You are here

AMD Trinity: A10-5800K & A8-5600K 2nd Gen APUs

A10-5800K & A8-5600K: AMD Trinity APUs

October 11, 2012 by Lawrence Lee

Product
AMD A10-5800K FM2 Processor
AMD A8-5600K
FM2 Processor
Manufacturer
Street Price
US$130
US$110

When AMD released Llano APUs for the desktop on the new FM1 socket last summer we were concerned about its future upgradeability. As the integrated graphics took up substantial amount of space on the processor package, we felt the physical restrictions would allow for only incremental updates. It was also notable that their long-awaited Bulldozer processor architecture was right around the corner and Llano was equipped with aging K10 processing cores, similar to those found in their Athlon II line. We are sad to report that we were right. Not only were there no appreciable APU upgrades, the short-lived FM1 socket has now been completely superseded with FM2.

The bright side of this development is new hardware to play with. Trinity, the codename for their new series of APUs, skips over the first incarnation of their Bulldozer CPUs which disappointed just about everyone in the hardware community. The flagship FX-8150 excelled only in heavily multithreaded workloads and didn't provide any improvements to AMD's dismal energy efficiency compared to Intel's offerings. Instead, Trinity is loaded with new Piledriver CPU cores, the next evolution of the Bulldozer architecture. The integrated GPU has also been updated of course, moving to the less complicated architecture found in the HD 6900 series and beefing up clock speeds.


Trinity die layout.

Trinity is manufactured using the same 32 nm fabrication process as Llano and is physically very similar in terms of layout. Besides the updated hardware under the hood, the biggest change is the arrangement of the CPU cores. Like Bulldozer, Piledriver cores are incorporated using modules with two cores each with 2MB of L2 cache being shared within rather than individual cores. The GPU portion now supports Eyefinity for three displays (or a 3+1 configuration if you daisy-chain through the DisplayPort). PCI Express 3.0 is nowhere to be seen but this is understandable given that Trinity is a budget platform.

Desktop Trinity Lineup
Model
Clock

(Max Turbo)
L2

Cache
GPU
GPU

Clock
Radeon

Cores
TDP
SSBP
A10-5800K
3.8 GHz

(4.2 GHz)
4MB
HD 7660D
800 MHz
384
100W
$122
A10-5700
3.4 GHz

(4.0 GHz)
760 MHz
65W
A8-5600K
3.6 GHz

(3.9 GHz)
4MB
HD 7560D
760 MHz
256
100W
$101
A8-5500
3.2 GHz

(3.7 GHz)
65W
A6-5400K

(dual core)
3.6 GHz

(4.2 GHz)
2MB
HD 7540D
760 MHz
192
65W
$67
A4-5300*

(dual core)
3.4 GHz

(3.6 GHz)
HD 7480D
724 MHz
128
$53
Athlon X4 750K
3.4 GHz

(4.0 GHz)
4MB
N/A
100W
$81
Athlon X4 740
3.2 GHz

(3.7 GHz)
65W
$71
*A4-5300 limited to DDR3-1600 and lacks dual graphics support.

The chips we're testing today, the A10-5800K and A8-5600K, are the fastest quad core models available at release with base speeds of 3.8 GHz and 3.6 GHz respectively. Unlike Llano, all members of the Trinity family are equipped with the latest version of Turbo Core, AMD's dynamic overclocking feature, so a few hundred extra MHz are also on the table for single-threaded applications. The 5800K and 5600K have TDPs of 100W and unlocked multipliers for easier overclocking. Their vanilla counterparts, the 5700 and 5600 have lower clock speeds and a less advanced GPU allowing them to operate within a 65W power envelope. It's also notable that Dual Graphics, the ability to pair the integrated GPU with an HD 6570 or 6670 in a kind of pseudo-CrossFire, is available on all the new APUs except for the entry-level A4-5300.


The AMD A10-5800K and A8-5600K.

The SSBP (suggested system builder prices) is US$122 for the 5800K and US$101 for the 5600K, though they're currently going for US$130 and US$110 respectively on Newegg. For those on a stricter budget, dual core options are available in the sub US$100 range. For purists, there are even a pair of Athlons which lack integrated graphics altogether.

Chipset & Motherboard


A85 chipset block diagram.

The new flagship chipset for FM2 is called A85 but it sports a very similar design to A75 for FM1. A75 incidentally, is also available for FM2, as is the barebones A55 chipset. The only notable difference in A85 is the inclusion of two extra SATA 6 Gbps ports.


The F2A85-M Pro.

Our test motherboard is the ASUS F2A85-M Pro, a feature-rich microATX model with a pair of PCI-E 16x slots (8x/8x in CrossFireX), seven SATA 6 Gbps ports (the eighth supported port is routed to the back for eSATA) and four USB 3.0 connectors. S/PDIF, HDMI, and DisplayPort outputs are also offered. The board retails for about US$130.


Trinity APU in FM2 socket of ASUS F2A85-M Pro motherboard

The layout of the socket hasn't changed, still parked very close to the memory slots. The plastic heatsink retention mechanism has also been carried over once again so while it won't take older processors, you can at least reuse the cooler.

When considering the cost of a system, the CPU/APU is only part of the equation as the price of motherboards varies greatly from platform to platform. In the chart above, we added the current street price of the chips compared today to those of an average compatible motherboard from Newegg. These criteria were used for the motherboards: retail versions, Asus/Intel/Gigabyte/MSI branded, microATX/ATX form factor, SATA 6 Gbps and USB 3.0 controllers (outrageously priced models were omitted).

The average motherboard price turned out to be US$142 for LGA1155, US$125 for AM3+, US$87 for FM1, and US$121 for FM2 (the FM2 figure will likely drop as more boards become available).

TEST METHODOLOGY

Common CPU Test Configuration:

Common IGP Test Configuration:

AMD AM3:

AMD AM3+:

AMD FM1:

AMD FM2:

  • AMD A10-5800K
    processor - 3.8 GHz, 32nm, 100W, integrated Radeon HD 7660D graphics
  • AMD A8-5600K
    processor - 3.6 GHz, 32nm, 100W, integrated Radeon HD 7560D graphics
  • ASUS F2A85-M Pro motherboard
    - A85 chipset

Intel LGA1155:

Discrete GPUs Compared: (using our 2012 GPU test system)

Measurement and Analysis Tools

Timed Benchmark Test Details

  • NOD32: In-depth virus scan of a folder containing 32 files of varying
    size with many RAR and ZIP archives.
  • WinRAR: Archive creation with a folder containing 68 files of varying
    size (less than 50MB).
  • iTunes: Conversion of an MP3 file to AAC.
  • TMPGEnc: Encoding a XVID AVI file with VC-1.
  • HandBrake: Encoding a XVID AVI file with H.264.
  • Photoshop: Image manipulation using a variety of filters, a derivation
    of Driver Heaven's Photoshop
    Benchmark V3
    (test image resized to 4500x3499).

3D Performance Benchmarks

Video Test Suite



1080p | 24fps | ~22 mbps

H.264/MKV: A custom 1080p H.264 encoded clip inside an Matroska container.




1080p | 24fps | ~2.3 mbps

Flash 1080p: The Dark Knight Rises Official Trailer #3, a YouTube HD trailer in 1080p.

Testing Procedures

Our main test procedure is a series of both CPU (timed tests of real-world applications) and GPU-centric (gaming tests and synthetics) benchmarks. System power consumption is measured during the CPU tests (an average of the first 10~15 seconds) and in various states including idle, H.264 and Flash playback and full CPU and GPU load using Prime95/CPUBurn and FurMark.

Certain services and features like Superfetch and System Restore are disabled
to prevent them from affecting our results. Aero glass is left enabled if supported.
We also make note if energy saving features like Cool'n'Quiet and SpeedStep
do not function properly.

Estimating DC Power

The following power efficiency figures were obtained for the
Seasonic SS-400ET used in our test system:

Seasonic SS-400ET Test Results
DC Output (W)
21.2
41.6
60.2
81.9
104.7
124.1
145.2
AC Input (W)
32.0
58.0
78.0
102.0
128.0
150.0
175.0
Efficiency
66.3%
71.7%
77.1%
80.3%
81.8%
82.8%
83.0%

This data is enough to give us a very good estimate of DC demand in our
test system. We extrapolate the DC power output from the measured AC power
input based on this data. We won't go through the math; it's easy enough
to figure out for yourself if you really want to.

