was that the hardware would be too slow for gameplay at acceptable frame-rates, which is why when it debuted RTX with Turing, it had carved out an entire lower end product stack—the GTX 16-series, which lacked ray tracing capabilities. So, what's changed now? We believe it's the growth in popularity of DLSS, and Intel's debut into the GPU market.



Intel in 2022 released its Arc Alchemist discrete graphics, and even its entry-level Arc A380 features the full DirectX 12 Ultimate feature set along with XeSS, a modern upscaling technology. The A380 is a really slow GPU, but later in 2023, the company released the Arc A580 for less than $200, which is based on a larger silicon, and capable of 1080p gaming, or gaming with ray tracing when you engage XeSS. Combine this with Intel's channel marketing muscle, and its ability to get DIY retailers to push combos of the A580 with entry-level Core i3 processors and H610 chipset motherboards, and the company was in a position to grab a slice of the entry-level e-sports gaming pie. We believe that NVIDIA is responding to exactly this trend—to give the entry-level a convincing new GeForce RTX option that features DirectX 12 Ultimate, including ray tracing; and NVIDIA's popular DLSS super resolution technology. NVIDIA's play with the RTX 3050 6 GB is to address the crowd that games on a 1080p display, but with DLSS enabled, which means the game is actually rendering at a lower resolution, such as 720p, but being upscaled and made to look close to 1080p native.

The GeForce RTX 3050 6 GB is a significantly different product from the original RTX 3050 that NVIDIA launched two years ago—January 2022 to be exact. The original RTX 3050, with 8 GB of memory, had released at a time when the gaming GPU market was reeling with acute shortages in supplies due to the double whammy of the pandemic and the GPU crypto-currency mining craze. It was the only GPU that was available at a relatively acceptable price, and was very much capable of 1080p native resolution gameplay. The RTX 3050 8 GB was initially based on a cut-down GA106 silicon—the same chip which NVIDIA used to build its popular RTX 3060—but later switched to the smaller GA107 silicon without any change in specs, as the RTX 3050 8 GB specs sheet exactly aligned with that of the GA107. In fact, these newer revisions of GA107-based RTX 3050 8 GB were more energy efficient, with a 10-20 W reduction in gaming power draw, compared to the GA106-based ones.

For the new RTX 3050 6 GB, NVIDIA stuck with the same 8 nm GA107 silicon, but lowered various specs—not only memory size. To begin with, the most obvious change with the RTX 3050 6 GB is the VRAM size, which sees a 25% reduction to 6 GB from 8 GB, and a proportionate reduction in memory bus width, to 96-bit from 128-bit. The memory bus ticks at 14 Gbps, which yields 168 GB/s in memory bandwidth, compared to the 224 GB/s on the original RTX 3050. That's not all, the core count is slightly reduced, too. While the original RTX 3050 maxes out all 20 streaming multiprocessors (SM) present on the GA107, the new RTX 3050 6 GB enables 18 out of those 20. This works out to 2,304 CUDA cores, 72 Tensor cores, 18 RT cores, 72 TMUs, and the chip's full 32 ROPs.

Perhaps the biggest impact of this reduction in SM has been on power—it comes with a TGP of just 70 W. This means that the card is optimized to work without any power connectors (the original RTX 3050 came with a 6-pin connector to meet the 115 W TGP). The spartan 70 W power reflects in thermals, and partners are allowed to use simple extruded aluminium fan-heatsinks that cost just a few dollars to make. It also vastly simplifies the power design of the GPU.

NVIDIA has no reference design for the RTX 3050 6 GB, and so we have with us the Gainward RTX 3050 6 GB Pegasus. We chose Gainward because its PCBs tend to be designed close to NVIDIA's internal reference design boards and their cards are sold at highly competitive pricing. This card offers all there is to have with this GPU, and even makes things easy for the entry-level crowd, with legacy display connectivity, such as a DVI connector. It features a simple aluminium monoblock fan-heatsink. Gainward is pricing the card at $180, which is the baseline price set by NVIDIA for the RTX 3050 6 GB.

NVIDIA GeForce RTX 3050 6 GB Market Segment Analysis
 PriceCoresROPsCore
Clock
Boost
Clock
Memory
Clock
GPUTransistorsMemory
RX 6500 XT$1401024322685 MHz2825 MHz2248 MHzNavi 245400M4 GB, GDDR6, 64-bit
Arc A580$1653072961700 MHzN/A2000 MHzACM-G1021700M8 GB, GDDR6, 256-bit
RTX 3050 6 GB$1802304321042 MHz1470 MHz1750 MHzGA107Unknown6 GB, GDDR6, 96-bit
RTX 3050$2302560321552 MHz1777 MHz1750 MHzGA10612000M8 GB, GDDR6, 128-bit
Arc A750$20035841122050 MHzN/A2000 MHzACM-G1021700M8 GB, GDDR6, 256-bit
RX 6600$1701792642044 MHz2491 MHz1750 MHzNavi 2311060M8 GB, GDDR6, 128-bit
RX 6600 XT$2002048642359 MHz2589 MHz2000 MHzNavi 2311060M8 GB, GDDR6, 128-bit
RTX 3060$2403584481320 MHz1777 MHz1875 MHzGA10612000M12 GB, GDDR6, 192-bit
RX 7600$2702048642250 MHz2625 MHz2250 MHzNavi 3313300M8 GB, GDDR6, 128-bit
RX 7600 XT$3302048642470 MHz2755 MHz2250 MHzNavi 3313300M16 GB, GDDR6, 128-bit
RTX 4060$3003072481830 MHz2460 MHz2125 MHzAD10718900M8 GB, GDDR6, 128-bit
Arc A770$28040961282100 MHzN/A2187 MHzACM-G1021700M16 GB, GDDR6, 256-bit
RTX 3060 Ti$2504864801410 MHz1665 MHz1750 MHzGA10417400M8 GB, GDDR6, 256-bit

Packaging

Package Front
Package Back


The Card

Graphics Card Front
Graphics Card Back
Graphics Card Front Angled

Gainward's card is very compact, with a black color theme on the main cooler. A backplate is not available.

