Short Review AMD A4-5000 APU "Kabini"

The review sample is a ‘reference design’ or ‘white book’ sent to us by AMD. This is a device that will not be available to the consumer from any channel and is to push OEMs in a particular direction when it comes to incorporating these chips. 

The simple design of the 14-inch notebook is has one stand-out feature that we hope some OEM would implement in a future model – a gorgeous IPS Full HD display. Otherwise there's nothing special about the notebook's configuration: 4 GB of RAM and a 320 GB hard drive are standard for an inexpensive notebook. Neither of those parts should have much influence on our measurements.

An overview of the specifications:

• AMD A4-5000, 4 cores at 1.5 GHz (no Turbo)
  
• Radeon HD 8330, 128 GCN shaders at 500 MHz, driver 12.105.0.0
• 1x 4 GB DDR3L-1600 @ 800 MHz, single-channel
• Toshiba MQ Series 320 GB (MQ01ABD032)
• Windows 8 64 Bit

Even though the A4-5000 barely has higher clock frequencies than the A6-1450 we recently reviewed, the differences between the two in terms of performance are in fact quite substantial. The Kabini chips have been designed to have higher TDPs that facilitate this jump in performance over the Temash line.  In order not to exceed its maximum power consumption of 8 watts (without "Turbo Dock"), the A6-1450 can reach 1.4 GHz in the turbo mode when a single core is under load, but under maximum load the frequency drops to about 1.0 GHz. Thanks to the constant 1.5 GHz available to the APU, the A4-5000 jumps ahead by almost 50 percent in some benchmarks, giving it a considerable lead in the performance stakes.

With all four cores under load, the APU performs at around the same level as the Core i3-2367M and even comes close to the newer Core i3-3217U in some areas. However, the APU's single thread performance is still low. Even Pentium 987 is at least 50 percent faster per core. Although the parallelization of modern applications has greatly improved in recent times, this facet of performance shouldn't be completely ignored.

Even so, in everyday use, the laptop feels quite speedy and quick to react to input. The APU's higher performance compared to the A6-1450 or the older E2-1800 is absolutely perceptible, though it could be even higher were there an additional turbo-mode. It is really a glaring omission by AMD but understandable as they wanted to hit the power envelope just right. It wouldn’t be a surprise if the revised version of the Kabini platform could reintroduce the Turbo mode in future chips with some design optimization. The turbo mode would have improved the core performance in single threaded applications significantly. But for office and multimedia applications, including Full HD videos, the APU performance is quite good.

Three factors are responsible for the superiority of the A4-5000 over the A6-1450 when it comes to 3D performance.  At 600 MHz, the Radeon HD 8330 clocks exactly 20 percent faster than the Radeon HD 8250.  Thanks to the APU's DDR3L-1600 memory, the GPU has more memory bandwidth at its command (A6-1450: DDR3L-1066). And finally, the higher CPU performance allows the graphics unit to function with fewer constraints.

That is still not enough for the GPU to attain to the level of Intel's HD Graphics 4000, but at least it beats the HD Graphics 3000 and HD Graphics (Ivy Bridge version) in most cases. For a very inexpensive and frugal entry-level solution, that's certainly not a bad result.

Unfortunately, for the time being we can't share the results from the popular 3DMark 11 benchmark with you. The driver installed by AMD, which is only a few weeks old, seems to be suffering from a few bugs and caused this benchmark to crash repeatedly. But our estimated score for the device in the performance preset is somewhere between 550 to 600 points.

As long as you're content with low graphics settings and an average frame rate between 25 and 40 fps, some less demanding titles from the last few years are playable with the A4-5000. The lead the APU has compared to the A6-1450 is quite surprising. Especially in Counter-Strike: Global Offensive and Dirt: Showdown, we observed drastic improvements, even though the two APUs should actually behave in a more similar manner. This can be explained with the loads that the games put on the APU. If the CPU is bogged down with AI and physics computations, then there is more strain on the GPU to keep up, leading to poorer performance from slower chips.

All in all, one should never lose sight of the fact that the A4-5000 is an inexpensive, low-end product. Its meager gaming capabilities are hardly surprising and can't be criticized. But if you want to play older LAN classics like Warcraft III, then the APU is certainly up to the task.

Idle: (Measurements on Windows 8 Desktop)

• Energy-saving mode, minimum brightness, WLAN off: 5.1 W
• Balanced, maximum brightness, WLAN off: 9.5 W
• High-performance, maximum brightness, WLAN on: 10.5 W

Load: (Measurements with high-performance, maximum brightness and WLAN on)

• Cinebench R11.5 Single (1.5 GHz CPU): 13.7 W
• Cinebench R11.5 Multi (1.5 GHz CPU): 17.0 W
• Prime95 large FFTs (1.5 GHz CPU): 20.5 W
• FurMark (1.5 GHz CPU): 20.7 W
• Prime95 + FurMark (1.5 GHz CPU): 26.6 W
• 3DMark 06: 19.9 W

In idle mode, the power consumption of the review sample is essentially equivalent to that of the smaller Acer Aspire V5-122P from our A6-1450 review. The power draw under load is interesting, though. Due to the higher clock frequency of 1.5 GHz, the maximum power requirement of the CPU cores in Prime95 rises from 6.6 W (A6-1450 at 1.0 GHz) to around 10 W and for a CPU with this kind of performance numbers, the values are very good.  The quicker graphics unit doesn't seem to have much of an effect on the APU's power draw, at least not enough to be apparent in our measurements.

Looking at the chip under full load with Prime95 and FurMark, it's clear that the chip is sticks to its TDP specification of 15 W. In this "worst case scenario", the consumption increased by 16.1 W. Subtracting the loss from the power supply and voltage converter, the number specified by AMD could be exactly correct. Games generally require about 4 to 6 W less.

(Note: Due to varying measuring devices and methods, the power consumption values we determined here aren't directly comparable to the values from our notebook tests)

A few more benchmarks without commentary: