|Introduction and Product Specifications|
With no major processor architecture updates in the immediate future and their DX10-capable graphics solutions still a few months away, the new AMD seems to be slowly picking up slack as we roll through the first two thirds of Q1 '07. Today, AMD is officially introducing their first new chipset since the ATI acquisition, the AMD 690G/V Series. Comprised of an RS690 Northbridge and SB600 Southbridge, the 690 series chipset's goal is to provide a full-featured motherboard backed by a solid integrated graphics solution, thanks to an integrated ATI Radeon x1250 with AVIVO support. What makes the 690G unique, at this point, is its support for VGA and DVI or HDMI, depending on the model, which enables the chipset to be attractive to both business and multimedia users looking for a quality video experience. With its integrated Radeon x1250 clocked at 400MHz and the ability to share up to 1GB of system memory, AMD looks to provide a versatile board that can fill a multitude of needs at a competitive price point.
Today we are going to size up the new 690G series from head-to-toe, giving an overview of its feature-set as well as providing a broad range of benchmark results, in an attempt to see just how well the new 690G compares to its main competition. In our tests, we will compare performance to two systems based on NVIDIA's nForce 430 chipset, one geared for 3D performance, the other for a better a multimedia experience.
|AMD 690 Series Chipset: Technical Overview|
The 690 series chipset comes in two flavors, the first is the AMD 690G with an ATI Radeon X1250 IGP clocked at 400MHz as well as integrated HDMI or DVI outputs. The second version is the AMD 690V which comes with an integrated ATI Radeon X1200 with no HDMI/DVI support.
Comprised of an RS690 Northbridge and SB600 Southbridge, AMD is positioning the 690 series as a well balanced Micro-ATX offering that can appeal to a broad range of personal and commercial consumers. Expect to see a number of mATX boards offered as well as barebones systems with 690 series boards inside. They've also jumped on the Microsoft Vista bandwagon, promising full support, and a premium experience, with the new OS.
Processor support for the 690G/V includes the AMD Sempron, AMD Athlon 64, AMD Athlon 64 X2 Dual-Core and AMD Athlon 64 FX. The RS690 Northbridge features one x16 PCI Express graphics interface and supports up to four additional x1 PCI Express connections. Additionally, one break-out feature of the chipset is that the Northbridge can also drive a High Definition audio controller that can feed audio over the integrated HDMI output.
The RS690's integrated Radeon X1250 sports a 400MHz, 128-Bit graphics engine, is Direct X9 compatible and has a maximum resolution of 2048x1536 with 32-Bit color. Both the VGA and HDMI (High-Definition Multimedia Interface) outputs can run independently for a dual monitor configuration. And with a supported discrete ATI graphic card installed, SurroundView can be enabled as well, allowing for up to three video outputs simultaneously. Its HDMI interface supports the 1.2 specification as well as HDCP (High-Digital Content Protection) 1.1. Lastly, this is the first chipset to integrate AVIVO, boasting 10-bit video quality for high quality video playback through hardware acceleration.
The SB600 Southbridge packs in the features as well. The chip can power up to 10 USB ports that are both USB 1.1 and 2.0 compliant. For storage needs, the SB600 supports a total of four SATA II 3Gb/s drives and can deliver RAID 0, 1 and 10. RAID 0 and 1 can work with any two drives while RAID 10 requires four. The SB600 also has a single PATA UDMA 133 connection that can connect up to two IDE devices. For expansion, the SB600 can drive up to six PCI slots.
For audio functionality, a 192kHz High Definition Audio controller is integrated offering separate input and output streams with up to 16 channels per output stream. Lastly, the SB600 handles power management based on ACPI 3.0 specifications with support for AMD On-Now, ASF 2.0, and it delivers complete hardware monitoring via the BIOS.
One might question why a chipset touted as offering a premium Vista experience wouldn't be DX10 ready. Considering DX10 is mainly for gaming, and Vista's Aero interface requires only a DX9 class GPU, this is really a non-issue in our opinion. Surely anyone who is serious about gaming knows that integrated graphics is not the way to go. For those looking for a premium DX10 gaming experience, you'll want to budget in a discreet video card upgrade into your plans. Regardless, as a sum total of its parts, the RS690 chipset makes for an excellent foundation of a very capable HTPC.
|Close Up: The MSI K9AGM2 690G Mainboard|
The board we received for review was an MSI K9AGM2 based on the 690G chipset with HDMI support. This was not a complete retail package, so we are going to zero in on the board's layout and features and skip the usual overview of what comes in the box along with the board. While the package was not representative of a retail product, the board itself is.
The MSI K9AGM2 comes with the AMD 690G chipset, obviously, and follows closely to the general specifications outlined earlier. The board is equipped with one x16 PCI Express graphic slot, a single x1 PCI Express slot and two PCI 2.3 slots. Two DIMM slots are provided, supporting up to 4GB of DDR2 memory. There are a total of four SATA II connections capable of RAID 0, 1 and 10 while the SB600 offers a single ATA133 IDE port and a legacy floppy port natively as well.
The board comes with one Gigabit Ethernet care of a Realtek 8111B. The board also comes with two IEEE1394 ports, one on the rear console and the other delivered through an optional header and bracket, both of which are driven by a VIA VT6308 chip. The High Definition onboard audio is managed by a Realtek ALC888 which meets the Azalia 1.0 specification and supports up to 8-channels. The Realtek ALC888 also provides jack sensing, so searching for the right port for each speaker is not needed.
The rear I/O panel provides PS/2 ports for legacy keyboard and mouse connections as well as a parallel port that supports SPP, EPP and ECP modes. There are four USB ports and 3 additional USB headers that can be used with optional brackets to bring the total to 10. This particular model comes with a VGA connector as well as an HDMI port, which may be traded off with a DVI port depending on the model.
Another option for the board is a TV-Out bracket, which was not included. This connects to the TV-Out header to provide S-Video and RCA composite outputs. An optional SPDIF In/Out bracket can also connect to the proper header for digital audio in and out.
When it comes to driver installation, ATI's method seems like a never ending cascade of installers that load one after the other, making the process seem much longer than it does with NVIDIA hardware. In fact, NVIDIA's installation process is far more elegant and completes much quicker, with less interaction, than the ATI Catalyst package. While improvements have been made to the load times of the Catalyst Control Center, we'd still love to see AMD strengthen this area so the installation process is smoother and faster for the end user.
When previewing the K9AGM2's BIOS screen below, don't expect to find anything extravagant in regards to performance or overclocking related settings. MSI covers the bases, but considering the target market, no advanced options for enthusiasts are present.
The main screen has a familiar layout found with many American Megatrends, Inc. BIOSes. When it comes to performance settings, the main place to go is the Advanced Chipset Features menu. Here, DRAM timings, shared video memory and other display settings are available. The DRAM timings offer CAS Latency, Min RAS# Active Time, RAS# Precharge Time and RAS# to CAS# Delay. CAS Latency options range from 3-6, Min RAS# Active Time from 5-18 CLKs and both RAS# Precharge Time and RAS# to CAS# Delay range from 3-6 CLKs respectively. What's missing? Most notably an adjustable command rate.
The VGA Share Memory Size was surprisingly large, allowing for up to 1GB of system memory to be allocated to the integrated graphics controller frame buffer. The Display Active Select lets you specify which device is the primary graphics output, although the auto setting should suffice in most cases.
The rest of the screens all yielded very standard options, with the Hardware Monitor screen offering insight on the current state of critical voltages, temperatures and fan speeds. Those looking for any advanced overclocking options will be disappointed, as this model comes with no clock register settings or voltage controls whatsoever.
|Test System Specifications and Notes|
Before we start the benchmarking segment, we wanted to make several notes about our setup and testing methodology. Currently, the closest hardware comparison we had available for this review were two nForce 430 boards, one with a GeForce 6100 integrated graphics controller, the other with a GeForce 6150.
To help interpret the results in each of our tests, we wanted to clarify the roles of the GeForce 6100 and 6150. The GeForce 6150 is a two chip design that incorporates NVIDIA's PureVideo engine. This is commonly found on system boards designed for media center applications. The GeForce 6100 is a second generation processor that is a single chip design, manufactured on a smaller process than its predecessor. NVIDIA also incorporated hardware z-cull for improved 3D performance with the 6100 while omitting an integrated PureVideo engine. This positions the 6100 more for reasonable game play rather than video quality. So when reviewing these results, we expect the 6100 series to top the 6150 in gaming tests, while we should see the inverse when in comes to our HQV video testing. In all integrated tests, each board was set to share 256MB of system memory, the maximum setting for the nForce 430 boards tested, providing an equal playing field. Virtually all tests were also run with discrete graphics as well to demonstrate the impact of sharing 256MB of main memory with the IGPs might have on performance.
You might also note that all testing was done in Windows XP Professional SP-2. While AMD is pushing the 690 series as offering a Premium Windows Vista Experience, we've opted to perform all testing on XP as Vista is still quite new and drivers for various hardware vendors need to mature a bit.
|SANDRA XI: CPU and Memory Performance|
We started testing with SiSoftware's SANDRA XI, the System ANalyzer, Diagnostic and Reporting Assistant, by running three built-in subsystem tests; CPU, Multimedia, and Memory. We ran each test module twice with each system, once with integrated graphics enabled and again with a discrete graphics card (GeForce 7600 GT) installed. First up are the CPU and Multimedia scores.
Both CPU and Multimedia modules revealed no surprises, reporting each test system to be on par with one another in each test. We saw no major impact on performance when comparing integrated and discrete graphics performance.
With the Memory Bandwidth test we saw more variation in the results. The AMD 690G sported the lowest scores of the three boards tested with integrated graphics enabled. Conversely, the 690G demonstrated the biggest jump in performance when discrete graphics was used, returning the highest overall scores when able to utilize the full system memory complement.
|Futuremark PCMark05 and 3DMark06 CPU|
In our next round of testing, we ran both CPU and Memory performance modules of FutureMark's PCMark05. Each results graph is prefaced by FutureMark's explanation of what each test does to arrive at its scoring metric. Once again, all tests were run twice, once with integrated graphics and again with discrete.
"The CPU test suite is a collection of tests that are run to isolate the performance of the CPU. The CPU Test Suite also includes multithreading: two of the test scenarios are run multithreaded; the other including two simultaneous tests and the other running four tests simultaneously. The remaining six tests are run single threaded. Operations include, File Compression/Decompression, Encryption/Decryption, Image Decompression, and Audio Compression" - Courtesy FutureMark Corp.
With integrated or discrete graphics, all three boards were on the same level, with the nForce 430's having a slight lead over the 690G.
"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing." - Courtesy FutureMark Corp.
Like we saw in the SANDRA memory test, the AMD 690G showed the broadest swings when tested with integrated and discrete graphics. In this test, the AMD 690G was competitive with a discrete graphics card installed, but was the slowest of the three when testing with the integrated X1250.
3DMark06's built-in CPU test is a multi-threaded "gaming related" DirectX metric that's useful for comparing relative performance between similarly equipped systems. This test consists of two different 3D scenes that are generated with a software renderer that is dependent on the host CPU's performance. This means that the calculations normally reserved for your 3D accelerator are instead sent to the central processor. The number of frames generated per second in each test are used to determine the final score. This test was run with both integrated and discrete graphics as well.
When focusing on 3DMark05's CPU test, the AMD 690G managed to show improved performance with integrated graphics enabled versus discrete, although the results are close enough to be considered insignificant. Overall, all three boards competed on similar levels with no clear dominant player.
|Lame MT and Cinebench R9.5 Performance Tests|
Our next two tests focus almost entirely on CPU performance, with graphics having minimal impact on the results. In fact, all tests were run with both integrated and discrete graphics and there was no difference in scores. As a result, the scores obtained with discrete graphics installed were omitted for brevity.
In this test, we created our own 223MB WAV file (a never-ending Grateful Dead jam) and converted it to the MP3 format using the multi-thread capable LAME MT application in single and multi-thread modes. Processing times are recorded below. Shorter times equate to better performance.
Being so CPU intensive, this test often shows identical results when the same CPU is used, and this is no different today. All three boards tested returned identical results in single and multi-threaded testing, demonstrating equal CPU efficiency.
The Cinebench R9.5 benchmark is an OpenGL 3D rendering performance test, based on the commercially available Cinema 4D application. This is a multi-threaded, multi-processor aware benchmark that renders a single 3D scene and tracks the length of the entire process. The time it took each test system to render the entire scene is represented in the graph below (listed in seconds).
With Cinebench R9.5, slight variations were reported, with the ASUS M2NPV-VM posting the best single-thread score, topping the MSI K9AGM2 by two seconds. Multi-thread testing was equal across all three systems tested.
|File System Testing With HDTach 3.0 and SANDRA XI|
In our next round of tests, we targeted drive performance to highlight the efficiency at which each chipset handles data flow to and from the hard drives. To do this, we used HDTach 3.0 and SANDRA XI's File System Test. As with the LAME MT and Cinebench testing, the tests were run with integrated and discrete graphics and no major differences were noted, so discrete graphics results were not included.
Using HDTach 3.0, we opted to run the Long Bench on the boot drive which utilizes 32MB Zones. Both Average Read and Burst Speeds were recorded.
The ASUS M2NPV-VM posted a slightly faster burst rate at 127MB/s, but the average reads were consistent with all three models tested.
Next we ran the File System Test in SANDRA XI which focuses on Random Access in ms and average reads, recorded as Drive Index.
With SANDRA XI's File System Testing, the results were even more consistent, with all boards returning identical results.
|World Bench 5.0: Office XP SP2 & Photoshop 7|
PC World Magazine's WorldBench 5.0 is a Business and Professional application benchmark suite which consists of a number of performance modules that each utilize one, or a group of, popular applications to gauge performance. In these test, both integrated and discrete graphics were used to help demonstrate the impact of sharing 256MB of system memory with the integrated graphics vs running no shared memory and a graphics card with 256MB of its own memory.
Below we have the results from WB 5's Office XP SP2 and Photoshop 7 modules, recorded in seconds. Lower times indicate better performance here, so the shorter the bar the better.
WorldBench 5's Office XP test had the MSI K9AGM2 finishing a fair amount beihnd than the nForce 430 systems when using their respective integrated graphics. Improvements were noted with discrete graphics, but the 690G based system still lagged the 430/6150 by roughly 20 seconds. Oddly, this test actually returned worse results when running discrete graphics with the 460/6100 board.
The Photoshop 7 module, on the other hand, was much closer, with no major differences between integrated and discrete graphics and system variations closer to 3 seconds overall.
|Worldbench 5.0: 3ds Max, WME 9 & Mozilla Multi-Tasking|
Up next, we have performance results of WB 5.0's 3ds Max (Direct 3D) test. A number of different 3D objects are rendered and animated in this test, and the entire time needed to complete the tasks is then recorded. As is the case with all of the individual Worldbench tests, a lower score here indicates better performance.
With discrete graphics, all three systems posted results within 4 seconds of each other on average. However, when each test machine's integrated graphics were used, the AMD 690 based board suffered the most by allocating 256MBs of system memory to the frame buffer, trailing by 10 seconds compared to the 430/6100 and 15 seconds compared to the 430/6150 series board.
In our next test, which is also part of the Worldbench 5.0 suite, a video is encoded using Windows Media Encoder, while an instance of the Mozilla browser is running and navigating through various cached HTML pages. Because the system is multi-tasking with two different applications, this test is more taxing of the system and its dual core processor.
In this test, the 430/6150 series boards enjoyed a lead of 25 seconds on average, while the 430/6100 based board weighed in as the slowest of the three systems. Neither system showed any ill effects from sharing memory with integrated graphics.
|Gaming Performance with 3DMark06 v1.1.0|
When comparing the GeForce 6100, 6150 and the Radeon x1250, the x1250 returned a better overall score at 288. As we established earlier, the 6150 isn't necessarily a fair fight as it is optimized for multimedia applications more so than for 3D. When we factor in the GeForce 6100, which is geared more toward 3D applications, the 690G with Radeon X1250 had its lead narrowed, but still maintained a 56 point edge. None of the IGPs performed very well in this benchmark, however. The fact that they can run 3DMark06 at all is noteworthy though.
With a discreet graphics card installed, all three of the test systems performed within a fraction of each other.
|Gaming Performance with Half Life 2: LC and F.E.A.R.|
Like with 3DMark05, the Radeon X1250 had no trouble topping the GeForce 6150. In this case, the lead was close to 12FPS overall. The GeForce 6100 did manage to narrow the gap, but still trailed the X1250 by just shy of 3 FPS.
Once again, the X1250 held the lead over the GeForce 6150, while the GeForce 6100 narrowed the margins. Regardless, neither system was capable of running F.E.A.R. at playable frame rates at 1024x768, even at medium quality settings. With the discreet card installed, however, performance shot way up and all platforms performed within 3 frames per second of each other.
|HQV Benchmark - DVD Video Quality|
HQV is comprised of a sampling of SD video clips and test patterns that have been specifically designed to evaluate a variety of interlaced video signal processing tasks, including decoding, de-interlacing, motion correction, noise reduction, film cadence detection, and detail enhancement. As each clip is played, the viewer is required to "score" the image based on a predetermined set of criteria. The numbers listed below are the sum of the scores for each section. We played the HQV DVD using the latest version of InterVideo's WinDVD 8 Platinum Suite, with hardware acceleration enabled.
While the tests are very specific, they require the viewer to deliver each score based on certain criteria, so they actually can be somewhat subjective in the end. However, every effort was made to be consistent with our scoring wherever possible.
In general, after scoring the R690G's performance, we found that our results were slightly higher than what AMD scored in internal testing, but we were in the same ballpark. The main difference was with the Picture Detail test, where we felt both the X1250 and NVIDIA's 6150 looked excellent. AMD gave themselves a 5 here, but according to HQV's specific directions on scoring this test we granted them (and NVIDIA) a score of 10. The rest of the test results with the X1250 were on target with AMD's own assessments.
As we stated in the beginning of the benchmarking segment, the GeForce 6150 has a PureVideo engine integrated, so this is the closest competitor to the Radeon X1250 with AVIVO technology here. With no PureVideo engine, the GeForce 6100 proved to be the worst performer, as expected.
In essence, we felt the NVIDIA GeForce 6150 offered a better video quality experience overall. Both Jaggie tests and the Flag test offered some degree of correction with the PureVideo engine and there seemed to be more activity in the noise levels in the noise reduction tests with the X1250. In the end, the Radeon x1250 with AVIVO did a fair job at improving image quality overall, but PureVideo appeared to be slightly better at this point in time.
|Overall Power Consumption: Idle and Loaded|
AMD has made their position on power consumption clear, with much attention being directed to performance per watt. With that said, it seemed appropriate to run a couple of tests to see the wattage draw of each system, as measured at the outlet. In each test, Cool'n'Quiet was enabled and the test beds were left to idle for 30 minutes before their wattage was recorded. Next, we ran two iterations of Prime95's torture test set for maximum power consumption and heat to pin the CPU to 100% on both cores. Again, each system was left in that state for a full 30 minutes before recording the wattage draw.
Running at full load, the MSI K9AGM2 consumed 10 less watts compared to the nForce 430/6150 series board and 2 watts below the 430/6100. When running idle, the 690G weighed in 6 watts less than the 430/6150 and 8 watts less than the 430/6100. This variance may seem small to some, but the best way to look at this is in percentages. When running idle, the 690G consumed close to 8% less power than the 430/6150 and 11% less than the 430/6100 based board. At full load the power consumption difference is closer to 7% less than that of the nForce 430/6150 based board, where the reduction was closer to 1.5% compared to the 430/6100. These aren't exactly make or break differentials obviously but notable none-the-less.
|Performance Summary and Conclusion|
Performance Summary: We found AMD's new 690G chipset to perform on roughly the same level as NVIDIA's 430 series chipsets. In synthetic testing and several other CPU intensive tests, all three boards tested delivered comparable results. Comparing synthetic memory scores, the 690G recorded the lowest scores when 256MB of system memory wall allocated to the IGP's frame buffer, but also showed the most gain when using discrete graphics. In WorldBench testing, the 690G series board tended to trail NVIDIA's offerings, but often benefited nicely from utilizing discrete graphics. Neither system offered any advantages over the other with regard to hard drive performance, but video image quality testing went best with the nForce 430/6150 and its PureVideo engine. The 690G's AVIVO video performance was only a few steps behind though. As promised, AMD delivered on their claims of a better gaming experience with the integrated X1250, with both gaming tests showing the Radeon X1250 as the top performer. Although higher comparable frame rates and playable frame rates are two different things entirely.
AMD's latest chipset offering looks to be a good balance in between the two current mATX offerings from NVIDIA. Comprised of the RS690 Northbridge and SB600 Southbridge, AMD brings a single solution that boasts an integrated X1250 Radeon graphics engine, AVIVO and native DVI/HDMI. OEM's may appreciate AMD's highly integrated solution, that offers competitive performance and a comprehensive feature list. AMD also has a leg-up with native DVI and HDMI support, as well as being a bit more energy efficient. Unlike NVIDIA, the only difference between the 690G and 690V is support for HDMI/DVI. Both offer AVIVO and the same functionality, unlike the 430/61xx series, where the GeForce 6150 comes with PureVideo and the GeForce 6100 does not but offers better 3D performance. We also liked AMD's larger frame buffer option. The nForce 430 line is limited to 256MB, compared to the 690G's 1GB capacity. Ultimately, the AMD 690 series chipset brings a balanced feature-set to the table, that is very competitive with what NVIDIA has to offer.
It has been a long time since we saw a chipset released under the AMD name and the 690 series appears to be a solid offering in the IGP market. However, it does appear that AMD is still playing catch-up with NVIDIA, as the nForce 430/61xx series has been out for quite a while and NVIDIA is poised to release their MCP68 in the not too distant future. With the promise of an integrated GeForce 7 class graphics engine in three variations, in addition to adding integrated DVI/HDMI support, the MCP68 should end up being a more apples to apples comparison to the 690G in the end. We should also consider that NVIDIA's future MCP72 may include DirectX 10 ready graphics. Though, as the saying goes, you can't compare unreleased products to a product that is here today.
Regardless what the future holds, the AMD 690 series stacks up well compared to what is currently available on the market. However, with new offerings on the horizon from NVIDIA, we hope AMD is already looking ahead to the next evolution from the 690 series to maintain and grow in the IGP space.
Look for all major motherboard manufacturers to roll out new models based on the 690 series, including MSI, ASUS, ECS, Shuttle and more. Pricing for motherboards based on the 690 series chipset is not clear at this time. However, our expectation is that they will be priced competitively.