Sapphire AMD Radeon RX 7900 XTX Nitro Plus Review
A Bigger, Faster, Cooler Version of AMD’s Fastest GPU
This…is big. Sure, AMD made some noise about how compact their reference design was – taking shots at NVIDIA for the size of their Founders Edition cards (and 12VHPWR connectors) – but Sapphire has embraced the giant GPU trend.
And why not? The Radeon RX 7900 XTX is AMD’s fastest card this generation, and Sapphire’s NITRO+ is an even faster version of it. And it’s a monster. Seriously, if you have been sharing memes about NVIDIA’s big, power hungry GPUs, you might feel a little foolish after buying this.
The full product name of this card is the NITRO+ AMD Radeon RX 7900 XTX Vapor-X 24GB. The size of the name is fitting, considering the card is over a foot long and will occupy the better part of four slots in your case. Oh, and it requires three 8-pin PCI-E power cables.
Have no fear, the Sapphire NITRO+ card offers ARGB lighting bars on the top and bottom of the card (less noticeable when viewed from the side)
Product Specifications
- SKU: 11322-01-40G
- GPU: AMD Radeon RX 7900 XT
- Engine Clock:
- Boost Clock: Up to 2680 MHz
- Game Clock: Up to 2510 MHz
- Boost Clock is the maximum frequency achievable on the GPU running bursty workload. Boost clock achievability, frequency, and sustainabilty will vary based on several factors, including but not limited to: thermal conditions and variation in application and workloads.
- Game Clock is the expected GPU clock when running typical gaming applications, set to typical TGP(Total Graphics Power). Actual individual game clock results may vary.
- Stream Processors: 6144
- Compute Units: 96 CU (with RT+AI Accelerators)
- Infinity Cache: 96MB
- Ray Accelerators: 96
- Memory:
- Memory Size/Bus: 24GB/384 bit GDDR6
- Memory Clock: 20 Gbps Effective
- Displays: Maximum 4 Displays
- Output: 2x HDMI, 2x DisplayPort
- Resolution:
- HDMI: 7680×4320
- DisplayPort 2.1: 7680×4320
- Interface: PCI-Express 4.0 x16
Pricing
Manufacturer Description
“At the core of the NITRO+ Graphics Cards series are the technologies of the SAPPHIRE PANTHEON Suite of permanent features. SAPPHIRE PANTHEON guarantees supreme cooling technology and other distinct features exclusive to NITRO+ graphics cards. Each NITRO+ iconic feature stands as strong and permanent as the pillars of the PANTHEON. Every NITRO+ Graphics Card guarantees the presence of the stable, reliable and innovative SAPPHIRE PANTHEON suite of features.”
The Sapphire NITRO+ Card
The Sapphire NITRO+ RX 7900 XTX (please ignore the Intel CPU clamshell – it was used as a kickstand)
I mentioned that this NITRO+ card is big, right? Well, it measures some 320 mm (12.6 inches) long, 133 mm (5.2 inches) tall, and 71.6 mm (2.8 inches) thick. The cooler extends far enough beyond the 3-slot backplate that a total of 4 slots are needed for installation. I also weighed the card on my kitchen scale and it came out to 1940 grams (about 4 lbs, 4.4 oz).
I personally like the styling of this card, which is a bit more understated than some, yet embraces ARGB for those that desire it. On a more practical note, you may have noticed the three PCIe 8-pin power connectors on the top edge of the NITRO+ card, and it definitely uses all of them (more on power draw later on).
The list of features outlined by Sapphire for the NITRO+ is extensive (to put it mildly):
Sapphire Features
- Dual BIOS
- TriXX Software Switch
- Premium Digital Power Design
- Ultra High Performance Conductive Polymer Aluminum Capacitors
- Fuse Protection
- Tri-X Cooling Technology
- Vapor-X Cooling
- High TG Copper PCB
- Optimized Composite Heatpipe
- Intelligent Fan Control
- Precision Fan Control
- Metal Backplate with ARGB
- Dual ARGB Light Bar
- Two-Ball Bearing Fans
- Angular Velocity Fan Blade
- Assistive System Fan Control
- Die Casted Aluminum-Magnesium Alloy Frame
- Frontplate Heatsink for VRM
- Wave Fin Design
- V Shape Fin Design for GPU Cooling
- TriXX Supported
- Fan Check
- Fan Quick Connect
- TriXX Boost
- NITRO Glow
- External RGB LED MB Synchronization
And if Sapphire’s kitchen sink approach to features for this new card isn’t enough, the NITRO+ also includes Sapphire’s “Graphics Card Supporter” metal bracket to prevent GPU sag, along with an ARGB extension cable for doing ARGB things.
Sapphire’s newest flagship brings back their Vapor-X Cooling system, as detailed by the company:
Vapor-X Cooling is BACK
The NITRO+ AMD Radeon RX 7900 Vapor-X Series Graphics Cards are fitted out with the iconic Vapor-X Cooling. The Vapor Chamber is mounted in contact with the surface of the GPU and GDDR6 Memory. The Vapor-X cooler is fine-tuned to work with the other cooling components such as the fins, heatpipes and fans to bring the best efficiency to cool the GPU and components. Since the entire area transfers heat at the same rate, the Vapor-X module is engineered to work more efficiently than a copper heat sink at carrying away heat. Upon gaining heat, the heat source is pushed to the Vaporization Wicks to begin the heat dissipation process. Due to extreme low pressure, working fluid and pure water are easily vaporized and moved through the vacuum until it reaches the Condensing Wick which is adjacent to the cooled surface. At this point, the vapor turns back to a liquid state whereby the liquid is then absorbed by the Transportation Wick by capillary action and moved back towards the Vaporization Wick. A recycled liquid system occurs when the heat source reheats the liquid and it becomes re-vaporized by the Vaporization Wick to restart the Vapor-X Cooling process. The Optimized Composite Heatpipes work in tandem with Vapor-X Cooling, fine-tuned for the cooling design, with optimal heat flow to efficiently and evenly spread out the heat to the entire cooling module.
You may already be familiar with much of the science behind this tech after a week of Radeon XTX vapor chamber news, but as the ongoing issue is limited to the “Made by AMD” (MBA) cards, Sapphire owners have nothing to worry about.
For more info – with all sorts of appealing visuals – visit Sapphire’s product page (link).
Performance Testing
To see how much of a difference this NITRO+ card can make over the reference version, it has been tested against some of the initial RX 7900 XTX review benchmarks. The exact same system was used, with the only difference being a move to the latest public driver, rather than the pre-release press driver.
| PC Perspective GPU Test Platform | ||||||||
|---|---|---|---|---|---|---|---|---|
| Processor | AMD Ryzen 9 7950X (Stock) | |||||||
| Motherboard | MSI MEG X670E ACE BIOS v1.25 Beta AGESA ComboPI 1.0.0.3 Patch A Resizable BAR Enabled |
|||||||
| Memory | 32GB (16GBx2) G.Skill Trident Z NEO DDR5-6000 CL30 | |||||||
| Storage | SK Hynix Platinum P41 2TB NVMe SSD | |||||||
| Power Supply | be quiet! Dark Power Pro 12 1500W | |||||||
| Operating System | Windows 11 Pro, 21H2 | |||||||
| Drivers | Adrenalin 22.40.00.57 – 22.12.2 GeForce Game Ready Driver 526.72 – 527.62 |
|||||||
3DMark Benchmarks
Beginning with 3DMark, here is a look at Time Spy Extreme – a standard DX12 raster test rendered at 3840×2160:
This is impressive! Out of the box the NITRO+ scored an average of 700 points higher than the reference card in Time Spy Extreme (GPU score, average of three runs). I wasn’t expecting this big of a jump right off the bat. Let’s see if it holds up.
Next we will have a look at Port Royal, the first real-time RT test in 3DMark, which is rendered at 2560×1440:
Well, it wasn’t just Time Spy Extreme; in Port Royal the Sapphire NITRO+ is an average of 754 points ahead of the reference RX 7900 XTX.
Next we will have a look at Speed Way, the latest DirectX 12 Ultimate test in 3DMark, which is also rendered at 2560×1440:
The AMD cards obviously don’t fare as well in Speed Way, which is heavy on DXR effects, but there is a notable improvement with the NITRO+ card again in this test. Sapphire’s card scores an average of 292 points higher than the reference card here.
3440×1440 Ultrawide Benchmarks
Now we will look at performance from a few games, beginning with Metro Exodus:
In Metro Exodus the gain over the reference card isn’t huge, but it’s enough to leapfrog the RTX 4080 and put the RX 7900 XTX NITRO+ in second place overall. No one is catching NVIDIA’s RTX 4090 this generation it seems, but the NITRO+ version of the RX 7900 XTX makes a compelling case against the RTX 4080 overall (at least when you aren’t running DXR titles, such as the Enhanced Edition of this game).
Next we will look at F1 22, a DXR title that features ray traced effects – particularly at the highest settings:
An increase of 4.77 FPS in this test equates to a nearly 6% performance improvement over the reference card (the biggest increase in this review), though, again, in ray traced titles the RTX 4080 is faster than both XTX versions.
Finally we look at Cyberpunk 2077, run at its highest non-RT preset (with no display scaling):
The reference XTX was already a bit faster than the RTX 4080 at these settings, but the NITRO+ tacks on another 1 FPS on average. Somehow the 1% lows got even worse, but I am moving on to a different GPU testbed this year and will be re-testing Cyberpunk to see if it was a platform issue.
Power and Thermals
And now we come to the portion of the review where those sensitive to high power draw are advised to proceed at their own risk. Am I being dramatic? Always, but this card does draw a lot of power. In fact, it actually pulls RTX 4090 FE levels of power under load:
With power reaching as high as 495.5 watts during hardware logging with this one test (3DMark Speed Way), the NITRO+ card is obviously making full use of those triple 8-pin PCIe power connectors (slot power did not exceed 60W). And while it only briefly flirted with 500 watts, during this test the card frequently pulled between 450W and 475W.
The Radeon RX 7900 XTX reference card already pulled more power than its rated TDP, but the Sapphire NITRO+ takes power draw with this 5nm GPU to the next level. Apparently Navi 31 doesn’t scale particularly well with power. We saw gains of 1% – 6% in the few tests presented above, but these obviously come at a high cost.
As to thermals, the card was exceptionally cool (and quiet) in operation. I’m talking load temps that barely touched 60 C, with hot spot temps not too far above this. In fact, during the highly unrealistic scenario of running Furmark for 15 minutes, the highest temps recorded were 73 C hot spot, with 61 C core. Now, this was a in a cool ~17 C room, but it’s still fantastic.
And no, before you ask, I didn’t test the card in different orientations. Just vertical on the test bench. And I sent it back already. Sorry.
Final Thoughts
Sapphire has done it again. Every generation they offer a premium take on AMD’s latest flagship GPU, and this NITRO+ version of the Radeon RX 7900 XTX is an outstanding example of this. The card offers better performance than the reference version of this GPU, with exceptionally good thermals and trivial noise output.
Unfortunately for Sapphire, even pumping up power draw well into the 450+ watt area has only resulted in performance that ranges from 1% to 6% above reference (in our limited testing). On the NVIDIA side of things the GeForce RTX 4080 continues to be a very efficient option, though AMD’s XTX out-performs it when ray tracing is taken out of the mix.
And now we must talk about pricing. The only “ships/sold by” listing from a major retailer I can find right now is on Newegg, where their official price is $1199.99. Third-party sellers on Newegg and Amazon are, of course, asking for a lot more than this. But assuming you can find it for $1199, I think it’s a pretty good deal by this generation’s standards, as it trades blows with the $1199 RTX 4080.
All pricing talk assumes that the product is readily available, and at list prices. We all know this isn’t the case, and that Radeon RX 7900 XTX GPUs in general are rare. The recent vapor chamber issues with AMD-made reference cards isn’t helping the scarcity problem, but buyers of custom cards obviously won’t have to worry about that.
Bottom line, the Sapphire NITRO+ might be the ultimate version of AMD’s fastest card this generation. Yes, it pulls a lot of power, but it does it quietly, and with ridiculously good thermals. It occupies nearly four slots, and you will need to provide three 8-pin power connectors to make it work, but it’s a solid, well-engineered card that will get you the very best that Radeon has to offer this generation.
Review Disclosures
This is what we consider the responsible disclosure of our review policies and procedures.
How Product Was Obtained
The product was on loan from Sapphire for the purpose of this review.
What Happens To Product After Review
The product was returned Sapphire.
Company Involvement
Sapphire had no control over the content of the review and was not consulted prior to publication.
PC Perspective Compensation
Neither PC Perspective nor any of its staff were paid or compensated in any way by Sapphire or this review.
Advertising Disclosure
Sapphire has not purchased advertising at PC Perspective during the past twelve months.
Affiliate Links
This article contains affiliate links to online retailers. PC Perspective may receive compensation for purchases through those links.
I do like these over-engineered big boy coolers – even on lower power cards like the 4070 ti and 4080. Digital Foundry were complaining about them on a recent podcast. They obviously don’t have my PTSD born out of a 390x.
The reference XTX was already a bit faster than the RTX 4080 at these settings, but the NITRO+ tacks on another 1 FPS on average.
RDNA3 is pretty decent (it has some chiplet kinks) but AMD’s stack is just weird for this generation. ‘XTX’ signifies some sort of ultra halo product, and ‘XT’ should be the sane normal high-end part which clearly isn’t the case in relative performance with their own products and nowhere even close where they stand with the competition.
Thats not Sapphire’s fault though and this cards looks like a great design and I’m really into the aesthetic. Power is out of control which is always the case when the GPU gets pushed to far out of its envelope so thats probably not Sapphire’s fault either. I hope they carry this design and look down to lower tier cards as this is not a price category I’m willing to play in for this hobby.
The Metro Exodus and F1 results cannot possibly be that low. when the Techpowerup review for this card shows way higher FPS in 4k resolution which is close to double the pixels of a 3440×1440 monitor.
how did you get such low frame rates, pushing nearly 3.3 million less pixels?
I am not digging at you by the way, I am genuinely curious because your fps just dont add up.
thank you in advance.
as a side note, yes my name is also Sebastian 😀
That’s easy: different settings. TPU does not indicate exactly what settings were used, and I have mine on the chart. The F1 22 test was the built-in benchmark, and the most intensive test (Singapore). I’ll specify the test parameters more completely going forward. For Metro I used the benchmark tool, with DX12, the preset, and the 3440×1440 resolution selected from there. No other changes.
Thank you for the response Sebastian! It’s Interesting, and TPU does state that all games are at their maximum allowed settings at each resolution unless otherwise noted. So that 7900xtx would be grossly underperforming at 3440×1440. so the fact that at 4k Metro Exodus and F1 get far more FPS is still weird. unless there is some other kind of issue. 3440×1440 at max settings vs 4k at max settings is literally impossible for the ultrawide to be more demanding. Something seems screwy not on your part like i said, but probably on your card, or something.
No, I’m just more transparent about testing. A blanket statement about how games are tested on the setup page is not adequate. They do not show settings for the game tested. I do. There is nothing “screwy”. Get a 7900 XTX and run with the 7950X on this version of Windows with these drivers at 3440×1440 and at the settings I use and with the built-in benchmark, and you will at least be within a couple of frames of these results.
And please stop extrapolation of performance at different resolutions. No one else is testing this card exclusively at 3440×1440 Ultra. And I only use built in benchmarks for consistency, they use some run they came up with. I show every setting. You can recreate the result with the same hardware and software. I run each test three times manually, allowing for a cool down to idle temp between each run, and throwing out any result that deviates by more than 1%. Other sites are mixing game versions, driver and AGESA versions, and making various other concessions to retain the big lists they like to publish. It’s folly. Every year I throw out all testing and start over, because it becomes obsolete.
I strive for accuracy, but I don’t make posts or videos about how deep I go into testing, so it goes unnoticed. I end up ranting about things no one knows or cares about in comments like this.
The last time I made an error in testing was the Cyberpunk 2077 results – all of them – in the RTX 4090 Founders Edition review. Every result, which was new as I have moved to a new testbed system, was wrong. The game, version 1.60, was not respecting the display scaling setting, and all results were at 80% instead of 100%. My frame rates were way too high. When version 1.61 was released I trashed all of the benchmarks up to that point and started over. Reviews published since are all on version 1.61, and all results are new. Mixing game versions is easier, but different versions can have their own performance characteristics.
Driver updates can alter performance enough to invalidate existing benchmarks, as well. It never ends. Comparing what I do to sites that insist on publishing results taken over a multi-year period is missing the point. But I don’t make that point very well and, again, it would be unnoticed anyway.
After just coming across this review I’ll admit my first thought was to ask wtf kind of review is this, but then reading your replies in the comments and understanding how you bench, your testing methodology makes sense—a lot more sense than other reviewers. Driver versions, Agesa updates, and adding in previously taken results from other cards etc, can really invalidate the true results as you mention. Also I’m glad someone benched the card in UW.