AMD Vishera FX-8350 Review

formerstaff - 2012-10-09 20:40:14 in CPU's
Category: CPU's
Reviewed by: formerstaff   
Reviewed on: October 22, 2012
Price: $195


About this time last year AMD unveiled what was the highly anticipated and first significant CPU architecture change in over a decade. Big things were expected from this bold new architecture approach from AMD, if for no other reason than Bulldozer had seemingly been sitting in the oven for so long and the expectations by AMD itself was set very high with the PR campaign accompanying it. The result was a resounding thud as the first incarnation of the K15 architecture did not live up to the press slides promising things like 1.6x the performance of CPUs that had an 'i' in front of their titles.

The release of the second generation codenamed "Vishera" has seemed to be much quieter than the first. This time around there will be only be four Vishera SKUs released: the FX-8350, FX-8320, FX-6300, and FX-4300. The part we are concerned with today is the new flagship of the Vishera lineup, the FX-8350 or "Piledriver" CPU. AMD had promised from the open volley of the new architecture that was Bulldozer, there will be a 10-15% performance increase with each of the four new generations. With the release of Vishera, AMD has said that there are improvements to the IPC, power efficiency, frequency, as well as architecture changes we will get into a little further along.

So let's see how all of these changes to the second generation have fared compared to Bulldozer and the Intel competition it is aimed at. And speaking of who this new flagship is aimed at, AMD makes no bones about it being aimed squarely at the high end enthusiast. As I am one of those, let's see if it hit the mark.











Closer Look:

Built on the same 32nm process as the previous generation, the FX-8350 is the enthusiast flagship of the Vishera offerings sporting the first ever 4.0GHz stock clocked CPU on the market. The FX-8350 is equipped with four Piledriver modules and eight cores in an 125W TDP. While the base clock speed is 4.0GHz, the Turbo will take this up to 4.2GHz when load dictates. The 315mm² die packs 1.2 billion transistors and an IMC that supports 1866MHz memory natively in dual channel mode. The FX-8350 has 8MB of L2 cache as well as 8MB of shared L3 cache



The chipset used for the Vishera is the same 990/970 used with the last generation. Good news for AMD fans in that your AM3+ 990FX motherboards will still provide a home for your new FX Vishera chip with only a BIOS update. The block diagram shows PCIE 2.0 lanes broken down as 2x16 or 4x8 +8x1+4x1 (2x1 on SB). Up to 14 USB ports are supported (up to four being USB 3.0) and six 6Gbps SATA ports and support for RAID configurations of 0, 1, 5, and 10. DDR3 officially supports up to 1866MHz of 1.5v, but supports up to 2400MHz (OC). I won't dwell on this too much as it's largely the same as the last release. So as a refresher course, here is a block diagram for your perusal.


AMD has made a number of improvements to the "Piledriver" cores it claims will add and enhance improvement to not only overall performance, but in the areas of IPC and operating frequency. Among them are a larger L1 TLB to improve virtual address translation speed, improved branch prediction, an improved HW data prefetch, and have improved upon the item that has been giving most of the blame for the slow single threaded performance of the last generation the FPU and INT scheduling. AMD has also claimed that it has made improvements to the efficiency of the shared L2 cache. The FX-8350 is identified as having C0 stepping as opposed to the FX-8150's B2 stepping. These and other improvements are adding up to some rather substantial improvements they claim.



Well some are saying that all this is a rework of the Piledriver core (module) some are saying it is nothing more than a stepping revision. The proof is in the numbers. So turn the page and have a look at the specifications and features, and then we will heat this chip up and see what it has to offer.



Processor Model
CPU Base
CPU Max Turbo
L2 Cache
L3 Cache
Max DDR 3





Well thats the Vitals on the hardware. Time to plug it in and pour the coals to AMD's newest enthusiast offering.


Testing the AMD FX-8350 Processor will involve running it and its comparison products through OCC's test suite of benchmarks, which include both synthetic benchmarks and real-world applications. The gaming tests will also consist of both synthetic benchmarks and actual game play, in which we can see if similarly-prepared setups offer any performance advantages. The system will receive a fully updated, fresh install of Windows 7 Professional 64-bit edition, in addition to the latest chipset drivers for each board and latest AMD Catalyst drivers for the XFX HD R7970. To ensure as few variables as possible, all hardware will be tested at their stock speeds, timings, voltages, and latencies, unless otherwise stated. Turbo Boost will be disabled on all processors to make a fair comparison without skewing results. After stock speed testing, each processor will then be over-clocked as much as possible, while still maintaining full stability.


Testing Setup: AMD 


Testing Setup: Intel Socket 2011


Testing Setup: Socket 1155


Comparison CPUs:



Overclocked settings:


Overclocking the FX-8350 was as easy as overclocking gets. I chose the "Ultra High" setting in the BIOS for the Load Line Control of the Gigabyte UD7, turned off Turbo Core and all energy management features (C1, Cool 'N' Quiet, C6 support etc), and I simply kept raising the unlocked multiplier of the FX-8350 with intermittent stress testing and voltage changes. It is apparent that CPUID needs an update for the Vishera line as the voltage did not match the BIOS readings. I ended up at 5200MHz at 1.520. Temperatures stayed in check throughout never getting above 65 °C according to HW monitor. For those of you who are not comfortable in the BIOS or would like to try out setting before committing to them, AMD has a brand new AOD version 4.2.3 for you to take advantage of or you can simply let AMD Turbo Core do it for you.




Maximum Core Clock Speed:

Each CPU has been tested for stability at the listed over-clocked speeds. These clock speeds will represent the level of performance shown by the over-clocked scores in the testing.




  1. Apophysis
  2. WinRAR
  3. Geekbench 2.1
  4. Office 2007 Excel Number Crunch
  5. POV-Ray 3.7
  6. Bibble 5
  7. Sandra 2012
  8. AIDA64 2.60
  9. HandBrake .9.8
  10. ScienceMark 2.02
  11. Cinebench 10 & 11.5
  12. HD Tune 4.60
  1. Metro 2033
  2. DiRT 3
  3. Battlefield 3
  4. 3DMark 11

Well we have our lineup and maximum stable overclocks. Time to see what results we get from the second generation Vishera


The first part of our testing will involve system-specific benchmarks.


Let's get started with Apophysis. This program is primarily used to render and generate fractal flame images. We will run this benchmark with the following settings:



The measurement used is time to render, in minutes, to completion.











Lower is Better


WinRAR is a tool that archives and compresses large files to a manageable size. Here, we will test the time needed to compress files of 100 MB and 500 MB. Time will be measured in seconds.




Lower is Better





Lower is Better



Geekbench 2.1 is a benchmark that tests CPU and memory performance in an easy-to-use tool. The measure used for comparison is the total suite average score.


Higher is Better


Bibble 5:

This test consists of converting one hundred 8.2 MP RAW images to JPEG format. The original file size is 837 MB. The measure used for comparison is time needed to convert the file, in seconds.


Lower is Better


We can start to see an emerging pattern of improvement over the last generation Bulldozer. Some of the improvements are running as high as 25%.


Office 2007 Excel Big Number Crunch: This test takes a 6.2 MB Microsoft Excel spreadsheet and performs about 28,000 sets of calculations that represent many of the most commonly used calculations in Excel. The measure of this test is the amount of time it takes to refresh the sheet.

















Lower Is Better


POV-Ray 3.7: This program features a built-in benchmark that renders an image using Ray Tracing. The latest versions offer support for SMP (Symmetric Multi-Processing), enabling the workload to be spread across several cores for quicker completion.


Higher Is Better


ProShow Gold: This program is used to take a collection of images and stitch them together in a slide show, using a variety of transitions and effects, to make a compelling show you can share with friends and family. The workload consists of 29 high-resolution images that are stitched into a 3 minute video file.


Lower Is Better


HandBrake .9.5: is an open source application used to transcode multiple video formats to an h.264 output format. The test file size is 128 MB in size and 43 seconds in length.


Lower Is Better


A strong showing for the FX-8350 coming in second only to the 3960 in PovRay and Handbrake.


SiSoft Sandra is a diagnostic utility and synthetic benchmarking program. Sandra allows you to view your hardware at a higher level to be more helpful. For this benchmark, I will be running a broad spectrum of tests to gauge the performance of key functions of the CPUs.
















Processor Arithmetic


Multi-Core Efficiency



Memory Bandwidth



Cache and Memory




Power Management Efficiency



AIDA64 Extreme Edition is a software utility designed for hardware diagnosis and benchmarking. I will be using the CPU Queen test that looks for the solution to the "Queens" problem on a 10x10 chessboard. This tests the branch prediction capabilities of the processor. The FPU Mandel test measures double precision floating point performance through the computation of several frames of the "Mandelbrot" fractal.


Higher is Better


Sandra benches are not kind to AMD CPUs, however the trend of improvement can be seen throughout the testing.


ScienceMark tests real-world performance instead of using synthetic benchmarks. For this test, we will run the benchmark suite and use the overall score for comparison.





















Higher is Better!




Cinebench 10 is useful for testing your system, CPU, and OpenGL capabilities using the software program, CINEMA 4D. We will be using the default tests for this benchmark.





Higher is Better

Cinebench 11.5



Higher is Better


HD Tune measures disk performance to easily make comparisons between drives or disk controllers.





Higher is Better





Lower is Better


PCMark 7 is the latest iteration of Futuremark's popular PCMark system performance tool. This latest version is designed for use on Windows 7 PCs and features a combination of 25 different workloads to accurately measure the performance of all PCs, from laptops to desktops.


Higher is Better


A nice 19% stock gain in Cinebench 10 single threaded performance, and the first 9+ score from a FX processor in Cinebench 11.5; not too bad at all.


Part first-person shooter, part survival horror, Metro 2033 is based on the novel of the same name, written by Russian author Dmitry Glukhovsky. You play as Artyom in a post-apocalyptic Moscow, where you'll spend most of your time traversing the metro system, with occasional trips to the surface. Despite the dark atmosphere and bleak future for mankind, the visuals are anything but bleak. Powered by the 4A Engine, with support for DirectX 11, NVIDIA PhysX, and NVIDIA 3D Vision, the tunnels are extremely varied – in your travels, you'll come across human outposts, bandit settlements, and even half-eaten corpses. Ensuring you feel all the tension, there is no map and no health meter. Get lost without enough gas mask filters and adrenaline shots and you may soon wind up as one of those half-eaten corpses, chewed up by some horrifying manner of irradiated beast that hides in the shadows just waiting for some hapless soul to wander by.













A trend noticed throughout the gaming benches, the new Vishera keeps up with the heavy hitters from the competition. Although to be fair, you will not find a more GPU dependent game than Metro 2033. It literally does not care if your CPU is running at 2.8GHz or 5GHz. In some of my own testing however, it does better with multiple GPU setups as well.


DiRT 3 is the third iteration of this series. Published and developed by Codemasters, this game uses the EGO 2.0 game engine and was released in the US on PC in May of 2011.
















The FX-8350 paired with an HD 7970 keeps pace with the top end Intel competition.


Battlefield 3 is a first-person shooter video game developed by EA Digital Illusions CE and published by Electronic Arts. Battlefield 3 uses the Frostbyte 2 game engine and is the direct successor to Battlefield 2. Released in North America on October 25, 2011, the game supports DirectX 10 and 11.


















Once again the FX-8350 keeps pace with the heavy hitters from Intel.


3DMark 11 is the next installment in Futuremark’s 3DMark series, with Vantage as its predecessor. The name implies that this benchmark is for Microsoft DirectX 11 and with an unintended coincidence, the name matches the year proceeding its release (which was the naming scheme to some prior versions of 3DMark nonetheless). 3DMark 11 is designed solely for DirectX 11, so Windows Vista or 7 are required along with a DirectX 11 graphics card in order to run this test. The Basic Edition has unlimited free tests on performance mode, whereas Vantage is only allowed for a single test run. The advanced edition costs $19.95 and unlocks nearly all of the features of the benchmark, while the professional edition runs $995.00 and is mainly suited for corporate use. The new benchmark contains six tests, four of which are aimed only at graphical testing; one to test for physics handling and one to combine graphics and physics testing together. The open source Bullet Physics library is used for physics simulation and although not as mainstream as Havok or PhysX, it still seems to be a popular choice.

With the new benchmark, comes two new demos that can be watched, both based on the tests. Unlike the tests, however, these contain basic audio. The first demo is titled "Deep Sea" and involves a few vessels exploring what looks to be a sunken U-Boat. The second demo is titled "High Temple" and presents a location similar to South American tribal ruins with statues and the occasional vehicle around. The demos are simple in that they have no story – they are really just a demonstration of what the testing will be like. The vehicles have the logos of the sponsors MSI and Antec on their sides – the sponsorships helping to make the basic edition free. The four graphics tests are slight variants of the demos. I will use the three benchmark test preset levels to test the performance of each card. The presets are used as they are comparable to what can be run with the free version, so that results can be compared across more than just a custom set of test parameters.













The FX-8350 does well here, leading the way in half of the tests. It would be nice to see the physics score brought up a bit more however.


When Bulldozer was released a year ago, it had a problem that was almost bigger than the single threaded performance and that was broken expectations. After five or so years of delays and hearing about the forthcoming new architecture from AMD, reviewers everywhere put the likes of SuperPi to it and it was all over but the crying. The debate went on about what was exactly wrong with the bold new architecture, and it has been largely written off by many "experts" as un-fixable.

Needless to say I have been eagerly anticipating getting to review the second incarnation of this new architecture to see for myself if it was viable and I have to say that I believe that it is a rousing success. The FX-8350 shows improvement on every front that I have tested it on. AMD has said from the onset that this is a "forward thinking " architecture and that idea is becoming increasingly more difficult to dismiss as subterfuge for poor single threaded performance. As it has always been, software coding follows hardware capability and the number of softwares that use multiple threads increases exponentially every year. It would not make sense from a competitive standpoint not to encode this way.

Whether the FX 8350 is considered to be merely a newly stepped version, or an architectural overhaul of the original Bulldozer release makes no difference to me. It's faster, runs lightly threaded apps better, uses less energy, and overclocks like a (insert your own expletive). As you paged through the benchmarks I ran for this review, you noticed that the new AMD flagship moved up the bar graphs a good bit, and in some cases coming in second only to the almighty i7-3960X. As software vendors start to code for this forward thinking architecture the performance disparity between Intel and AMD's latest should be minimized as it does perform better in software that is able to utilize the strengths of the processor.

AMD has taken a distinctly different path with the release of the Piledriver than it did with its predecessor. Last time there was no shortage of good looking company slides promising that the performance would be 1.6x the performance of some processor with an i7 in front of it, and it left people scratching their heads as to what that reference was to, and trying to locate that specific benchmark program. This time AMD seems to have done things a lot differently. Relatively little was heard or seen in the way of promises, but what showed up in that little green clam shell definitely delivered the performance promise of a year ago.