Biostar TF560 - Motherboard Based on
NVIDIA nForce 560 (Socket AM2+) Chipset
It is very nice to see how active the Biostar company has been lately. To begin with, it releases a board based on a new chipset ahead of its competitors - Biostar TF7050-M2. Now it produces a board suited for a new platform even prior to its official announcement (which is expected soon, though). Fortunately, Socket AM2+ is fully compatible with existing processors. Therefore, users will have no problems with finding parts for this board even now. Support of certain technologies that could be required in the future allows building a more "lasting" computer. For us it is an excellent reason to discuss what the future holds and clarify a few issues.
First of all, we should point out that Socket AM2+ platform's connector is not any different from Socket AM2's, the number of contacts and locations of keys are the same. One will be able to install CPUs based on K10 core, which are designed for Socket AM2+ platform, onto modern boards with Socket AM2 and still use their full potential. It is possible, because they support the same DDR2 memory. For AMD replacing a processor socket with an incompatible one can only be caused by modifications in the memory controller built into the CPU - change of memory type or change in the number of channels.
However, besides memory type there are many other parameters characterizing a platform, and it so happens that now is the time for a planned revision of system interfaces:
The HyperTransport bus is used on AMD platform for connecting CPU with the chipset and processors with one another (for a multi-processor system). We are accustomed to thinking (and rightly so) that for a single-processor configuration the current bandwidth of this bus is more than enough. With HT 1.0 at 1 GHz it totals 16 GB/s for both directions of data flow. It is quite possible that if we include three video cards into a single configuration making one of them directly interact with the processor, for example to perform physics calculations in games, we will generate amount of traffic through the bus comparable to its maximum bandwidth value. So it wouldn't be stretching the truth to say that increasing the bandwidth of this bus (converting to HT 3.0) should be expected and will provide an advantage only for the multi-processor server systems, while very few of the desktop PC users will notice any difference. At 2.6 GHz, the maximum frequency according to specifications, total bandwidth will reach 41.6 GB/s. Moreover, even among the server systems "version three" support will be introduced gradually, so as to accommodate the interests of existing systems' users, who want to upgrade them with newer processors. Such "miracles" as ability to increase high-speed channel length up to 1 meter and "on the fly" connectivity aren't really applicable to desktop PCs (unless ECS or ASRock decides to make a board with processor socket on an external module).
Nevertheless, practice shows that trying to prove and demonstrate prematurity of a certain technical decision is only reasonable if its implementation causes unjustified expenditures or lowers benchmark scores (as could have been the case with quick transition to DDR3, for example). Otherwise it is more logical to implement such technology and add a corresponding good-looking line to the specifications. For example, having 2000 MHz (which in terms of "effective frequency" results in 4000 MHz) is by far better than 1333 MHz or even 1600 MHz ;). In other words, introduction of HT 3.0 support to Socket AM2+ platform from the point of view of a uniprocessor PC user can be considered as a nice side effect of this bus' expected physical implementation, but of course not as a reason for upgrading.
PCI Express bus. Support of version 2.0, strictly speaking, is not a definitive characteristic of Socket AM2+ as a platform and depends on the chipset. High-end chipsets coming out this fall, first of all AMD RD790, which are designed to accompany the release of processors based on K10 core, will certainly provide such support. From the practical standpoint though, the effect from doubling the bandwidth of this bus is even less clear than in the case mentioned above. Video cards (the main consumers of this bus' resources) are not in any hurry to get rid of their own memory. On the contrary, they are having more of it and as our tests show are, to some degree, independent devices. For professional systems, tasks concerning high definition format video encoding etc. increased bandwidth of this bus can be useful. Also, raising maximum available power for a video card installed into such port from 75 to 150 W might help. Again, one shouldn't expect any increase in performance, especially in the beginning when even the most progressive users would have to make do with boards of the previous standard. Thankfully, backwards compatibility is provided.
However, if PCI SIG has set this standard on the industry level and switching to it won't make things worse, then postponing its implementation would only give unsubstantiated weight to the support issue, which would distract from the objective comparison of competing platforms based on practical tests.
Adding (separating, to be exact) supply lines for independent power supply of the memory controller and CPU cores will allow putting all CPU cores into sleep mode during idle state, while keeping the ability of peripheral devices to access memory. For example, a chipset integrated graphics core will be able to refresh frame buffer, and adjacent cores in a multi-core system will be able to read memory arrays linked to other processors. It's not hard to guess that in this situation the advantage is once again expected to be enjoyed by servers (or notebooks - in the future), while desktop systems only get such ability "as ballast".
Support of DDR2-1066 is expected to be offered by the new processors. However, the owners of boards based on Socket AM2 may rejoice, because AM2+ introduces no exclusive features to support this standard. Meaning that everyone will be able to enjoy the increased memory bandwidth. The opportunity to witness the operability of such a memory in overclocked mode has been available for quite some time by installing off-the-shelf memory modules with increased frequency. Correspondingly, the change is in adding a standard multiplier to the CPU and the board's BIOS (as it was unofficially done before for DDR-500, only this time official backing by JEDEC is expected). Much more interesting than the fact of support itself is discovering how much of such memory's potential will be utilized by the new integrated controller. According to our numerous researches up till now the bottleneck of memory subsystem has not been the memory itself by rather the platform (true for both platforms, though for different reasons). Of course the answer to this question may only come from testing new processors, while dependency on the board being used traditionally promises to remain minimal.
Summing it up one can make the obvious conclusion that there is no need to look for a board supporting Socket AM2+ even if you want to prepare the platform for switching to new processors ahead of everyone else. If you have to make a decision about upgrading or buying a new computer today it will be enough to purchase a board with Socket AM2, but with functionality meeting your possible demands for the near future. If you fit it with Athlon 64 X2 with enough computing power for the current needs, then it will be very inexpensive by present standards and will serve quite well as a "transitional" platform. In due time you will be able to install Socket AM3 processors on it, including 8-core, because their memory controller will support both DDR2 and DDR3 memory. The boards themselves will, in contrast, only support DDR3 memory and, therefore, will not be able to work with current CPUs, except maybe some rare combined models. As a visual aid we present a table of processor and board mutual compatibility.
Nevertheless, if the Biostar board turns out to be interesting in the rest of parameters, then the Socket AM2+ support criterion should also be considered when choosing a board (given that alternatives are otherwise equal). You might be wondering: why was this particular chipset chosen if a new platform usually attracts those who love everything powerful, whereas nForce 560 is in the economy category? It's simple: Biostar has probably decided to produce at least one model on the AMD platform for each NVIDIA chipset. Thus, it seemed logical to use nForce 560 that came as addition to the series this summer and conforms to the new requirements. This has served two goals - to become the first manufacturer of the Socket AM2+ platform boards and to add a board based on a new chipset to the line of products. Even such a dandy company as Biostar can't afford to treat the enthusiasts ahead of time with a board based on AMD RD790 or NVIDIA MCP72. Well, enough with the opening speeches, let's move on to examining the board itself.
The layout is in accordance with contemporary "style" of Biostar TForce-series boards. In comparison to Biostar TForce550 that we have examined a year ago, the most notable differences at first glance are the passive chipset heat sink and a smaller number of expansion slots. Nevertheless, having three PCI slots is still not bad. In everything else the design has only been improved. In a year Biostar has learned to place power supply connectors near the board edges and to group peripheral ports more conveniently. Alas, again for reasons unknown they have "missed" FireWire, which we have become used to seeing supported on majority of full-size ATX boards. Clearly, it wouldn't have been difficult to find room on the board for such a controller.
The heat sink copes with cooling the chipset, although overclocking fans can benefit from paying attention to the CPU cooler design and finding a model capable of auxiliary chipset cooling.
The three-channel impulse CPU supply voltage stabilizer uses 3 field transistors per channel, 6 solid-state Panasonic capacitors of 820 microfarad each and 3 "regular" United Chemi-Con of 1500 microfarad each. Usage of polymer capacitors is still new to Biostar, so the number of such components is limited. Voltage stabilizer of chipset and memory uses plenty of regular 1000 microfarad OST capacitors in keeping with Biostar's traditions. There are no laid out but unsoldered elements on the board. Board's size is 305x245 mm (a full-size ATX). It is mounted using 9 screws with all corners being secured.
System monitoring (ITE IT8716F-S, from BIOS Setup data):
- Voltage of CPU core, memory, chipset, HT bus, +3.3, +5, +12 V, +5 V Standby and battery;
- Rotation frequency of the 3 fans (CPU, system and chipset);
- Temperature of CPU (by built-in CPU sensor);
- Smart Fan - automatic rotation frequency control mode for the CPU and system coolers. There is an ability to modify the standard algorithm for the CPU fan - to set 4 temperature thresholds:
- CPU Fan Off - stop the fan if temperature is below the set value;
- CPU Fan Start - start the fan if the temperature rises (in other words, one can set different values for start and stop in order to keep fan from switching on and off frequently);
- CPU Fan Full Speed - temperature corresponding to maximum rotation frequency;
- Start PWM Value - a temperature threshold, above which gradual rotation frequency adjustment takes place.
