AMD’s Radeon HD 5850 arrives

At this point, we know AMD’s new Radeon HD 5870 graphics processor is very fast—roughly as quick as two previous-gen GPUs put together. We know the 5870 introduces DirectX 11 support, improved anisotropic filtering algorithms, and reborn supersampled antialiasing. We know the card actually draws less power than the older Radeon HD 4890, both at idle and under load.

We shared all of those discoveries last week in our review of the Radeon HD 5870.

If you’ve read that review, you’ll also know about the 5870’s formidably long circuit board and its relatively onerous $379 price tag. Those aren’t surprising attributes for a top-of-the-line graphics card, but they definitely call for a more mainstream derivative. That’s where the Radeon HD 5850 comes in. With a $259 suggested retail price and a shorter board, the 5850 serves up a slightly diluted version of the 5870’s potent cocktail. Let’s see if it goes down any easier.

Cypress sheds a couple of branches
Lift the 5850’s cooler, and you’ll find the exact same 40-nm Cypress graphics processor that powers the 5870. AMD has made a few adjustments to go along with the 5850’s lower price tag and smaller footprint, however. First, it’s disabled two of Cypress’ 20 SIMD arrays and two of the texture units that accompany each array. Since every SIMD array includes 80 ALUs, or stream processors, and each texture unit can churn out four texels per clock, this change takes us down from 1600 SPs and 80 texels/clock on the Radeon HD 5870 to 1440 SPs and 72 texels/clock on its little brother.

AMD has also reduced clock speeds to keep the 5850 from flying too close to the sun. Where the 5870’s GPU ticks away at 850 MHz with its memory at 1200 MHz (for an effective 4.8 Gbps data rate), the 5850 runs at 725 MHz with 1000 MHz (or 4 Gbps) RAM. Both cards have the same amount and type of memory, though: 1GB of GDDR5. Considering the resolutions and quality options available in the latest PC games, 1GB seems like the bare minimum for an enthusiast graphics solution. Even the aptly named Radeon HD 4870 1GB comes with that amount by default, and it’s selling for less than $150 these days.

Viewed from the outside, the Radeon HD 5850 looks pretty much like a shorter 5870. Makes sense, right? The new card has a similar Batmobile-style cooler dressed in black and red hues, but its PCB is a more manageable 9.5″—an inch shorter than the 5870 and the same length as the old 4870. No need to take a hacksaw to your hard-drive cage with this one. You’ll still need a pair of six-pin PCI Express power connectors, though.

The reference 5850 also has the same port configuration as its big brother; a pair of vertically stacked DVI outputs share the ride with DisplayPort and HDMI 1.3a connections. These ports should allow you to use up to three displays simultaneously, just like on the 5870. As we explained last week, AMD’s Eyefinity technology presents such triple-monitor configs to the operating system as one ultra-wide display, so many existing games will happily stretch across.

What kind of competition is this newcomer facing? Technically speaking, the 5850 has no direct rivals, since it’s one of the only two DirectX 11 graphics processors on the market today (the other being the 5870). However, Nvidia offers two DX10 cards in the same neighborhood. There’s the GeForce GTX 285, which has recently dropped to $295.99 after rebate at Newegg, and the slower GeForce GTX 275, which can be nabbed for as little as $209.99. We’ll be contrasting the 5850’s performance with the faster of those two GeForces today.

Before we move on, readers unfamiliar with AMD’s new DirectX 11 graphics processor would do well to peruse our initial look at the Radeon HD 5870. That piece includes all the details and diagrams you’ll need to help wrap your head around the Cypress GPU. If you’re already well-versed in the particulars of AMD’s latest graphics processor, read on for our look at how it performs in the Radeon HD 5850.

Our testing methods
As ever, we did our best to deliver clean benchmark numbers. Tests were run at least three times, and we’ve reported the median result.

Our test systems were configured like so:

Thanks to Corsair for providing us with memory for our testing. Their quality, service, and support are easily superior to no-name DIMMs.

Our test systems were powered by PC Power & Cooling Silencer 750W power supply units. The Silencer 750W was a runaway Editor’s Choice winner in our epic 11-way power supply roundup, so it seemed like a fitting choice for our test rigs.

Unless otherwise specified, image quality settings for the graphics cards were left at the control panel defaults. Vertical refresh sync (vsync) was disabled for all tests.

We used the following versions of our test applications:

• Crysis Warhead 1.1 64-bit
• Far Cry 2 1.03
• Left 4 Dead build 3939
• Sacred 2: Fallen Angel 2.40.0 build 1551
• Tom Clancy’s HAWX 1.01
• Wolfenstein 1.1
• FRAPS 2.9.8

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.

Far Cry 2
We tested Far Cry 2 using the game’s built-in benchmarking tool, which allowed us to test the different cards at multiple resolutions in a precisely repeatable manner. We used the benchmark tool’s “Very high” quality presets with the DirectX 10 renderer and 4X multisampled antialiasing.

The Radeon HD 5850 is off to a promising start here, edging out Nvidia’s GeForce GTX 285, its most direct competitor, at all resolutions. We’re also getting our first glimpse at how Cypress performs after a little pruning. At resolutions of 1680×1050 and up, the 5850 lies almost exactly half-way between the 4890 and 5870. Product segmentation at its finest, folks.

Wolfenstein
We recorded a demo during a multiplayer game on the Hospital map and played it back using the “timeNetDemo” command. At all resolutions, the game’s quality options were at their peaks, with 4X multisampled AA and 8X anisotropic filtering enabled.

