ASUS M4A78T-E 790GX AM3 Motherboard Review

[MAC]

Memory Benchmarks

Memory Benchmarks

Lavalys Everest Ultimate v5.00​

Everest Ultimate is the most useful tool for any and all benchmarkers or overclockers. With the ability to pick up most voltage, temperature, and fan sensors on almost every motherboard available, Everest provides the ability to customize the outputs in a number of forms on your desktop. In addition to this, the memory benchmarking utility provides a useful tool of measuring the changes to your memory sub-system.

After getting used to the humongous numbers that the Core i7 puts out, it was rather humbling to see the Phenom II's memory bandwidth. For both the stock and overclocked configurations, the system was running in dual-channel, unganged mode and with the memory frequency set to DDR3-1600. Likewise, the CPU/NB frequency was locked at 2000Mhz for all configurations, so there was no performance differential to be had from that area. As a result, it is no surprise that the bandwidth numbers are so similar. The only noteworthy improvement is the roughly 6% difference in copy speeds, but this is largely attributable to the higher CPU frequency.

Due to the aforementioned similarity in the memory sub-system settings, the memory latency is effectively identical across the board. However, there was an unusual sweet spot at 3.2Ghz (16 x 200). There is no rhyme or reason for this anomaly, and it is only 1% lower than the other results, but we were able to repeat the results on numerous occasions.

ScienceMark v2.0​

Although last updated almost 3 years ago, and despite its rudimentary interface, ScienceMark v2.0 remains a favorite for accurately calculating bandwidth on even the newest chipsets.

​

ScienceMark reveals a minor 4% increase in overall memory bandwidth between the stock and overclocked configurations. One way or another, despite benefiting from an integrated memory controller (IMC) and new DDR3 memory, no one is going to be breaking any memory-related records on this new AM3 platform. The M4A78T-E is obviously not to blame, it is simply a characteristic of the Phenom II's.

 

[MAC]

System Benchmarks

System Benchmarks

SuperPi Mod v1.5​

When running the SuperPI 32MB benchmark, we are calculating Pi to 32 million digits and timing the process. Obviously more CPU power helps in this intense calculation, but the memory sub-system also plays an important role, as does the operating system.

An enthusiast favourite, SuperPi excels at revealing how the processing and memory sub-systems are performing and it can demonstrate the slightest performance variances. In this case, our overclocked configuration really shines in this benchmark, reducing the overall time by over 5 minutes. This represents an almost 26% performance improvement from a 32% CPU overclock, which is strong scaling result. In a benchmark in which every second matters, this a massive gain.

PCMark Vantage x64​

The latest iteration of the popular system benchmark is PCMark Vantage from the Futuremark crew. The PCMark series has always been a great way to either test specific areas of a system or to get a general overview of how your system is performing. For our results, we simply run the basic benchmark suite which consists of a wide range of tests involving all the sub-systems of the computer.

Almost not quite as impressive as above, the roughly 11% performance gain achieved in this PCMark Vantage is quite respectable since it represents an improvement in <i>overall</i> system performance. Despite the fact that this benchmark tests all critical sub-systems, it is evident that nearly all the gains are isolated in the CPU-dependent Communication and Productivity sections.

Cinebench R10​

Developed by MAXON, creators of Cinema 4D, Cinebench 10 is designed using the popular Cinema software and created to compare system performance in 3D Animation and Photo applications. There are two parts to the test; the first stresses only the primary CPU or Core, the second, makes use of up to 16 CPUs/Cores. Both are done rendering a realistic photo while utilizing various CPU-intensive features such as reflection, ambient occlusion, area lights and procedural shaders.

Cinebench demonstrates the multi-threading processing power our triple-core X3 720 BE chip . In this test, we see a roughly 28% performance improvement, which is a quite linear to the 32% increase in CPU clocks.

x264 HD Benchmark​

Tech Arp's recent development of the x264 HD Benchmark takes a 30 second HD video clip and encodes it into the x264 codec with the intention of little to no quality loss. The test is measured using the average frames per second achieved during encoding, which scales with processor speed and efficiency. The benchmark also allows the use of multi-core processors so it gives a very accurate depiction of what to expect when using encoding application on a typical full length video.

Although not really an encoding powerhouse, our Phenom II X3 chip achieves a 32% performance boost through overclocking, which in this case equates to a decent 3.5 frames per second improvement.

Lame Front End​

Unlike the DivX conversion program we just looked at, Lame Front End is not multi-threaded and only utilizes a single processor core. This will obviously limit performance but we should still achieve significant time savings going from the stock to the overclocked settings. We will be encoding a WAV rip of Santana’s Supernatural album and converting it to MP3 using the VBR 0 quality preset.

LFE is a single-threaded application that allows the ripping of a CD in a matter of minutes. It reveals nearly perfect performance scaling, with a 31.5% gain from a 32% overclock. In practical terms, we were able to shave approximately 1 minute off of this task through overclocking.

Photoshop CS3​

For the image editing portion of this review, we will use Photoshop CS3 in coordination with Driver Heaven’s Photoshop Benchmark V3, which is an excellent test of CPU power and memory bandwidth. This is a scripted benchmark that individually applies 15 different filters to a 109MB JPEG, and uses Photoshop’s built-in timing feature to provide a result at each test stage. Then it’s simply a matter of adding up the 15 results to reach the final figure.

Impressive. At 3.7Ghz, our overclocked configuration was able to complete the Photoshop test a full minute and half faster than at stock. This represents an almost 40% improvement in image-editing performance. This is greater than the 32% CPU overclock, but this is could simply attributed to benchmarking variances, which are fairly significant with Photoshop.

WinRAR 3.8.0​

The last of our real-life tests will be with the highly popular & multi-threaded WinRAR 3.8.0 tool, in which we take a 1GB batch of assorted files and archive them, timing the task until completion.

WinRAR is a tool that we use countless times per day, so any performance improvement can provide some welcome time savings. Our overclocked Phenom II X3 system finished the compression task a solid 42 seconds faster than at stock, which is a worthwhile 20% performance improvement.

 

[MAC]

Gaming Benchmarks

Gaming Benchmarks

Futuremark 3DMark06​

The Futuremark 3DMark series has been a part of the backbone in computer and hardware reviews since its conception. The trend continues today as 3DMark06 provides consumers with a solid synthetic benchmark geared for performance and comparison in the 3D gaming realm. This remains one of the most sought after statistics, as well as an excellent tool for accurate CPU comparison, and it will undoubtedly be used for years to come.

In 3DMark 06, the overclocked configuration boasts a mighty 21% lead over the stock clocked system. This is a very impressive improvement when you consider that our GeForce GTX 280 graphics card was not overclocked. Let's see if the gains are as prominent on Futuremark's latest gaming benchmarking...

Futuremark 3DMark Vantage​

You asked for it, so we have finally included 3DMark Vantage, Futuremark’s latest release in their renowned line of 3D benchmarking software. This latest DX10-only 3DMark comes with a variety of presets, but for our tests will be use the standard Performance preset which is suitable for a much greater range of system specifications than the other more demanding presets.

In 3DMark Vantage, despite a proper 30% gain in the CPU score, the overall score increases a relatively meager 8%.

Can we expect similar results in actual games? Let's find out...

Far Cry 2​

Far Cry 2 is the hot new new first-person shooter from Ubisoft's Montreal studio, and the first game to utilize the new visually stunning Dunia Engine, which will undoubtedly be used by numerous future games. Using the included Benchmarking Tool, we ran the Long Ranch demo in DX10 mode at 1680x1050 with all settings set to very high.

At these settings, the GeForce GTX 280 appears to be limiting gaming performance, since between 3.2Ghz and 3.7Ghz the frames per second are effectively identical.

Company of Heroes: Opposing Fronts​

This test consists of actual gameplay using a single mission (Authie: Boudica's Boys from the British Campaign) since it holds a bit of everything the game has to offer: vehicle battles, artillery barrages and house-to-house fighting. We recorded frame rates using FRAPS up until the 15 minute mark of the mission and excluding any in-game briefings / cut scenes. The game was run in DirectX 10 mode at 1680x1050 with all settings maxed out.

Clearly, Company of Heroes: Opposing Fronts is still a graphical force to be reckoned with when all its graphical settings are maxed out. Our high-end GeForce GTX 280 is just barely able to keep the minimum frame rate at the vital 30FPS limit across the board. Ignoring the large increase in maximum FPS, we do notice a very impressive 17% gain in the average frame rate.

Left 4 Dead​

Left 4 Dead is the latest disorienting, fast-paced zombie apocalypse mega-hit from Valve. L4D uses the latest version of the Source engine with enhancements such as multi-core processor support and physics-based animation. We test here at 1680x1050 with in-game details set to their highest levels, with MSAA 4X and AF 8X. For benching, we used a pre-recorded 20 minute timedemo taken on the No Mercy campaign during The Apartments mission.

Last but certainly not least is Valve's ground-breaking new multi-player title. From stock to overclocked, the average frame rate increased by a significant 17%, which demonstrates the Source engine's strong CPU scaling.

 

[MAC]

Gaming Benchmarks - Radeon HD 3300 IGP

Gaming Benchmarks - Radeon HD 3300 IGP

As previously mentioned, one of the M4A78T-E's key features is the Radeon HD 3300 integrated graphics processor (IGP) that is part of the 790GX northbridge. Based on the RV610 core, this model features a 700Mhz core clock, 4 ROPs, 40 Shaders, and 128MB of DDR3-1333 SidePort memory built onto the motherboard, as well as a UMA option to access up to 512MB of system memory. Although recently surpassed by NVIDIA's GeForce 9300/9400 series, the Radeon HD 3300 was a breakthrough in IGP performance when it was first released, and it remains a surprisingly competent performer. Let's take a closer look below.

Far Cry 2​

Far Cry 2 is the impressive new first-person shooter from Ubisoft's Montreal studio, and the first game to utilize the new visually stunning Dunia Engine, which will undoubtedly be used by numerous future games. Using the included Benchmarking Tool, we ran the Long Ranch demo in DX9 mode at both 1024x768 and 1680x1050 and with all settings set to low.

Much to our surprise, the Radeon HD 3300 did quite well in Far Cry 2 at 1024x768. It is certainly playable, with only very few sub-30 FPS moments. Sure the game doesn't look great, but in a pinch this IGP could be used for some light gaming sessions. As expected, the GPU runs out of steam at the higher resolution and the frame rates are significantly lower than at 1024 x 768. Having said that, we cannot overlook the significant improvement that occurs at 1680 x 1050 once UMA is enabled, at the expense of 512MB being borrowed from the system memory. Clearly, this is a worthwhile trade-off if you are using the IGP as your main graphics adapter.

Crysis​

With its lush landscapes and intense action, Crysis is renowned for being extremely GPU dependent, and it remains a terrific gaming benchmarks and system stress test. We utilized the gpu_benchmark demo in 64-bit DX9 mode and with all detail levels set to low.

Yes, that's right...an IGP can play Crysis above 30FPS. As with Far Cry 2, the game looks fairly poor but it <b>is</b> playable and that is what matters when you're in a bind. At resolutions above 1024 x 768, the HD 3300 starts suffering from its small memory buffer. This is somewhat rectified when you enable the UMA option, and allow the IGP to utilize some system memory, but the performance improvements still don't allow a playable frame rate at 1680 x 1050.

World in Conflict​

Developed by Massive Entertainment and published by Ubisoft, World in Conflict, commonly known was WiC, is easily the best and most visually appealing real-time tactical video game in recent memory. We ran the in-game benchmark in DirectX 9, and using the low quality setting.

At 1024 x 768, WiC is very playable at the low setting, but it looks absolutely hideous with graphics reminiscent of games circa 1999-2000. However, at least it will run and at a playable frame rate. It is a better experience than a blank screen or an error message. This is also a great example of how useful the UMA option can be, as it has made the 1680 x 1050 resolution borderline playable.

Left 4 Dead​

Left 4 Dead is the latest disorienting, fast-paced zombie apocalypse mega-hit from Valve. L4D uses the latest version of the Source engine with enhancements such as multi-core processor support and physics-based animation. We test here at 1680x1050 with in-game details set to their highest levels, with MSAA 4X and AA 8X. For benching, we used a pre-recorded 20 minute timedemo taken on the No Mercy campaign during The Apartments mission.

Once again demonstrating the unparalleled flexibility of the Source engine, L4D runs at a surprising average of 50 frame per second. Most importantly, the lesser graphics quality really doesn't diminish the gaming experience very much, which is something that cannot be said for the other three games. While enabling UMA doesn't make 1680 x 1050 playable, we still see a more than 100% performance improvement.


In summary, the Radeon HD 3300 is...surprisingly capable in the gaming department, and it is a very welcome addition to this motherboard, especially when you consider that it's not really costing you anything and it can bail you out if/when something happens to your primary graphics card. Also, if you have 4GB+ of system memory definitely enable the "SidePort + UMA" option, because as you can see above the performance gains can be significant.

 

[MAC]

Voltage Regulation / Temperature Testing / Power Consumption

Voltage Regulation

Our voltage regulation testing will focus on the various voltages and the differences encountered between what is selected in the BIOS and what is reported by a digital multi-meter (DMM). We have found five voltage read points on the motherboard so the vCORE, CPU/NB, NB, HT, SB, and vDIMM will be recorded with our DMM. The vCORE and CPU/NB will be read from the underside of the motherboard at the output side of the corresponding inductor. The remaining voltages will be read from points on the top of the motherboard. vDIMM will be read directly from a DIMM slot's VDD pin. The ground point used for all readings will be a screw hole. Here are a few photos showing the various read points used.

Click on image to enlarge​

Now that we have established where the read points are, let’s have a look at the results. These measurements were taken at stock system speeds and with Cool N' Quiet and C1E disabled in the BIOS. So without further ado, here are our extensive findings:

Taken as a whole, what you see is what you get with the M4A78T-E. It has very good voltage output. There is relatively little variance between what you select in the BIOS and what the board actually outputs. The HT, NB, SB, and DRAM voltages are effectively perfect. It should be noted that when the DRAM voltage is set to AUTO with memory speeds of DDR3-1333 and above, it defaults to a hefty 1.92V, so keep that in mind if you are using the new generation of low voltage DDR3 modules. The critical CPU/NB voltage, which feeds the integrated memory controller, overvolts by approximately 0.3-0.5V once you manually set 1.25V or above. As with the DRAM voltage, when left on AUTO and a memory frequency of DDR3-1333 and above is selected, the CPU/NB defaults to roughly ~1.41V, which is appropriate. While the vCore undervolts by approximately 0.3-0.4V, this is a relatively minimal droop and it won't affect stability nor overclocking. There is no Load-Line Calibration (LLC) feature (yet) for Phenom II processors. Although we would usually take a closer look at the vCORE's characteristics with a one hour OCCT stress test, the current version's of the program does not seem to recognize this motherboard's vCore line.</b>

Temperature Testing

Now it's time to see if the M4A78T-E's simplistic cooling system works as well it as looks. Although this motherboard has a temperature sensor for the northbridge its accuracy seemed slightly suspect to us. Therefore, we relied on our trusty digital thermometer for all temperature measurements. We set the system to its overclocked configuration, and then ran Prime 95 Blend for one hour. In order to test stress the IGP, we subsequently ran a Crysis timedemo on loop for one hour. In both cases, the temperatures were recorded at ten minute intervals throughout the one hour tests and the results averaged out. The ambient temperature was 20°C/68°F

Considering our substantial overclock and liberal use of voltage, we were impressed with these temperature results. Even with its relatively dinky heatsink, having the IGP enabled only increases the northbridge temperatures by a relatively minor 8°C. None of the above components benefited from any direct air cooling, so all these figures essentially represent a worst-case scenario. In a regular case with some half-decent airflow, temperatures should be even lower. As a result, we can definitely state that the M4A78T-E's cooling system is capable of handling all but the most extreme overclocking attempts, and we can also commend AMD for making such cool running chipsets.

Power Consumption

In order to test this system's default power consumption, we enabled both Cool'n'Quiet and C1E in the BIOS, then let the system idle for 1 hour and measured the KW/h power consumption through our UPM EM100 power meter. We repeated the same procedure for 1 hour of Prime 95 Blend load. We then installed the EPU software, set it to maximum energy saving, and repeated the tests. Here are the results:

As you can see, we did not measure any decrease in power consumption when EPU was installed. In both cases, the Phenom II X3 downclocked to 800Mhz (4 x 200Mhz) when idle, so clearly AMD's power savings technologies are doing their functions well. The system with the GeForce GTX 280 idled in the 110W range and reached a peak of 172W when running Prime 95 Blend. The IGP-enabled system idled at a frugal 65W and maxed out at 128W under load, which is quite impressive for modern system.

 

[MAC]

Conclusion

Conclusion

Our very first AMD motherboard review is coming to a close and it has been an interesting experience. On the one hand, it is a whole lot of fun to overclock Phenom II's, especially those with unlocked multipliers like our X3 720 Black Edition. <i>However</i>, the DDR3 aspect of this platform was largely frustrating. The platform is picky when it comes to DDR3 modules, it doesn't have much memory overclocking headroom, and it is generally accepted that Phenom II processors do not really benefit from the extra bandwidth that DDR3 provides. As a result, the logical question is why pay a premium for a DDR3 motherboard/memory combo when the DDR2 equivalent will provide roughly the same performance for less money.

None of this is a slight against the M4A78T-E though. We would easily qualify this model as a "Jack of all Trades, Master of Most". It is a very well designed motherboard with terrific features and connectivity, it ran like a champ throughout our battery of tests. It proved to be a very good overclocker, and the only hiccups we encountered were the slight "FSB Hole" during our CPU bus overclocking efforts, and the fact that the CPU/NB frequency would revert back to stock when selecting a CPU/NB frequency above 2900Mhz. This was an annoyance, but not one that truly detracted from the overall experience. Although most gamers will scoff at it, we couldn't help but be impressed at the capabilities of the Radeon HD 3300 IGP for rudimentary gaming, and it is also one of the most capable HTPC-centric IGP's on the market.

Frankly, the only thing that prevents us from giving this product our Dam Good Award is the DDR3 memory format, and the knowledge that it has a fraternal twin that supports DDR2 and is a good $20-25 cheaper, the M4A78-E. Otherwise, it is a universally good motherboard and a model that we have no qualms recommending to those hell-bent on making the transition to DDR3.


Pros

- Top-notch layout.
- Good CPU overclocking capabilities.
- Impressive IGP performance.
- Runs cool & power efficiently.
- Excellent BIOS.
- Superior connectivity.
- Impressive software suite.
- Decent Bang for the Buck...


Cons

- ...but the M4A78-E DDR2 model provides even better Bang for Buck.
- BIOS-related "FSB Hole" & CPU/NB frequency limit.
- Not enough voltage and temperature readouts (as always).

Our thanks to ASUS for making this review possible!​

ASUS M4A78T-E Motherboard Review Comment Thread​