Bulldozer is AMD’s latest CPU architecture featuring an overhauled 32nm chip design, updated instruction sets, and up to eight processing cores. The FX-8150 sports a maximum clock speed of 4.2 Ghz while retaining the 125W TDP of most of the Phenom II X4/X6 line.
October 19, 2011 by Lawrence Lee
Product | AMD FX-8150 AM3+ Processor |
Manufacturer | |
Street Price | US$245 |
AMD has always been the CPU underdog, with Intel holding a substantial lead in performance except in the days of the Athlon 64 and Athlon 64 X2. Once Intel countered with Core 2 Duo line, they took the reigns and never looked back. We’ve been waiting for AMD to at least reach parity ever since.
AMD’s latest processor is Bulldozer, a revamped architecture, the first major redesign since Phenom II was released in 2009. Bulldozer has a restructured layout that AMD claims reduces some redundancies in traditional multi-core designs, creating a more efficient chip. Bulldozer uses a smaller 32 nm manufacturing process, allowing more cores to fit on the die. The desktop models (codename “Zambezi”) have up to eight cores and over two billion transistors, more than double that of the fastest Phenom II’s.
AMD Zambezi Lineup | |||||||
Model | CPU Base | Turbo Core | Max Turbo | TDP | Cores | L2 Cache | Price |
FX-8150 | 3.6 GHz | 3.9 GHz | 4.2 GHz | 125W | 8 | 8MB | $245 |
FX-8120 | 3.1 GHz | 3.4 GHz | 4.0 GHz | 95W / 125W | 8 | 8MB | $205 |
FX-8100 | 2.8 GHz | 3.1 GHz | 3.7 GHz | 95W | 8 | 8MB | N/A |
FX-6100 | 3.3 GHz | 3.6 GHz | 3.9 GHz | 95W | 6 | 6MB | $165 |
FX-4170 | 4.2 GHz | – | 4.3 GHz | 125W | 4 | 4MB | N/A |
FX-B4150 | 3.8 GHz | 3.9 GHz | 4.0 GHz | 95W | 4 | 4MB | N/A |
FX-4100 | 3.6 GHz | 3.7 GHz | 3.8 GHz | 95W | 4 | 4MB | $115 |
Note: all CPUs have 8MB L3 cache |
AMD is launching four Zambezi chips, the quad core FX-4100, the six core FX-6100, and the eight core FX-8120 and FX-8150. These CPUs use AMD’s latest AM3+ socket, an updated version of AM3 with the same pin count. An AM3+ motherboard or an AM3 model with listed AM3+ support is required; see your motherboard manufacturer website for details. If you have an AM3 chip, they are forwards-compatible — if for some reason you would like a new motherboard but do not wish to upgrade your processor at this time.
AMD provided us with one sample just paltry couple days before the embargo date, the top-of-the-line FX-8150 which will sell for US$245, putting in somewhere between the Intel Core i5-2500K and i7-2600K in pricing. AMD managed to keep the TDP at a reasonable 125W despite the eight cores and higher clock speeds. The FX-8150 is rated as a 3.6 GHz processor, but like Intel’s Sandy Bridge processors, always runs slightly faster on load due to its dynamic overclocking feature (Turbo Core for AMD, TurboBoost for Intel).
Die layout. |
The FX-8150 isn’t a “true” octa core processor but rather has four Bulldozer modules packed with a pair of processing cores each — this is the biggest change from the older Phenom II architecture. The CPU has plenty of cache, 8MB of L2 (2MB per core) and 8MB of L3 (2MB per module). Bulldozer has also been updated with FMA, XOP, AES, AVX, and SSE 4.2 instruction sets giving it a boost in applications that support them. The total package has an amazing 2+ billion transistors, more than twice that of latest Phenom II X4/X6’s, fit on a ~315 mm2 die.
CPU-Z screenshot: idle. |
The lowest CPU state has a clock speed of 1400 MHz, up from 800 MHz, primarily to speed up the time it takes to reach the 3.6+ GHz performance states offered by the FX-8150. While the idle frequency has been increased, the core voltage is noticeably lower at 0.864V according to CPU-Z; Phenom II X4/X6’s typically idle at around ~1.000V.
CPU-Z screenshot: two core load. |
Turbo Core is more aggressive on the FX-8150 than the Phenom II X6 line, with a maximum clock speed of 4.2 GHz. On load, idle cores are put into sleep states while active cores get pumped up clock speeds and voltages depending on how many cores are stressed, provided it does not exceed the 125W TDP. When all eight cores are stressed, they should all run at 3.6 GHz in optimal conditions.
Absurdly overpackaged sample set. |
As usual, our review kit was a bulbous, well-padded package containing a high-end motherboard, in this case, the Asus Crosshair V Formula. However, we didn’t expect the unnecessary tin box holding the CPU and a ridiculously oversized AMD FX belt buckle. It seems that AMD has taken some really odd steps to make reviewers pay attention to Bulldozer; even teens couldn’t be much impressed with the package.
PACKAGE & MOTHERBOARD
AMD’s high-end chipset for AM3+, 990FX, is not much of an upgrade over 890FX for AM3, updated slightly for higher speed memory support. |
The Asus Crosshair V Formula is a high-end board with multiple PCI-E 16x slots and some heavy duty VRM cooling. |
AM3+ has the same pin count as AM3, so the socket is not physically any different. The heatsink retention frame is essentially unchanged as well so any heatsink with AM3 compatibility will work. |
The FX-8150. |
Intel’s flagship six-core i7-980X ships with a tower cooler but the FX-8150 ships with the same dual heatpipe cooler as most of AMD’s older Phenom II X4/X6 chips. The FX-8150 will also be sold in a bundle with an Asetek water cooler similar to the Corsair Hydro H50. |
For reference, the stock heatsink is 62 mm tall, weighs 360 grams, and sports a 70 mm fan. We didn’t use it during testing for obvious reasons. |
TEST METHODOLOGY
Common Test Platform:
- Corsair
XMS3 memory 2x2GB, DDR3-1600 @ 1333MHz, 9-9-9-24 - Asus EN9400GT Silent Edition
graphics card – 512MB - Western Digital VelociRaptor
hard drive – 300GB, 10,000RPM, 16MB cache - Seasonic
SS-400ET ATX power supply - Microsoft
Windows 7 operating system – Ultimate, 64-bit - Nvidia
Forceware 266.58 graphics driver - Scythe Kabuto
CPU cooler – stock fan at 800RPM
Intel LGA1155:
- Intel Core i3-2100
processor – 3.1 GHz, 32nm, 65W (dual core) - Intel Core i5-2400
processor – 3.1 GHz, 32nm, 95W - Intel Core i5-2500K
processor – 3.3 GHz, 32nm, 95W - Intel Core i7-2600K processor
– 3.4 GHz, 32nm, 95W - Asus P8P67 motherboard
– P67 chipset
AMD AM3:
- AMD Phenom II X4 955 Black Edition
processor – 3.2 GHz, 45nm, 125W - AMD Phenom II X4 965 Black Edition
processor – 3.4 GHz, 45nm, 125W - AMD Phenom II X4 975 Black Edition
processor – 3.6 GHz, 45nm, 125W - AMD Phenom II X6 1100T Black Edition
processor – 3.3 GHz, 45nm, 125W - Asus M4A78T-E motherboard
– 790GX chipset
AMD AM3+:
- AMD FX-8150
processor – 3.6 GHz, 32nm, 125W - Asus Crosshair V Formula motherboard
– 990FX chipset
Measurement and Analysis Tools
- CPU-Z
to monitor CPU frequency and voltage. - SpeedFan
to monitor CPU temperatures. - Real Temp
to monitor CPU temperatures. - CPUBurn
processor stress software. - Prime95
processor stress software. - Media Player
Classic – Home Cinema to play x264/MKV video using DXVA. - Adobe
Photoshop CS2 as an image manipulation benchmark. - Eset NOD32 as
an anti-virus benchmark. - WinRAR as an
archiving benchmark. - iTunes
an audio encoding benchmark. - TMPGEnc
Xpress as a video encoding benchmark. - HandBrake as a
video encoding benchmark - Seasonic
Power Angel AC power meter, used to measure the power consumption
of the system. - Custom-built, four-channel variable DC power supply, used to power
and regulate the CPU fan.
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 4500×3499).
x264 1080p: Spaceship is a 1080p x264 clip encoded from the |
Testing Procedures
Our main test procedure is a series of benchmarks, timed tests of real-world applications. System power consumption (AC) is measured with a Seasonic Power Angel during these tests (an average of the first 10~15 seconds) as well as at idle, during playback of a 1080p H.264 encoded clip, and during full CPU load. To stress the CPU we use either Prime95 (large FFTs setting) or CPUBurn depending on which
produces higher system power consumption. The AC system power is then later converted to DC.
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.
Operating Voltages
|
Before we jump into our test results, please note the operating voltages
of the processors tested today as sort of a disclaimer. Different samples of
the same processor often run at slightly different operating voltages which
can affect energy efficiency. Different motherboard models do not apply the exact same core voltage either.
TEST RESULTS
System Power
|
Under light load the FX-8150 is a marked improvement over the X6 1100T, with a power savings of 7W when idle and 4W during H.264 video playback. It was close to X4 levels, but didn’t hold a candle to Intel’s Sandy Bridge lineup.
|
The FX-8150 consumed 25W more than 1100T when running our synthetic CPUBurn/Prime95 stress tester and 31W more when encoding video with HandBrake. This may be considered impressive considering it’s an eight core processor, but it’s staggering too see it use almost double the energy of Sandy Bridge.
PERFORMANCE
To test performance, we pit the processors against one another in a short series of timed tasks using real world applications. Power consumption was measured when applicable.
|
The FX-8150 produced a 9% gain in Photoshop compared to the 1100T but the driving force behind this is likely the higher clock speed. Despite the performance boost, it still trailed Intel’s latest Core i5’s and even Core i3’s.
|
A surprising result came in our NOD32 test with the FX-8150 coming in dead last. Even with Turbo Core pumping up clock speed in non-threaded applications, it still ended up 17% slower than the lowly X4 955.
|
The biggest improvement thus far was in WinRAR, where the FX-8150 posted a nice 17% improvement over the 1100T and with a slight power savings as well. Sandy Bridge still beat out all comers, but Zambezi narrowed the gap substantially.
PERFORMANCE (Continued)
|
The FX-8150 was struck by another setback in our iTunes encoding test, falling to X4 levels of performance and consuming about 10W more.
|
TMPGEnc is a multithreaded video encoding program, so if an eight core processor would shine, it would be here. The good news is that the FX-8150 reaches parity with Sandy Bridge here, matching the Core i5-2400’s encoding time. The bad news is it used more than twice the juice to achieve the same result.
|
HandBrake, which supports some of the new instruction sets included in Bulldozer’s architecture, is the only test where the FX-8150 clearly scored a win, beating out the i7-2600K by 13 seconds. However again, the power consumption was off the charts.
Overall Performance
We arrived at our overall performance score by weighing each test equally (each composing 1/6 of the total). Mathematically, a processor that finishes first in every single test receives a score of 100. As the Core i7-2600K only lost one test by a small margin, it tops the chart with 99 points.
|
The FX-8150’s overall performance in our test suite was lackluster compared to Intel’s Sandy Bridge quad core offerings. It was only competitive in our threaded video encoding tests. In the simpler tests, we got Jekyl and Hyde numbers which brought down the score though it was still an improvement over the 1100T.
|
The total power consumed by the FX-8150 running our test suite was more than 100% higher than the Core i5’s. Obviously the slower bench times inflate these numbers, but even in our video encoding tests where Zambezi drew even with Sandy Bridge in performance, the power draw was terrible. We had hoped that even if Zambezi wasn’t as fast as Sandy Bridge, it would at least be more energy efficient but sadly it’s just more of the same.
Power, Performance and Cost Analysis
We derived our average system power consumption numbers to consider both heavy and light usage. Heavy usage is defined as using the system running 75% of the time on high load (an average of the power consumption of our five measured benchmarks), while low usage is defined as running 75% of the time on light load (an average of the power consumption when sitting idle and playing H.264 video).
|
The standings of each processor do not vary depending on the usage pattern, but the relative differences are significant. The FX-8150 for example uses 45% more power than the i7-2600K in our light load scenario but this figure jumps to 67% on heavy load.
|
We arrive at our performance per watt scores by dividing the overall CPU performance by the average power consumption and adjusting the results so the best CPU scores 100 points. With better overall performance and superior energy efficiency, Sandy Bridge is firmly on top for both light and heavy usage systems. The AMD processors (the FX-8150 in particular) do slightly better on light load as the power consumption differences are lower.
|
Note: Motherboard pricing data was collected from Newegg’s catalog with the following criteria: retail versions, US$0~$250 after rebates if applicable, Asus/Intel/Gigabyte/MSI branded, microATX/ATX form factor, SATA 6 Gbps and USB controller. H61/H67/Z68 chipset for Core i3-2100, P67/Z68 chipset for Core i5/i7’s, AM3/AM3+ socket for Phenom II’s, AM3+ socket for FX-8150.
AMD has always had one advantage in its battle against Intel: Lower CPU and motherboard prices. The FX-8150 breaks this trend as its US$245 price is $35 more than the Core i5-2500K, and AM3+ motherboards also carry a slight premium over the older AM3 stock. Even with a more expensive P67/Z68 motherboard, the average i5-2500K system comes in slightly cheaper.
|
Focusing on performance:price alone, the FX-8150 falls short once again. The only way it reaches parity with Sandy Bridge is if only highly threaded applications are used.
FINAL THOUGHTS
Based on the tests we’ve conducted, the only end-users who could benefit from the new AMD chip are PC professionals that deal with sophisticated audio and video encoding, rendering, virtualization, and/or heavy workstation/server loads which max out CPU cores and system memory on a regular basis. If your needs are that demanding, you’ll probably want to run the system for longer durations than the average home user; undoubtedly this will drive up your electricity costs considerably. Already behind Intel in energy efficiency, Zambezi, or at least the eight core version, delivers no improvement in this regard.
For lighter usage patterns, the argument for the FX-8150 is fragile. High power consumption, albeit to a lesser degree is still a major factor and performance is lukewarm. We noted some improvements in a couple of simpler non-threaded benchmarks but in another two, the FX-8150 turned out to be slower than the X6 1100T. As the FX-8150 has a 300 MHz advantage, it seems that as a result of restructuring the chip architecture and/or splitting resources between cores/modules, per clock cycle, Bulldozer seems slower than its predecessor.
It is unfortunate for AMD that the chip lives up to its name. “Bulldozer” appears to be a big, lumbering beast that takes a lot of power to run and is only great at a few specific things. The FX-8150 excels mainly in multithreaded applications so you could argue it’s the CPU of the future, but with AMD playing catch-up they needed a hit today, not something that might outperform Sandy Bridge in a couple of years when more programs are coded to use multiple cores. Currently an Intel Core i5-2500K based system offers better, more well-rounded performance and far superior power efficiency at a slight discount compared to the FX-8150, so even AMD’s price advantage has evaporated. The less expensive FX processors may fare better, but only time will tell. Still struggling to fight in the desktop space, AMD faces further pressure as Intel’s newer, faster Sandy Bridge-E chips will debut in a month’s time.
Our thanks to AMD
and Asus for the FX-8150 and Crosshair V Formula samples used in this review.
* * *
Articles of Related Interest
AMD A8-3850 Quad Core Desktop APU (updated July 10)
Intel Core i3-2100T & Core i5-2400S Low Power CPUs
Intel Core i3-2100 vs. AMD Phenom II X2 565
Asus E35M1-M Pro: AMD Fusion Motherboard
Core i5-2400, i5-2500K and i7-2600K CPUs
AMD Athlon II X3: Affordable Compromise
* * *