The tables turn here, with the GeForce GTX 285 scoring a small win over the 5850. Since Wolfenstein uses a modified version of the id Tech 4 engine, we can probably chalk up Nvidia’s edge to its traditionally better OpenGL implementation.

Left 4 Dead
We also used a custom-recorded timedemo with Valve’s excellent zombie shooter, Left 4 Dead. We tested with 4X multisampled AA and 16X anisotropic filtering enabled and all of the game’s quality options cranked

The Radeon HD 5850 is back in full force in Left 4 Dead, although this test is a bit like the Yankees crashing a little league game. At 2560×1600 with AA and AF cranked, even the slowest card here runs Left 4 Dead well beyond the 60 Hz refresh rate of most LCD monitors.

Tom Clancy’s HAWX
We used the built-in benchmark tool in HAWX, which seems to do a good job of putting a video card through its paces. We tested this game in DirectX 10 mode with all of the image quality options either turned on or set to “High”, along with 4X multisampled antialiasing. Since this game supports DirectX 10.1 for enhanced performance, we enabled it on the Radeons. No current GeForce GPU supports DX10.1, though, so we couldn’t use it with them.

We’re back in the major leagues here, and the Radeon HD 5850 continues to outpace its GeForce rival at higher resolutions.

In surprising twist, the Radeon HD 4890 pulls ahead of those two two cards at 1680×1050 and 1920×1200. If we were the betting sort, we’d pin the odd result on the 5850’s lower core speed (725 MHz vs. 900 MHz for our 4890), driver immaturity, or some obscure bottleneck in the new Cypress design.

Sacred 2: Fallen Angel
A little surprisingly for an RPG, this game is demanding enough to test even the fastest GPUs at its highest quality settings. And it puts all of that GPU power to good use by churning out some fantastic visuals.

We tested at 2560×1600 resolution with the game’s quality options at their “Very high” presets (typically the best possible quality setting) with 4X MSAA.

Given the way this game tends to play, we decided to test with fewer, longer sessions when capturing frame rates with FRAPS. We settled on three five-minute-long play sessions, all in the same area of the game. We then reported the median of the average and minimum frame rates from the three runs.

This game also supports Nvidia’s PhysX, with some nice GPU-accelerated add-on effects if you have a GeForce card. Processing those effects will put a strain on your GPU, and we’re already testing at some pretty strenuous settings. Still, we’ve included results for the GeForce GTX 295 in two additional configurations: with PhysX effects enabled in the card’s default multi-GPU SLI configuration, and with on-card SLI disabled, in which case the second GPU is dedicated solely to PhysX effects. It is possible to play Sacred 2 with the extra PhysX eye candy enabled on a Radeon, but in that case, the physical simulations are handled entirely on the CPU—and they’re unbearably slow, unfortunately.

The Radeon HD 5850 scores another win against the GeForce GTX 285 here, and it puts the 4890 back in its place, as well.

Crysis Warhead
Although we’ve had a bit of a tough time finding games that will really push the limits of the new Radeons, this game engine is certain to do it. In a true test of GPU power, we turned up all of the quality settings in Warhead to the highest settings using the cheesily named “Enthusiast” presets. The game looks absolutely gorgeous at these settings, but few video cards will run it smoothly. In fact, we chose to test at 1920×1200 rather than 2560×1600 because it appears at least some of the cards have serious trouble at the higher resolution, almost as if they were running out of video RAM. Anyhow, this is a pretty brutal test, tough enough to challenge even our fastest multi-GPU setups.

For this game, we tested each GPU config in five 60-second sessions, covering the same portion of the game each time. We’ve then reported the median average and minumum frame rates from those five runs.

The 5850’s average frame rate lies almost equidistant between that of the 5870 and 4890 yet again. Its minimum frame rate is only slightly higher than 4890’s, though. A 23 FPS average isn’t really playable to begin with, of course, but this test gives us a rare glimpse at how these GPUs scale when pushed to their limits by a DirectX 10 game.

Power consumption
We measured total system power consumption at the wall socket using an Extech power analyzer model 380803. 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 Left 4 Dead at a 2560×1600 resolution, using the same settings we did for performance testing.

Even the Radeon HD 5870 consumes less power than the GeForce GTX 285, so there’s really no contest between the Nvidia card and the new 5850. AMD’s DirectX 11 offerings both have about the same idle power draw—a testament to Cypress’ power efficiency—but the 5850 draws a good 36W less than its big brother under load.

Noise levels
We measured noise levels on our test system, sitting on an open test bench, using an Extech model 407738 digital sound level meter. The meter was mounted on a tripod approximately 8″ from the test system at a height even with the top of the video card. We used the OSHA-standard weighting and speed for these measurements.

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.

The 5850’s lower power consumption pays off, allowing for much lower noise levels under load. The Radeon HD 5850 is nevertheless a tad louder than the 5870 at idle, although it still does better than the GTX 285 across the board.

GPU temperatures
For most of the cards, we used GPU-Z to log temperatures during our load testing. In the case of multi-GPU setups, we recorded temperatures on the primary card. However, GPU-Z didn’t yet know what to do with the 5870, so we had to resort to running a 3D program in a window while reading the temperature from the Overdrive section of AMD’s Catalyst control panel.

The Radeon HD 5850 runs considerably cooler than either the 5870 or the GeForce GTX 285—an unexpectedly strong showing, seeing as the 5850 has a smaller cooler than the 5870 and lower noise levels under load. What a refreshing change from the blistering-hot temperatures of previous high-end cards.