Introduction
Llano is AMD's first realization of the long awaited fusion concept. It is also the first time a truly competent GPU has ever been integrated into a processor before; intel and AMD have designed other APU processors, but none with this kind of horse power or these kind of ambitions.
It definitely needs to be stated before any of the benchmarks in the review; llano is a platform concept. It does not break any performance records compared to other processors or other discrete graphics. It is easy in a community obsessed with performance to scoff at the idea of playing games at low graphic settings, or at sub 100 frames per second. Llano is absolutely not a performance part. What makes llano so interesting and exciting is its potential in small form factor, low power systems. The first mini ITX FM1 motherboards have already been released, and manufactures like Asus have promised full featured mini ITX motherboards with VRM power phases, USB 3.0, Sata 6, and more. With a mini itx or a proprietary form factor motherboard, llano systems can find their way into all-in-one style pcs, ultra small form factor computers, and even tablets and televisions.
If you take a look at the technical specifications of AMDs highest performance part:
Core
Llano
Multi-Core
Quad-Core
Name
A8-3850
Operating Frequency
2.9GHz
L2 Cache
4MB
Manufacturing Tech
32 nm
Integrated Graphics
AMD Radeon HD 6550D
Thermal Design Power
100W
There is a lot of promise in this kind of package. The allure is a unique combination of features and size thus far not seen by any other product from AMD or intel. The llano processor offers support for 64 bit system, virtualization, and nearly all of the modern accelerations and instructions available in any CPU today. The integrated 6550D graphics support D11 features including tesselation, hardware acceleration for media and applications, as well as eyefinity support for up to 3 displays. In both the processor and graphics department llano has all the features you would expect from any modern system, all packed neatly into one CPU. This all sounds very impressive, but the question is can llano deliver the performance needed for all of today's most demanding applications and games? This is what I attempt to answer in my review. Llano is not fastest part available by a long shot either for processing or graphics, and as such all the comparisons between it and sandy bridge processors as well as AMD phenom hexacores is all but useless and unhelpful. Does llano have enough power for people to consider ditching their bulky and noisy towers for a simpler and more efficient system? Read the rest of the review and decide for yourself.
Test System and Methodology
In all tests the following system was used:
AMD A8 3850 Processor with 6550D integrated graphics
Asus FM1 A75M-Pro micro atx motherboard
2 x4gb G skill memory 1866mhz 8-10-7-27
2 x 120gb Intel 320 SSDs in RAID 0
Apex DM-387 Micro ATX Slim Case
Apex TFX power supply
Coolermaster Gemini II HSF with low profile fan
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks%20llano/CPU-Z.png
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CPU-Zmainboard.png
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CPU-Zmemory.png
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Gpu-Z.png
Methodology: Each benchmark was run one time. Any comparisons between other processors or graphics were generated by the benchmark used.
Several game benchmarks were used, but the game itself was not installed on the system. These benchmarks have been included as “synthetic benchmarks”. In the case were the game is had been installed and run as well, the in game benchmarks are posted along side my own time trials.
Each games settings were adjusted to try and achieve playable framerates. I did the best I could not to fine tune the settings to much make my results as reproducible as possible. The time trials were run with FRAPS benchmarking tool. I intentionally only tested for small periods of time and always in high intensity moments; like in a gun fight or during a battle. My results represent the performance of llano during peak activity, in something close to worse case scenario. Nominal performance is often better than described by the charts. The reason I did this is because I didn't want average fps to be inflated while the character is staring at the ceiling, or walking through a hallway on the way to the next bit of action. Performance during a intensive gameplay is what really matters.
The screenshots of each game are taken at the settings used to test them. Each game is given a rating of “playability” on a scale of 0-4:
4 -flawless gameplay
3 -some slow downs but its difficult to notice
2 -choppy enough to be distracting at times
1 -very choppy and near unplayable
0 -unplayable performance
Temperatures and Overclocking:
Overclocking on a FM1 motherboard is tricky, and in my case a complete failure. AMD has taken a page out of intel's book, and combined south bridge, PCIE, and system bus to one system clock. When you overclock a llano computer you are forced to overclock every component with it. In my testing I was able to get my motherboard to post at a system clock of 125mhz, an overclock of 25%. Unfortunately, at anything over 105mhz the internal raid controller would fail so I was unable to boot into an operating system and test these kinds of settings for stability. Someone using a single HDD or SSD as their boot drive would very likely be able to boot into their OS with these same settings. Unfortunately RAID controllers in general are extremely temperamental.
I did do some undervolting and underclocking. I was able to get the system mostly stable at 2.5ghz CPU/600mhz GPU with a vcore of 1.1v. I ended up settling on stock clocks with an undervolt. At 1.2v, 1.88v accounting for vdroop, the system ran stable for more then 8 hours with a prime95 blend test and a furmark torture test. Running either the prime95 blend test or the furmark torture test, the CPU would peak at about 43c. Running the combination of the 2 of them together my temperatures peaked at 51c. For each of these tests, the system was run at a 1.225v core, less than that at load accounting for v droop.
In general overclocking with llano is limited and difficult. It is possible for USB ports to stop working properly at very high system bus frequencies, as well as for other components, other then the CPU or the graphics, to behave abnormally or become unstable. Other reviews have shown though that llano scales very well with overclocking in both the processor at the graphics, however. Some have even shown more then a 100% scaling with increases in speed.
CPU Benchmarks – Synthetic
Windows Experience Index
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/WindowsExperienceIndex.png
Sisoft Sandra Arithmetic
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraArithmeticTest.png
From Wikipedia:
Dhrystone is a synthetic computing benchmark program developed in 1984 by Reinhold P. Weicker intended to be representative of system (integer) programming. The Dhrystone grew to become representative of general processor (CPU) performance until it was superseded by the CPU89 benchmark suite from the Standard Performance Evaluation Corporation, today known as the "SPECint" suite. The name is a pun on a different benchmark algorithm called Whetstone.
The Whetstone benchmark is a synthetic benchmark for evaluating the performance of computers.[1] It was first written in Algol 60 in 1972 at the National Physical Laboratory in the United Kingdom and derived from statistics on program behaviour gathered on the KDF9 computer, using a modified version of its Whetstone Algol 60 compiler. The program's behavior replicated that of a typical KDF9 scientific program and was designed to defeat compiler optimizations that would have adversely affected the accuracy of this model. The Whetstone Compiler was built at the Atomic Power Division of the English Electric Company in Whetstone, Leicestershire, England,[2] hence its name.
The Whetstone benchmark primarily measures the floating-point arithmetic performance. A similar benchmark for integer and string operations is the Dhrystone.
Sisoft Sandra Multimedia
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraMultimediaTest.png
From Sisoft Sandra's User Manual:
Benchmark the (W)MMX(2), SSE(2/3/4), AVX processor units. Shows how your processors handle multi-media instructions and data in comparison to other typical processors.
Such operations are used by more specialised software, e.g. image manipulation, video decoders/encoders, games.
The test involves the generation of Mandelbrot Set fractals that are used to realistically describe and generate natural objects such as mountains or clouds. By using various multi-media extensions better performance is achieved.
Sisoft Sandra Cryptography
http://i303.photobucket.com/albums/...Benchmarks llano/SisoftSandraCryptography.png
From Sisoft Sandra's User manual:
Measures the cryptography efficiency of the processor units: encryption, decryption and hashing. Shows how your processors handle cryptographic operations in comparison to other typical processors.
Sisoft Sandra Multicore Efficiency
http://i303.photobucket.com/albums/...arks llano/SisoftSandraintercorebandwidth.png
From Sisoft Sandra's User Manual:
Benchmark the multi-core efficiency of the processors. Shows how efficient the processor cores and their inter-connects are in comparison to other types to other typical processors.
The ability of the cores to process data blocks and pass them to another core for processing (producer-consumer paradigm) of different sizes and different chain sizes is measured. The efficiency of the inter-connect between cores is thus benchmarked; however, the number of cores (and processors) also counts as more data buffers can be processed simultaneously (aka "in flight").
True multi-core processors that have shared L2/L3 caches will thus perform much better than cores that have separate caches and are connected by the traditional FSB.
Sisoft Sandra Power Management and Efficiency
http://i303.photobucket.com/albums/.../SisoftSandrapowermanagementandefficiency.png
From Sisoft Sandra's User Manual:
Benchmark the power management efficiency of the processors. Shows how efficient the power management of your processors is in comparison to other typical processors.
The ability of the processors to step-down in frequency and voltage at different workloads is measured. The more a processor steps down in both frequency and voltage the better the score at the specific workload. The test stops when the workload is too great the processor even at 100% efficiency.
The ALU/FPU score is a geometric mean based on the whole range of workloads; thus the power of the processors does matter in obtaining a higher score.
The Power Efficiency score is a geometric mean based on the supported workloads only. Thus the power of the processors does not matter.
AIDA64 CPU Queen
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUQueen.png
This simple integer benchmark focuses on the branch prediction capabilities and the misprediction penalties of the CPU. It finds the solutions for the classic "Queens problem" on a 10 by 10 sized chessboard
CPU Queen test uses integer MMX, SSE2 and SSSE3 optimizations. It consumes less than 1 MB system memory and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 CPU Zlib
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUZlib.png
This integer benchmark measures combined CPU and memory subsystem performance through the public ZLib compression library
CPU ZLib test uses only the basic x86 instructions, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 CPU AES
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUAES.png
This integer benchmark measures CPU performance using AES (a.k.a. Rijndael) data encryption. It utilizes Vincent Rijmen, Antoon Bosselaers and Paulo Barreto's public domain C code in ECB mode.
CPU AES test uses only the basic x86 instructions, and it's hardware accelerated on VIA PadLock Security Engine capable VIA C3 and VIA C7 processors. The test consumes 48 MB memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 CPU Hash
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUHash.png
AIDA64 PhotoWorxx
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64PhotoWorxx.png
This integer benchmark peforms different common tasks used during digital photo processing.
CPU PhotoWorxx test uses only the basic x86 instructions, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 FPU Julia
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUJulia.png
This benchmark measures the single precision (also known as 32-bit) floating-point performance through the computation of several frames of the popular "Julia" fractal. The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x87, 3DNow!, 3DNow!+ or SSE instruction set extension.
FPU Julia test consumes less than 1 MB system memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 FPU Mandel
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUMandel.png
This benchmark measures the double precision (also known as 64-bit) floating-point performance through the computation of several frames of the popular "Mandelbrot" fractal. The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x87 or SSE2 instruction set extension.
FPU Mandel test consumes less than 1 MB system memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 FPU V8
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUV8.png
AIDA64 FPU Sinjulia
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUsinjulia.png
This benchmark measures the extended precision (also known as 80-bit) floating-point performance through the computation of a single frame of a modified "Julia" fractal. The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing trigonometric and exponential x87 instructions.
FPU SinJulia test consumes less than 1 MB system memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
SuperPi
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Superpi.png
from wikipedia:
Super PI is a computer program that calculates pi to a specified number of digits after the decimal point - up to a maximum of 32 million. It uses Gauss–Legendre algorithm and is a Windows port of the program used by Yasumasa Kanada in 1995 to compute Pi to 232 digits.
Prime95
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Prime95.png
Prime95 is a torture test/benchmark/distributed computing project all in one. Using Mersenne's Prime Search algorithm it searches for fantastically large prime numbers. It is a great multithreaded benchmark that stresses processor cache and measures floating point arithmetic performance.
Cinebench 11 CPU
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CinebenchCPU.png
From Maxon.net:
CINEBENCH is a real-world cross platform test suite that evaluates your computer's performance capabilities. CINEBENCH is based on MAXON's award-winning animation software CINEMA 4D, which is used extensively by studios and production houses worldwide for 3D content creation. MAXON software has been used in blockbuster movies such as Spider-Man, Star Wars, The Chronicles of Narnia and many more.
CINEBENCH is the perfect tool to compare CPU and graphics performance across various systems and platforms (Windows and Mac OS X).
LinX
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/LinX.png
From benchmarkHD.ru:
LinX is a CPU stability tester based on the Intel Linpack technology. The main point of Linpack is to solve systems of linear equations. It is designed as a benchmark to test the performance of a system in GFlops - billions of floating point operation per second. But it is also the most stressful CPU testing program to date and is a great tool to determine CPU stability. One and the same system of equations is solved repeatedly; if all results match each other - the CPU is stable, otherwise the instability is obvious, since the same sytem cannot produce different solutions. Main features include: support for both Intel and AMD CPUs, both 32 and 64-bit Linpack support, HyperThreading support, Real-time error checking, simple and intuitive interface.
3D Mark Vantage
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3Dmarkvantage.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
3D Mark 11
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3dmark11.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
GPU Benchmarks – Synthetic
3D Mark Vantage
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3Dmarkvantage.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
3D Mark 11
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3dmark11.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
Unique Haven
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/UniqueHavenSettings.png
http://i303.photobucket.com/albums/...hmarks llano/Unigine2011-08-0218-22-07-24.jpg
Sisoft Sandra GP (GPU/CPU/APU) Processing
http://i303.photobucket.com/albums/...s llano/SisoftSandraCPU-GPU-APUProcessing.png
Sisoft Sandra GP (GPU/CPU/APU) Bandwidth
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/SisoftSandraGPbandwidth.png
Sisoft Sandra GP (GPU/CPU/APU) Cryptography
http://i303.photobucket.com/albums/...hmarks llano/SisoftSandraGPGPUcyptography.png
Sisoft Sandra Video Rendering
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraVideoRendering.png
Sisoft Sandra Video Bandwidth
http://i303.photobucket.com/albums/...ks llano/SisoftSandraVideoMemoryBandwidth.png
Sisoft Sandra Media Transcode
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraMediaTranscode.png
Furmark
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/furmark.png
Cinebench OpenGL
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CinebenchOpenGL.png
Aliens Vs Predator Benchmarking Tool
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Capture.png
Street FighterIV Benchmarking Tool
1920x1080
max settings
no AA
http://i303.photobucket.com/albums/...tFighterIV_Benchmark2011-08-0314-19-53-85.jpg
http://i303.photobucket.com/albums/...tFighterIV_Benchmark2011-08-0314-25-24-61.jpg
Resident Evil 5 Benchmarking Tool
1920x1080
max setttings
no AA
http://i303.photobucket.com/albums/...hmarks llano/RE5DX102011-08-0313-34-13-36.jpg
variable benchmark
http://i303.photobucket.com/albums/...hmarks llano/RE5DX102011-08-0313-39-13-36.jpg
fixed benchmark
http://i303.photobucket.com/albums/...hmarks llano/RE5DX102011-08-0313-41-12-03.jpg
S.T.A.L.K.E.R. Call of Pripyat Benchmarking Tool
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Salkerbenchmark.png
HAWK 2 Benchmarking Tool
http://i303.photobucket.com/albums/...rks llano/HAWX2_DX112011-08-0314-11-07-19.jpg
http://i303.photobucket.com/albums/...rks llano/HAWX2_DX112011-08-0314-11-35-93.jpg
GPU Benchmarks – Gaming
Battle Field Bad Company 2
1680x1050
low detail
low shadows
http://i303.photobucket.com/albums/...arks llano/BFBC2Game2011-08-0213-41-39-64.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/BF2.png
Playability: 3.
Call of Duty: Black Ops
1920x1080
low detail
dynamic shadows disabled
http://i303.photobucket.com/albums/...marks llano/BlackOps2011-08-0213-47-59-63.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/COD.png
playability: 3.
Civilization V DX11
1920x1080
(default presets)
high detail
medium tesselation
medium shadows
medium water
2xmsaa
http://i303.photobucket.com/albums/...o/CivilizationV_DX112011-08-0312-42-45-97.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CivV.png
Playability: 3.5 . For me it was a 4 on playability. The framerates were low but it is really difficult to notice in a turn based strategy game.
Crysis 2
1280x720
gamer quality
http://i303.photobucket.com/albums/...hmarks llano/Crysis22011-08-0210-41-14-15.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/crysis.png
playability: 3.5
Elderscrolls: IV Oblivion
1920x1080
2xAA
4xAF
Max details and viewing distance
HDR disabled
http://i303.photobucket.com/albums/...marks llano/Oblivion2011-08-0121-53-39-24.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Oblivion.png
Playability: 3.
Dirt 3 DX11
1680x1050
medium preset
2xmsaa
http://i303.photobucket.com/albums/...rks llano/dirt3_game2011-08-0212-26-21-22.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Dirt3.png
http://i303.photobucket.com/albums/...rks llano/dirt3_game2011-08-0212-33-30-33.jpg
playability: 3.5 .
Fallout New Vegas
1920x1080
high preset
2xAA
4xAF
bloom + HDR
http://i303.photobucket.com/albums/...arks llano/FalloutNV2011-08-0216-20-58-28.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Falloutnv.png
playability: 3.9. The vegas strip in this game is really quite a framerate sink, and that is where I ran my time trial. The rest of the game is really very much smoother than this.
Grand Theft Auto IV
1280x1024
medium details
low shadows
AF x4
viewing distance 15
http://i303.photobucket.com/albums/...nchmarks llano/GTAIV2011-08-0222-36-42-24.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/GTAIV.png
playability: 2. The game didn't stutter, but the slow downs are hard to ignore.
Lost Planet 2 DX11
1600x900
low detail
low tesselation
motion blur on
http://i303.photobucket.com/albums/...hmarks llano/lp2dx112011-08-0309-16-08-91.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/LostPlanet2.png
Benchmark A
http://i303.photobucket.com/albums/...hmarks llano/lp2dx112011-08-0311-25-31-51.jpg
Benchmark B
http://i303.photobucket.com/albums/...hmarks llano/lp2dx112011-08-0310-10-46-47.jpg
playability: 4. I played the single player for about an hour and couldn't make frames drop. I am not a big l2p player, but my experience from playing the game is that benchmark B is not very representative if actual in game performance.
Mafia 2 DX11
1920x1080
medium shadows
medium geometry
1x af
no aa
ambient occlusion disabled
http://i303.photobucket.com/albums/...chmarks llano/mafia22011-08-0312-23-31-34.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/mafia2.png
http://i303.photobucket.com/albums/...chmarks llano/mafia22011-08-0313-20-51-09.jpg
playability: 2.5
Metro 2033 DX9
1280x1024
low detail
http://i303.photobucket.com/albums/...arks llano/metro20332011-08-0308-18-15-77.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/metro2033.png
playability: 3.5.
Starcraft II Wings of Liberty
medium preset
http://i303.photobucket.com/albums/...Benchmarks llano/SC22011-08-0214-07-48-89.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/sc2.png
playability: 4.
Conclusion
So all things considered, did AMD deliver on fusion? I would say just barely yes, the A8 3850 model of llano packs just enough performance. The llano processor outpaces the quadcores of the past like the Q6600 and Phenom I quads, and it keeps up with more modern CPUs like the phenom II and intel i5 processors. A more powerful intel or AMD CPU might be able to encode movies or crunch numbers nearly twice as fast, but does it really matter to most of us if our favorite tv shows take just a few hours to encode vs over night? I can't honestly remember the last time I was frustrated by my CPUs lack of performance... there are certainly those who require a lot of processing power for certain production grade applications, but for me and nearly everyone else llano is all the processor we need right now.
On the GPU side of things, the integrated 6550D was really quite impressive. In every game and application I tested the integrated graphics was able to provide an acceptable level of performance with at least low settings. This is really a first for integrated graphics in computing history. It is kind of a surreal feeling to be playing crysis smoothly on a system that doesn't have the familiar hum of powerful discrete graphics card spinning inside it. I can say with confidence that someone with a AMD A8 3850 processor has everything they need to get in on the action with any modern PC title.
To be critical though there were a few of the games, GTA IV and Mafia II in particular, that suffered noticeably from the lack of more robust hardware. Crysis was playable, and looked great too, but only at the gamer settings on a measly 1280x720 resolution. That doesn't leave a lot of wiggle room.
As I said in the introduction llano is a platform part, and as a platform it has achieved a funny level of performance. The highest performing part, the AMD A8 3850 tested here, just barely made the cut in a few instances; a lower clocked variant like the A8 3650 or a mobile llano part would likely not provide acceptable performance in some of these games. Cut the performance of the 3850 so it can fit a 65 watt, 45 watt, or 35 watt power envelope and llano “just enough” performance is not longer there. With these parts you would have an integrated GPU that is clearly vastly superior to any other integrated graphics solution out there, but it would fall just short of providing the performance you would need for an entry level gaming system. Ideally fusion should have the flexibility and muscle to fit a variety of different systems while still providing all the power you really need. With llano I don't think AMD is quite there yet. The AMD A8 3850 is a great CPU and I would highly recommend it, but I would have to consider carefully before recommending a lower performance part from the llano line up.
Fortunately this is AMDs first Fusion part. It has been rumored that AMD is having some issues on the production end with die yields, so if AMD can get that all sorted out we might see higher clocked parts and better low wattage options. AMD already has the next generation of fusion in the works; code named Trinity is promises to use processing cores based on the bulldozer architecture with an integrated 7000 series graphics. These processors are supposed to come out in 2012 and offer as much as 50% more performance. If it can do that with a corresponding increase in power efficiency the full concept of fusion may yet be realized.
Llano is AMD's first realization of the long awaited fusion concept. It is also the first time a truly competent GPU has ever been integrated into a processor before; intel and AMD have designed other APU processors, but none with this kind of horse power or these kind of ambitions.
It definitely needs to be stated before any of the benchmarks in the review; llano is a platform concept. It does not break any performance records compared to other processors or other discrete graphics. It is easy in a community obsessed with performance to scoff at the idea of playing games at low graphic settings, or at sub 100 frames per second. Llano is absolutely not a performance part. What makes llano so interesting and exciting is its potential in small form factor, low power systems. The first mini ITX FM1 motherboards have already been released, and manufactures like Asus have promised full featured mini ITX motherboards with VRM power phases, USB 3.0, Sata 6, and more. With a mini itx or a proprietary form factor motherboard, llano systems can find their way into all-in-one style pcs, ultra small form factor computers, and even tablets and televisions.
If you take a look at the technical specifications of AMDs highest performance part:
Core
Llano
Multi-Core
Quad-Core
Name
A8-3850
Operating Frequency
2.9GHz
L2 Cache
4MB
Manufacturing Tech
32 nm
Integrated Graphics
AMD Radeon HD 6550D
Thermal Design Power
100W
There is a lot of promise in this kind of package. The allure is a unique combination of features and size thus far not seen by any other product from AMD or intel. The llano processor offers support for 64 bit system, virtualization, and nearly all of the modern accelerations and instructions available in any CPU today. The integrated 6550D graphics support D11 features including tesselation, hardware acceleration for media and applications, as well as eyefinity support for up to 3 displays. In both the processor and graphics department llano has all the features you would expect from any modern system, all packed neatly into one CPU. This all sounds very impressive, but the question is can llano deliver the performance needed for all of today's most demanding applications and games? This is what I attempt to answer in my review. Llano is not fastest part available by a long shot either for processing or graphics, and as such all the comparisons between it and sandy bridge processors as well as AMD phenom hexacores is all but useless and unhelpful. Does llano have enough power for people to consider ditching their bulky and noisy towers for a simpler and more efficient system? Read the rest of the review and decide for yourself.
Test System and Methodology
In all tests the following system was used:
AMD A8 3850 Processor with 6550D integrated graphics
Asus FM1 A75M-Pro micro atx motherboard
2 x4gb G skill memory 1866mhz 8-10-7-27
2 x 120gb Intel 320 SSDs in RAID 0
Apex DM-387 Micro ATX Slim Case
Apex TFX power supply
Coolermaster Gemini II HSF with low profile fan
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks%20llano/CPU-Z.png
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CPU-Zmainboard.png
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CPU-Zmemory.png
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Gpu-Z.png
Methodology: Each benchmark was run one time. Any comparisons between other processors or graphics were generated by the benchmark used.
Several game benchmarks were used, but the game itself was not installed on the system. These benchmarks have been included as “synthetic benchmarks”. In the case were the game is had been installed and run as well, the in game benchmarks are posted along side my own time trials.
Each games settings were adjusted to try and achieve playable framerates. I did the best I could not to fine tune the settings to much make my results as reproducible as possible. The time trials were run with FRAPS benchmarking tool. I intentionally only tested for small periods of time and always in high intensity moments; like in a gun fight or during a battle. My results represent the performance of llano during peak activity, in something close to worse case scenario. Nominal performance is often better than described by the charts. The reason I did this is because I didn't want average fps to be inflated while the character is staring at the ceiling, or walking through a hallway on the way to the next bit of action. Performance during a intensive gameplay is what really matters.
The screenshots of each game are taken at the settings used to test them. Each game is given a rating of “playability” on a scale of 0-4:
4 -flawless gameplay
3 -some slow downs but its difficult to notice
2 -choppy enough to be distracting at times
1 -very choppy and near unplayable
0 -unplayable performance
Temperatures and Overclocking:
Overclocking on a FM1 motherboard is tricky, and in my case a complete failure. AMD has taken a page out of intel's book, and combined south bridge, PCIE, and system bus to one system clock. When you overclock a llano computer you are forced to overclock every component with it. In my testing I was able to get my motherboard to post at a system clock of 125mhz, an overclock of 25%. Unfortunately, at anything over 105mhz the internal raid controller would fail so I was unable to boot into an operating system and test these kinds of settings for stability. Someone using a single HDD or SSD as their boot drive would very likely be able to boot into their OS with these same settings. Unfortunately RAID controllers in general are extremely temperamental.
I did do some undervolting and underclocking. I was able to get the system mostly stable at 2.5ghz CPU/600mhz GPU with a vcore of 1.1v. I ended up settling on stock clocks with an undervolt. At 1.2v, 1.88v accounting for vdroop, the system ran stable for more then 8 hours with a prime95 blend test and a furmark torture test. Running either the prime95 blend test or the furmark torture test, the CPU would peak at about 43c. Running the combination of the 2 of them together my temperatures peaked at 51c. For each of these tests, the system was run at a 1.225v core, less than that at load accounting for v droop.
In general overclocking with llano is limited and difficult. It is possible for USB ports to stop working properly at very high system bus frequencies, as well as for other components, other then the CPU or the graphics, to behave abnormally or become unstable. Other reviews have shown though that llano scales very well with overclocking in both the processor at the graphics, however. Some have even shown more then a 100% scaling with increases in speed.
CPU Benchmarks – Synthetic
Windows Experience Index
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/WindowsExperienceIndex.png
Sisoft Sandra Arithmetic
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraArithmeticTest.png
From Wikipedia:
Dhrystone is a synthetic computing benchmark program developed in 1984 by Reinhold P. Weicker intended to be representative of system (integer) programming. The Dhrystone grew to become representative of general processor (CPU) performance until it was superseded by the CPU89 benchmark suite from the Standard Performance Evaluation Corporation, today known as the "SPECint" suite. The name is a pun on a different benchmark algorithm called Whetstone.
The Whetstone benchmark is a synthetic benchmark for evaluating the performance of computers.[1] It was first written in Algol 60 in 1972 at the National Physical Laboratory in the United Kingdom and derived from statistics on program behaviour gathered on the KDF9 computer, using a modified version of its Whetstone Algol 60 compiler. The program's behavior replicated that of a typical KDF9 scientific program and was designed to defeat compiler optimizations that would have adversely affected the accuracy of this model. The Whetstone Compiler was built at the Atomic Power Division of the English Electric Company in Whetstone, Leicestershire, England,[2] hence its name.
The Whetstone benchmark primarily measures the floating-point arithmetic performance. A similar benchmark for integer and string operations is the Dhrystone.
Sisoft Sandra Multimedia
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraMultimediaTest.png
From Sisoft Sandra's User Manual:
Benchmark the (W)MMX(2), SSE(2/3/4), AVX processor units. Shows how your processors handle multi-media instructions and data in comparison to other typical processors.
Such operations are used by more specialised software, e.g. image manipulation, video decoders/encoders, games.
The test involves the generation of Mandelbrot Set fractals that are used to realistically describe and generate natural objects such as mountains or clouds. By using various multi-media extensions better performance is achieved.
Sisoft Sandra Cryptography
http://i303.photobucket.com/albums/...Benchmarks llano/SisoftSandraCryptography.png
From Sisoft Sandra's User manual:
Measures the cryptography efficiency of the processor units: encryption, decryption and hashing. Shows how your processors handle cryptographic operations in comparison to other typical processors.
Sisoft Sandra Multicore Efficiency
http://i303.photobucket.com/albums/...arks llano/SisoftSandraintercorebandwidth.png
From Sisoft Sandra's User Manual:
Benchmark the multi-core efficiency of the processors. Shows how efficient the processor cores and their inter-connects are in comparison to other types to other typical processors.
The ability of the cores to process data blocks and pass them to another core for processing (producer-consumer paradigm) of different sizes and different chain sizes is measured. The efficiency of the inter-connect between cores is thus benchmarked; however, the number of cores (and processors) also counts as more data buffers can be processed simultaneously (aka "in flight").
True multi-core processors that have shared L2/L3 caches will thus perform much better than cores that have separate caches and are connected by the traditional FSB.
Sisoft Sandra Power Management and Efficiency
http://i303.photobucket.com/albums/.../SisoftSandrapowermanagementandefficiency.png
From Sisoft Sandra's User Manual:
Benchmark the power management efficiency of the processors. Shows how efficient the power management of your processors is in comparison to other typical processors.
The ability of the processors to step-down in frequency and voltage at different workloads is measured. The more a processor steps down in both frequency and voltage the better the score at the specific workload. The test stops when the workload is too great the processor even at 100% efficiency.
The ALU/FPU score is a geometric mean based on the whole range of workloads; thus the power of the processors does matter in obtaining a higher score.
The Power Efficiency score is a geometric mean based on the supported workloads only. Thus the power of the processors does not matter.
AIDA64 CPU Queen
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUQueen.png
This simple integer benchmark focuses on the branch prediction capabilities and the misprediction penalties of the CPU. It finds the solutions for the classic "Queens problem" on a 10 by 10 sized chessboard
CPU Queen test uses integer MMX, SSE2 and SSSE3 optimizations. It consumes less than 1 MB system memory and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 CPU Zlib
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUZlib.png
This integer benchmark measures combined CPU and memory subsystem performance through the public ZLib compression library
CPU ZLib test uses only the basic x86 instructions, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 CPU AES
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUAES.png
This integer benchmark measures CPU performance using AES (a.k.a. Rijndael) data encryption. It utilizes Vincent Rijmen, Antoon Bosselaers and Paulo Barreto's public domain C code in ECB mode.
CPU AES test uses only the basic x86 instructions, and it's hardware accelerated on VIA PadLock Security Engine capable VIA C3 and VIA C7 processors. The test consumes 48 MB memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 CPU Hash
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64CPUHash.png
AIDA64 PhotoWorxx
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64PhotoWorxx.png
This integer benchmark peforms different common tasks used during digital photo processing.
CPU PhotoWorxx test uses only the basic x86 instructions, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 FPU Julia
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUJulia.png
This benchmark measures the single precision (also known as 32-bit) floating-point performance through the computation of several frames of the popular "Julia" fractal. The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x87, 3DNow!, 3DNow!+ or SSE instruction set extension.
FPU Julia test consumes less than 1 MB system memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 FPU Mandel
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUMandel.png
This benchmark measures the double precision (also known as 64-bit) floating-point performance through the computation of several frames of the popular "Mandelbrot" fractal. The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing the appropriate x87 or SSE2 instruction set extension.
FPU Mandel test consumes less than 1 MB system memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
AIDA64 FPU V8
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUV8.png
AIDA64 FPU Sinjulia
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/AIDA64FPUsinjulia.png
This benchmark measures the extended precision (also known as 80-bit) floating-point performance through the computation of a single frame of a modified "Julia" fractal. The code behind this benchmark method is written in Assembly, and it is extremely optimized for every popular AMD and Intel processor core variants by utilizing trigonometric and exponential x87 instructions.
FPU SinJulia test consumes less than 1 MB system memory, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.
SuperPi
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Superpi.png
from wikipedia:
Super PI is a computer program that calculates pi to a specified number of digits after the decimal point - up to a maximum of 32 million. It uses Gauss–Legendre algorithm and is a Windows port of the program used by Yasumasa Kanada in 1995 to compute Pi to 232 digits.
