The Nvidia GeForce RTX 4080’s slow-paced announcement and eventual launch has been… an interesting highway. From its mis-named little brother getting “unlaunched” to the RTX 4080 Founders Edition being the identical dimension as the massive RTX 4090 for some motive, to the melting 12VHPWR power connectors—that are nonetheless on this card, sure. Nvidia has fairly the PR battle with the $1,200 GeForce RTX 4080, however when you ignore all of the noise, does this new card ship the products for content material creators?
Thankfully, performance-wise it does largely dwell as much as the hype; I simply discover myself wishing we had increased VRAM capacities on such costly graphics playing cards.
This overview is concentrated on work, as an alternative of play, just like my recent RTX 4090 content creation analysis. Photo enhancing, video enhancing, encoding, AI artwork era, 3D and VFX workflows—stuff I take advantage of day-after-day for my EposVox and analog_dreams YouTube channels. Look for PCWorld’s gaming-focused RTX 4080 overview to land tomorrow, or try our high-level holistic take a look at 5 must-know RTX 4080 facts from our intensive gaming and creator testing proper now.
Let’s dig in.
Our take a look at setup
Most of my benchmarking was carried out on this take a look at bench:
- Intel Core i9-12900okay CPU
- 32GB Corsair Vengeance DDR5 5200MT/s RAM
- ASUS ROG STRIX Z690-E Gaming Wifi Motherboard
- EVGA G3 850W PSU
- Source recordsdata saved on a PCIe gen 4 NVMe SSD
This time, my targets had been to see how properly the brand new RTX 4080 stood its floor in opposition to the last-generation GeForce RTX 3080 and RTX 3090. At $1,200, the RTX 4080 wants to essentially out-shine the RTX 3090 (which may now be had for $1,000 to $1,200 on common)—however then if it does, the query turns into whether or not the RTX 4090 is warranted in any respect.
For some extra hardcore testing later, as I’ll point out, testing was accomplished on this take a look at bench:
- AMD Threadripper Pro 3975WX CPU
- 256GB Kingston ECC DDR4 RAM
- ASUS WRX80 SAGE Motherboard
- BeQuiet 1400W PSU
- Source recordsdata saved on a PCIe gen 4 NVMe SSD
- Each take a look at featured every GPU in the identical config as to not combine outcomes.
Photo enhancing
I’m going to maintain testing this simply because there’s fixed “little surprises” in efficiency in photograph apps—even when, realistically, you’re not going to get any main features upgrading from one high-end graphics card to a different.
Adam Taylor/IDG
Adam Taylor/IDG
The GeForce RTX 4080 tops the charts for many scores in Adobe Photoshop (apart from the AMD Radeon RX 6600’s weirdly excessive efficiency… maybe we have to get some extra AMD playing cards in the home for an AMD photograph enhancing deep dive). Photoshop was examined utilizing the PugetBench take a look at suite, offered by the superior workstation builders over at Puget Systems.
Adam Taylor/IDG
Adam Taylor/IDG
In Lightroom Classic (additionally utilizing PugetBench), the RTX 4080 beats out even the 4090’s general rating, however for many duties, you’ll see marginal features over the RTX 3080 and 3090, and proper up in opposition to the 4090’s efficiency.
Adam Taylor/IDG
Adam Taylor/IDG
Adam Taylor/IDG
Adam Taylor/IDG
Benchmarking Affinity Photo (I do know V2 just came out… let’s hold it constant for now) the RTX 4080 slots completely between the RTX 3090 and 4090, because it ought to.
Adam Taylor/IDG
Video enhancing
Adam Taylor/IDG
Adam Taylor/IDG
Adam Taylor/IDG
Using PugetBench for Adobe Premiere Pro sees the brand new RTX 4080 land a good bit increased than the RTX 3090, slightly below the RTX 4090. Interestingly, export and playback scores had been barely increased than the RTX 4090, in the meantime the Effects and GPU scores had been decrease. If you’re doing GPU-heavy workloads at 4K and better, you’ll completely need a 4090, however most video editors will likely be simply positive with a 4080.
Adam Taylor/IDG
Adam Taylor/IDG
This carries over into BlackMagic DaVinci Resolve. Resolve simply obtained a brand new replace to model 18.1—a huge replace bringing AI audio cleanup, new fusion options and lastly fixing fractional scaling points on Windows, together with efficiency boosts.
Adam Taylor/IDG
Running PugetBench, we see the GeForce RTX 4080 put up fairly substantial features—features you’d really feel—over final gen’s RTX 3080 and even the 3090. This is severely spectacular stuff. If you had been contemplating a 3090 or RTX Titan for video enhancing earlier than across the $1,000 value vary, I’d actually contemplate throwing in an additional couple hundred {dollars} to get this increase plus AV1 encoding! If you had been eying a 3080 or 3070 as an alternative, that’s a a lot harder promote given the upper price.
Adam Taylor/IDG
Adam Taylor/IDG
Adam Taylor/IDG
Adam Taylor/IDG
Adam Taylor/IDG
Exporting a pattern 4K venture, the RTX 4080 and RTX 4090 run laps across the older playing cards in H265 and AV1 in comparison with Arc. They’re nonetheless quicker than older playing cards in H.264 too, however by a lot smaller margins.
Adam Taylor/IDG
Throwing the cardboard in my Threadripper Pro construct to textual content my intense 8K venture filled with 8K RAW footage, Resolve SuperScaled 4K footage, plugins and results… sadly the 4080 couldn’t export this, identical to the earlier 80-tier playing cards I examined. 16GB of VRAM nonetheless isn’t sufficient, I suppose—plugins error out, colour graded footage takes longer than the 4090, and so forth. 8K is hard, however for $1,200 I’d hope at this level to begin to see increased VRAM capacities when Intel is handing out $350 Arc graphics cards with 16GB. Strange occasions.
One of the large enhancements with Resolve 18.1 is pace will increase to AI-based Face Tracking and object masking. The beta construct of Resolve I had entry to for my RTX 4090 overview was alleged to have efficiency boosts for this, too—however from what I can inform the Face Tracking updates weren’t working proper as a result of it’s considerably fasterin 18.1 in comparison with even that preview construct.
The time taken to face observe on 8K RAW footage in a 4K timeline dropped by 45 % throughout all examined GPUs. Even the older Ampere and Turing playing cards moved quicker for this sometimes very gradual job—however Lovelace simply runs away with it.
Adam Taylor/IDG
The RTX 4080 tracks simply barely slower than the 4090, however each go away the Ampere playing cards within the mud. Weirdly, the RTX Titan is in an in depth third behind the brand new playing cards—maybe some added AI magic within the Titan {hardware}, or higher floating level efficiency benefits? I’m unsure.
Adam Taylor/IDG
Object masking didn’t enhance as a lot with older generations, however the GeForce RTX 4080 matches the 4090 right here and places up 17 % quicker speeds than older playing cards. Very rad. Huge shouts out to the Resolve staff for nailing such an enormous replace, and when you do video enhancing day-after-day like I do, this new era goes to earn your time again quick. I’m so bored with ready on issues.
AI wizardry
Upscaling video to 4K utilizing Topaz Labs’ Video Enhance AI (720p upscaled to 4K utilizing Artemis Low Quality), the GeForce RTX 4080 is a tad slower than the 4090, however nonetheless nearly a minute quicker than the RTX 3090. It’s nonetheless a bit of too gradual for me to be comfy utilizing it on the common for quick-turnaround initiatives, however we’re getting there!
Adam Taylor/IDG
Similar outcomes are present in Gigapixel. Upscaling 8 pictures reveals the RTX 4080 operating 10 seconds slower than the 4090, however nonetheless a good bit quicker than the rest.
Adam Taylor/IDG
PhotoAI wanted to be utterly re-tested. Remember how horrible Nvidia playing cards carried out in my RTX 4090 review? Apparently there was an enormous bug inflicting that, and Topaz has now mounted it. The RTX 4080 performs the identical because the 4090 right here, however each playing cards take lower than a third of the time the older playing cards take. This is the type of AI efficiency enhancements I used to be anticipating to see with Lovelace!
Adam Taylor/IDG
Generating AI artwork with Stable Diffusion reveals the RTX 4080 being a bit of slower than the RTX 3090 usually, however nonetheless beating out the RTX Titan and 3080. I’m informed by folks in these communities that there’s most likely some optimization that may be accomplished within the apps for Stable to optimize particularly for Lovelace, so hopefully we see that quickly. (Who do I ship a card to?)
Adam Taylor/IDG
Adam Taylor/IDG
Using MoveFrames to AI interpolate a 60FPS video into 120FPS for slowmo, the RTX 4080 is just a bit quicker than the RTX 3090. Nice to see, however Lovelace has {hardware} built-in to do body interpolation in real-time with DLSS3 Frame Generation—there merely isn’t any software program help for this outdoors of video games. (For video, this can be a characteristic referred to as FRUC.) I’d love to see integrations with DaVinci Resolve for making use of this to footage on the timeline, in addition to in OBS Studio for smoothing out stuttery captures. A man can dream, I assume.
Adam Taylor/IDG
3D & VFX
If you’re a 3D artist or VFX artist utilizing 3D rendering, Nvidia’s GeForce RTX 4080 isn’t any slouch, both. Now, the larger, badder RTX 4090 put up insane numbers, doubling the 3090’s efficiency in methods which you could’t absolutely count on with this step-down GPU—but it surely’s shut!
Adam Taylor/IDG
In Blender, benchmarking the Monster scene with the RTX 4080 nets about 60 % quicker pace over the RTX 3090 and 85 % quicker than the 3080. The Junkshop scene benched about 29 % quicker than the 3090 and 50 % quicker than the 3080, whereas the basic Classroom scene reveals 55 % quicker efficiency than the 3090 and 82 % quicker than the 3080. So whereas the GeForce RTX 4080 was by no means going to be as absurdly quick because the 4090, these are some superb features.
Adam Taylor/IDG
Adam Taylor/IDG
The development continues into Octane, the place the RTX 4080 performs 42 % quicker than the 4090 and 67 % quicker than the 3080. And only for kicks, when you’re nonetheless on the RTX Titan… almost 3X the efficiency. I’m so sorry that card as soon as price $2500.
Adam Taylor/IDG
Testing the V-Ray engine with the VRAY 5 benchmark is an oddity. The VRAY RTX take a look at reveals the 4080 nonetheless trailing behind the RTX 4090 a good bit, however leaping considerably over the older era GPUs, however in CUDA? For some motive I’m persistently scoring beneath these older playing cards in efficiency. I can’t clarify this, a bug, given all the opposite efficiency outcomes we’re seeing right here.
Adam Taylor/IDG
Streaming
All GeForce RTX 40-series playing cards have twin video encoder chips to permit for quicker and better high quality 4K and 8K recording, in addition to 8K 60FPS seize (Ampere caps out at 8K 30FPS). You can learn extra about this within the streaming section of my RTX 4090 review, however this has been implausible for my workflow and has me placing 40-series playing cards in all of my seize and streaming rigs.
I’m recording most issues to AV1 now to avoid wasting on area. Resolve helps enhancing it with out difficulty now, and the latest Nvidia driver update just improved AV1 decode performance, too! I’m using the additional general headroom to document my display captures in H.265 in 4:4:4 chroma subsampling for lossless punch-ins for my tutorials and for lovely 8K upscales on export. You can view a demo of this here—display seize by no means seemed so good, and I couldn’t do that with out risking body drops and encoder lag even on my RTX 3090.
As with the RTX 4090, recording AV1 in OBS Studio 28.1 (which now supports NVENC AV1 within the public construct) has very minimal efficiency impression in your gameplay, a mere fraction of the hit that H.264 causes. I’ll have extra deeper testing on this coming later./y
Discord can be getting AV1 streaming help for Go Live later this month. I examined this and it seems nice, however there’s some quirks to how Discord calls work. Since it’s a “video call,” video is distributed instantly from host to viewers, with no transcoding server or a lot of a middle-man as you’d have with providers comparable to Twitch and YouTube. This implies that each events should have the identical help.
So when you’re streaming AV1 to Discord and a viewer joins who doesn’t have the fitting driver or {hardware} (presently I’m unclear as as to whether it is going to require Nvidia playing cards for this replace, or simply GPU AV1 decode help usually, which would come with AMD and Intel), the stream will likely be pressured to revert to H.264. This could possibly be complicated and messy for group calls however remains to be an enormous win to have it built-in. I simply hope we see Arc and Radeon RDNA3 AV1 support added sooner slightly than later—and ideally SVT-AV1 CPU encoding help, because it’s gotten fairly quick.
Bottom Line: Nvidia’s product stack is complicated
While it felt like a win for Nvidia to “unlaunch” the 12GB 4080—because it by no means ought to have been referred to as a 4080 within the first place—doing so makes it a lot more durable to grasp Nvidia’s upcoming product stack in context. The GeForce RTX 4080 out-performs the last-gen RTX 3090 in lots of instances, which looks as if a win for $1,200, however you’re getting shorted on VRAM in ways in which do begin to matter for high-end work.
Just wanting as much as the $1,600 GeForce RTX 4090, the 4080’s efficiency checks out. But the cheaper 12GB… 4070? 4070 Ti? 4080 mini?… that may ultimately hit the streets is perhaps largely simply pretty much as good for this sort of work for less expensive, on condition that the RTX 4080’s 16GB of VRAM nonetheless limits super-high-end work. But we received’t know till Nvidia’s subsequent merchandise break cowl.
It’s laborious to counsel shopping for something proper now till this painfully gradual season is over—however ready places you in battle with scalpers and shortages and FOMO. It’s powerful. I used to be actually hoping the “unlaunch” situation would drive down the worth of each RTX 4080 playing cards a bit, however I assume that’s not taking place. Either manner, in case your explicit content material creation workflows don’t require greater than 16GB of onboard VRAM, Nvidia’s GeForce RTX 4080 delivers some beautiful efficiency enhancements over its predecessors.