Sapphire ITX Compact R9 380 Review


Sapphire’s ITX Compact R9 380 is designed to be used in small form factor cases. Many enthusiasts are looking to compact gaming rigs for living room PCs, and portable LAN boxes will only increase in demand as esports grow. Does Sapphire's ITX-oriented card have the performance to meet those needs?

We tend to review a lot of high-end graphics cards. You know, the type that everyone wants, but often can't justify purchasing. AMD’s Radeon R9 380 is a more affordable option for gamers who don’t need a Fury X or Titan X. It's positioned as a solution for smooth frame rates at 1920x1080, and derived from last year’s Radeon R9 285 based on the Tonga GPU. AMD sets the processor to a base clock rate of 970MHz and complements it with either 2GB or 4GB of GDDR5 on a 256-bit memory interface. Otherwise, the technical details are largely the same as what we covered in our launch article.

The Radeon R9 380 does support technologies like FreeSync, which matches the refresh rate of compatible monitors to the frame rate output of the graphics card within a specified range, and Eyefinity. It also supports TrueAudio technology, Virtual Super Resolution, bridgeless CrossFire and LiquidVR technology, intended to help render the output for virtual reality headsets. As with AMD's other GCN-based GPUs, the R9 380 also supports DirectX 12.

The sample on our bench today is Sapphire’s implementation optimized for mini-ITX enclosures. With its relatively short PCA, single-fan cooling solution and compact dimensions, it will be interesting to see how the card handles elevated thermal loads.

Product 360

Sapphire’s ITX Compact R9 380 is a tiny card. As you can tell from the name, it's designed to drop into mini-ITX motherboards, and as such takes up a very small footprint. The whole thing, with the full length of the shroud accounted for, barely extends past the PCI Express slot. The exact dimensions are 171mm long, 116mm tall and 32mm wide, easily fitting within the bounds of two expansion slots. The irony, of course, is that mini-ITX-based platforms typically include just one slot along the bottom edge. So, a two-slot card like this one is going to hang over.

Sapphire didn't really spend a lot of time dressing up the 380's aesthetics. There’s nothing inherently wrong with its looks, but the grey/silver with dark grey contrast isn’t exactly eye-catching, particularly for a company known for adding color to its products.

Due to its size, the ITX Compact R9 380 only includes a single 85mm dual ball bearing fan. It's situated directly in the center of the card, and blows air across the fins of the heat sink. Unlike many of the graphics cards we've reviewed recently, this one does not incorporate a semi-passive mode, though it does slow down to 20% duty cycle when more aggressive cooling isn't needed.

Sapphire crams four rather large copper heat pipes into R9 380 ITX's cooler. Coming out of the contact surface is a pair of 8mm pipes, flanked by two 6mm ones. The quartet starts at the bottom and passes vertically through the aluminum heat sink fins. At the top, they spread in both directions and wrap back down through the fins on the outer edges.

Sapphire utilizes the same high-polymer aluminum 16K-hour capacitors found on the R9 390 Nitro that we recently reviewed. It also employs its black diamond choke, which it claims operates 10% cooler and 25% more efficiently than a normal choke.

Unlike many full-length cards, Sapphire’s ITX Compact R9 380 does not have a top-mounted power connector. Instead you’ll find the eight-pin connector at the back of the card. This shouldn’t be an issue, but it’s possible that the plug's position may get in the way. Then again, it's difficult to anticipate where a mini-ITX build is going to run out of room.

Along the top edge of the card you'll find a toggle button used for switching between firmwares. Sapphire includes a legacy BIOS as well as a modern UEFI. The button lights up to indicate the mode it's in. When the light is on, the legacy BIOS is active. The UEFI is digitally signed by Microsoft to support secure boot on Windows 8.1 and 10, and should speed up boot times when it's enabled. Sapphire defaults to the legacy BIOS for compatibility reasons.

The video outputs on this card are not exactly typical. Sapphire opts for the same layout as its Radeon R9 285 ITX from last year. You get a single DVI-D port, one HDMI output and two mini-DisplayPort connectors.

Sapphire thankfully includes an adapter to convert one of those mini-DP ports to a full-size interface, so you won't have to buy a special cable to attach your 4K display. A DVI to VGA adapter is also bundled in the box.

Along with those adapters, Sapphire includes a dual six-pin-to-single-eight-pin PCIe cable in case your power supply doesn’t have the requisite leads. There's also a driver installation DVD, a product registration guide and a quick-start booklet.

How We Test

Test System Components

All of the tests run on this card were performed using our reference system. Each reviewer for Tom’s Hardware has the same setup, which consists of an Intel Core i7-5830K running at 4.2GHz, paired with 16GB of Crucial Ballistix Sport DDR4 and a pair of 500GB MX200 SSDs. The board we use is an MSI X99S Xpower AC, and the system is powered by a Platinum-rated 850W power supply from be quiet!. All of the components are installed into a Lian-Li PC-T80 test bench.

Software & Drivers

By the time you read this, Windows 10 and AMD’s Catalyst 15.7.1 driver will have been out for a little while. But at the time of testing, neither was available. All of the tests on Sapphire’s R9 380 ITX were performed running Windows 8.1 with Catalyst 15.7.

The closest comparison with AMD’s R9 380 GPU is Nvidia’s GeForce GTX 960. Our Zotac GeForce GTX 960 AMP! served as our example, and we used the GeForce 353.30 WHQL driver with it.

DirectXDirectX 11
Graphics DriverSapphire Radeon R9 390: AMD Catalyst 15.7
Sapphire ITX Compact R9 380: AMD Catalyst 15.7
Zotac GeForce GTX 960 AMP!: Nvidia 353.30 WHQL driver
EVGA GeForce GTX 970 SSC: Nvidia 353.30 WHQL driver

Comparison Products


To compare this card to the next price tier, an EVGA GeForce GTX 970 SC and Sapphire R9 390 Nitro were used. Each card leveraged the same drivers our subject and its principal competition.


Sapphire's ITX Compact R9 380 was run through the same suite of benchmarks as the Fury Tri-X and R9 390 Nitro reviews. Tests were run in Battlefield 4, Far Cry 4, Tomb Raider, Metro: Last Light, Middle Earth: Shadow of Mordor and Grand Theft Auto V at FHD and QHD resolutions.

Battlefield 4Custom THG Benchmark, 100-sec Fraps, Ultra preset
Far Cry 4Version 1.9.0, Custom THG benchmark, 60-sec Fraps, Ultra preset
Grand Theft Auto VBuild 350, Online 1.26, In-game benchmark sequence #5, 110-sec Fraps, FXAA: On, MSAA: 2x, Texture Quality: Very High, Shader Quality: Very High, Shadow Quality: High, Reflection Quality: Very High, Water Quality: High, Particles Quality: Very High, Grass Quality: High, Soft Shadows: Softer, Post FX: Very High, Anisotropic Filtering: 16x
Metro Last LightBuilt-in benchmark, 145-sec Fraps, Very High preset, 16x AF, Normal motion blur
Middle-earth: Shadow of MordorBuilt-in benchmark, 40-sec Fraps, Ultra preset
Tomb RaiderVersion 1.01.748.0, Custom THG Benchmark, 40-sec Fraps, Ultimate preset


My sound measurements were taken in the same manner as the GeForce GTX 960 reviews that I’ve written this year, along with the recent R9 390 Nitro review. The test isn’t as detailed or accurate as Igor's, but I also don't have the same kind of equipment. Instead, I use a hand-held dB meter. It's only capable of detecting sound as quiet as 35 dB. Anything lower than that and the meter reads 0 dB. In the graphs, if there is audible noise that is not registered, the graph will show 34 dB to represent that there is some sound, but an unknown level. Seeing 0 dB on the graph means the card made no discernible sound.


To test power consumption using our reference platform, a bit of creative math is needed. Since Haswell-E processors don’t have integrated GPU cores, we can’t boot the system without a discrete board installed to get a baseline. We are able to estimate consumption based on the approximate power draw of the test bench, though. In our observations, we’ve found that the approximate power draw from everything other than the GPU is 120W. By deducting that from the recorded wattage reported on our in-line power meter, we can calculate the approximate draw of the GPU.

Battlefield 4

Sapphire’s ITX Compact R9 380 is designed for gaming at 1080p. In Battlefield 4, which is not a new title anymore, the 380 wasn’t quite able to sustain 60 FPS using the Ultra preset. These settings are playable, but you might want to tone back the details to get a higher frame rate.

The performance we recorded is on par with the competition, falling slightly behind Zotac’s GTX 960 AMP!, but surpassing it after our highest stable overclock.

With the performance at the bottom end of what most people would accept at 1920x1080, QHD was bound to be a bloodbath. Cards with 2GB of memory are really not suited for QHD.

To my surprise, the ITX Compact R9 380 sustained higher than 30 FPS on average, but dipped significantly lower at times. To game at 2560x1440, you’ll need to drop the settings significantly.

Far Cry 4

In Far Cry 4, Sapphire's ITX Compact R9 380 struggles. There’s something about the sequence we test that causes the frame rate dip to ~30. Even with the overclocked settings, it barely stays above that threshold. Even still, dipping this low results in a jarring experience, so we wouldn't recommend it.

Predictably, at QHD, Far Cry 4 flirts with the edge of playability. The ITX Compact R9 380 averages more than 35 FPS, but dips as low as 24 FPS.

Performance is slightly better after overclocking, but not enough to make these settings playable.

Grand Theft Auto V

Sapphire’s ITX Compact R9 380 keeps GTA at a pretty good frame rate. The lowest dip takes performance down to 35 FPS, but the average is closer to 50. The GeForce GTX 960 sustains a slightly higher average in this game.

With the resolution cranked up to 1440p, the ITX Compact R9 380 keeps its average frame rate above 30 FPS. It takes overclocking to push the card beyond the GeForce GTX 960.

Metro: Last Light

Metro: Last Light is a much better fit for the 380 at 1920x1080. Before overclocking, the card was already outpacing Zotac's GeForce GTX 960 AMP! by nearly 10 FPS, maintaining an average of nearly 75 FPS. Overclocking yielded an additional 6 FPS on average.

With the resolution cranked up to QHD, surprisingly, the ITX Compact R9 380 delivers fairly good performance. The frame rate never dropped below 30 FPS and it averaged 43. After a bit overclocking, the average jumped to almost 50 FPS.

You could get away with playing Metro at this resolution with an ITX Compact R9 380, but you’d likely want to drop the detail settings to keep the minimum frame rate higher.

Shadow of Mordor

Playing Shadow of Mordor with these settings is pretty much untenable. The Ultra preset requires more performance than the R9 380 can muster.

The card couldn’t even average 30 FPS (then again, neither did the GTX 960). This game requires significantly lower settings to make it viable.

Curiously, the performance dip when upping the resolution was negligible. It seems the biggest variable holding the card back in this game is its 2GB of GDDR5.

Tomb Raider

Although it's the oldest game in our suite, Tomb Raider still brings many cards to their knees. In my testing, I found the ITX Compact R9 380 handles 1080p well in this game, but still ends up dipping into the 30 FPS range. The average frame rate is well above 60 FPS, and an overclock pushes it up to around 70 FPS most of the time. Pull back a bit on the settings for smoother game play.

Considering how hard FHD was on the R9 380, I didn’t expect much from it at 2560x1440.

The frame rate dipped too low to be enjoyable all of the time, though the average stayed around 45 FPS.


We like to test how far graphics cards can be pushed. Overclocking can damage your hardware, so please be careful. Do not use the clock rates we achieve as an indicator of what you'll see; no overclock is guaranteed, and your results will most certainly vary from ours.

To adjust this card's settings, I used MSI Afterburner. It supports almost every card out there, and Sapphire doesn’t bundle its own branded tool.

The first thing I always do is adjust the power limit all the way up. For this card, that limit is 20% above stock. From there, I began to increase the core frequency in increments of 10MHz. After each change, stability is tested by running 3DMark FireStrike, and the results of each run are recorded. This process continues until the test either starts reporting lower results or artifacts start to appear on-screen.

Sapphire ships this card with a stock core clock of 980MHz. Impressively, though, despite its compact frame and single fan, we managed to get 1100MHz out of ours.

Once the GPU's maximum clock was established, the memory was adjusted. Bumping it up by 20MHz at a time (from 1400), the maximum stable frequency ended up being 1500MHz. The test did complete at 1510MHz, but our test results slid back the other way.

With the GPU set at 1100MHz and memory clocked at 1510MHz, we realized decent gains in 3DMark. More important, in every game other than Shadow of Mordor, the overclocked card managed to improve average frame rates by between three and almost 10. 


The measurements for noise were taken using a hand-held dB meter positioned two inches from the rear of the card. Readings are recorded while the system is at idle and after 10 minutes of Battlefield 4 game play.

Considering this card is a compact single-fan design, I was really expecting it to make more noise than it did. Under full load, it only registered 41.5 dB(A), which is practically the same level of noise as the GTX 970 SC at idle. From inside the case, the meter recorded 50 dB(A), which is certainly audible, but not over speakers.

You’ll notice the graph shows 34 dB(A) at idle for Sapphire's ITX Compact R9 380. That's because the meter did not report a dB rating, indicating it was below 35 dB(A), which is the instrument's lower threshold. The fan never stopped spinning, though, and I was able to hear it when very close to the test rig. It’s not silent, but it’s quieter than most offices.


Using GPU-Z to record the temperature over time, Battlefield 4 runs for 10 minutes on each card. The ambient temperature of the room where the tests were done was sitting at 25 degrees. As you can see from the graph, the ITX Compact R9 380 manages to stay cool for longer than the other cards tested.

The temperature spiked a few times throughout the first five minutes of the test, before gradually ramping up. Only in the seventh minute did the card reach its peak temperature of 70 degrees. While overclocked, the card gets hotter a lot faster, but it leveled off just above stock at 73 degrees.

Power Consumption

Using an in-line power meter, measurements are taken at three intervals: while the system is at idle, at the end of the 10-minute Battlefield run, and while running FurMark. These three measurements cover idle, load and torture test situations.

Sapphire’s ITX Compact R9 380 draws quite a bit of power from the wall, especially after overclocking. Compared to its direct competition (Zotac’s GeForce GTX 960), load power consumption is almost twice as high on the R9 380. It even draws more power than the significantly faster GTX 970.


Sapphire’s ITX Compact R9 380 is the first card I’ve tested based on AMD hardware in this price class. Previously, I tested four GTX 960s, which are all direct competition in both price and performance. Given Sapphire's compact form factor, I was really expecting a loud card that ran hot. I was pleasantly surprised to find that just isn’t the case.

Sapphire did a wonderful job on this card's cooling solution. The four heat pipes effectively dissipate heat, and the dual ball bearing fan proved quiet, even under load. It would be nice to have the fan shut off completely at idle. But even at its lowest setting, our audio equipment didn't pick it up.

As for performance, I was happy with the results. Some of the tests settings I chose were too much for the GPU, but in every case, you could dial those details back for a more enjoyable experience. At the beginning of this story, I asked if Sapphire’s ITX Compact R9 380 could be used in a living room gaming PC, and the answer is yes. Most enthusiasts would be happy to have it for gaming at 1080p.

I also wondered if it would be a suitable choice for gamers attending LAN parties. While I didn’t test any popular esports titles, today's top competitive games don't require top-tier graphics cards. This is more than enough rendering power to drive those games at 1080p and potentially even higher, though on-board memory will almost certainly become a limiting factor.

I was pleasantly surprised by this little card's overclocking headroom, too. Despite the tiny stature of its heat sink, Sapphire managed to cram in four copper heat pipes and they proved their worth. The GPU jumped by more than 100MHz, yet the temperature only increased by a few degrees. Performance was also noticeably improved in many tests.

Sapphire’s ITX Compact R9 380 delivered reasonable performance in most of our suite, yielding playable frame rates in all but one of our tests. Shadow of Mordor proved too much for this card to handle with the Very High preset enabled. That may be due to its 2GB of on-board memory though, since the GeForce GTX 960 suffered a similar fate. GTA V was also hampered by available memory. The settings used exceed what we had available, according to the game (though you can't trust the built-in indicator; it's not entirely accurate). The rest of our tests sometimes dipped into the 30 FPS range, so many gamers will want to scale back on detail or buy a higher-end model.

MORE: Best Graphics Cards For The Money
MORE: All Graphics Content
MORE: How To Build A PC

Kevin Carbotte is an Associate Contributing Writer for Tom's Hardware, covering Graphics.

Follow us @tomshardware, on Facebook and on Google+.

This thread is closed for comments
    Your comment
  • chaosmassive
    please do review on Nano card, it has been released almost a week ago
  • Cryio
    "[Being an AMD card] I was really expecting a loud card that ran hot. I was pleasantly surprised to find that just isn’t the case."

    The very definition of everyone's perception on the Internet of AMD's GPUs, unfortunately.

    Don't know why the author expected this, when we found out the from the beggining that the 380 is both faster and more efficient than the 960.
  • Yuka
    Great article guys. Thanks for giving a little more light to this little card.

    I'm only missing one piece of information (unless I missed it somehow): the speed graph for the card. I have an intuition that the big dips you see in the FPS charts are attributed to the card throttling back due to increased GPU usage. I'd like to confirm that.

    In any case, it seems like a pretty capable card.

  • Joe Black
    Personally I like the practical, functional look.
  • Haravikk
    Looks like a great option for my upcoming (much delayed) Mini-ITX build. Though I really wish on cards this size they'd ditch the other interfaces and just switch to mini-displayport; DVI connectors are a big space waster, and with the card's size I can't help but think that more room for the exhaust grille would be more valuable than an interface that I could easily just buy an adapter for.

    With the card supporting FreeSync it also makes a lot of sense to get a monitor that supports it anyway to get the full benefit from the card, in which case you can get one with mini display port of at least mini HDMI (not sure if FreeSync works on mini display port?)

    But yeah, it's time to start getting rid of DVI connectors that take up half a slot.
  • wtfxxxgp
    "[Being an AMD card] I was really expecting a loud card that ran hot. I was pleasantly surprised to find that just isn’t the case." The very definition of everyone's perception on the Internet of AMD's GPUs, unfortunately. Don't know why the author expected this, when we found out the from the beggining that the 380 is both faster and more efficient than the 960.

    Your quote is misrepresentative of what the author said in every way. He referred directly to the small footprint as being the reason for his initial expectation.

    That said, I don't know how anyone can say that it is more efficient than the 960. That's simply not true - it draws TWICE the power but is not even twice as fast, so efficiency being BETTER than 960? I want what you're smoking. Please.
  • wtfxxxgp
    @ TOMS - when is the GPU chart being updated - last one was JUNE guys...

    Isn't it a monthly event?
  • rgd1101
    964583 said:
    please do review on Nano card, it has been released almost a week ago

    Nano was a paper launch.
  • kcarbotte
    964583 said:
    please do review on Nano card, it has been released almost a week ago

    The Nano was only announced last week.
    Some reviewers may have cards right now, but the card is not actually out until the 11th, and reviews won't be up anywhere until that date.
  • dorsai
    I just can't see buying any 2gig card in today's gaming environment even at's not a matter of IF but When you'll want to play a game and end up disappointed in the performance a 2gig card delivers. It's just makes some much more sense to spend another 25-50 bucks and get a card with 4gig...even an older 280x with 3gig makes a lot more sense to me.
  • InvalidError
    "the biggest factor holding back performance is 2GB GDDR5"

    Not quite. When comparing a 380 vs a 390, you also are also pitting a 256bits product against a 512bits one.

    According to results here:
    There is very little difference between a R9-380 with 2GB vs 4GB: most results are within 3% with the biggest gains being just shy of 10%.

    A fairly similar conclusion gets reached here with the GTX960:
    4GB only benefits some games but the GPU lacks the processing power required to deliver playable frame rates in the cases where the extra memory does significantly improve performance, so the extra RAM is mostly pointless.
  • timetravelingtrevor
    Thank you for the review. I've been contemplating a very low power mini-itx build that was centered around using this card. This is good information for my research.
  • timetravelingtrevor
    Thank you for the review. I've been planning a low power mini-itx gaming rig around this card. This is very good information for my research.
  • Mousemonkey
    414569 said:
    "[Being an AMD card] I was really expecting a loud card that ran hot. I was pleasantly surprised to find that just isn’t the case." The very definition of everyone's perception on the Internet of AMD's GPUs, unfortunately. Don't know why the author expected this, when we found out the from the beggining that the 380 is both faster and more efficient than the 960.

    Who is this "we" you refer to? I own both a GTX960 and an R9 380 and the 380 is not "faster" than the 960 as they produce about the same numbers in benchmarks and it sure as heck is not more efficient as it uses 70w more power to produce those very similar results. You really need to come up with some evidence to back up your claims.
  • George Phillips
    The card appears to be built extremely well. A lot of heat pines and high quality components. AMD should have built 380s like this from the beginning.
  • cd000
    please do review on Nano card, it has been released almost a week ago

    Actually the launch date is Sept 10th.
  • littleleo
    "[Being an AMD card] I was really expecting a loud card that ran hot. I was pleasantly surprised to find that just isn’t the case." The very definition of everyone's perception on the Internet of AMD's GPUs, unfortunately. Don't know why the author expected this, when we found out the from the beggining that the 380 is both faster and more efficient than the 960.

    Who is we? Me, myself & I? I think when you see a small card like this you are concerned about the amount of cooling it can get so yes there is concern about the heat, and since it is a smaller card with less area for a cooling solution you'd expect the fans to run at a higher RPMs creating more noise. In this case Sapphire was able to handle both well thereby the surprising results. It had nothing to do with the internet.
  • littleleo
    @ TOMS - when is the GPU chart being updated - last one was JUNE guys... Isn't it a monthly event?

    I agree if they are doing it quarterly it is about time, perhaps they are waiting on the Fury Nano launch to they can add it.
  • kcarbotte
    414569 said:
    "[Being an AMD card] I was really expecting a loud card that ran hot. I was pleasantly surprised to find that just isn’t the case." The very definition of everyone's perception on the Internet of AMD's GPUs, unfortunately. Don't know why the author expected this, when we found out the from the beggining that the 380 is both faster and more efficient than the 960.

    I never said that, nor did I imply that.

    Don't go around intentionally misrepresenting what people say/write.
  • Tzn
    the fan looks sick, the rest of plastic part is ugly.
  • wh3resmycar
    ahh the bullcrap about an r9 380 vs a gtx 960... gtx960 wins. being faster in AVG FPS means crap when your screen tears up here and there.

    boosting to 1400mhz+ on a gtx960 will make you forget about that r9 380.
  • Sensei Gamer
    Love this card! Small, Compact, quiet, perfect for small builds. I wish the power consumption could be lowered more and that it has a 0db fan technology like other brands to maintain silent operations. Good Job Sapphire for coming up with this Mini ITX solution! AMD too! :)
  • InvalidError
    2022321 said:
    I wish the power consumption could be lowered more

    Wait for the 16nm 960/380 equivalents and TDP for the same performance level might drop by more than half. It would be nice to see single-slot mid-range GPUs make a comeback.
  • Joshv8
    Only problem with the 380 is the price. here in NZ a GTX 980 will set you back $900 NZD and the 380 at the moment is $450 NZD. Bummer.