Prime95
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Prime95.png
Prime95 is a torture test/benchmark/distributed computing project all in one. Using Mersenne's Prime Search algorithm it searches for fantastically large prime numbers. It is a great multithreaded benchmark that stresses processor cache and measures floating point arithmetic performance.
Cinebench 11 CPU
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CinebenchCPU.png
From Maxon.net:
CINEBENCH is a real-world cross platform test suite that evaluates your computer's performance capabilities. CINEBENCH is based on MAXON's award-winning animation software CINEMA 4D, which is used extensively by studios and production houses worldwide for 3D content creation. MAXON software has been used in blockbuster movies such as Spider-Man, Star Wars, The Chronicles of Narnia and many more.
CINEBENCH is the perfect tool to compare CPU and graphics performance across various systems and platforms (Windows and Mac OS X).
LinX
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/LinX.png
From benchmarkHD.ru:
LinX is a CPU stability tester based on the Intel Linpack technology. The main point of Linpack is to solve systems of linear equations. It is designed as a benchmark to test the performance of a system in GFlops - billions of floating point operation per second. But it is also the most stressful CPU testing program to date and is a great tool to determine CPU stability. One and the same system of equations is solved repeatedly; if all results match each other - the CPU is stable, otherwise the instability is obvious, since the same sytem cannot produce different solutions. Main features include: support for both Intel and AMD CPUs, both 32 and 64-bit Linpack support, HyperThreading support, Real-time error checking, simple and intuitive interface.
3D Mark Vantage
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3Dmarkvantage.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
3D Mark 11
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3dmark11.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
GPU Benchmarks – Synthetic
3D Mark Vantage
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3Dmarkvantage.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
3D Mark 11
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/3dmark11.png
3DMark uses an AI and a CPU computed physics tests to measure processor performance for gaming applications.
From futuremark.com:
The 3DMark score is the overall measure of a systems 3D gaming performance, based on comprehensive individual real-time 3D graphics tests and processor tests. Each 3DMark product produces a 3DMark score as its outcome, and the 3DMark scores are comparable within the same product.
Unique Haven
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/UniqueHavenSettings.png
http://i303.photobucket.com/albums/...hmarks llano/Unigine2011-08-0218-22-07-24.jpg
Sisoft Sandra GP (GPU/CPU/APU) Processing
http://i303.photobucket.com/albums/...s llano/SisoftSandraCPU-GPU-APUProcessing.png
Sisoft Sandra GP (GPU/CPU/APU) Bandwidth
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/SisoftSandraGPbandwidth.png
Sisoft Sandra GP (GPU/CPU/APU) Cryptography
http://i303.photobucket.com/albums/...hmarks llano/SisoftSandraGPGPUcyptography.png
Sisoft Sandra Video Rendering
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraVideoRendering.png
Sisoft Sandra Video Bandwidth
http://i303.photobucket.com/albums/...ks llano/SisoftSandraVideoMemoryBandwidth.png
Sisoft Sandra Media Transcode
http://i303.photobucket.com/albums/...nchmarks llano/SisoftSandraMediaTranscode.png
Furmark
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/furmark.png
Cinebench OpenGL
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CinebenchOpenGL.png
Aliens Vs Predator Benchmarking Tool
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Capture.png
Street FighterIV Benchmarking Tool
1920x1080
max settings
no AA
http://i303.photobucket.com/albums/...tFighterIV_Benchmark2011-08-0314-19-53-85.jpg
http://i303.photobucket.com/albums/...tFighterIV_Benchmark2011-08-0314-25-24-61.jpg
Resident Evil 5 Benchmarking Tool
1920x1080
max setttings
no AA
http://i303.photobucket.com/albums/...hmarks llano/RE5DX102011-08-0313-34-13-36.jpg
variable benchmark
http://i303.photobucket.com/albums/...hmarks llano/RE5DX102011-08-0313-39-13-36.jpg
fixed benchmark
http://i303.photobucket.com/albums/...hmarks llano/RE5DX102011-08-0313-41-12-03.jpg
S.T.A.L.K.E.R. Call of Pripyat Benchmarking Tool
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Salkerbenchmark.png
HAWK 2 Benchmarking Tool
http://i303.photobucket.com/albums/...rks llano/HAWX2_DX112011-08-0314-11-07-19.jpg
http://i303.photobucket.com/albums/...rks llano/HAWX2_DX112011-08-0314-11-35-93.jpg
GPU Benchmarks – Gaming
Battle Field Bad Company 2
1680x1050
low detail
low shadows
http://i303.photobucket.com/albums/...arks llano/BFBC2Game2011-08-0213-41-39-64.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/BF2.png
Playability: 3.
Call of Duty: Black Ops
1920x1080
low detail
dynamic shadows disabled
http://i303.photobucket.com/albums/...marks llano/BlackOps2011-08-0213-47-59-63.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/COD.png
playability: 3.
Civilization V DX11
1920x1080
(default presets)
high detail
medium tesselation
medium shadows
medium water
2xmsaa
http://i303.photobucket.com/albums/...o/CivilizationV_DX112011-08-0312-42-45-97.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/CivV.png
Playability: 3.5 . For me it was a 4 on playability. The framerates were low but it is really difficult to notice in a turn based strategy game.
Crysis 2
1280x720
gamer quality
http://i303.photobucket.com/albums/...hmarks llano/Crysis22011-08-0210-41-14-15.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/crysis.png
playability: 3.5
Elderscrolls: IV Oblivion
1920x1080
2xAA
4xAF
Max details and viewing distance
HDR disabled
http://i303.photobucket.com/albums/...marks llano/Oblivion2011-08-0121-53-39-24.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Oblivion.png
Playability: 3.
Dirt 3 DX11
1680x1050
medium preset
2xmsaa
http://i303.photobucket.com/albums/...rks llano/dirt3_game2011-08-0212-26-21-22.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Dirt3.png
http://i303.photobucket.com/albums/...rks llano/dirt3_game2011-08-0212-33-30-33.jpg
playability: 3.5 .
Fallout New Vegas
1920x1080
high preset
2xAA
4xAF
bloom + HDR
http://i303.photobucket.com/albums/...arks llano/FalloutNV2011-08-0216-20-58-28.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/Falloutnv.png
playability: 3.9. The vegas strip in this game is really quite a framerate sink, and that is where I ran my time trial. The rest of the game is really very much smoother than this.
Grand Theft Auto IV
1280x1024
medium details
low shadows
AF x4
viewing distance 15
http://i303.photobucket.com/albums/...nchmarks llano/GTAIV2011-08-0222-36-42-24.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/GTAIV.png
playability: 2. The game didn't stutter, but the slow downs are hard to ignore.
Lost Planet 2 DX11
1600x900
low detail
low tesselation
motion blur on
http://i303.photobucket.com/albums/...hmarks llano/lp2dx112011-08-0309-16-08-91.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/LostPlanet2.png
Benchmark A
http://i303.photobucket.com/albums/...hmarks llano/lp2dx112011-08-0311-25-31-51.jpg
Benchmark B
http://i303.photobucket.com/albums/...hmarks llano/lp2dx112011-08-0310-10-46-47.jpg
playability: 4. I played the single player for about an hour and couldn't make frames drop. I am not a big l2p player, but my experience from playing the game is that benchmark B is not very representative if actual in game performance.
Mafia 2 DX11
1920x1080
medium shadows
medium geometry
1x af
no aa
ambient occlusion disabled
http://i303.photobucket.com/albums/...chmarks llano/mafia22011-08-0312-23-31-34.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/mafia2.png
http://i303.photobucket.com/albums/...chmarks llano/mafia22011-08-0313-20-51-09.jpg
playability: 2.5
Metro 2033 DX9
1280x1024
low detail
http://i303.photobucket.com/albums/...arks llano/metro20332011-08-0308-18-15-77.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/metro2033.png
playability: 3.5.
Starcraft II Wings of Liberty
medium preset
http://i303.photobucket.com/albums/...Benchmarks llano/SC22011-08-0214-07-48-89.jpg
http://i303.photobucket.com/albums/nn153/mothergoose729/Benchmarks llano/sc2.png
playability: 4.
Conclusion
So all things considered, did AMD deliver on fusion? I would say just barely yes, the A8 3850 model of llano packs just enough performance. The llano processor outpaces the quadcores of the past like the Q6600 and Phenom I quads, and it keeps up with more modern CPUs like the phenom II and intel i5 processors. A more powerful intel or AMD CPU might be able to encode movies or crunch numbers nearly twice as fast, but does it really matter to most of us if our favorite tv shows take just a few hours to encode vs over night? I can't honestly remember the last time I was frustrated by my CPUs lack of performance... there are certainly those who require a lot of processing power for certain production grade applications, but for me and nearly everyone else llano is all the processor we need right now.
On the GPU side of things, the integrated 6550D was really quite impressive. In every game and application I tested the integrated graphics was able to provide an acceptable level of performance with at least low settings. This is really a first for integrated graphics in computing history. It is kind of a surreal feeling to be playing crysis smoothly on a system that doesn't have the familiar hum of powerful discrete graphics card spinning inside it. I can say with confidence that someone with a AMD A8 3850 processor has everything they need to get in on the action with any modern PC title.
To be critical though there were a few of the games, GTA IV and Mafia II in particular, that suffered noticeably from the lack of more robust hardware. Crysis was playable, and looked great too, but only at the gamer settings on a measly 1280x720 resolution. That doesn't leave a lot of wiggle room.
As I said in the introduction llano is a platform part, and as a platform it has achieved a funny level of performance. The highest performing part, the AMD A8 3850 tested here, just barely made the cut in a few instances; a lower clocked variant like the A8 3650 or a mobile llano part would likely not provide acceptable performance in some of these games. Cut the performance of the 3850 so it can fit a 65 watt, 45 watt, or 35 watt power envelope and llano “just enough” performance is not longer there. With these parts you would have an integrated GPU that is clearly vastly superior to any other integrated graphics solution out there, but it would fall just short of providing the performance you would need for an entry level gaming system. Ideally fusion should have the flexibility and muscle to fit a variety of different systems while still providing all the power you really need. With llano I don't think AMD is quite there yet. The AMD A8 3850 is a great CPU and I would highly recommend it, but I would have to consider carefully before recommending a lower performance part from the llano line up.
Fortunately this is AMDs first Fusion part. It has been rumored that AMD is having some issues on the production end with die yields, so if AMD can get that all sorted out we might see higher clocked parts and better low wattage options. AMD already has the next generation of fusion in the works; code named Trinity is promises to use processing cores based on the bulldozer architecture with an integrated 7000 series graphics. These processors are supposed to come out in 2012 and offer as much as 50% more performance. If it can do that with a corresponding increase in power efficiency the full concept of fusion may yet be realized.
Last edited: