AMD have revised Bulldozers architecture, and now Piledriver is ready for the public. Will it be the improvement we were looking for?
Product on Review: AMD Piledriver FX-8350
Manufacturer and Sponsor: AMD
Street price: £150 inc. VAT (At time of review)
Today, we have our hands on the latest CPU from AMD – Piledriver. It is a revision of Bulldozer which promises to bring increased performance. Since Bulldozer was released, AMD have struggled to match Intel for their performance, even though Intel were only using half the number of cores on the chips. AMD have quoted that there will be an upwards of 20% increase in performance and we will be putting that to the test today to find out whether that is the case or not. We certainly hope so, AMD need to regain their lost ground that Intel have increasingly gained over the past number of years with their Core family of processors.
Piledriver is meant to bring us better performance over Bulldozer, but the thermal side of things have not been improved just yet. It is still rated at 125w which is a lot of power to be chewing through but if the performance/watt ratio increases, it may be justifiable. We will have to see how much of a gain there is in terms of performance before we jump to any conclusions. Without any further ado, lets dive in.
AMD’s Overview of Piledriver
We call it the new AMD FX 8-Core Processor Black Edition and it’s unlocked for your overclocking pleasure. Experience unmatched multitasking and pure core performance with the 32nm 8-core desktop processor. Get the speed you crave with AMD Turbo CORE Technology to push your core frequencies to the limit when you need it most. Go beyond the limits of maximum speed with easy-to-use AMD OverDrive™ and AMD Catalyst Control Center™ software suites. But the best part of all? You’ll get all this impressive performance at an unbelievable price. You’ll be asking yourself “what competition?” in no time.
Features & Specifications
Tech/Package
32nm / AM3+
TDP Configs
95W, 125W
Processor Core
“Piledriver” (up to 8 Cores), 8MB L2 cache, 256-bit FPUs, 8MB L3 cache
Memory
DDR3, 1333-1866, 1.5V
Graphics Core
N/A – Pair with AMD Radeon™ HD 7000 Series Graphics
Performance Management
Core Level: CC6 Power State
Module Level: L2 Cache power gating via CC6
The Motherboard(s): 990FX + SB950
Tech/Package
65nm / FC BGA
TDP Configs
10W for typical configurations
ALink III
x4 Gen2
SATA
6 Ports, 6Gbps
USB
12 USB 2.0 Ports, 2 USB 1.1 Internal Ports
PCIe 2.0 Lanes
2×16 or 4×8 + 8×1 + 4×1 + (2×1 on SB)
RAID
0, 1, 5, 10
Software
Windows® 7, Windows® 8, Linux
32nm / AM3+
TDP Configs
95W, 125W
Processor Core
“Piledriver” (up to 8 Cores), 8MB L2 cache, 256-bit FPUs, 8MB L3 cache
Memory
DDR3, 1333-1866, 1.5V
Graphics Core
N/A – Pair with AMD Radeon™ HD 7000 Series Graphics
Performance Management
Core Level: CC6 Power State
Module Level: L2 Cache power gating via CC6
The Motherboard(s): 990FX + SB950
Tech/Package
65nm / FC BGA
TDP Configs
10W for typical configurations
ALink III
x4 Gen2
SATA
6 Ports, 6Gbps
USB
12 USB 2.0 Ports, 2 USB 1.1 Internal Ports
PCIe 2.0 Lanes
2×16 or 4×8 + 8×1 + 4×1 + (2×1 on SB)
RAID
0, 1, 5, 10
Software
Windows® 7, Windows® 8, Linux
The current generation of Bulldozer CPUs are being replaced with four models at release. There may be more models released in the future with revised steppings but for now, we will have these on the shelves:
There will be no new chipset released along with Piledriver as the current generation of 990FX motherboards support the technology. After all, it is just a revised Bulldozer architecture with improved performance. Lets take a look at the basics of the design anyway, just to refresh our memory:
In the chart below, AMD have demonstrated where they feel their competition against Intel lies with their pricing schemes. Unfortunately for them, it appears that their top of the line product will be compared to a mid-ranged Intel product and Intel still have the i7 3770K card to play but AMD have not responded to this.
Test Setup & Benchmarks
Our test benches consist of the following components:
AMD Piledriver FX-8350 Test Bench
CPU – AMD Piledriver FX-8350 Eight-Core Processor
Cooling – Noctua NH-D14
Motherboard – ASUS M5A99X EVO
Memory – G.SKILL RipJawX DDR3-1866MHz CL9
Graphics – ZOTAC GTX 570 1280MB
Storage – Corsair Force F120
PSU – FSP Aurum CM Series 750W 80PLUS Gold
CPU – AMD Piledriver FX-8350 Eight-Core Processor
Cooling – Noctua NH-D14
Motherboard – ASUS M5A99X EVO
Memory – G.SKILL RipJawX DDR3-1866MHz CL9
Graphics – ZOTAC GTX 570 1280MB
Storage – Corsair Force F120
PSU – FSP Aurum CM Series 750W 80PLUS Gold
AMD Bulldozer FX-8120 Test Bench
CPU – AMD Bulldozer FX-8120 Eight-Core Processor
Cooling – Noctua NH-D14
Motherboard – ASUS M5A99X EVO
Memory – G.SKILL RipJawX DDR3-1866MHz CL9
Graphics – ZOTAC GTX 570 1280MB
Storage – Corsair Force F120
PSU – FSP Aurum CM Series 750W 80PLUS Gold
CPU – AMD Bulldozer FX-8120 Eight-Core Processor
Cooling – Noctua NH-D14
Motherboard – ASUS M5A99X EVO
Memory – G.SKILL RipJawX DDR3-1866MHz CL9
Graphics – ZOTAC GTX 570 1280MB
Storage – Corsair Force F120
PSU – FSP Aurum CM Series 750W 80PLUS Gold
Before we start any overclocking, we will be setting the BIOS to the following defaults before beginning any overclocking.
Overclocking Settings
RAM speed – 1066MHz
CPU/NB and HT Links – 1600MHz (8x CPU bus)
RAM speed – 1066MHz
CPU/NB and HT Links – 1600MHz (8x CPU bus)
For all of our AMD CPU reviews, we will be running the following benchmarks to give you a varied array of results to best portray the CPU(s).
AIDA64 – CPU benchmarks
X264 HD 5.0.1 – 1st and 2nd pass encoding
SiSoftware Sandra 2012 – CPU and Memory benchmarks
CineBench 11.5 – CPU score
3DMark 11 – DX11 3D Performance and Extreme preset
Games – Crysis 2, DiRT Showdown, Skyrim and Sleeping Dogs
X264 HD 5.0.1 – 1st and 2nd pass encoding
SiSoftware Sandra 2012 – CPU and Memory benchmarks
CineBench 11.5 – CPU score
3DMark 11 – DX11 3D Performance and Extreme preset
Games – Crysis 2, DiRT Showdown, Skyrim and Sleeping Dogs
Overclocking
In terms of overclocking, it is difficult to expect high overclocks from such a power hungry chip such as this one. It is rated at 125w TDP at stock and that means that more voltage and clock speed is going to push up thermals fast. However, having said that, our AMD Bulldozer FX-8120 which we are pitting the AMD Piledriver FX-8350 of ours against was capable of 4.875GHz benchmark stable, so we are hoping to see our Piledriver sample hit similar clock speeds. Perhaps it will overtake its older brother, Bulldozer.
AMD Piledriver FX-8350 chips are clocked at 4GHz with a 4.2GHz turbo profile. That means at stock speeds, it is already 900MHz faster without turbo than the AMD Bulldozer FX-8120 which we are comparing it against, and that is no mean feat to achieve on the same wattage and voltage. However, the FX-8120 CPUs were capable of these speeds with the same voltages applied so it is any ones guess why AMD did not set them to this speed to begin with.
Anyway, this is part of the review is not about how much quicker it is at stock speeds compared to the older generation of CPUs. It is about finding out how far we can push it and maintain a clock speed whereby our Noctua NH-D14 can keep it cool and run through our benchmark suite. So, lets crack on!
Initially, I began overclocking with the voltage set to what I had used for our AMD Bulldozer FX-8120 review sample and used that as a starting point. Pleasantly surprised, the AMD Piledriver overclocked to the same speeds and even surpassed its older brother. Temperatures were also lower which meant that I could comfortably push the voltage up to 1.5 volts, which was not a viable option on Bulldozer. At the end, I was able to boot into Windows at 5.2GHz with all eight cores enabled. Disabling all but the first two cores (1 Module) did not help so all eight were left on. Unfortunately, as soon as you put any kind of stress on it from a benchmark, it would freeze up and restart, and occasionally throw a BSOD in our face.
A maximum boot into Windows is all fine and dandy, but if you cannot use it, then it is pretty useless. So, the clock speed was knocked back enough until it was finally stable enough to complete benchmarks. A final and stable overclocked speed which was usable was 5040MHz. Anything higher and it would occasionally throw a wobbly which meant that our results would be unreliable. At 5040MHz, it was able to complete our entire benchmark suite without freezing up or rebooting itself once which is what we are after.
Power Consumption
It’s interesting to note the power consumption across all motherboards because energy results can fluctuate between manufacturers due to the features and specifications that are integrated. First of all, for the idle tests the system is cold booted into Windows and no additional applications are executed – the watts are noted on the external energy monitor we use. Then the system is loaded with Intel Burn Test to stress the CPU – watts are again monitored throughout the period of testing and the highest and lowest rated watts are then noted down. Below are the results of this energy test.
Power consumption is one of AMDs big downfalls at the moment. Just take a look at these figures, and you will see why.
Benchmarks: AIDA64
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. At the same clock speed theoretically the processor with the shorter pipeline and smaller misprediction penalties will attain higher benchmark scores. For example – with HyperThreading disabled – the Intel Northwood core processors get higher scores than the Intel Prescott core based ones due to the 20-step vs. 31-step long pipeline. The CPU Queen test uses integer MMX, SSE2 and SSSE3 optimizations.
As you can see, the stock results are comparably higher due to the fact that the clock speeds are 900MHz higher. However, when you overclock the system, the playing field soon levels out.
Benchmarks: CINEBENCH 11.5
CINEBENCH can measure systems with up to 64 processor threads. This test scene contains approximately 2,000 objects which in turn contain more than 300,000 polygons in total, and uses sharp and blurred reflections, area lights, shadows, procedural shaders, antialiasing, and much more. The result is displayed in points (pts). The higher the number, the faster your processor is.
Whilst there is a performance increase, this again is down to the increased clock speed but it is also down to the improved IPC rate. If we take a look at the 1 Module mode, we can see that there is more or less a 30% increase in performance.
Benchmarks: SiSoftware SANDRA 2012
SANDRA isn’t always a benchmark that is included in hardware reviews but we believe it shouldn’t be ignored. SANDRA provides a vigorous package that tests your system in a rather large array of benchmarks. There are many aspects of benchmark that can be executed. We will be considering the CPU and memory tests.
In some areas of the SiSoftware SANDRA 2012 benchmarks which we ran, there are obvious gains in performance but some of the results are within a margin of error – however, the improvement is still there.
Benchmarks: x264HD Version 4
Simply put, this test measures how fast your machine can encode a short, DVD quality MPEG-2 video clip into a high-quality x264 video clip. What’s x264, you ask? It’s more or less the next-generation Xvid/DivX codec. We think it’s ideal for a benchmark because the application (x264.exe) reports fairly accurate compression results (in frames per second) for each pass of the video encoding process and it uses multi-core processors very efficiently
There is definite improvement to be seen here in both the first and second passes of this benchmark.
3D Benchmarks: 3DMark 11
Designed to measure your PC’s gaming performance 3DMark 11 makes extensive use of all the new features in DirectX 11 including tessellation, compute shaders and multi-threading. Trusted by gamers worldwide to give accurate and unbiased results, 3DMark 11 is the best way to consistently and reliably test DirectX 11 under game-like loads.
Improvement in this benchmark is minimal but that is due to this being a heavily graphical orientated benchmark rather than one which tests the CPU. Granted the fact that there is a slight improvement – but it is not much.
3D Benchmarks: Unigine Heaven 3.0
Unigine’s benchmark is to showcase the benefits of DX 11 and the use of tessellation as opposed to normal rendering techniques. The demo has impressive visuals that give an insight into what we can expect from gaming in the future.
As with 3DMark 11, this is a very graphically intense benchmark and therefore the CPU does not have much of an effect on the overall outcome unless. There is a tiny improvement but this is all within tolerances that could be considered to be a margin of error.
Gaming Benchmarks: Crysis 2 & DiRT Showdown
It’s important for us to perform a number of game benchmarks as well as the synthetic tests because it allows us to test the 3D performance. So here we have some popular titles, critically acclaimed and action-packed. FRAPS and in-house benchmarking tools were used to capture average FPS.
Gaming Benchmarks: Skyrim & SleepingDogs
It’s important for us to perform a number of game benchmarks as well as the synthetic tests because it allows us to test the 3D performance. So here we have some popular titles, critically acclaimed and action-packed. FRAPS and in-house benchmarking tools were used to capture average FPS.
Conclusion
AMD have certainly raised the game with the revised version of Bulldozer, known now as Piledriver. However, we have not seen the quoted ‘up to 23%’ improvement which AMD said there should be. Obviously, that is in an ideal world it may be that big of a difference between the two CPUs. In some of our benchmarks where we benchmarked only one module (2 cores/2 threads), the gain was there – but when you leave them all enabled, there is nowhere near such a gain to be seen.
Pricing on these chips is very competitive and it is clear what AMD knows where their CPUs lie when they are compared against Intels lineup of processors.
Unfortunately, AMD have not been able to bring down the TDP ratings, again. They are still rated at 125W which is a lot of power when compared against Intel’s offerings. As this CPU is pitted against the Intel Core i5 3570K, according to AMD and their pricing strategy, when you compare Intel at 77w to AMD at 125w, the difference is large. Granted that Intel has four cores whereas this chip has eight, it still needs improving.
As of yet, AMD have not been able to get past the 32nm manufacturing process for the processors yet, whereas Intel is already on 22nm which means both lower voltages can be used and the TDP can therefore be lowered as well. However, it is not all bad. AMD managed to get the clock speeds of their CPUs up a considerable amount without going over the 125w limit.
When you compare the Intel Core i5 3570K, the AMD Piledriver FX-8350 certainly does give it a good run for its money. However, this is only in multi-threaded applications. When it comes down to a single threaded application, the Intel CPU(s) continue to trounce all over the AMD chips. It is a step in the right direction for AMD though. It may be a case of too little too late but hopefully this will put them back in the game where they can regain some of the consumers trust and continue to push them towards something better.
To cut a long story short. Would this be the choice of processor to go for if you were building a new gaming rig from the ground up? In short, probably not as an Intel CPU continues to beat AMD in the performance department and they use considerably less energy – 77w versus 125w, at stock speeds. Not only will you need more cooling power to tame the beast, you will also see increased electricity bills. Whilst the near 50w difference may not seem like a lot, if you leave your system on all day, every day, you will certainly feel it in your pocket.
This is a competitive CPU when pitted against the Intel Core i5 3570K, and it definitely deserves a better reward from us than Bulldozer got. Although it still is not a gem in the rough, a silver medal is definitely in order here for the improved performance and competitive pricing against Intel.
Pros
+ Definite clock for clock improvement over Bulldozer
+ Increased clock speeds whilst keeping the same TDP
+ Competes against the Intel Core i5 3570K in lots of tasks
+ Good performance for the money
Cons
– High power consumption is still an issue
– Single threaded performance still needs improvement
– 32nm technology, Intel are on 22nm
+ Definite clock for clock improvement over Bulldozer
+ Increased clock speeds whilst keeping the same TDP
+ Competes against the Intel Core i5 3570K in lots of tasks
+ Good performance for the money
Cons
– High power consumption is still an issue
– Single threaded performance still needs improvement
– 32nm technology, Intel are on 22nm
