Full Review of AMD Radeon HD 7790 Graphics Card

This article is a review of the AMD Radeon HD 7790 graphics card. It includes graphs, images, information, data, costs, and game testing.

This post is sponsored by DVwarehouse where you can buy refurbished/used computer products for fantastic prices. See the great prices for refurbished apple desktop computers on their website. Buying DVWarehouse computer products helps support us at Techreport!

Introduction

Some of us were expecting AMD to unleash next-generation Radeons this spring, but it was not to be. We learned last month that the company intends to keep its Radeon HD 7000 series around through much of 2013. A completely new product series is in the works, but it’s not due out until very late this year—likely just before Christmas.

As we hung our heads listening to the news, we learned that AMD’s plans didn’t preclude new releases long before the holidays. In fact, we were told that the Radeon HD 7000 series would soon be expanded with fresh cards featuring new silicon. We’d soon have some previously unseen hardware to sink our teeth into.

A New GPU

True to its word, AMD has now introduced the Radeon HD 7790, a $149 graphics card powered by a new GPU called Bonaire. This addition offers an interesting middle ground between the Radeon HD 7770 and the Radeon HD 7850, not to mention a potentially compelling alternative to Nvidia’s GeForce GTX 650 Ti.

The new Bonaire GPU

Buckle up, folks, because AMD’s code names get a little bumpy here. Bonaire is officially part of the Sea Islands product family. Sea Islands no longer implies a next-gen graphics architecture as it once did, however; in AMD’s words, the name now encompasses “all products we’re producing in 2013.” Bonaire, despite being a completely new ASIC, is actually based on the exact same Graphics Core Next graphics architecture as the Radeon HD 7000 series (which was itself code-named Southern Islands).

Bonaire also happens to be the name of an island in the planet’s northern hemisphere. And Northern Islands code name refers to the Radeon HD 6000 series. But I digress.

All this code-name mumbo jumbo aside, Bonaire is an exciting addition to AMD’s GPU lineup. While it features the same 128-bit memory interface and ROP arrangement as Cape Verde, the chip that powers the Radeon HD 7770, it has four more compute units and one additional geometry engine. That means the ALU count has gone up from 640 to 896, the number of textures filtered per clock has increased from 40 to 56, and the number of triangles rasterized per clock cycle has risen from one to two.

	ROP pixels/ clock	Texels filtered/ clock (int/fp16)	Shader ALUs	Rasterized triangles/ clock	Memory interface width (bits)	Estimated transistor count (Millions)	Die size (mm²)	Fabrication process node
Cape Verde	16	40/20	640	1	128	1500	123	28 nm
Bonaire	16	56/28	896	2	128	2080	160	28 nm
Pitcairn	32	80/40	1280	2	256	2800	212	28 nm
GF114	32	64/64	384	2	256	1950	360	40 nm
GK104	32	128/128	1536	4	256	3500	294	28 nm
GK106	24	80/80	960	3	192	2540	214	28 nm

Translation: Bonaire is rigged to offer higher floating-point math performance, more texturing capability, and better tessellation performance than Cape Verde. Also, as you’ll see on the next page, AMD equips Bonaire with substantially faster GDDR5 RAM, which gives it a bandwidth advantage despite its identical memory controller setup.

In addition to the different unit mix, Bonaire has learned a trick from Trinity and Richland, AMD’s mainstream APUs. That trick takes the form of a new Dynamic Power Management (DPM) microcontroller, which enables Bonaire to switch between voltage levels much quicker than Cape Verde or other members of the Southern Islands family. Behold:

The diagram above shows the different DPM states available to Bonaire. You can click the buttons under the image to switch between the first diagram, which shows Bonaire’s capabilities, and the second diagram, which shows the states available to a “Boost”-equipped version of Tahiti, as found in the Radeon HD 7970 GHz Edtion.

In Tahiti, there are four discrete DPM states, each with its own voltage and clock speed. The GPU can switch between clock speeds very rapidly—in as little as 5-10 ms—but voltage changes require “several hundred milliseconds.” In order to stay within its power and thermal limits at the High and Boost states, the chip attempts to reduce its clock speed without lowering the voltage level. AMD call these reductions “inferred” states. They enable the GPU to respond quickly to load increases in order to prevent power consumption from going over the limit. If lowering the clock speed isn’t enough, then the chip falls back to a lower discrete state, which involves a voltage cut—and therefore takes longer than a simple clock-speed adjustment.

That’s not a bad approach. However, it means the GPU may often find itself with more voltage than it needs to operate at a given clock speed. As a result, power consumption may be higher than it should be, while the core clock speed (and thus performance) might be lower than it needs to be.

How does Bonaire Improve the Balance

How does Bonaire improve on this formula? Well, it has a total of eight discrete DPM states, each with a different clock speed and voltage. Bonaire can switch between those states as quickly as every 10 milliseconds, which removes the need for the “inferred” states seen in Tahiti—that is, clock speed reductions without corresponding voltage cuts. This means the GPU can very quickly select the optimal clock speed and voltage combination to offer the best performance at the predefined power envelope.

Although it lacks support for the Boost power state, the Cape Verde chip in the Radeon HD 7770 otherwise behaves much like Tahiti, whose DNA it shares. Thus, the additional power states in Bonaire give the Radeon HD 7790 an advantage in power efficiency over the 7770.

Here are the cards:

The Radeon HD 7790 is slated to be available for purchase on April 2 at a suggested e-tail price of $149. At “participating retailers,” the card will be sold with a free copy of BioShock Infinite. Since the Radeon HD 7750 and 7770 were left out of the Never Settle Reloaded bundle, that’s good to know.

AMD sent us a list of some of the Radeon HD 7790 variants its partners have in store. Here it is:

Officially, the AMD Radeon HD 7790 is meant to run at an even 1GHz with 6Gbps memory. As you can tell from the list above, however, retail cards with above-reference clock speeds will be commonplace—possibly more so than standard designs. When briefing us about the 7790, AMD suggested that it gave partners an exceptional amount of leeway in designing their cards. The chipmaker also said we’ll see versions of the 7790 with 2GB of GDDR5 memory, up from the default 1GB. None of those 2GB models are in the list above, though, and we weren’t given a timetable for their arrival.

AMD eschewed sending us a reference board for our review. Instead, the company sent us Asus’ Radeon HD 7790 DirectCU II, which is one of the nicer offerings coming in April. Asus expects to sell it for around $155, but exact pricing hasn’t yet been set.

That big, heatpipe-laden dual-slot cooler almost dwarfs the stubby circuit board, which measures only 6.8″ in length. Cooler included, the card is about 8.5″ long. The DisplayPort, HDMI, and dual DVI outputs can drive up to six displays (provided you use a DisplayPort hub), and the card takes power from a single six-pin PCI Express connector. (Note that the PCIe connector is rotated, so that the clip faces the back of the circuit board. If it were in the usual position, the heatsink fins would be in the way.)

Unfortunately, we didn’t get the DirectCU II until Tuesday, which gave us too little time to benchmark it. By then, we’d already started testing Sapphire’s Radeon HD 7790, which FedEx delivered the day before.

The Sapphire card has the same 1075MHz core speed and 6.4Gbps memory speed as the Asus. It features a similar dual-fan cooler, albeit without conspicuous heat pipes, and it has the same 8.5″ overall length and display output arrangement. The circuit board spans the whole length of the cooler, however, and the PCI Express power connector sits at the top of the card with the clip facing the front—a pretty common arrangement.

A farewell to the Radeon HD 7850 1GB

AMD’s Radeon HD 7850 1GB came out in October. Since its launch, the card has wooed value-conscious gamers by delivering much of the performance of its 2GB namesake at a lower price—often as little as $160. We’ve recommend it in several of our system guides.

Now, sadly, the 7850 1GB is about to disappear from retail listings forever. The Radeon HD 7790 will be its de facto successor.

According to AMD, the 7850 1GB is going away because memory makers have stopped producing the 128MB GDDR5 chips it requires. The card has four 64-bit dual-channel memory controllers that must each be fed by two memory chips; it therefore needs eight 128MB chips to achieve a 1GB capacity. The 7790 doesn’t have that problem. With only two 64-bit memory controllers, it can deliver the same 1GB capacity using larger, 256MB GDDR5 chips, which are still being made.

This disappearing act gives the 7790 some pretty big shoes to fill. GPUs with 128-bit memory interfaces don’t often match the performance of their 256-bit siblings, especially when they’re based on the same architecture. If the 7790 fails to deliver, folks could be forced to splurge for a Radeon HD 7850 2GB, which would set them back at least $180.

Comparing the AMD Radeon HD 7790 

The Radeon HD 7790 won’t just be trying to live up to the 7850 1GB’s legacy. It will also face competition from higher-clocked versions of Nvidia’s GeForce GTX 650 Ti, which are available in the same price range, as well as some of the old GeForce GTX 560 cards that remain on the market. We’ll look at real-world benchmarks very soon, but before we do, let’s take a quick look at theoretical numbers. The table below includes peak rates for both reference cards and the souped-up variants we’ve got in our labs.

	Base clock (MHz)	Boost clock (MHz)	Peak ROP rate (Gpix/s)	Texture filtering int8/fp16 (Gtex/s)	Polygon throughput (Mtris/s)	Peak shader tflops	Memory transfer rate (GT/s)	Memory bandwidth (GB/s)
Radeon HD 7770	1000	N/A	16	40/20	1000	1.3	4.5	72
Radeon HD 7790	1000	N/A	16	56/28	2000	1.8	6.0	96
Sapphire Radeon HD 7790	1075	N/A	17	60/30	2150	1.9	6.4	102
Radeon HD 7850 1GB	860	N/A	28	55/28	1720	1.8	4.8	154
GeForce GTX 650 Ti	928	N/A	15	59/59	1856	1.4	5.4	86
Zotac GeForce GTX 650 Ti 2GB AMP!	1033	N/A	17	66/66	2066	1.6	6.2	99
GeForce GTX 560	810	N/A	26	45/45	1620	1.1	4.0	128
MSI GeForce GTX 560 Twin Frozr II	870	N/A	28	49/49	1760	1.2	4.2	134

Compared to the Radeon HD 7850 1GB, the 7790 in theory has similar texture filtering and shader performance, and it should offer even higher tessellation throughput. However, the 7790 has only three fifths the ROP rate, which means less resolve power for multisampled anti-aliasing, and two thirds the memory bandwidth. Those limitations may or may not affect real-world gaming performance, depending on the nature of the graphics workload.

The 7790 is more comparable to the GTX 650 Ti. On paper, these two cards have roughly equivalent ROP rates, polygon throughput, and memory bandwidth. The 7790 enjoys an advantage in shader throughput, while the 650 Ti promises better texture filtering performance, especially for fp16 texture formats. This contest is probably too close to call at this stage.

As for the old GTX 560, that card has the same advantages as the Radeon HD 7850 1GB—higher memory bandwidth and ROP rates—but it trails the 7790 in key rates like texture filtering, shader arithmetic, and polygon rasterization. The 7790 may come out ahead more often than not in newer games, especially those that use shader-based antialiasing techniques instead of MSAA.

A quick word about our guinea pigs

We had an unusually short time window to review the Radeon HD 7790, and AMD didn’t reveal the 7790’s pricing until Tuesday evening. We did our best to estimate the card’s positioning and obtain a comparable GeForce GTX 650 Ti from Nvidia’s graphics card partners, but we were unable to get one in time.

The card you’ll see tested alongside the 7790 over the next few pages is a Zotac AMP! Edition offering, which has 2GB of onboard memory and somewhat higher clock speeds than most other GTX 650 Ti variants. It currently retails for $181 at Newegg, or about $20 more than what Sapphire expects to charge for its Radeon HD 7790 at launch.

Now, there’s nothing particularly wrong with comparing these two cards. They’re both genuine retail offerings, and the performance comparison should be enlightening. That said, we’d ask that you please keep the price difference in mind as you peruse our benchmarks. GTX 650 Ti variants priced around the $160 mark are likely to be a little slower than our sample. Also, please stay tuned. Very soon, we’ll have another article with more benchmarks that include another version of the GTX 650 Ti.

We were, however, able to get a new model of the Radeon HD 7770 GHz Edition in time for the review: Diamond’s version of the card, which is a good representative of vanilla offerings available out there. It runs at the reference 1000MHz core and 4500MT/s memory speeds, and it has a stubby dual-slot cooler with a large, quiet fan. This seems to be a stubbier version of the model selling at Newegg for $135.99 (before a $20 mail-in rebate) right now.

Game Testing the AMD Radeon HD 7790

Our testing methods

As ever, we did our best to deliver clean benchmark numbers. Tests were run at least three times, and we reported the median results. Our test systems were configured like so:

Processor	Intel Core i7-3770K
Motherboard	Gigabyte Z77X-UD3H
North bridge	Intel Z77 Express
South bridge
Memory size	4GB (2 DIMMs)
Memory type	AMD Memory DDR3 SDRAM at 1600MHz
Memory timings	9-9-9-28
Chipset drivers	INF update 9.3.0.1021 Rapid Storage Technology 11.6
Audio	Integrated Via audio with 6.0.01.10800 drivers
Hard drive	Crucial m4 256GB
Power supply	Corsair HX750W 750W
OS	Windows 8 Professional x64 Edition

Thanks to AMD, Corsair, and Crucial for helping to outfit our test rig. Asus, Diamond, MSI, Sapphire, XFX, and Zotac have our gratitude, as well, for supplying the various graphics cards we tested.

Image quality settings for the graphics cards were left at the control panel defaults, except on the Radeon cards, where surface format optimizations were disabled and the tessellation mode was set to “use application settings.” Vertical refresh sync (vsync) was disabled for all tests.

We used the following test applications:

• Battlefield 3
• Borderlands 2
• Crysis 3
• Sleeping Dogs
• The Elder Scrolls V: Skyrim
• Tomb Raider
• LuxMark 1.0 Win64
• ShaderToyMark 0.2.0
• TessMark 0.3.0
• Fraps 3.5.9
• GPU-Z 0.6.4

Some further notes on our methods:

• We used the Fraps utility to record frame rates while playing a 90-second sequence from the game. Although capturing frame rates while playing isn’t precisely repeatable, we tried to make each run as similar as possible to all of the others. We tested each Fraps sequence five times per video card in order to counteract any variability. We’ve included frame-by-frame results from Fraps for each game, and in those plots, you’re seeing the results from a single, representative pass through the test sequence.
• We measured total system power consumption at the wall socket using a P3 Kill A Watt digital power meter. The monitor was plugged into a separate outlet, so its power draw was not part of our measurement. The cards were plugged into a motherboard on an open test bench.The idle measurements were taken at the Windows desktop with the Aero theme enabled. The cards were tested under load running Skyrim at its High quality preset.
• We measured noise levels on our test system, sitting on an open test bench, using a TES-52 digital sound level meter. The meter was held approximately 8″ from the test system at a height even with the top of the video card.You can think of these noise level measurements much like our system power consumption tests, because the entire systems’ noise levels were measured. Of course, noise levels will vary greatly in the real world along with the acoustic properties of the PC enclosure used, whether the enclosure provides adequate cooling to avoid a card’s highest fan speeds, placement of the enclosure in the room, and a whole range of other variables. These results should give a reasonably good picture of comparative fan noise, though.
• We used GPU-Z to log GPU temperatures during our load testing.

The tests and methods we employ are generally publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Synthetic testing

Yes, yes, I know you’re dying to get into game benchmarks. However, synthetic benchmarks set the stage for everything else, helping to demonstrate how well the theoretical peak throughput numbers we’ve discussed translate into delivered performance.

Texture filtering

	Peak bilinear filtering (Gtexels/s)	Peak bilinear FP16 filtering (Gtexels/s)	Memory bandwidth (GB/s)
Radeon HD 7770	40	20	72
Sapphire Radeon HD 7790	56	28	102
Radeon HD 7850 1GB	55	28	154
MSI GeForce GTX 560 Twin Frozr II	49	49	134
Zotac GeForce GTX 650 Ti 2GB AMP!	66	66	99

In the real world, memory bandwidth plays a part in texturing performance. That probably explains why the Radeon HD 7790 falls behind the 7850 1GB here.

Tessellation

	Peak rasterization rate (Mtris/s)	Memory bandwidth (GB/s)
Radeon HD 7770	1000	72
Sapphire Radeon HD 7790	2150	102
Radeon HD 7850 1GB	1720	154
MSI GeForce GTX 560 Twin Frozr II	1760	134
Zotac GeForce GTX 650 Ti 2GB AMP!	2066	99

Wow. The 7790’s two geometry engines do wonders for tessellation performance, especially when paired with a 1075MHz core clock speed, as on the Sapphire card.

Shader performance

	Peak shader arithmetic (TFLOPS)	Memory bandwidth (GB/s)
Radeon HD 7770	1.3	72
Sapphire Radeon HD 7790	1.9	102
Radeon HD 7850 1GB	1.8	154
MSI GeForce GTX 560 Twin Frozr II	1.2	134
Zotac GeForce GTX 650 Ti 2GB AMP!	1.6	99

The 7790 comes out on top here.

We’ll have to stop our theoretical explorations here, unfortunately. I normally include LuxMark, an OpenCL-accelerated ray-tracing benchmark, in this set, but it refused to run using the drivers AMD provided for the 7790. Oh well; moving on…

Tomb Raider

Developed by Crystal Dynamics, this reboot of the famous franchise features a more believable Lara Croft who, as the game progresses, sheds her fear and vulnerability to become a formidable killing machine. I tested Tomb Raider by running around a small mountain area, which is roughly 10% of the way into the single-player campaign.

This is a rather impressive-looking game that’s clearly designed to take full advantage of high-end gaming PCs. The Ultra and Ultimate detail presets were too hard on these cards, so I had to settle for the High preset and leave the game’s TressFX hair physics disabled. Testing was done at 1080p.

Frame time (ms)	FPS rate
8.3	120
16.7	60
20	50
25	40
33.3	30
50	20

Let’s preface the results below with a little primer on our testing methodology. Along with measuring average frames per second, we delve inside the second to look at frame rendering times. Studying the time taken to render each frame gives us a better sense of playability, because it highlights issues like stuttering that can occur—and be felt by the player—within the span of one second. Charting frame times shows these issues clear as day, while charting average frames per second obscures them.

To get a sense of how frame times correspond to FPS rates, check the table on the right.

We’re going to start by charting frame times over the totality of a representative run for each system. (That run is usually the middle one out of the five we ran for each card.) These plots should give us an at-a-glance impression of overall playability, warts and all. You can click the buttons below the graph to compare our protagonist to its different competitors.

Right away, it’s clear that the Radeon HD 7790 is much closer to the 7850 1GB than to the 7770, whose plot shows frequent spikes above 30 ms. However, the 7790’s plot is still a little higher than that of the 7850 and the more expensive GTX 650 Ti 2GB AMP! Edition, which suggests that it’s not quite as fast.

We can slice and dice our raw frame-time data in several ways to show different facets of the performance picture. Let’s start with something we’re all familiar with: average frames per second. Average FPS is widely used, but it has some serious limitation. Another way to summarize performance is to consider the threshold below which 99% of frames are rendered, which offers a sense of overall frame latency, excluding fringe cases. (The lower the threshold, the more fluid the game.)

The average FPS and 99th-percentile results confirm our appraisal of the frame time plots. However, the performance difference between the 7790 and its faster rivals isn’t that big, especially in the 99th-percentile metric, which gives us a better indication of seat-of-the-pants smoothness and playability than average FPS.

Now, the 99th percentile result only captures a single point along the latency curve, but we can show you that whole curve, as well. With single-GPU configs like these, the right hand-side of the graph—and especially the last 5% or so—is where you’ll want to look. That section tends to be where the best and worst solutions diverge.

Finally, we can rank the cards based on how long they spent working on frames that took longer than a certain number of milliseconds to render. Simply put, this metric is a measure of “badness.” It tells us about the scope of delays in frame delivery during the test scenario. Here, you can click the buttons below the graph to switch between different milisecond thresholds.

None of the cards spend much time beyond our most important threshold of “badness” at 50 milliseconds—that means none of them dip below the relatively slow frame production rate of 20 FPS for long. In fact, except for the Radeon HD 7770, none of our cards spend a significant amount of time working on frames that take longer than 33.3 ms to render. That should mean pretty fluid gameplay from each of them.

Crysis 3

Yep. This is the new Crysis game. There’s not much else to say, except that this title has truly spectacular graphics. To test it, I ran from weapon cache to weapon cache at the beginning of the Welcome to the Jungle level for 60 seconds per run.

I tested at 1080p using the medium detail preset with high textures and medium SMAA antialiasing.

We’re seeing a bit more variability in frame times here than we did in Tomb Raider. Variability by itself isn’t necessarily bad; it’s frame time spikes that truly impair gameplay. Except for the 7770, which struggles at these settings, the cards we tested have surprisingly similar plots.

The 7790 may not have the highest average frame rate, but its 99th-percentile frame times are lower than those of all the other cards. Given the choice, we’d pick the 7790 over the cards with higher FPS averages.

Frame times for the GeForces and the 7850 1GB start spiking around the 95th percentile, but the 7790 holds largely steady up until the 97th or 98th percentile. In other word, it stays smoother throughout a larger chunk of the run. (The 7770 shows a similar progression to the 7790, but its frame times are far higher on average. In practice, it feels very sluggish and choppy in this game.)

If the percentile line graph above didn’t make it clear, this will. The Radeon HD 7790 spends a negligible amount of time working on frames that take longer than 50 ms to render, and it also spends less time beyond 33.3 ms than the other cards. Fewer spikes, smoother gameplay.

The 16.7-ms graph doesn’t show the 7790 in as positive a light, but none of these cards are quick enough for that metric to matter very much. Even the 7850 1GB spends almost nine full seconds, or about one seventh of the run, above that threshold. (For reference, a 16.7 ms frame time works out to a 60 FPS frame rate.)

Borderlands 2

For this test, I shamelessly stole Scott’s Borderlands 2 character and aped the gameplay session he used to benchmark the Radeon HD 7950 and GeForce GTX 660 Ti. The session takes place at the start of the “Opportunity” level. As Scott noted, this section isn’t precisely repeatable, because enemies don’t always spawn in the same spots or attack in the same way. We tested five times per GPU and tried to keep to the same path through the level, however, which should help compensate for variability.

I tested at 1920×1080. All other graphics settings were maxed out except for hardware-accelerated PhysX, which isn’t supported on the Radeons.

The Radeon HD 7790 does a much better job of keeping frame times steady than the other Radeons in this game. In fact, the other Radeons don’t look like they’re benefiting from the Borderlands 2 latency optimizations AMD first rolled out in the Catalyst 13.2 beta. Perhaps the beta driver AMD sent us with the 7790 doesn’t include those optimizations for other cards, somehow, or maybe AMD’s optimizations somehow don’t apply to the 7850 and 7770. We’ve asked AMD to clarify and are awaiting a response.

In any event, the 7790 looks to be about neck-and-neck with the pricier GeForce GTX 650 Ti AMP! Edition here. Not a bad showing at all.

Sleeping Dogs

I haven’t had a chance to get very far into Sleeping Dogs myself, but TR’s Geoff Gasior did, and he got hooked. From the small glimpse I’ve received of the game’s open-world environment and martial-arts-style combat, I think I can see why.

The game’s version of Hong Kong seems to be its most demanding area from a performance standpoint, so that’s what I benchmarked. I took Wei Shen on a motorcyle joyride through the city, trying my best to remember I was supposed to ride on the left side of the street.

I benchmarked Sleeping Dogs at 1920×1080 using a tweaked version of the “High” quality preset, with vsync disabled and SSAO bumped down to “Normal.” The high-resolution texture pack was installed, too.

Again, we have a nice, smooth plot for the Radeon HD 7790, and spiky plots for the other Radeons. Hmm. Whatever AMD’s doing, the 7790 performs very well, displaying even fewer spikes than the GTX 650 Ti 2GB AMP! and the GTX 560.

Yep. The 7790 hits a home run here.

The Elder Scrolls V: Skyrim

Here, too, I borrowed Scott’s test run, which involves a walk through the moor not far from the town of Whiterun—and perilously close to a camp of Giants.

The game was run at 1920×1080 using the “Ultra” detail preset. The high-resolution texture pack was installed, as well.

The 7770 and 7850 1GB fare poorly here, too, even though AMD addressed frame latency spikes in Skyrim in recent Catalyst beta drivers. By contrast, the 7790 appears to perform better; its plot has fewer, smaller frame time spikes than its fellow Radeons’ plots. Odd.

Although it has a higher FPS average, the 7790 generally trails the GTX 650 Ti AMP! in Skyrim. It fares worse in the 99th percentile frame time, and it spends more time beyond our 50- and 33-ms thresholds.

Battlefield 3

I tested Battlefield 3 by playing through the start of the Kaffarov mission, right after the player lands. Our 90-second runs involved walking through the woods and getting into a firefight with a group of hostiles, who fired and lobbed grenades at us.

I kept things simple, using the game’s “High” detail preset at 1080p.

The 7790 shadows both the 7850 1GB and GTX 650 Ti 2GB AMP! Edition in these frame-by-frame plots.

Our average FPS and percentile results confirm our initial observation. The 7790, 7850 1GB, and GTX 650 2GB AMP! are all neck-and-neck.

The 7790 pulls ahead ever so slightly in the “time spent beyond 33.3 ms” graph, but not by much. These three top contenders are about equally playable in Battlefield 3.

Power consumption

The Radeon HD 7790 actually draws a touch more power than the 7850 1GB at idle. However, it’s substantially more power-efficient under load, where it doesn’t consume much more than our reference-clocked Radeon HD 7770.

Noise levels and GPU temperatures

Sapphire’s dual-fan cooler keeps the 7790 both quiet and very cool.

A note about our noise levels: I live on the eighth floor of a tall building, and it was unusually windy both times I tried to take noise readings for this review. I attempted to alleviate the problem by taking the lowest reading from a one-minute recording for each card at each setting, so occasional wind gusts shouldn’t have impacted the numbers substantially. These cards really are all very quiet—except for the 7850, whose cooler whines a little more than the others under load. (I also heard a faint mechanical chirping from the 7850 and the GTX 650 that wasn’t present on the 7790.)

Well, well. Despite being thrown into the ring with a more expensive GeForce GTX 650 Ti card with twice as much memory, the Radeon HD 7790 more than holds its own overall. In fact, it’s quicker on average according to our 99th-percentile plot, which we think offers the best summation of real-world performance. The 7790 is negligibly slower in the average FPS plot—but it’s still a better deal considering the lower price.

Based on these numbers, I’d expect the 7790 to perform even better compared to a lower-clocked, like-priced version of the GeForce GTX 650 Ti. We’ll have to run the numbers to be sure, but this is hardly an outlandish extrapolation to make.

The 7790 also manages to outdo the 7850 1GB overall, proving its worth as a successor to that product. Sure, as the tests on the previous pages show, the 7790 doesn’t always outmatch its predecessor. Nevertheless, the fact that it does so overall should certainly be of some comfort to those saddened by the 7850 1GB’s departure.

Add to that the Radeon HD 7790’s power efficiency, its low noise levels, and the free copy of BioShock Infinite in the box, and it looks like we have a winning recipe from AMD.