Graphics Card Dimensions

Dimensions of the card are 16.0 x 11.5 cm, and it weighs 411 g.

Graphics Card Height
Graphics Card Back Angled

Installation requires two slots in your system. We measured the card's width to be 40 mm.

Monitor Outputs, Display Connectors

Display connectivity options include one HDMI 2.1, one DisplayPort 1.4a and one DVI port (without analog support).

There is one NVENC and one NVDEC unit in the chip, which supports hardware-accelerated decode for H.264, H.265 and AV1. For encoding H.264 and H.265 are supported, but not AV1.

Graphics Card Power Plugs

The card requires no additional power cables, it runs with just PCIe slot power.

Teardown

Graphics Card Cooler Front
Graphics Card Cooler Back

The fan assembly can be removed independently of the heatsink, which makes it easy to clean both. It also lets you replace the fan should it die after a few years.


While we usually see heatpipes here, the Gainward card uses just a very basic heatsink—which is good enough for the card's low heat output. The heatsink provides cooling for the three memory chips, too, but not for the VRM, which is cooled by the airflow of the cooler.

High-resolution PCB Pictures

These pictures are for the convenience of volt modders and people who would like to see all the finer details on the PCB. Feel free to link back to us and use these in your articles, videos or forum posts.

Graphics Card Teardown PCB Front
Graphics Card Teardown PCB Back

High-resolution versions are also available (front, back).

Circuit Board (PCB) Analysis

GPU Voltage, VRM Configuration
GPU Chip Voltage Controller

GPU voltage is an 3-phase design, managed by a Richtek RT8845A.


Sinopower SM4503+ SM4507 MOSFETs are used for GPU voltage; they are rated for 80 A of current each.

Memory Voltage, VRM Configuration

Memory voltage is a two-phase design, managed by a Richtek RT8237 (not pictured)


For memory, SM4507NH/4503NH are used, too.

Graphics Card Memory Chips

The GDDR6 memory chips are made by Hynix and carry the model number H56G42AS4DX014. They are specified to run at 2000 MHz (16 Gbps GDDR6 effective).

Graphics Chip GPU

NVIDIA's GA107 graphics processor is built using a 8 nanometer process at Samsung, with a transistor count of 8.7 billion and a die size of 200 mm².

Test System

Test System - GPU 2024.1
Processor:Intel Core i9-14900K
Raptor Lake, 6.0 GHz, 8+16 cores / 32 threads
PL1 = PL2 = 330 W
Motherboard:EVGA Z790 Dark
BIOS 1.13
Resizable BAR:Enabled on all supported cards
(NVIDIA, AMD & Intel)
Memory:Thermaltake TOUGHRAM XG
2x 16 GB DDR5-7200 MHz 36-46-46-96
Cooling:Arctic Liquid Freezer II
280 mm AIO
Thermal Paste:Arctic MX-6
Storage:2x 2 TB M.2 NVMe SSD
Power Supply:Seasonic Vertex GX 850 W
ATX 3.0 / 16-pin 12VHPWR
Case:darkFlash DLX4000 Mesh
Operating System:Windows 11 Professional 64-bit 23H2
VBS enabled (Windows 11 default)
Drivers: RTX 3050 6 GB: 551.23 WHQL
RTX 4080 Super: 551.22 Press Driver
RTX 4070 Super & 4070 Ti Super: 551.23 WHQL
NVIDIA: 546.17 WHQL
AMD: 23.11.1 WHQL
Intel: 101.4953 WHQL
Arc A580: 101.5252 Beta
Date of Retest
Benchmark scores in other reviews are only comparable when this exact same configuration is used.

  • All games and cards are tested with the drivers listed above—no performance results were recycled between test systems. Only this exact system with exactly the same configuration is used for all results in this review.
  • All graphics cards are tested using the same game version.
  • All games are set to their highest quality setting unless indicated otherwise.
  • AA and AF are applied via in-game settings, not via the driver's control panel.
  • Before starting measurements, we heat up the card for each test to ensure a steady state is tested. This ensures that the card won't boost to unrealistically high clocks for only a few seconds until it heats up, as that doesn't represent prolonged gameplay.
  • For better real-life applicability, all game tests use custom in-game test scenes, not the integrated benchmarks
  • All cards used for comparison are reference designs. When a reference design does not exist, we go the extra mile and buy the closest possible match, using reference clocks and default power limit.
Each game is tested at these screen resolutions:
  • 1920x1080: Most popular monitor resolution.
  • 2560x1440: Intermediary resolution between Full HD and 4K, with reasonable performance requirements.
  • 3840x2160: 4K Ultra HD resolution, available on high-end monitors.
Our Patreon Silver Supporters can read articles in single-page format.
Nov 18th, 2026 13:12 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts