Quick answer

For 4K raster gaming, the RTX 5080 averaged 77.4 FPS to the RTX 5070 Ti’s 68.6 in Tom’s Hardware’s test. That is 12.8 percent more frames. If the delivered 5080 price is more than 12.8 percent above a comparable 5070 Ti, the Ti costs less per frame in that suite. Both cards have 16GB.

Contents
Jump to a decision

The RTX 5080 makes sense when the extra frames at 4K change how you play. In the 4K raster suite below it averaged 77.4 FPS; the RTX 5070 Ti averaged 68.6. Divide each board’s delivered price by that board’s frame rate. At similar feature support and the same 16GB capacity, that is a better starting point than comparing either card with its launch price. My 4.4 out of 5 rating is the same as in my ROG Astral RTX 5080 OC review.

The reservation is price. NVIDIA announced the RTX 5080 at $999 on January 6, 2025 for a January 30 launch. At Newegg on October 7, 2026 the lowest of the six new listings sold and shipped by Newegg on the search page I checked was the ASUS PRIME-RTX5080-16G at $1,599.99, about 60 percent over that launch price (derived from 1,599.99 against 999). The silicon did not change. The price did.

This page reviews the RTX 5080 as a GPU family across several partner boards. The readings for my own card come from the ASUS ROG Astral RTX 5080 OC that I benchmarked over June 8 to 11 of 2026, with a June 28 telemetry log, and published in the ROG Astral review and its open dataset. The separately normalized comparison index uses cards from my retired test bench; the other board results come from the named reviews cited inline.

I add no new measurements here.

The version and fit to check

NVIDIA’s RTX 5080 specification page, read October 7, 2026, prints the reference figures below. The right-hand column is where partner boards diverge.

ItemNVIDIA specificationWhy I would check it
GPUBlackwell GB203, 10,752 CUDA coresSame die as the RTX 5070 Ti; the 5080 needs a fully enabled chip (Tom’s Hardware)
Clocks2.62 GHz boost, 2.30 GHz baseASUS lists 2,760 MHz default for the Astral OC
Memory16GB GDDR7, 256-bitSame capacity as the RTX 4080 Super and RTX 5070 Ti
Total graphics power360 WMy Astral OC ships at a 400 W limit
Required system power850 WASUS recommends 850 W for the Astral OC too
Power connector3x PCIe 8-pin via the bundled adapter, or one 450 W-or-greater PCIe Gen 5 cableCheck the connector, adapter and PSU guidance for the board you buy
Founders Edition size304 mm long, 137 mm wide, 2 slotsThe Astral OC is 357.6 x 149.3 x 76 mm, 3.8 slots
Maximum GPU temperature88 °CThe ceiling for every thermal figure below
Outputs3x DisplayPort, 1x HDMI; 4K at 480 Hz or 8K at 165 Hz with DSCThe Astral adds a second HDMI 2.1b
Interface and codecsPCIe Gen 5; DLSS 4; 2x ninth-gen NVENC, 2x sixth-gen NVDECDual encoders matter for streaming

KitGuru’s spec recap fills in the SM count NVIDIA’s page omits. The RTX 5080 is a full implementation of the GB203 die with 84 SMs and 84 RT cores plus 336 Tensor cores, and its 30 Gbps GDDR7 delivers 960 GB/s, a 30 percent increase over the RTX 4080 Super.

Capacity stays at 16GB. Only the bandwidth moved.

Two fit facts decide whether the card goes in at all. Length and slot count belong to the board: the Founders Edition is a 2-slot card at 304 mm, while my Astral OC is about 54 mm longer (derived from ASUS’s 357.6 mm) and hangs vertically on the HYTE Y60’s riser. The connector is the other. For the Founders Edition, follow NVIDIA’s 450 W-or-greater native cable or three 8-pin adapter instructions; check the individual partner board and PSU manual. The slot generation matters less; my card links at PCIe 4.0 x16.

How fast it is

The generational gain is small. TechSpot’s 17-game suite (Ryzen 7 9800X3D, driver 566.36, January 29, 2025) put the RTX 5080 4 percent ahead of the RTX 4080 Super at 1440p; at 1080p it only matched the 4080 Super. At 4K the gap opened to 11 percent over the 4080 Super and 8 percent over the RX 7900 XTX. With ray tracing and quality upscaling TechSpot measured 5 percent at 1440p and 9 percent at 4K.

Tom’s Hardware’s 16-game rasterization suite (January 29, 2025) found 9 percent over the 4080 Super at 4K ultra. KitGuru’s 12 games (9800X3D, 64GB DDR5-6000 CL30) landed at 9 percent at 1440p and 13 percent at 4K; KitGuru called that gain “clearly disappointing”. GamersNexus (January 31, 2025) measured 112 FPS in FFXIV Dawntrail at 4K, about 15 percent better than the 4080 Super, against 139 for the RTX 4090 and 182 for the RTX 5090, and called the card “a boring product generationally”.

Three suites, and every answer sits between 9 and 13 percent at 4K.

The ceiling above it is far away. Tom’s Hardware’s conclusion says the RTX 5090 is “over 50% faster on average” at 4K; KitGuru put the 5080 15 percent behind the RTX 4090 and 34 percent behind the RTX 5090 at 4K. The site’s RTX 5090 vs RTX 5080 page works through that pair. The floor below it is close. Tom’s Hardware’s RTX 5070 Ti review (February 19, 2025) explains that the 5070 Ti uses the same GB203 with 83 percent of the cores enabled and “may be up to 17% slower than the 5080, give or take, for 25% less money”, with the gap “only 5~10 percent” in memory-dependent tasks. Read the RTX 5070 Ti vs RTX 5080 comparison before paying the difference.

My own card confirms the tier. On my hand-tuned profile the Astral ran Cyberpunk 2077 Ultra raster at 4K at 58 FPS average with a 49 FPS 1 percent low, and Alan Wake 2 RT Ultra with DLSS Quality at 49 and 38. In the site’s 4K raster index, with the RTX 4080 Super at 100, the RTX 5080 FE scores 114 and the Astral on my manual profile 123; the FE and the Ada cards were measured on my retired 2024 to 2025 bench and normalized as the dataset documents.

Multi Frame Generation separates the RTX 5080 from the RTX 4080 Super in a way no raster chart shows. In path-traced Cyberpunk 2077 at 4K on my tuned card, native rendering gave 19 FPS at 95 ms of PC latency, while DLSS Quality alone gave 40 FPS at 45 ms. With DLSS Quality still on, MFG 4x raised the displayed frame rate to 135 FPS and added 17 ms of latency, to 62 ms. TechPowerUp’s DLSS 4 table for the Founders Edition in Cyberpunk 2077 with ray tracing runs 35 FPS at 55 ms native to 120 FPS at 74 ms with MFG x4, against 54 and 196 for the RTX 5090; the table does not state its resolution (archive capture of March 20, 2026 after a live 403).

I did not repeat those five 4K FPS tests at stock settings or measure the noise with a meter. I did record stock gaming power and temperatures separately; the readings follow below.

Power, thermals and noise

The Founders Edition drew less than its rating in these gaming tests. Tom’s Hardware’s PCAT v2 measurements averaged 297 W at 4K ultra against 299 W for the RTX 4080 Super, far below the 360 W TGP. KitGuru’s 12-game average was 242.3 W at 1080p, and at 4K eight of twelve games drew under 300 W. GamersNexus’s interposers read 298 W in FFXIV at 4K and 330 W in F1 24 at 4K with ray tracing.

Temperatures follow. TechSpot’s Founders Edition peaked at 63 °C after an hour of The Last of Us Part I at 4K maxed in an enclosed ATX case, at 266 W average. GamersNexus saw 65 to 66 °C on the GPU and 72 to 75 °C on memory. KitGuru’s 30-minute stress test held the GPU just below 65 °C, 2 dBA quieter than the RTX 5090 FE with the meter at one foot, and heard no coil whine.

TechPowerUp’s temperature and noise table (open bench, microphone at 50 cm; archive capture of September 15, 2026 after a live 403) puts the boards side by side. The Founders Edition ran 67 °C at 36.8 dBA. The ASUS Astral OC ran 62 °C at 36.3 dBA on its performance BIOS and 70 °C at 25.8 dBA on its quiet BIOS, and MSI’s Suprim SOC managed 60 °C at 25.5 dBA. My Astral matches that picture. At stock it games at 266 to 340 W and 63 to 72 °C with load noise near-silent. My June 28 HWiNFO64 log recorded the hand-tuned profile in two windows of about 16 and 10 minutes, separated by nearly an hour without readings. Across 308 samples taken about five seconds apart within those windows, the card averaged 261 W, 56 °C and 3,202 MHz. Every sampled throttle flag was zero. The file cannot rule out throttling between samples or during the gap.

Coil whine on my card scores 1 in the site’s database, audible only in uncapped game menus.

NVIDIA and ASUS specify 850 W for the boards above. Check the PSU recommendation and supported cable for your exact board and system; the RTX 50-series PSU guide sizes the rest of the system. If the stock fan curve is louder than you want, the RTX 5080 undervolt guide is the next step.

Price and the alternatives

The launch price was $999 on January 6, 2025. The October sample was higher. On October 7, 2026 the untagged Newegg search for an RTX 5080 graphics card showed 36 first-page listings, six of them sold and shipped by Newegg, from the ASUS PRIME-RTX5080-16G at $1,599.99 to the ASUS TUF OC at $1,899.99; the Sources list carries every item number. Tom’s Hardware’s price tracker, modified September 28, 2026, lists the best US RTX 5080 price at $1,638 against a $929 lowest-ever, and its September 1, 2026 report put the card at $1,549, 55 percent over MSRP, blaming component supply diverted to enterprise AI customers.

At the two October quotes, the 5070 Ti offered more frames per dollar. A Gigabyte Eagle RTX 5070 Ti sold and shipped by Newegg cost $1,179.99 on October 7, 2026, so the 5080 premium was $420, or 35.6 percent (derived), for 12.8 percent more 4K raster speed in Tom’s Hardware’s GPU hierarchy updated June 24, 2026 (77.4 against 68.6 fps, 42.1 against 36.7 with ray tracing, 14.7 percent; the percentages are mine), or up to 17 percent by Tom’s launch-review estimate.

At an RTX 5070 Ti quote of $1,179.99, a 5080 would have had to cost about $1,331 to match its cost per frame in the cited raster test ($1,179.99 × 77.4 ÷ 68.6). With a different quote, plug that price into the same calculation. On November 5 the exact ASUS Prime listing showed Newegg’s own offer as out of stock. A marketplace offer or another 5080 is a different comparison; check the seller, delivered price and return terms again. In its November 2 pricing survey, TechSpot recorded different U.S. October prices: $2,200 for the 5080 and $1,250 for the 5070 Ti. Those were model-level samples, not quotes for the two boards above.

The next tier was much more expensive in the checked sample. The untagged Newegg RTX 5090 search on October 7, 2026 started at $6,885.90 for a Gigabyte Windforce OC; I did not read the sellers of those cheapest listings. Tom’s Hardware’s tracker listed $6,995 as the best US RTX 5090 price on September 28, 2026, more than four times the cheapest RTX 5080 (derived). At those quotes, paying several times more for a 5090 changes the budget as well as the performance target. Waiting has no date attached either. TechPowerUp reported on February 5, 2026, citing The Information, that NVIDIA postponed the RTX 50 Super refresh, including a planned 24GB RTX 5080 Super, because 3GB GDDR7 is going to AI products (archive capture of July 13, 2026 after a live 403), and Notebookcheck’s June 8, 2026 report, citing Benchlife, points at CES 2027 while noting contradictory earlier reports. In a statement the previous year, NVIDIA told Tom’s Hardware on September 15, 2025 that Founders Editions “continue to be in production”, and told Wccftech in January 2026 that it still ships all GeForce SKUs while memory supply is constrained.

One launch-era condition matters for a used purchase. Tom’s Hardware reported on February 8, 2025 that driver 572.16 caused black screens on the RTX 5080 and 5090, and TechRadar on March 6, 2025 covered the 572.70 driver’s claimed fix, with users still reporting problems then. I found no 2026 firmware or hardware revision.

The 4.4 rating reflects the 5080’s 4K performance and the 16GB limit. For a 1440p or budget-limited buyer, the October comparison favored the 5070 Ti on cost per frame. If today’s quotes are close, use the 4K numbers for the games and settings you actually run.

Who should buy it

  • You game at 4K on a 120 Hz or faster display and want ray tracing or path tracing with DLSS 4 Multi Frame Generation, which the 5070 Ti handles with less headroom.
  • You want more 4K headroom than the 5070 Ti delivers in your games, but a 5090 at its actual delivered price is outside your budget.
  • Your workload scales with CUDA cores or the two NVENC encoders, such as streaming or paid compute.
  • Your case clears the exact board (304 mm and 2 slots for the FE, 357.6 mm and 3.8 slots for the Astral OC) and you have checked the PSU and cables for that board and the rest of your system.

Who should skip it

  • You play at 1440p or 1080p. TechSpot measured just 4 percent more frames at 1440p than the RTX 4080 Super. At the October 7 board quotes the 5070 Ti cost $420 less; compare the gap on the offers you can buy.
  • You need more than 16GB for 4K path tracing with frame generation, or local LLMs above the 16GB tier; Tom’s Hardware found Indiana Jones and the Great Circle at 4K with full RT and frame generation overflows it.
  • You own an RTX 4080 Super. The upgrade is 9 to 13 percent at 4K in the named suites, and Tom’s Hardware calls the benefits “pretty thin”.
  • You can wait for official specifications. TechSpot’s November 2 report describes a 24GB successor as a rumor; neither that report nor the earlier accounts confirm a release date.

Before checkout, measure the board you picked against the case rather than the Founders Edition dimensions on NVIDIA’s page.

Sources

Frequently asked questions

Is the RTX 5080 worth it in 2026?
At 4K, TechSpot measured the RTX 5080 11 percent ahead of the RTX 4080 Super across 17 games; its 1440p lead was just 4 percent. For a new purchase, compare the delivered price with the extra frames your games need. The $1,599.99 ASUS Prime 5080 and $1,179.99 Gigabyte Eagle 5070 Ti were two October 7 offers. On November 5 Newegg listed its own ASUS Prime offer as out of stock.
How much faster is the RTX 5080 than the RTX 5070 Ti?
Both use NVIDIA’s GB203 chip; the 5070 Ti has 83 percent of the cores enabled. Tom’s Hardware’s RTX 5070 Ti review estimated the 5070 Ti at up to 17 percent slower than the 5080 for 25 percent less money at launch prices, with only a 5 to 10 percent gap in memory-bound tasks. At Newegg on October 7, 2026 Newegg sold the ASUS Prime RTX 5080 for $1,599.99 and the Gigabyte Eagle RTX 5070 Ti for $1,179.99: a $420 difference.
Is 16GB of VRAM enough on the RTX 5080?
For most 4K gaming in 2026, yes, but it is the card’s known limit. The RTX 5080 carries 16GB of GDDR7 on a 256-bit bus, the same capacity as the RTX 4080 Super and RTX 5070 Ti. Tom’s Hardware found Indiana Jones and the Great Circle at 4K with full ray tracing, DLSS Quality and frame generation needed more than 16GB and locked up. A reported 24GB successor has no confirmed release date in the cited reporting.
What PSU does the RTX 5080 need?
NVIDIA lists 360 W total graphics power and an 850 W minimum system power supply, fed by one 450 W-or-greater PCIe Gen 5 16-pin cable or three 8-pin cables through the adapter in the box. Measured draw sits well below the rating. Tom’s Hardware averaged 297 W at 4K ultra on the Founders Edition and KitGuru saw eight of twelve games under 300 W at 4K. Factory-overclocked boards ship with higher limits; ASUS also recommends 850 W for its ROG Astral OC.
Should I wait for the RTX 5080 Super?
There is no confirmed release date in the sources below. TechSpot reported on November 2 that a 24GB successor was rumored for the first quarter of 2027, but had no confirmation. If your work already needs more than 16GB, compare cards that meet that requirement and are available to order.

Sources and corrections

Last updated
Methodology
See our methodology for research and review standards.
Corrections
Spotted an error? Email a correction with the affected line and what you found. The draft includes this page’s address. Verified changes are recorded here. Contact details.

About the author

Written by Lowell K. Wood IV, who builds and runs TechFuelHQ from St. Louis, Missouri.