AMD FX-8350 review: is AMD back? - BeHardware
>> Processors

Written by Marc Prieur

Published on October 23, 2012

URL: http://www.behardware.com/art/lire/880/


Page 1

CMT, high clocks and Piledriver



On its introduction in October 2011, Zambezi, an AM3+ rollout of Bulldozer architecture, disappointed on several fronts. One year on AMD is back with Piledriver, a development of its CMT architecture, for a new AM3+ processor, Vishera. In this report, we’ll be analysing the performance of the flagship processor from this new range, the FX-8350. Does it mark AMD’s return to form in the world of CPUs?

CMT, high clocks and Piledriver
Back in May 2011, we devoted a report to the Bulldozer architecture. The Piledriver and Bulldozer architectures have a common base, Cluster Multi-threading (CMT) technology. This technology means an 8-core processor is in fact made up of 4 modules. Within a module, the two cores share a certain number of components:

- the front-end which groups the fetch unit and instruction decoding as well as the L1 instruction cache which is supplied by these units;
- the floating point unit;
- the L2 cache.

AMD claims 80% of the performance of two full cores for major efficiencies in terms of silicon area and energy consumption. Many other changes were also made, both to the processing units themselves and the memory sub-system, in particular to allow the architecture to clock higher.


In terms of x86 instruction sets Bulldozer already supported the latest versions of SSE4 (4.1 and 4.2), AES-NI instructions enabling acceleration of encryption and AVX, introduced by Intel with Sandy Bridge, and its 256-bit variants. In addition, it added its own instructions, grouped under the names XOP, FMA4 and CVT16. XOP operates mainly on integer operands, FMA4 on 128-bit floating point numbers and CVT16 groups high precision floating point conversion instructions to medium and low precision floating points. FMA4, which allows the processing of a multiplication and addition in a single cycle, should among other things enable gains when used by applications.

Piledriver adds FMA3 (Fused Multiply/Add on 3 operands, a = a * b +c ) in addition to FMA4 (a = b * c + d) which was already previously supported (Intel will use FMA3 as of Haswell), as well as the F16C (16/32 bit floating point conversion) instructions introduced by Intel in Ivy Bridge. For the rest, the changes are mostly small touches at all levels, with the branch prediction mechanisms and schedulers announced as more efficient and gains announced for divisions.


Page 2
Piledriver & AM3+ = Vishera, the range

Piledriver & AM3+ = Vishera

Piledriver already made its appearance a few months ago on the Trinity and Virgo APUs, designed for laptops and desktops respectively. Virgo uses a new Socket FM2 while Vishera, the ‘CPU’ rollout of Piledriver, is still an AM3+ type. The good news is that a simple bios update should make any AM3+ board compatible. The CPU worked on our ASUS M5A99X EVO with bios 901 dated December 2011, though we used a more recent bios, the 1503, for our test.


The Vishera and Zambezi specs are similar. Both are manufactured on the 32nm SOI process from GlobalFoundries and have a surface area of 315mm˛ for 1.2 billion transistors – to recap Zambezi "lost" 800 million transistors. The total number of transistors communicated can be significantly variable and Intel carried out the opposite operation, increasing the number of transistors for Sandy Bridge.

The Vishera and Zambezi cache sizes are also identical, though the DDR3 memory controller is now officially DDR3-1866, a mode which in practice already functioned on Zambezi.

The Vishera and Zambezi dies are significantly bigger than the Intel offer, which integrates HD Graphics, and this of course has an impact on production costs.
The range
For this launch, AMD is bringing out four new references:


To recap, there are two Turbo Core clocks. ‘Max Turbo’ can be achieved when half the modules are in use and 'Turbo’ even when all modules and cores are in use, within TDP limitations.

It’s not easy to compare the old and new generations. Looking at the FX-4s, the FX-4300 costs the same as the FX-4170. It is of course based on Piledriver but has a lower clock and Turbo with an L3 cache that has been halved. The TDP has however been revised down.