INTEGRATED GRAPHICS TESTING

Our first set of tests focuses on the graphics portion of the APU against other integrated graphics platforms as well as discrete graphics. All of the processors included in these comparisons were paired with DDR3-1600 memory while the A10-5800K was also tested with DDR3-1333 to see what effect slower RAM would have on the platform. Memory speed is particularly important in gaming performance.

Energy Efficiency


With Llano, AMD brought their idle power consumption down to Intel's level and Trinity continues this trend. The A10-5800K's energy draw was in line with the older A8-3850 and at idle and beat out Intel's Ivy Bridge i7-3770K by 3W. The A8-5600K was the most efficient of all, being noticeably more thrifty than the faster A10-5800K when playing video.

On more demanding loads, the efficiency of the Intel combination drew clearly
ahead, while the new APUs left the A8-3850 in the dust, particularly with a
synthetic CPU load. The addition of a synthetic GPU load resulted in a 24W and
20W increase respectively the A10-5800K and A8-5600K, compared to 26W for the
old A8-3850.

3D Performance

According to 3DMark Vantage and 3DMark11, the 7660D is substantially faster than the 7560D, which in turn is marginally better than the 6550. When equipped with slower DDR3-1333 memory, the 7660D's performance dropped off to the midway point between the 7660D and 7560D running DDR3-1600. Memory frequency is definitely an important factor to consider for a Trinity based PC if gaming is on the agenda.

When it comes to actual frame rates, it's really hit or miss depending on the game. With a challenging title like Aliens vs. Predator, none of the compared platforms could get to 30 fps even at 1366x768 with all details set to low. Lost Planet 2 is an easier test, with all of AMD's APUs passing the 40 fps mark, even at 1440x900. The 7660D naturally led the pack but interestingly, slower memory really hindered its capability, lowering its average framerate below that of the 7560D.

GPU Performance vs. Discrete Graphics

Performing well against older integrated graphics chips isn't exactly a great challenging task, so now we pit the new APUs against discrete graphics cards using a more extensive battery of tests. We also tried out the dual graphics mode featuring, pairing up the 5800K's 7660D with an HD 6570.

Note: Discrete GPUs were tested on our GPU testing platform which uses a Core i3-2100. CPU scaling shouldn't be an issue; we tried out the HD 6570 with both the i3-2100 and A10-5800K and the difference was less than one frame per second in almost all of our gaming benchmarks.

In our synthetic tests, the HD 7560D traded blows with the HD 5570, while the HD 7660D trailed the HD 6570 (equipped with fast GDDR5 memory) somewhat. The 7660D + 6570 combination produced a nice 26~43% improvement over the 6570 on its own.

Both APUs performed decently with Lost Planet 2 and Crysis at modest resolutions with medium detail with the HD 7560D and HD 7660D edging out the HD 5550 and HD 5570D respectively. The Dual Graphics configuration gave a substantial leap in framerate in both games but didn't offer any advantage to running the HD 6570 on its own.

Dual Graphics gave us a much bigger improvement in Sniper Elite V2 and Aliens vs. Predator, about 50%, giving it the edge over a GTS 450 with GDDR5 (an approximately US$100 value). On their own, the HD 7660D and HD 7560D, once again were neck and neck with the HD 5570 and HD 5550 respectively.

CPU TESTING

Our CPU testing is conducted with a discrete graphics card (a GeForce 9400 GT) to eliminate integrated graphics as a variable, most notably with regards to power consumption. It's also necessary for comparing CPUs that do not have an onboard graphics such as Bulldozer and the older Phenom and Athlons from AMD.

Performance

The last AMD CPU we reviewed, the FX-8150, was a big disappointment as it was slower in many of our tests than older products, despite being a flagship product. Thankfully the new Trinity APUs are clearly faster than the Llano-based A8-3850. Both the A10-5800K and A8-5600K delivered some impressive gains in Photoshop, WinRAR, iTunes, and TMPGEnc. In many of the tests, the two new APUs performed on par with the Phenom II X4 955/965, only with considerably lower power consumption.

Energy Efficiency

Energy efficiency is a very important and often neglected aspect to consider as it determines overall operating costs. It also affects the amount of heat produced, which in turn affects how much noise is needed to maintain adequate cooling.

Under light load, the two Trinity chips fared similarly to its older Llano brother, the A8-3850. All three APUs were on par with a Intel's Sandy Bridge offerings in this regard, a noticeable improvement over AMD's old school Phenom-based CPUs and Bulldozer.

While Trinity is quite thrifty when lightly taxed, the same cannot be said on heavy load. The A10-5800K and A8-5600K used about 30W and 15W more, respectively, than the i5-2500K when encoding video using HandBrake.

One odd thing we noticed was the A10-5800K's Prime95 power draw started off at about 158W, then slowly tapered off until it remained steady at 130W. According to CPU-Z, the CPU frequency was fluctuating the entire time, varying anywhere between 2.9 and 4.1 GHz.

For some extra context, we've determined what we call the "average power consumption" which assumes the system is used half the time for light load activities (an average of idle and H.264 playback) and the remaining half for heavy load (an average of the power consumption used running our five benchmarks). We believe this is a very common usage scenario for an average PC — they are often left on for long periods of time, doing little to no work.

In this scenario, Trinity isn't really that far behind Intel's offerings. The
A8-5600K for example, consumes just 4W more than the Core i5-2500K.

For users with heavy workloads, the total power consumed while running our benchmark suite is of pertinent interest. The total power takes into account the energy efficiency of each CPU while running our benchmark tests as well as how quickly they complete each task. This simulates the power draw of a machine that is purely for doing work and shuts down when its job is finished.

When it comes to getting things done, Intel has a massive advantage, with the Core i5-2500K using almost half the total power as many of AMD's multi-core processors. Keep in mind the 2500K is a US$200+ CPU, though the i3-2100 which is much closer in price as the new Trinity APUs, still hands them both a good beating.

Performance Analysis

We arrived at our overall performance score by giving each CPU a proportional score in each real world benchmark with each test having an equal weighting. The scale has been adjusted so that the A10-5800K is the reference point with a score of 100.

Overall, the A10-5800K and A8-5600K appears to be an incremental upgrade over the A8-3850 in terms of CPU performance, providing a Phenom II-like CPU experience while the older Llano platform is more in line with the Athlon II X4. Like the older quad core AMD chips, Trinity is really only competitive with Intel when you take into account multithreaded applications (or multitasking) which isn't heavily weighted in our tests. This allowed the lowly dual core Intel Core i3-2100 to finish ahead of them but we believe our benchmark suite is a fair representation of general use.

To determine performance per watt, we divided the overall performance score by the average power consumption calculated earlier and re-scaled with the A10-5800K as our reference. Neither APU comes close to Intel here though they're both considerable improvements over AMD's previous offerings.

Dividing the overall performance by the platform street cost (CPU plus an average priced motherboard) gives us the performance per dollar, again re-weighted with the A10-5800K at 100 points.

With the FM1 socket being superseded by FM2, the Llano platform is a terrific value at the moment if you don't mind the lack of upgradeability. The Phenom II X4's are actually a good deal cheaper than the A10-5800K currently, so they also offer some significant bang-for-your-buck. The A10-5800K lands near the bottom of the chart but the A8-5600K delivers reasonable value.

GPU-Adjusted Performance Analysis

Now we incorporate the graphics portion of the APUs into the equation, adjusting the non-APU processors for the power consumption and cost of a dedicated graphics card of similar capabilities. Performance-wise, the HD 7660D is faster than the HD 5570 but slower than the HD 6570 with GDDR5, so we'll say it's on par with a HD 6570 equipped with DDR3 (US$45). The 7560D and 6550D are closer to the HD 5550 which isn't widely sold any longer, so we'll call it US$40 and adjust the cost of the A8-5600K/3850 accordingly. For power, we'll be using the HD 5570's 9W idle draw as the reference.

-+

The added power draw of a discrete graphics card knocks down Sandy Bridge chips by about 20 points a piece but they still have a substantial edge in performance per watt.

The extra cost of a US$45 video card causes of all the CPUs to fall in our value chart, allowing the APU to rise to the top. Even still, the A10-5800K and A8-5600K hold only a marginal advantage over the older Phenom II's in this respect.

FINAL THOUGHTS

Compared to Llano, or more specifically, the A8-3850,
the high-end Trinity offerings deliver a boost in performance but more on the
CPU side of the equation. The A8-5600K and A10-5800K were about 10% and 15%
faster respectively in our CPU tests, lifting them up to the level of the Phenom
II X4 955 and 965. In a typical usage scenario, the 5600K also had a modest
edge in power consumption of about 6% over the A8-3850 while the 5800K was slightly
worse, though if you factor in its overall performance, the newer APU comes
out ahead per Watt.

The 5800K's 7660D and 5600K's 7560D graphics chips are also a bit faster than
the A8-3850's 6550D, which makes them far more capable than any of the integrated
graphics in Intel CPUs. They're capable of running most games smoothly, albeit
with less than full eye candy and at lower than 1080p, but they're budget solutions
comparable to sub-US$50 discrete graphics cards. If you're on a strict
budget and desire a general purpose PC with some gaming capability, Trinity
is definitely the way to go rather than similarly priced dual core Intel machines
with their much weaker integrated HD graphics. It's also a splendid choice for
a home theater PC, though the quad core models are overkill unless video encoding/transcoding
is on the menu; the cheaper dual core A4 and A6 should suffice for most. We
hope to examine the 65W TDP chips in the near future, as they are more apropos
for HTPC applications.

If you prefer to use discrete graphics, Trinity presents a bit of a quandary.
For these users, purchasing an APU that allocates half of its hardware to graphics
may be a bit of a waste unless you're thinking of running a Dual Graphics configuration
with an HD
6570/6670
. That leaves the Athlon X4 750K and 740 as your best value
for money, but neither offers the speed of the flagship 5800K; the 750K's unlocked
multiplier can help make up for the lower clock speeds through overclocking
though. Still, the processing power of the Piledriver CPU core is pretty competitive
with Phenom II X4 discrete chips, so a graduated upgrade path with the unlocked
5800K and 5600K to a mid-range or higher-end graphics card is not an unreasonable
plan for Trinity buyers with gaming interests. (KitGuru explored this idea seriously
in their examination of an
A10-5800K system with an AMD HD7970 discrete graphics card
and reached positive
conclusions.)

One question is whether FM2 will offer much in the way of upgrades down the
line. For DIYers, you'll most likely pop in a CPU/APU and never take it out
again and only change peripherals through the rest of its lifetime, essentially
making it a one-and-done system. Enthusiasts are often wary of investing in
platforms that don't have a long shelf life. It doesn't sound like a PC many
enthusiasts would build for themselves but rather for a friend or family member
with general needs and a smaller budget. In any case, competitive pricing of
both the APUs and motherboards will definitely be a factor in how much penetration
Trinity achieves among DIYers.

For the value and office market, Trinity's much improved efficiency at idle
and low power modes combined with its improved overall performance makes it
competitive with Intel offerings. It's not hard to imagine lots of Trinity-based
value PCs being released by smaller system integrators, and by AMD's big brand
partners in the very near future. Let's face it, we all know that the big numbers
for PCs is in the value market well under $1,000. For DIYers, it works well
for HTPC and as a general purpose PC for less serious gamers. Trinity is a significant
evolutionary improvement on the Llano platform, and launch pricing is definitely
more attractive than FM1. AMD has raised the bar a rung or two for what
can be expected from an APU.

Our thanks to AMD
and ASUS for the A10-5800K,
A8-5600K, and F2A85-M Pro samples used in this review.

* * *

Articles of Related Interest

Intel Core i7-3770 Ivy Bridge CPU

Intel Sandy Bridge Extreme: Core i7-3960X LGA2011 Processor

AMD FX-8150 8-Core Bulldozer Processor

AMD A8-3850 Quad Core Desktop APU (updated July 10)

Intel Core i3-2100T & Core i5-2400S Low Power CPUs

Core i5-2400, i5-2500K and i7-2600K CPUs


* * *

Discuss
this article in the SPCR forums.

Google

www SPCR