Original Link: http://www.anandtech.com/show/8316/amds-5-ghz-turbo-cpu-in-retail-the-fx9590-and-asrock-990fx-extreme9-review

While AMD’s FX-9590 CPU has been in systems for over a year, it suddenly comes to market as a retail package for end-users to buy with a bundled liquid cooling system. This 220W CPU that has a turbo speed of 5.0 GHz still sits at the top of AMD’s performance stack, despite subsequent improvements in the architecture since. We have decided to grab ASRock’s 990FX Extreme9 and an FX-9590 for a review to see if it still is the AMD performance CPU champion.

Spot the CPU

The story behind AMD’s fastest ever x86 CPUs is slightly odd. Two models, the FX-9590 and FX-9370, were both launched into OEM channels in June 2013. Being an OEM component, the only way to get one was in a pre-built system through a retailer, or through a bulk system integrator that had a model around one of these CPUs. Typically this is a process that is only exhibited with server class processors: from a range of CPUs being produced, only several will be available for end-users at retail because server CPUs usually go through a system builder. At the time, it seemed that AMD concerned that the high TDP of this CPU, at 220W listed, is too much for most cooling setups within a home user system and the best way to get it to consumers would be if a system builder chose the appropriate cooling for them.

As a result of this orientation of sales, AMD did not sample the media with review units. We review an AMD product typically though an AMD sourced sample. It was also noted that the OEM price for the CPU was near $900 for the FX-9590, which seemed like an excruciating amount for what was essentially a good overclocking version of the FX-8350. Several media websites were able to collaborate with system builders in order to get a chance to review the CPU, and AMD was confident in their promotion and handling of the new CPU.

Anecdotally, in my field of vision, the promotion of this CPU was relatively limited. The price was the main factor, resulting in comparative AMD/Intel systems being more power hungry on the AMD side, and substantially more expensive when put up against the latest mainstream i7 at the time. As a result, while some retailers were selling the OEM CPU at full price, some retailers decided to sell their OEM stock with a severe price cut directly to consumers, down from $900 to $390, in order to get rid of units (this is when I picked up our sample).

Due to the OEM nature of these sales to end-users, each CPU had either no warranty with AMD or a limited warranty. For the user interested in a 3-year system cycle without the fear of a bad egg, the OEM route is never a positive one.

AMD subsequently released, relatively silently, a proper package and retail version of the FX processors. It was apparent that this was in response to the OEM sales, with the retailers list ‘heatsink and fan not included’ alongside the specifications.

AMD FX CPU Comparison
Release Date April
October 2012 October 2012 October 2012 June 2013 June 2013
Modules 2 3 4
L1 Cache (Code) 128 KB 192 KB 256 KB
L1 Cache (Data) 64 KB 96 KB 128 KB
L2 Cache 4 MB 6 MB 8 MB
L3 Cache 8 MB
TDP 125W 220 W
Base Frequency 4200 3900 3500 4000 4400 4700
Turbo Frequency 4300 4200 4000 4200 4700 5000
Core Name Vishera
Microarchitecture Piledriver
Socket AM3+
Memory Support DDR3-1866
Price (US) $140 $140 $160 $190 $230 
$300 CLC
$370 CLC

Since that release, AMD has not upgraded their enthusiast processor line with the latest architecture. The FX line has stayed where it is, perhaps for a number of reasons. One could speculate that releasing the next generation of FX-85xx might put them behind the FX-9590 in performance, or that the fabrication process was not suitable for a quad-module CPU with the new architecture improvements. The FX line for desktops, as far as we know, is staying at 32nm with no improvements.

Now Available

Fast forward twelve months to June 20th 2014 and Roy Taylor, AMD’s VP of Global Channel Sales tweets this innocuous picture:

Speculation was rife as to what this was. Here is a large box for an FX processor with the words ‘with Liquid Cooling System’ underneath. AMD supplied liquid cooling to the media when we reviewed the FX-8350 CPUs, the main CPU that sits underneath the FX-9590 and FX-9370, so there was an expectation that was something new.

On June 26th, the @AMDFX twitter account posted the following, confirming that this was the older FX-9590 but in a retail box with retail cooling:

The AMD FX Processor page has been updated accordingly, showing the same render of the new box. Here we see that the liquid cooler is supplied by Cooler Master, and uses a wide range PWM fan as part of the package.

Of course, this leaves several questions unanswered: how much, when is it on sale, where is it on sale, and is it still any good?  Well for the US at least, it is on sale today from Newegg at $370 with the water cooling kit, or $330 without. NCIX has it listed for CAD$500, although this is currently in ‘back stock’ mode.

The SKU to look for is the FD9590FHHKWOX, which in the UK does not seem to be on the shelves as of yet. Amusingly, when this is typed in to Google, the search engine asked me if I meant FD9590FHHKWOF, the non-CLC version.

This Review

Back when the FX-9590 was originally released alongside the FX-9370, we were unable to secure a sample from AMD and the limited availability made us feel the CPU had a fairly limited scope for testing. However, now the landscape has changed. There has been no new FX CPUs on the market from AMD, and this subsequent release of a retail version piques the interest as to how relevant AMD still sees their high-frequency part. Because I now have a FX-9590 all of my own to test from when the OEM stock was sold, I felt it was worth revisiting to see if it can be considered an investment.

Alongside testing this CPU, the 220W TDP requires a substantial motherboard to match. Due to the age of the platform, the AM3+ socket and the old 990FX chipset, finding a motherboard can be rather tricky. Many of the AM3+ motherboards that were launched were only suited for the FX-8350 processors, which had a 125W TDP. This is yet another reason that AMD wanted the FX-9590 in the hands of system builders who would chose high end motherboards that could cope.

Two of the newest motherboards to be released for 990FX were the ASRock 990FX Killer and the ASRock 990FX Extreme9. We reported the release of the Killer in December 2013, but the Killer is unsuitable here as the specification sheet lists processors up to 125W only. The Extreme9 is ASRock’s high-end AM3+ motherboard, and more suited to the task. Luckily I had requested a sample almost a year ago for some regression testing, so we will be reviewing this motherboard as part of this article. 

For new users to the PC industry, or those that migrated towards newer APU platforms, it is worth going back and recalling the AM3+ socket with the 990FX chipset. When the platform was released, it offered several advantages that Intel lacked at the time: a full set of SATA 6 Gbps ports was the main advantage which took Intel another two generations to offer. The chipset, with the right CPU, also offered substantially more PCIe lanes than the mainstream Intel parts which were similarly priced. While the user could have sixteen PCIe 2.0 lanes from an Intel CPU for graphics coupled with eight PCIe 2.0 lanes from the chipset, AMD users had 32 PCIe 2.0 lanes from the CPU for graphics, another six PCIe 2.0 x1 lanes for controllers and four PCIe 2.0 x1 lanes from the chipset. This gave the AMD motherboard manufacturers more bandwidth to add extra ports or adjust their PCIe layout for graphics. Note that this is the latest AMD platform to support SLI, rather than the newer FM1/FM2 platforms that do not.

There are a few limitations on the 990FX chipset worth mentioning. When this motherboard we are testing today was released, PCIe 3.0 was gaining momentum. The only way to add PCIe 3.0 to these motherboards was to integrate a PLX chip between the Northbridge and the GPUs which gave PCIe 3.0 capabilities between the GPUs, but it still limited data transfer between the PLX chip and the CPU to PCIe 2.0. The other limitation was one of cost. AMD platforms have historically been low cost markets, at least for end users, which correlates to a reluctance to expand spending on motherboards. This reduces the market for high end motherboard solutions which might incorporate extra features and controllers, and as a result many AM3+ motherboards were aimed at price/performance rather than feature set.

The ASRock 990FX Extreme9 sits near the top of the stack for feature set, and currently retails for $170. To put that into perspective, we discuss $170 motherboards for Intel’s latest chipsets as a mid-range point rather than the high end.

Also worth noting that because our last 990FX reviews were with the FX-8150 processor, in order to compare to historical data we also used the Extreme9 with the FX-8150 for comparison points.

ASRock 990FX Extreme9 Overview

Visual Inspection

Taking the motherboard out of the box for the first time and there are several items worth noting. Firstly the extended heatsink which covers the 12+2 phase power delivery to the side of the socket and the North Bridge just below the socket. ASRock has placed all the power delivery chokes in a line, and uses a CHIL8328 IC for a digital design. This PWM controller powers 6-8 phases, and thus the system uses multiplexing to get the desired 12 for the CPU voltage.

The socket area has four fan headers within immediate reach – two CPU and one chassis header directly above the socket, and a 3-pin PWR header to the top right of the DRAM slots. The other two fan headers on the motherboard are located at the bottom, one to the left of the 2-digit debug and the other to the right of the power/reset buttons. The socket area uses a low heatsink profile combined with a gap to the DRAM that should allow for large air coolers to be used.

The DRAM slots use double sided latch mechanisms and there is ample space to the first major PCIe slot. At the bottom of the DRAM slots, to the right, are two USB 3.0 headers from an Etron EJ188H controller. At the time this motherboard was made, a chassis may have had one USB 3.0 header, and thus ASRock also includes a USB 3.0 panel in the box.

Underneath this are eight SATA 6 Gbps ports, the top two from an ASMedia ASM1061 controller followed by six from the south bridge. In this situation, with an additional controller, I would have preferred if ASRock had used a different color for the ASMedia ports. Below these is a fan header, the power and reset buttons, and the two-digit debug.

At the bottom of the motherboard is an IEEE1394 header, a COM header, two USB 2.0 headers, an IR header and the front panel headers. Above this is the odd PCIe layout, which combines the 32 PCIe 2.0 lanes from the north bridge with another PCIe 2.0 x4.

The layout is such that the top PCIe slot is an x16, followed by an x1. The second full length slot is a PCIe 2.0 x4, followed by another PCIe 2.0 x16. This slot shares bandwidth with the bottom PCIe slot, whereby if the bottom is populated, they both go to PCIe 2.0 x8. There is also a PCI slot near the bottom.

This means:

Configuration 1: x16/x1/x4/x16/PCI/-
Configuration 2: x16/x1/x4/x8/PCI/x8

Having this layout allows the user to equip the board with three GPUs in the first, third and fourth full length slots. If they are double slot cards, the PCIe 2.0 x4 is left vacant for a sound card, network card, RAID card or other PCIe device. It is worth noting that in terms of audio, ASRock has equipped this motherboard with a Realtek ALC898 codec.

The rear panel uses two PS/2 ports, one for mouse and one for keyboard, followed by a ClearCMOS button and SPDIF outputs. The four blue USB 3.0 ports are powered by another Etron EJ188H controller, and the panel also has four USB 2.0 ports with two eSATA 6 Gbps ports. The IEEE1394 port on the rear is perhaps one of the last consumer oriented motherboard to have this port pre-installed. The Intel NIC on the rear IO might seem a little strange on an AMD motherboard, but this is one of the top line 990FX solutions. The audio jacks round off the set.

Board Features

Board Features
Price US (Newegg)
Size ATX
CPU Interface Socket AM3+
Chipset 990FX + SB950
Memory Slots Four DDR3 DIMM slots supporting up to 32 GB
Up to Dual Channel, 1066-2450 MHz
Video Outputs None
Onboard LAN Intel 82583V
Onboard Audio Realtek ALC898
Expansion Slots 3 x PCIe 2.0 x16 (x16/x16/- or x16/x8/x8)
1 x PCIe 2.0 x4
1 x PCIe 2.0 x1
1 x PCI
Onboard SATA/RAID 6 x SATA 6 Gbps (Chipset), RAID 0,1,5,10
2 x SATA 6 Gbps (ASMedia ASM1061)
USB 3.0 4 x Rear USB 3.0 (Etron EJ188H)
2 x USB 3.0 Headers (Etron EJ188H)
Onboard 8 x SATA 6 Gbps Ports
2 x USB 3.0 Headers
2 x USB 2.0 Headers
6 x Fan Headers
1 x COM Header
Power/Reset Switches
Dr. Debug LED
Front Panel Connector
Front Audio Connector
Power Connectors 1 x 24-pin ATX
1 x 8-pin CPU
Fan Headers 2 x CPU (4-pin, 3-pin)
3 x CHA (4-pin, 2x 3-pin)
1 x PWR (3-pin)
IO Panel 1 x PS/2 Keyboard
1 x PS/2 Mouse
4 x USB 2.0
4 x USB 3.0
2 x eSATA 6 Gbps
1 x IEEE1394
1 x Intel NIC
Clear CMOS Switch
Audio Jacks
Warranty Period 3 Years
Product Page Link

If we were making a high end AMD motherboard for 2014, the extra lanes would be perfect for PCIe storage. Pile on a SATA Express and M.2 x4 slot without losing lanes to other functions. Some native USB 3.0 would be nice, or 3.1 via controllers. PCIe 3.0 support would be a must of course, along with a Realtek ALC1150 or more advanced audio codec. ASRock’s latest motherboards have featured a water-proof/superhydrophobic coating, or small LCD panels to aid in overclocks, which might be something in AMD meets 2014.

ASRock 990FX Extreme9 BIOS

For those that have followed our Z87 and Z97 BIOS coverage, the ASRock BIOS for the 990FX Extreme9 falls very much on the side of Z87, which makes sense as the platform will probably not be seeing many updates, if any, for the foreseeable future. The main screen uses white text on a starry background, with the selected option very clearly defined.

In this main screen we have details of the motherboard name, the BIOS version, the CPU installed, the CPU Speed, the DRAM installed, the per-module density and the current speed of each module. Other information that would be good to have includes the main system voltages, temperatures and fan speeds. Note that ASRock has not yet implemented a system that contains this information across every BIOS screen, unlike some of their competition. On this screen is also an ‘Active Page on Entry’, useful for overclockers that want the BIOS to skip straight into the OC Tweaker menu.

The OC Tweaker menu, as the name suggests, has the overclocking options. ASRock like to add in some automatic overclock options, and here we get the choice from a 5% to a 50% overclock:

Typically the higher overclocks would be outlined in red to indicate an extreme overclock, but ASRock does not do that here. The rest of the OC Tweaker menu has the CPU configuration settings, DRAM Timing options and Voltage configuration. Load Line Calibration is a single option in the voltage section, and the DRAM Timing Control is a separate menu item for memory sub-timings:

The Advanced tab contains options for enabling/disabling controllers, adjusting SATA modes and implementing some power saving features. The Tool menu next is where ASRock has their more interesting features, such as System Browser that gives an overview of the motherboard and tells the user what is installed:

The Online Management Guard feature displays a timetable for the week and allows the user to disable the network features. One of the major purposes of this feature is for adults to restrict their children’s use of the internet at odd hours of the day.

The Dehumidifier function keeps fans enabled after the system has switched off in order to equilibrate the temperature inside the case with the temperature outside the case. In the event of warm, moist air in the case that cools in the evening, in humid climates this may cause condensation, hence this feature. There are also menu options to save user profiles within the BIOS underneath this setting.

The Fan Controls are how ASRock used to do them, in terms of ‘levels’, with the higher the level indicating how quickly the CPU fan power would ramp up. Some of the headers have the option of tying the fan power to the CPU temperature, meaning that if the CPU temperature rises above a value, the fan is placed on full speed until the target temperature is reached.

The BIOS finishes up with boot options and security. One missing feature from ASRock BIOSes is a boot override feature, allowing for a one-time boot from within the BIOS.

ASRock 990FX Extreme9 Software

The software for the Extreme9 is centered around the ASRock Extreme Tuning Utility, or AXTU for short. AXTU used to be the main interface, until A-Tuning came about for Intel’s 8-series motherboards, but due to the age of the motherboard (or perhaps incompatibilities) this software has not been upgraded.

The screen that pops up first with AXTU is the Hardware Monitor interface. This shows the CPU speeds, base frequencies and multipliers, as well as the fan speeds, temperatures and voltages. Two key bits of information missing are the CPU installed and the motherboard which is being used.

AXTU also is part of the fan control:

The fan controls here are similar to that in the BIOS, relying in the ‘level’ system to determine the gradient at which the fan speed is ramped up. Users can select a target temperature which will put the fan at full speed when the CPU is above the target.

The overclocking interface offers several simple options, although no automatic overclocking setups. Here we can adjust the base frequencies, the multipliers and the voltages, although nothing along the lines of load-line calibration.

AXTU also provides an interface to save the BIOS and some power saving features, but also XFast RAM. This feature allows the user to partition off some of the memory into a RAMDisk, and AXTU will also manage some caching options in order to speed up some of the operating system functions.

While ASRock has dropped XFast USB with the latest Intel 9-series motherboards, the software is here on the Extreme9 and implements newer USB protocols to increase peak transfer speeds at the expense of latency. In our copy tests, the XFast feature does provide a tangible benefit, although the feature is limited to one USB device at a time.

XFast LAN is a licensed and skinned version of cFos, which allows for software manipulation of network priority.

ASRock 990FX Extreme9 In The Box

When I started reviewing motherboards for AnandTech, the motherboard industry was on the verge of USB 3.0 being this new feature, only available through controllers. At the time, due to the difference between USB 2.0 and USB 3.0 headers, cases had not implemented USB 3.0 and the motherboard manufacturers had to add in extra bits and bobs to their motherboards so users could feel the benefit of internal connectors. As time progressed, these extra add-in interfaces migrated into the chassis industry, with most chassis supporting one USB 3.0 header. This relegated any USB 3.0 add-on in the motherboard to those motherboards with two USB 3.0 headers. It is at this point that the 990FX Extreme9 was made, and due to the Extreme9 nomenclature, this bundle is meant to epitomize ASRock’s in-the-box offerings.

In the box we get:

Driver DVD
Rear Panel Shield
USB 3.0 front panel
Six SATA Cables
Two SLI Cables

The Extreme9 is also one of the last AMD motherboards to support SLI, so the motherboard manufacturers also had to include SLI bridges. I still think the USB 3.0 panel is a good idea for motherboards with two USB 3.0 headers, even today.

ASRock 990FX Extreme9 Overclocking

Experience with ASRock 990FX Extreme9

Overclocking with AM3+ CPUs brought back a small wave of nostalgia. Here we are back at 200 MHz base frequency, and have to deal with 0.5x multipliers. For our testing, we overclocked the FX-8150 CPU similar to our previous 990FX motherboards in order to get consistent data.

Overclocking the 990FX Extreme9 in the BIOS is relatively straightforward, although the level of automatic options in the software is disappointing. We enabled a high load-line calibration, started at 20x200 MHz (4.0 GHz) with 1.200 volts set in the BIOS. Our sample hit 4.6 GHz at 1.325 volts, with +132W power draw over stock.


Our standard overclocking methodology is as follows. We select the automatic overclock options and test for stability with PovRay and OCCT to simulate high-end workloads. These stability tests aim to catch any immediate causes for memory or CPU errors.

For manual overclocks, based on the information gathered from previous testing, starts off at a nominal voltage and CPU multiplier, and the multiplier is increased until the stability tests are failed. The CPU voltage is increased gradually until the stability tests are passed, and the process repeated until the motherboard reduces the multiplier automatically (due to safety protocol) or the CPU temperature reaches a stupidly high level (100ºC+). Our test bed is not in a case, which should push overclocks higher with fresher (cooler) air.

Manual Overclock:

Overclocking on the FX-9590 proved less fruitful.  While 5.0 GHz on all cores was stable at stock voltages, at 5.2 GHz temperatures on our setup were already high and caused throttling of the CPU below stock performance levels.

Many thanks to...

We must thank the following companies for kindly providing hardware for our test bed:

Thank you to OCZ for providing us with PSUs and SSDs.
Thank you to G.Skill and ADATA for providing us with memory kits.
Thank you to Corsair for providing us with an AX1200i PSU, Corsair H80i CLC and DRAM.
Thank you to ASUS for providing us with the AMD HD7970 GPUs and some IO Testing kit.
Thank you to MSI for providing us with the NVIDIA GTX 770 Lightning GPUs.
Thank you to Rosewill for providing us with PSUs and RK-9100 keyboards.
Thank you to ASRock for providing us with some IO testing kit.

Test Setup

Test Setup
Processor AMD FX-9590
4 Modules, 8 Threads, 4.7 GHz, 5.0 GHz Turbo
Motherboards ASRock 990FX Extreme9
Cooling Corsair H80i
Thermalright TRUE Copper
Power Supply OCZ 1250W Gold ZX Series
Corsair AX1200i Platinum PSU
Memory G.Skill RipjawsZ 4x4 GB DDR3-1866 9-11-9 Kit
Memory Settings DDR3-1866 8-9-9
Video Cards MSI GTX 770 Lightning 2GB (1150/1202 Boost)
Video Drivers NVIDIA Drivers 337
Hard Drive OCZ Vertex 3 256GB
Optical Drive LG GH22NS50
Case Open Test Bed
Operating System Windows 7 64-bit SP1
USB 2/3 Testing OCZ Vertex 3 240GB with SATA->USB Adaptor
WiFi Testing D-Link DIR-865L 802.11ac Dual Band Router

System Benchmarks

Power Consumption

Power consumption was tested on the system as a whole with a wall meter connected to the OCZ 1250W power supply, while in a single MSI GTX 770 Lightning GPU configuration. This power supply is Gold rated, and as I am in the UK on a 230-240 V supply, leads to ~75% efficiency > 50W, and 90%+ efficiency at 250W, which is suitable for both idle and multi-GPU loading. This method of power reading allows us to compare the power management of the UEFI and the board to supply components with power under load, and includes typical PSU losses due to efficiency. These are the real world values that consumers may expect from a typical system (minus the monitor) using this motherboard.

While this method for power measurement may not be ideal, and you feel these numbers are not representative due to the high wattage power supply being used (we use the same PSU to remain consistent over a series of reviews, and the fact that some boards on our test bed get tested with three or four high powered GPUs), the important point to take away is the relationship between the numbers. These boards are all under the same conditions, and thus the differences between them should be easy to spot.

Power Consumption - Long Idle

Power Consumption - Idle

Power Consumption - OCCT

The added power draw of the FX-9590 is quite clear, showing 350W for the full system during a heavy CPU load. However, the idle power draw of the CPU is similar to that of the FX-8150.

Windows 7 POST Time

Different motherboards have different POST sequences before an operating system is initialized. A lot of this is dependent on the board itself, and POST boot time is determined by the controllers on board (and the sequence of how those extras are organized). As part of our testing, we are now going to look at the POST Boot Time - this is the time from pressing the ON button on the computer to when Windows 7 starts loading. (We discount Windows loading as it is highly variable given Windows specific features.)  These results are subject to human error, so please allow +/- 1 second in these results.

POST (Power-On Self-Test) Time - Single MSI GTX 770

The FX-9590 afforded a shorter POST time than the FX-8150, although both are north of 16 seconds.

Rightmark Audio Analyzer 6.2.5

The premise behind Rightmark:AA is to test the input and output of the audio system to determine noise levels, range, harmonic distortion, stereo crosstalk and so forth. Rightmark:AA should indicate how well the sound system is built and isolated from electrical interference (either internally or externally). For this test we connect the Line Out to the Line In using a short six inch 3.5mm to 3.5mm high-quality jack, turn the OS speaker volume to 100%, and run the Rightmark default test suite at 192 kHz, 24-bit. The OS is tuned to 192 kHz/24-bit input and output, and the Line-In volume is adjusted until we have the best RMAA value in the mini-pretest. We look specifically at the Dynamic Range of the audio codec used on board, as well as the Total Harmonic Distortion + Noise.

Rightmark: AA, Dynamic Range, 24-bit / 192 kHz

Rightmark: AA, THD+N, 24-bit / 192 kHz

USB Backup

For this benchmark, we run CrystalDiskMark to determine the ideal sequential read and write speeds for the USB port using our 240 GB OCZ Vertex3 SSD with a SATA 6 Gbps to USB 3.0 converter. Then we transfer a set size of files from the SSD to the USB drive using DiskBench, which monitors the time taken to transfer. The files transferred are a 1.52 GB set of 2867 files across 320 folders – 95% of these files are small typical website files, and the rest (90% of the size) are the videos used in the WinRAR test. In an update to pre-Z87 testing, we also run MaxCPU to load up one of the threads during the test which improves general performance up to 15% by causing all the internal pathways to run at full speed.

USB 2.0 Copy Times

USB 3.0 Copy Times

The increase in CPU speed of the FX-9590 gives a measured improvement in standard USB copy times – saving 4.3 seconds over the FX-8150 for USB 2.0 transfers. However using XFast USB eliminated that difference completely. For USB 3.0 transfers, using the Etron controller put a consistent advantage on the side of the FX-9590.

DPC Latency

Deferred Procedure Call latency is a way in which Windows handles interrupt servicing. In order to wait for a processor to acknowledge the request, the system will queue all interrupt requests by priority. Critical interrupts will be handled as soon as possible, whereas lesser priority requests, such as audio, will be further down the line. So if the audio device requires data, it will have to wait until the request is processed before the buffer is filled. If the device drivers of higher priority components in a system are poorly implemented, this can cause delays in request scheduling and process time, resulting in an empty audio buffer – this leads to characteristic audible pauses, pops and clicks. Having a bigger buffer and correctly implemented system drivers obviously helps in this regard. The DPC latency checker measures how much time is processing DPCs from driver invocation – the lower the value will result in better audio transfer at smaller buffer sizes. Results are measured in microseconds and taken as the peak latency while cycling through a series of short HD videos - less than 500 microseconds usually gets the green light, but the lower the better.

DPC Latency Maximum

In a surprising twist, it would seem the FX-9590 has a disadvantage for DPC latency. I would have expected the faster processor to have the lower result, but one might speculate that the BIOS is tuned more for the mainstream processors such as the FX-8150. The chipset might also be geared more to the FX-8150 as it was the earlier generation compared to the FX-9590. 

CPU Benchmarks: Comparing the AMD FX-9590

The dynamics of CPU Turbo modes, both Intel and AMD, can cause concern during environments with a variable threaded workload. There is also an added issue of the motherboard remaining consistent, depending on how the motherboard manufacturer wants to add in their own boosting technologies over the ones that Intel would prefer they used. In order to remain consistent, we implement an OS-level unique high performance mode on all the CPUs we test which should override any motherboard manufacturer performance mode.

It is worth noting that the FX-9590, against the A10-7850K for example, has double the threads and +1 GHz of core speed, despite the generational gap. Considering only these factors, the FX-9590 is most likely similar if not better at single threaded performance, but also 2x for multithreaded results.

HandBrake v0.9.9: link

For HandBrake, we take two videos (a 2h20 640x266 DVD rip and a 10min double UHD 3840x4320 animation short) and convert them to x264 format in an MP4 container. Results are given in terms of the frames per second processed, and HandBrake uses as many threads as possible.

HandBrake v0.9.9 LQ Film

HandBrake v0.9.9 2x4K

The eight threads of the FX-9590 give it almost another third of performance over the A10-7850K for small frame videos and nearly 2x for the larger frame conversion.

Agisoft Photoscan – 2D to 3D Image Manipulation: link

Agisoft Photoscan creates 3D models from 2D images, a process which is very computationally expensive. The algorithm is split into four distinct phases, and different phases of the model reconstruction require either fast memory, fast IPC, more cores, or even OpenCL compute devices to hand. Agisoft supplied us with a special version of the software to script the process, where we take 50 images of a stately home and convert it into a medium quality model. This benchmark typically takes around 15-20 minutes on a high end PC on the CPU alone, with GPUs reducing the time.

Agisoft PhotoScan Benchmark - Total Time

The FX-9590 puts in the best AMD CPU performance, similar to that of the i5-2500K.

Dolphin Benchmark: link

Many emulators are often bound by single thread CPU performance, and general reports tended to suggest that Haswell provided a significant boost to emulator performance. This benchmark runs a Wii program that raytraces a complex 3D scene inside the Dolphin Wii emulator. Performance on this benchmark is a good proxy of the speed of Dolphin CPU emulation, which is an intensive single core task using most aspects of a CPU. Results are given in minutes, where the Wii itself scores 17.53 minutes.

Dolphin Emulation Benchmark

The FX-9590 beats almost everything pre-Haswell, showing the strength of a 5.0 GHz turbo mode. Note that it compares to the 3.5 GHz, 4MB L3 cache Haswell, which is 1.5 GHz slower and has half the L3.

WinRAR 5.0.1: link

WinRAR 5.01, 2867 files, 1.52 GB

PCMark8 v2 OpenCL on R7 240

A new addition to our CPU testing suite is PCMark8 v2, where we test the Work 2.0 and Creative 3.0 suites in OpenCL mode.

PCMark8 v2 Work 2.0 OpenCL with R7 240 DDR3

PCMark8 v2 Creative 3.0 OpenCL with R7 240 DDR3

Hybrid x265

Hybrid is a new benchmark, where we take a 4K 1500 frame video and convert it into an x265 format without audio. Results are given in frames per second.

Hybrid x265, 4K Video

The FX-9590 holds up very well in the Hybrid x265 conversion, which makes me wonder how well an 8-thread Kaveri CPU would perform.

Cinebench R15

Cinebench R15 - Single Threaded

Cinebench R15 - Multi-Threaded

The FX-9590 still has the single thread edge over the newer AMD CPUs due to the high frequency, but is easily overtaken by the modern cheap Intel CPUs. For multithreaded competition, the 8 threads needs an 8 threaded Intel CPU to compete.

3D Particle Movement

3DPM is a self-penned benchmark, taking basic 3D movement algorithms used in Brownian Motion simulations and testing them for speed. High floating point performance, MHz and IPC wins in the single thread version, whereas the multithread version has to handle the threads and loves more cores.

3D Particle Movement: Single Threaded

3D Particle Movement: MultiThreaded

Although the FX-9590 again performs as the best AMD CPU, the floating point nature of the calculations in this test seems not to be favored by this architecture.

FastStone Image Viewer 4.9

FastStone is the program I use to perform quick or bulk actions on images, such as resizing, adjusting for color and cropping. In our test we take a series of 170 images in various sizes and formats and convert them all into 640x480 .gif files, maintaining the aspect ratio. FastStone does not use multithreading for this test, and results are given in seconds.

FastStone Image Viewer 4.9

Similarly to 3DPM, for basic photo conversion any base 3.2 GHz Haswell CPU can outperform a 5.0 GHz AMD.

CPU Benchmarks: Comparing Motherboards

Readers of our motherboard review section will have noted the trend in modern motherboards to implement a form of MultiCore Enhancement / Acceleration / Turbo (read our report here) on their motherboards. This does several things – better benchmark results at stock settings (not entirely needed if overclocking is an end-user goal), at the expense of heat and temperature, but also gives in essence an automatic overclock which may be against what the user wants. Our testing methodology is ‘out-of-the-box’, with the latest public BIOS installed and XMP enabled, and thus subject to the whims of this feature. It is ultimately up to the motherboard manufacturer to take this risk – and manufacturers taking risks in the setup is something they do on every product (think C-state settings, USB priority, DPC Latency / monitoring priority, memory subtimings at JEDEC). Processor speed change is part of that risk which is clearly visible, and ultimately if no overclocking is planned, some motherboards will affect how fast that shiny new processor goes and can be an important factor in the purchase.

Point Calculations – 3D Movement Algorithm Test: link

3DPM is a self-penned benchmark, taking basic 3D movement algorithms used in Brownian Motion simulations and testing them for speed. High floating point performance, MHz and IPC wins in the single thread version, whereas the multithread version has to handle the threads and loves more cores.

3D Particle Movement: Single Threaded

3D Particle Movement: MultiThreaded

Compression – WinRAR 5.0.1: link

Our WinRAR test from 2013 is updated to the latest version of WinRAR at the start of 2014. We compress a set of 2867 files across 320 folders totaling 1.52 GB in size – 95% of these files are small typical website files, and the rest (90% of the size) are small 30 second 720p videos.

WinRAR 5.01

Image Manipulation – FastStone Image Viewer 4.9: link

FastStone is the program I use to perform quick or bulk actions on images, such as resizing, adjusting for color and cropping. In our test we take a series of 170 images in various sizes and formats and convert them all into 640x480 .gif files, maintaining the aspect ratio. FastStone does not use multithreading for this test, and results are given in seconds.

FastStone Image Viewer 4.9

The FX-8150 result for the Extreme9 compared to other motherboards shows one of two things - the latest Windows SP1 with core parking updates has an effect, or the BIOS is more efficient at handling turbo modes than our older reviews.

Video Conversion – Handbrake v0.9.9: link

For HandBrake, we take two videos (a 2h20 640x266 DVD rip and a 10min double UHD 3840x4320 animation short) and convert them to x264 format in an MP4 container. Results are given in terms of the frames per second processed, and HandBrake uses as many threads as possible.

HandBrake v0.9.9 Film CPU Only

HandBrake v0.9.9 2x4K CPU Only

Rendering – PovRay 3.7: link

The Persistence of Vision RayTracer, or PovRay, is a freeware package for as the name suggests, ray tracing. It is a pure renderer, rather than modeling software, but the latest beta version contains a handy benchmark for stressing all processing threads on a platform. We have been using this test in motherboard reviews to test memory stability at various CPU speeds to good effect – if it passes the test, the IMC in the CPU is stable for a given CPU speed. As a CPU test, it runs for approximately 2-3 minutes on high end platforms.

PovRay 3.7 beta

Synthetic – 7-Zip 9.2: link

As an open source compression tool, 7-Zip is a popular tool for making sets of files easier to handle and transfer. The software offers up its own benchmark, to which we report the result.

7-Zip MIPS

Gaming Benchmarks

F1 2013

First up is F1 2013 by Codemasters. I am a big Formula 1 fan in my spare time, and nothing makes me happier than carving up the field in a Caterham, waving to the Red Bulls as I drive by (because I play on easy and take shortcuts). F1 2013 uses the EGO Engine, and like other Codemasters games ends up being very playable on old hardware quite easily. In order to beef up the benchmark a bit, we devised the following scenario for the benchmark mode: one lap of Spa-Francorchamps in the heavy wet, the benchmark follows Jenson Button in the McLaren who starts on the grid in 22nd place, with the field made up of 11 Williams cars, 5 Marussia and 5 Caterham in that order. This puts emphasis on the CPU to handle the AI in the wet, and allows for a good amount of overtaking during the automated benchmark. We test at 1920x1080 on Ultra graphical settings.

Discrete SLI, Average FPS, F1 2013

The FX-9590 seems to match the i3-4360, indicating that even more cores, more frequency and more PCIe lanes is not always a good thing. The i3-4360 is using PCIe 3.0 x8/x8, compared to PCIe 2.0 x16/x16 for the FX-9590, which should put them both equal in bandwidth.

Bioshock Infinite

Bioshock Infinite was Zero Punctuation’s Game of the Year for 2013, uses the Unreal Engine 3, and is designed to scale with both cores and graphical prowess. We test the benchmark using the Adrenaline benchmark tool and the Xtreme (1920x1080, Maximum) performance setting, noting down the average frame rates and the minimum frame rates.

Discrete SLI, Average FPS, Bioshock Infinite

Again, the FX-9590 is trading around the Haswell i3 margin.

Tomb Raider

The next benchmark in our test is Tomb Raider. Tomb Raider is an AMD optimized game, lauded for its use of TressFX creating dynamic hair to increase the immersion in game. Tomb Raider uses a modified version of the Crystal Engine, and enjoys raw horsepower. We test the benchmark using the Adrenaline benchmark tool and the Xtreme (1920x1080, Maximum) performance setting, noting down the average frame rates and the minimum frame rates.

Discrete SLI, Average FPS, Tomb Raider

Tomb Raider has historically been CPU agnostic, with all the latest CPUs performing similarly.

Sleeping Dogs

Sleeping Dogs is a benchmarking wet dream – a highly complex benchmark that can bring the toughest setup and high resolutions down into single figures. Having an extreme SSAO setting can do that, but at the right settings Sleeping Dogs is highly playable and enjoyable. We run the basic benchmark program laid out in the Adrenaline benchmark tool, and the Xtreme (1920x1080, Maximum) performance setting, noting down the average frame rates and the minimum frame rates.

Discrete SLI, Average FPS, Sleeping Dogs

The FX-9590 loses 4-6 FPS on average to the latest Intel cohort, which is not bad considering the release date difference.

Company of Heroes 2

Company of Heroes 2 also can bring a top end GPU to its knees, even at very basic benchmark settings. To get an average 30 FPS using a normal GPU is a challenge, let alone a minimum frame rate of 30 FPS. For this benchmark I use modified versions of Ryan’s batch files at 1920x1080 on High. COH2 is a little odd in that it does not scale with more GPUs with the drivers we use.

Discrete SLI, Average FPS, Company of Heroes 2

Company of Heroes 2 is usually relatively CPU agnostic except for the older dual core CPUs, but the FX-9590 gets a win here.

Battlefield 4

The EA/DICE series that has taken countless hours of my life away is back for another iteration, using the Frostbite 3 engine. AMD is also piling its resources into BF4 with the new Mantle API for developers, designed to cut the time required for the CPU to dispatch commands to the graphical sub-system. For our test we use the in-game benchmarking tools and record the frame time for the first ~70 seconds of the Tashgar single player mission, which is an on-rails generation of and rendering of objects and textures. We test at 1920x1080 at Ultra settings.

Discrete SLI, Average FPS, Battlefield 4

Similar to Sleeping Dogs, the FX-9590 does not lose much considering the release date difference of the architectures.

The AMD FX-9590

The analysis in this review shows that even a year after the OEM release of the FX-9590, and almost two years from the architecture coming to market it remains AMD’s performance part. If power consumption is not a concern, as a CPU compute and an AMD gaming CPU (especially when considering SLI) the FX-9590 is the best choice at stock speeds. On that basis alone, it makes sense that AMD should actually release it as a retail part, assuming they have enough stock. One might argue that a user could buy an FX-8350 and overclock, but if our sample CPUs were anything to go by, a user needs a fair bit of luck. The FX-9590 guarantees a 5.0 GHz turbo with a warranty.

With the retail release of the CPU, that warranty might be based on using the water cooling provided for the lifetime of the CPU. One might argue that AMD had trouble finding enough dies that could reach the frequencies and voltages for the FX-9590, and hence the delay combined with selling the SKU in select markets only.

The FX-9590 is the same Piledriver architecture as the FX-8350, which in turn was used in the A10-5800K/A10-6800K APUs, codename ‘Trinity’ and ‘Richland’ respectively. Since then, AMD has launched the Steamroller architecture modifications in the form of Kaveri APUs. The difference between a PIledriver APU and a similar frequency Steamroller APU, if we put aside the move from 32nm SOI to 28nm SHP, is around 10% for CPU performance. If that was shifted into a four-module, eight-thread CPU, it would surely be AMD’s performance part. The issue here is that AMD has almost discarded the high CPU performance arena in favor of integrated graphics. From Trinity to Kaveri, the IGP inside those APUs has improved considerably, indicating where AMD is investing its research dollars.

AMD clearly still cares about the performance market, otherwise this retail FX-9590 with water cooling would have never been pushed through to retailers. The high power consumption, the lack of a modern chipset, and the comparison to Intel CPUs in single threaded benchmarks are the main barriers to adoption. If AMD is to return to the performance market, the power consumption has to be comparable to Intel, or if it is slightly higher, the chipset has to offer something Intel cannot. Any suggestions for what that feature should be should be submitted on a postcard/in the comments.

ASRock 990FX Extreme9 Conclusion

One of the big issues surrounding AMD motherboards is their price sensitive nature. With an Intel based product, a $250-$400 motherboard is common enough to signify the expense in research or extra features. Because the AMD ecosystem, even in the high performance segment, is a cost sensitive market there is little room to move. For example, this year sees the first overclocking based motherboard for AMD APUs since the AM3+ era. So at $170, the Extreme9 could arguably be described as ‘limited’ compared to Intel standards.

The motherboard itself has specified support for 220W CPUs, something other motherboards either fail to mention or advise against completely. The native SATA 6 Gbps ports were ahead of Intel at the time, plus ASRock adds in another SATA 6 Gbps controller for good measure.

The eight USB 3.0 ports makes the Extreme9 have more USB 3.0 ports than almost every other 990FX/AM3+ motherboard ever released. This is combined with plenty of legacy support, such as separate PS/2 connectors, a PCI slot, an IEEE1394 port and an IEEE1394 header. The Intel NIC is paired with a Realtek ALC898 codec, with the PCIe layout aimed at 3-way GPU users for both Crossfire and SLI.

Aside from an updated chipset, if we were building a high-end AM3+ motherboard in 2014, I would insist on WiFi support and an upgraded audio codec to the ALC1150 at the minimum. We cannot get around the lack of PCIe 3.0 support, although moving the CPU modules from Piledriver to Steamroller along with the IO support might help with that. If we are being greedy with what we would like, I would add in M.2 support as well.

There is plenty to speculate if AMD had kept updating their high-end performance CPU line, even if the socket was not updated. As it stands, users who want SLI either look back to 990FX or invest in Intel. Users who want high multithreaded CPU performance either look back to 990FX or invest in Intel. Users who do not want processor graphics either look back to 990FX, buy an APU with the graphics disabled, or invest in Intel. AMD clearly does care about the performance market, or at least someone senior in the company does. 

Log in

Don't have an account? Sign up now