The FX-6300 also has a lower TDP than the FX-6200, which is priced at the same level. While its base clock is lower, in Turbo mode the clocks are the same.

For the FX-8s the FX-8320 is positioned between the 8120 and 8150, though it is closer to the second in clocks. The FX-8350’s base clock is 500 MHz higher than that of the FX-8150 but only 200 MHz higher in Turbo and the same in Turbo Max.


Page 3
The AMD FX-8350, test protocol

The AMD FX-8350

For this test, before, we hope, sending us other models, AMD supplied us with an AMD FX-8350. With a base clock of 4 GHz, it is configured this way in terms of frequency / voltage:

- 4.0 GHz at 1.3375v (Base)
- 4.1 GHz at 1.4125v (Turbo)
- 4.2 GHz at 1.4250v (Turbo max)


To give you a comparison, our FX-8150 has VIDs of 1.2375v, 1.375v and 1.4125v with clocks of 3.6, 3.9 and 4.2 GHz respectively. As we’ll see further on, there’s not as much of an impact on energy consumption as might have been feared, which shows that improvements have been made with respect to current leakage.
Test protocol
For the test protocol, we used the one set up for the review of the AMD FXs, which is detailed on this page.

On the applications side first of all, in place of build 2085 of x264 we used build 2216 which includes various optimisations using among other things the instructions introduced with Bulldozer. In practice, Intel processors also benefit from gains in performance with gains varying between 3 and 5% depending on the processor. While we were at it, we also updated the MainConcept H.264 codec, going from Reference 2.2 to TotalCode 2.5, the new name of the software. Here there was no real change in performance.


All the gaming performance measures were taken with a GeForce GTX 680 in place of a GeForce GTX 580. All the games were updated with the latest patches. We changed the scene used in Rise Of Flight, the one we were using before being too demanding and posing some issues for the latest version of the engine. F1 2011 has been replaced by F1 2012, Anno 1404 by Anno 2070 and Skyrim has been included as our eighth game. As before, we tested these games at 1920 x 1080 with all options (except antialiasing) at maximum, and searched out demanding scenes which put us in a position of having performance limited by the processor.

Because of these changes and due to time pressures we were only able to include LGA 1155 and AM3+ processors here (others to come in forthcoming articles!). The tests were carried out on the following platforms:

- Intel DP67BG (LGA1155)
- ASUS M5A99X EVO (AM3+)
- 2x4 GB DDR3-1600 9-9-9
- GeForce GTX 680 + GeForce 306.97
- SSD Intel X25-M 160 GB + SSD Intel 320 120 GB
- Corsair AX650 Gold power supply

The tests were carried out in Windows 7 SP1, with the two patches designed to make best use of the AMD FX processors used.


Page 4
Energy consumption and efficiency

Energy consumption and efficiency
For the energy consumption test we tried to use a test which is more or less representative for all architectures of what we get in applications in terms of performance and energy consumption. In the end we opted for Fritz Chess Benchmark once again. In addition this application has the advantage of allowing us to fix the number of threads to be used.

The energy consumption readings therefore shouldn't be taken as absolute maximum values but rather as typical of a heavy load - applications specialised in processor stress such as Prime95 can consume up to 20% more. All energy economy features, including those on motherboards such as the ASUS EPU, were turned on for this test, as long as they didn't have a negative impact on performance.

Remember we give two types of readings, the first at the 220V wall socket using a wattmeter for the whole test configuration and the second at the ATX12V via a clip-on ammeter. This reading allows us to isolate the energy consumption of the processor for the most part, but unfortunately it isn’t exactly comparable from one platform to another because in certain cases a small proportion of the energy consumption of the CPU comes from the standard 24-pin ATX socket.


[ 220V socket ]  [ ATX12V ]

The AM3+ platform draws quite a lot of power at idle in comparison to the LGA1155 platform, an issue that AMD has resolved on FM2. In this domain the FX-8350 does a bit better than the FX-8150 and is the most economical AM3+ CPU. In load the energy consumption of the FX-8350 is notably lower than the FX-8150 with one thread but moves back slightly above it with eight threads. All in all, these readings are reassuring given the higher voltage required by the FX-8350, proof of better handling of current leakage.

There’s a huge gap to the Intel processors and the low energy consumption on the last Core i3-3240 is particularly impressive: less than 23w in load on the ATX12V!

We now move on to our representation of energy efficiency. To get this you have to divide the performance levels obtained in Fritz Chess Benchmark by CPU energy consumption. The only problem is however that it’s impossible to get an exact reading of CPU consumption: the readings at the ATX12V aren’t 100% comparable from one platform to another and the reading at the socket doesn’t allow us to isolate CPU consumption entirely.

We therefore decided to use two methods to look at processor consumption:

- Energy consumption at the ATX12V
- 90% of the difference in energy consumption between load and idle at the socket

We took this at 90% so as to exclude power supply yield. Note that while the first reading favours processors that draw a small proportion of power from the standard ATX socket, the second favours those with high energy consumption at idle. Unfortunately no method is perfect.



[ 220V socket ]  [ ATX12V ]

Even if Fritz is the application that benefits least from the Piledriver architecture as we'll see a little further on, the higher clock combined with energy consumption that remains close to the FX-8150 allows the FX-8350 to be more efficient whether with one or eight threads. With multithreading it is however less efficient than the Phenom II X6s.

The Intel offer is completely out of reach. A Core i5 or i7 Sandy Bridge is around 2x more efficient than the AMD FX-8350 and the Ivy Bridges go even further!


Page 5
Overclocking and undervolting

Overclocking and undervolting
We did of course try to overclock the AMD FX-8350. To recap, the previous AMD FXs are very much at ease with high clocks and hold the record for an x86 processor with 8.7 GHz obtained on an FX-8150. The only thing is that this overclocking was carried out in extreme conditions with just one active module, liquid nitrogen for cooling and no load other than display of a CPU-Z!


Under normal conditions we stopped our tests at 4.6 GHz at 1.34v on the FX-8150, due to excessive energy consumption levels. What do you get out of the FX-8350, still supplied with an ASUS M5A99X EVO and cooled with a Noctua NH-U12P SE2?


For this test we turned Turbo off and checked stability in Prime95 with voltage increased or reduced in 0.05v steps. Its official voltage at 4 GHz is 1.3375v and the motherboard supplies in reality 1.24v when the ASUS EPU energy economy mode is on because of the usual vDrop. For our tests we limited this vDrop by turning the EPU off and putting the CPU Load Line Calibration on High and then Extreme and CPU Current Capability at 130%. We also turned the HPC Mode on in the ASUS bios so as to cater for any drop in clock linked to excessive energy consumption.


Still at a clock of 4 GHz, we first tried to stabilise the processor at a lower voltage than the 1.24v supplied by default in load. With our bios and VID settings configured at 1.25v, a voltage of 1.21v was supplied by the motherboard, which allowed us to reduce very slightly the energy consumption in load in Prime95 at the socket.


We then managed to stablise the clock in Prime95 up to 4.4 GHz with a reasonable increase in voltage and energy consumption (+0.11v and +45w at the socket). Clocking up to 4.5 GHz was however more delicate with energy consumption at +90w and we were only able to ‘stabilise’ the processor at 4.6 GHz by going from LLC to High, giving us a voltage of 1.44v (for reasons unknown when we used a VID that was higher than 1.4125v, half the cores dropped to 1.4 GHz). We use speech marks here because energy consumption was really very high and we didn’t manage more than five minutes in Prime95, with energy consumption starting at 340 Watts and climbing little by little to 363 Watts, with the ATX12V measured at 273 Watts on its own, which is double the TDP of the processor: we chose to cut the test short rather than risk damaging the processor or motherboard.

Based on this test with the FX-8350, overclocking doesn’t really seem to be better with Vishera in normal conditions than it was before and it very much looks as if higher voltages and power stage cooling on the motherboard will be needed. This won’t necessarily be the most economical solution and if you’re after a more practical overclocking you may as well start with a lower clocked model such as the FX-8320. Of course, we’ll have to wait for more overclocking attemps to come in to be sure of elaborating from these results but it is likely that we were using a particularly uncooperative sample, as AMD is talking about clocks more in the order of 5 GHz!


Page 6
Bulldozer vs Piledriver at 4 GHz

Bulldozer vs Piledriver at 4 GHz
Before looking at the performance of the processors in their configurations as you’ll get them from stockists, we wanted to evaluate the gains offered by the Piledriver architecture in comparison to Bulldozer, released a year ago. To do so, we clocked an AMD FX-8150 Zambezi and an AMD FX-8350 Vishera at the same clock of 4 GHz.

We started with the performance of the caches and the memory controller, with readings taken in AIDA64:


While the latency of the caches hasn’t changed, certain speeds are slightly up, especially for L2 cache reads. With the memory, both architectures use controllers running at 2.2 GHz and there has been a slight improvement across the board.
Moving onto more practical figures:


In applications there was an average gain of 7.7% at equal clocks, with an improvement of up to 16.7% in V-Ray but only 1.3% in Fritz Chess Benchmark 4.3. The gain is most significant in games, with an average of no less than 13.5%! In the worst case Vishera has an 8.1% advantage (Skyrim, F1 2012) with a maximum gain of 20.8% in Crysis 2.

Coming on releases just a year apart, these gains are impressive, comparable to what Intel offers with a ‘Tock’ (as was the case between Lynnfield and Sandy Bridge) and more than we had with the Ivy Bridge ‘Tick’ (3-4%).


Page 7
3D rendering: Mental Ray and V-Ray

3d Studio Max 2011 - Mental Ray

We now move on to the practical tests, firstly with a 3D rendering in 3d Studio Max 2011 using the Mental Ray rendering engine on an Evermotion scene. The rendering was carried out at 600*375 so as to retain a reasonable test time.


In this first test the FX-8350 was 18.5% faster than its predecessor, putting the new arrival just behind the Core i7-3770K.
3d Studio Max 2011 - V-Ray 2.0

Still in 3d Studio Max 2011, we changed the engine for the more popular third party engine, V-Ray 2. We used another version of the same scene prepared by Evermotion for this engine, still with a 600*375 rendering. Rendering times are a good deal faster but of course we’re not carrying out a comparison of the engines themselves or the quality of the final files.


With V-Ray the gain was even bigger than in Mental Ray, the FX-8350 giving an additional 23.8% over the FX-8150. This puts AMD at the head of the field, in front of the whole LGA 1155 range.


Page 8
Compilation: Visual Studio and MinGW/GCC

Visual Studio 2010 SP1

We compiled the source code of the 3D Ogre engine in Visual Studio 2010 SP1.


With a gain of 13.8% over the FX-8150, the FX-8350 is positioned between the Core i7-3770K and 2600K.
MinGW / GCC 4.5.2

The same source code was compiled in MinGW / GCC 4.5.2.


In MinGW the gain was lower at 12%, but the FX-8350 is still positioned between the two Core i7s.


Page 9
Compression: 7-zip and WinRAR

7-zip 9.2

7-zip has been added to our test protocol. In contrast to WinRAR, this application is highly multithreaded if its highest performance algorithm, LZMA2, is used. We measured the time required to compress a large volume of files.


Piledriver has less of an impact in 7-zip, though at 8.1% the FX-8350 gain is by no means shabby. The new AMD CPU is slightly faster than the Core i7-2600K, with the i7-3770K retaining a small advantage.
WinRAR 4.01


The same files were compressed in WinRAR using the most demanding RAR algorithm ("Best").


WinRAR doesn’t really use more than two cores for compression and therefore, in spite of a performance gain of 9.1%, the FX-8350 is only able to outdo the i5-3330 and trails the faster Intel processors.


Page 10
Encoding: x264 and MainConcept H.264

StaxRip - x264 build 2085

For video encoding we retained the popular x264, here in build 2216. We used the StaxRip interface to transcode a 1080p file taken from the Avatar Blu-ray using two passes in fast mode with a bitrate of 10 Mbits /s. We’ve posted the times for both passes, the first being less multithreaded than the second and only really exploiting three or four cores.


[ Total ]  [ 1st pass ]  [ 2nd pass ]

Piledriver does have an impact on x264 encoding and the FX-8350 gave a 15.9% improvement over the FX-8150 overall and 16.4% on the second pass. The FX-8350 is faster than the Core i7-2600K and is positioned between the i5 and i7 Ivy Bridge ‘K’ processors. It is in fact handicapped by its performance on the less multithreaded first pass, in contrast to the second where it’s the fastest of the CPUs tested here, something that will be an important consideration for those who encode using just a single pass.
Rovi TotalCode 2.5 H264 Pro

We then moved on to another H.264 codec, from Rovi (previously MainConcept). We used the Rovi TotalCode H.264 interface to carry out the same type of transcoding as in x264. Note that the first pass is more multithreaded here and we have only given the overall score.


As in x264 the FX-8350 makes a big difference, giving a gain of 14.9% over the FX-8150. This puts it second overall, just behind the Core i7-3770K.


Page 11
Photo processing: Lightroom and Bibble

Adobe Lightroom 3.4

We have now introduced photo processing by lot to our protocol. We started by exporting a lot of 96 RAW photos from a 5D Mark II as JPEGs in Lightroom, applying various effects such as colour and lens correction or noise processing.


Lightroom is one of the applications that benefits the least from the FX-8350 with a gain of just 10.5% over the FX-8150. It is pretty much on a par with the Core i7-2600K and Core i5-3570K but is behind the i7-3770K.

In Bibble we processed a lot of 48 RAW photos. Note that Bibble is slower than Lightroom but, as with the rendering engines, we didn’t carry out this test to compare the applications with each other - this would imply comparing the quality of results as a slower export may also be of higher quality.


In contrast to Lightroom, Bibble sees a big boost of 20.9% with the FX-8350. Here AMD had a big deficit to make up on Intel and has only done so partially with a positioning between the Ivy Bridge i5 and i7.


Page 12
Chess AI: Houdini and Fritz

Houdini 2.0 Pro

We finished up our survey of applications with quite a particular choice, namely artificial intelligence algorithms designed for chess. We started with Houdini Pro 2, via the Arena 3 interface. Version 1.5 dominated the top of the chess engine classifications and Version 2 seems destined to do the same. We left the engine running until the 24th move at the beginning of a game and noted the speed in kilo nodes per second.


The FX-8350 is at the head of the field here with a performance gain over the FX-8150 of 18.5%.
Fritz Chess Benchmark 4.3

We now move on to Fritz Chess Benchmarking from Chess Base. Here again the figures are expressed in kilo-nodes per second.


In Fritz, Intel retains the lead however with the difference between the FX-8150 and 8350 of 10.1%. The 8350 comes in just behind the i7-2600K.


Page 13
3D gaming: Crysis 2 and Arma II: OA

Crysis 2 v1.9

The 3D gaming part of this comparative begins with Crysis 2. We used the latest version 1.9 in DirectX 11 and measured the framerate obtained at 1920*1080 Ultra at a precise point in the game during a shoot-out.


In spite of moving over to a GeForce GTX 680, we continue to hit a wall of around 50 fps in our test scene. This shouldn't therefore be due to a GPU limitation, especially as during the Ivy Bridge test we saw gains linked to moving over to DDR3-2133.

The Piledriver architecture works miracles here and the FX-8350 is 24.4% faster than the FX-8150. The new AMD processor moves a long way ahead of what was the fastest AMD processor up until now, the Phenom II X4 980. This puts the FX-8350 more or less at the level of the i5 and i7 processors and a good way in front of the i3s.
Arma II: Operation Arrowhead

In Arma II: Operation Arrowhead we measured the framerate when crossing a village in the first solo mission, still at 1920*1080 and with all options pushed to a maximum, including visibility.


The gain between the FX-8150 and 8350 is less marked in Arma II, but still 13.9%. While this is enough to make the FX-8350 the fastest AMD processor, it can’t stand up to the Intel processors. It is at Core i3-3240 level, some way behind the i5s and i7s.


Page 14
3D gaming: Rise of Flight and F1 2012

Rise Of Flight v1.021b

We used Rise Of Flight, a First World War fighter plane simulator, at 1920*1080 at high graphics settings. In this test we launched a fast mission with a fifteen against fifteen dogfight, with the framerate measured with the back-facing view on the fight between our acolytes and adversaries.


The FX-8350 gives a 15.7% gain over the FX-8150. In general AMD processors struggle in Rise of Flight and the FXs do particularly badly with the Core i3s markedly faster than the FX-8350.
F1 2012

We ran the brand new F1 2011 at 1920*1080 with settings pushed to a maximum. We measured the framerate at the start of the Abu Dhabi GP.


With a gain of 15.7% over the FX-8150, the FX-8350 does better in F1 2012, falling between the Core i3-3240 and the Core i5-3330. It has a more than notable advantage over the Phenom IIs.


Page 15
3D gaming: Total War Shogun 2 and Skyrim

Total War: Shogun 2

For Total War: Shogun 2 we used the huge battle of the 'DX9 CPU' test modified for DX11 at 1920*1080 and with high graphics settings.


The FX-8350 is by far the fastest AMD processor in Shogun 2, with a gain of 20.6% over the FX-8150. Once again, this allows AMD’s latest to compete between the Core i3-3240 and Core i5-3330.
Skyrim

In Skyrim we used a saved game near Faillaise (Riften) and measured the framerate in a place that is quite demanding for the CPU. Graphics options were pushed to max at 1920*1080 (except for anti aliasing) with uGridsToLoad at 7.


In Skyrim the gain is more contained but still worth having: +11.7% for the FX-8350. For the first time, the 8350 isn’t the fastest AMD processor, with the Phenom II X4 980 out front. As in Rise Of Flight, the Intel offer is very definitely superior here.


Page 16
3D gaming: Starcraft II and Anno 2070

Starcraft II

For Starcraft II a major attack during a replay was generously donated by some forum users (thanks!). This replay contained a very (very) full-on attack and we measured its framerate at a resolution of 1920*1080 with all graphics settings pushed to a max.


The FX-8350 gives an impressive 21.7% advantage over the FX-8150 to take over as the no.1 AMD processor. Unfortunately this only puts it on a par with the Core i3-2130.
Anno 2070

Lastly in Anno 1404 we loaded a saved game with a city of 46,600 inhabitants that we partly visualize from a distance. The resolution was 1920*1080 and all graphics settings were pushed to a maximum.


Once again, there was a very big gain over the FX-8150, to the tune of +21.3% here. The FX-8350 is by far AMD's highest performing processor but this only puts it on a par with the Core i3-3240.


Page 17
Performance averages

Performance averages
Although individual app results are worth looking at, we have also calculated a performance index based on all tests with the same weight for each test. We have included two averages, one that’s applied across all the tests with the exclusion of 3D games and the other specific to 3D games.


[ Standard ]  [ By performance ]

While the gain at equal clocks between Vishera and Zambezi is 7.7% on average in applications, between the FX-8350 and FX-8150 it increases to 15% when the difference made by the higher clock is taken into account. This puts the new arrival above the Core i7-2600k, though the i7-3770K is still faster on average. This is very good indeed when you consider that the FX-8350 is cheaper than the i5-3570K at the same time as being a good deal faster in applications!



[ Standard ]  [ By performance ]

In 3D games, there's an even bigger gain (18.1%) for the FX-8350 over the FX-8150. This time the clock has less of an impact because games tend to be less multithreaded and moreover there’s a higher Turbo margin on the FX-8150. To recap, the gain in clock is 13.5%.

Unfortunately for AMD, while the FX-8350 now heads up its range in games with the FX-8150 behind the Phenom II X4 980, it struggles to match the Intel offer - the FX 8350 is only positioned between the Core i3-2130 and 3240. The Core i5s are clearly out of reach.

Of course we looked for cases in our tests where gaming performance was limited by the CPU rather than the GPU, this in spite of using high graphics options and a 1920*1080 resolution. Naturally, by using scenes that are less demanding for the CPU and/or pushing graphics settings up further to impact solely on the GPU, so as no longer to allow us to attain framerates made possible by the highest performance CPUs (via AA or resolution… or by reducing GPU power!), it would be possible to reduce the gaps between CPUs.


Page 18
Conclusion

Conclusion
Does, the AMD FX-8350, flagship of the new AMD FX range, mark AMD’s return to form? There has been much progress since the launch of the FX-8150, though it has to be said that this isn’t saying all that much. In terms of performance first of all, there have been gains across the board, particularly so in games where last year’s FX range struggled to outdo the Phenom IIs. Price wise, AMD has positioned itself more advisedly. While the FX-8150 was launched at $245, the FX-8350 comes in at $195.


Compared to the Intel offer and taking applications performance into account, AMD is thus very aggressively placed in terms of price/performance ratios. When it comes to applications, the FX-8350 is on a par with the Core i7 LGA 1155 processors (the LGA 2011s are out of reach), which cost a minimum of $294 in their standard version and $317 for the K versions on the Intel price list! This pricing has moreover been frozen since the launch of the Sandy Bridges at the beginning of 2011, in spite of the fact that the Ivy Bridges are smaller as a result of Intel’s considerable technological fabrication process advances.

Unfortunately, while the FX-8350 does give notable gains over the previous FX range, it is still quite some way behind the Intel processors in games, with the Intels able to do without the eight heavy threads required for the proper use of the FXs. While on a level with an i7 in applications, the FX-8350 is more on a level with an i3 in games, which is certainly enough most of the time but makes this processor a good deal less attractive if your main usage of the machine is for games.

The other main black spot for the FX range is of course energy efficiency, which is between 2 and 2.5x higher with the Intel range. AMD hasn’t made much progress here since the Phenom IIs, which really is poor when you consider that it has reduced its manufacturing process from 45 to 32nm over this period! Intel’s advantage here isn’t only –marginally— an electricity bill issue but is also a factor in allowing a quiet configuration in load.


There will however be widespread approval for the fact that AMD is offering a range with the multiplier unlocked across the board, this while with Intel's it is entirely locked on the i3s, partially unlocked on the i5s/i7s and unlocked on the ‘K’ versions, which remain expensive. An unlocked multiplier doesn’t necessarily translate to a good overclocking margin and given the modest clock reached on our processor (4.5 GHz stable), it will be better to look to processors with a lower starting clock than 4 GHz, such as the FX-8320 and 6300. Note however, our test sample was probably particularly uncooperative.

Finally, we come to the question of platforms. On the Intel side we know already that LGA1155 will make way for LGA 1150 and the new Haswell architecture in the second half of 2013, no doubt increasing Intel’s energy efficiency advantage. But what about AM3+? AMD hasn’t yet developed the platform with a chipset supporting PCI-Express 3.0 and USB 3 and hasn’t yet said officially whether Steamroller, the next development of its architecture, will be rolled out on anything other than an FM2 Socket APU. Shame!


Copyright © 1997-2014 BeHardware. All rights reserved.