Ports, connectors and sockets on board surface
- Processor socket (Socket AM2+, declared support of all AMD Athlon 64/X2/FX/Sempron processors produced for Socket AM2 and soon to be released for Socket AM2+, after a corresponding BIOS update);
- 4 DDR2 SDRAM DIMM slots (up to 8 GB DDR2-400/533/667/800, dual channel mode of operation);
- 1 PCIEx16 video accelerator slot;
- 2 PCIEx1 slots;
- 3 PCI slots;
- Power supply connectors: standard ATX 2.2 (24 pins), 4-pin ATX12V for CPU supply and 4-pin "peripheral" for additional video card supply;
- FDD slot;
- 1 "chipset" IDE (Parallel ATA) slot for 2 ATA133 devices;
- 4 "chipset" SATA-II (Serial ATA II) slots for 4 SATA300 devices, disks connected to them can be combined into a RAID of level 0, 1, 0+1 and 5;
- 2 headers for 4 extra USB ports;
- A header for an extra LPT port;
- CD/DVD-drive audio-out header;
- Block of analog audio-in/out connectors for the computer's front panel;
- 1 S/PDIF-Out header;
- 3 fan connectors with ability to control revolution rate, CPU (4-pin) and system (3-pin, JSFAN1 connector) ones have "intelligent" BIOS rotation frequency control. In case of using a 3-pin CPU cooler full compatibility including gradual rotation frequency regulation within a temperature range of PWM, which is set in BIOS, is provided.
Board's rear panel (left to right, by blocks)
click to view the board in 3/4 perspective from the rear panel side
- PS/2 mouse and keyboard ports;
- 1 COM;
- 2 USB ports;
- 2 USB ports;
- 2 USB ports and 1 RJ-45 (Gigabit Ethernet) port;
- 6 analog audio sockets (Center/Sub, Rear-Out, Side-Out, Line-In, Front-Out, Mic-In).
- Packing: a standard size box with renovated design for the "T-Series" and a picture of some kind of optical fiber dragon;
- Documentation: instructions sheet (main section in English, but with specifications in other languages);
- Cables: 2 SATA cables with a supply adapter for 1 SATA device, 1 ATA66 cable and a cable for connecting a floppy drive;
- Rear panel plug for corresponding connectors;
- Compact-disk with drivers and brand Biostar utilities.
Besides components listed above the user's guide mentions a rear panel bracket with S/PDIF-Out port and a bracket with 2 USB ports as an option. It's worth noting that it isn't uncommon for Biostar to supply different packages for the same model, so please pay attention to the specified package contents before purchasing.
The set of brand utilities comes in the form of T-Utility suite, including the following applications that one can install separately:
- Over Clock - provides acceleration, including ability to adjust frequency, multiplier and voltage of the CPU core, frequency and voltage of memory and frequency of PCI-Express bus (the latter ability, by the way, is not available in the current BIOS versions);
- Hardware Monitor - monitoring system parameters;
- Fan Control - monitoring cooler rotation frequency and temperature of CPU and system, as well as turning Smart Fan mode on and setting constant rotation frequency for both fans;
- BIOS Live Update - automatic BIOS update with ability to search for a new version on the manufacturer's web-site.
Besides that the package includes Bullguard Internet Security 6.0 antivirus suite (a 90-day trial version).
- Audio controller is based on "chipset" HDA support and Realtek ALC888 codec. It provides ability to connect 7.1 audio systems, connectors for front audio- ins and outs and an S/PDIF-Out;
- Network Gigabit Ethernet controller supports 10/100/1000 Mbit/s and is based on Realtek RTL 8110SC PCI controller.
We have evaluated quality of the integrated audio in 16-bit 44-KHz mode using RightMark Audio Analyzer 5.5 testing suite and ESI Juli@ sound card:
|Frequency response (40 Hz to 15 KHz), dB:
|Noise level, dB (A):
|Dynamic range, dB (A):
|IMD + N, %:
|Channel crosstalk, dB:
|IMD at 10 KHz, %:
Overall rating: Very good. For Biostar this result (typical by measures of the majority of boards with the same HD codec) can be viewed as significant progress, considering that in previous tests the integrated audio controller displayed poor results. We are very glad that Biostar engineers have found a solution to this problem.
Brand technologies and features
- CMOS Reload Program - ability to save BIOS settings profiles in the CMOS area;
- Self Recovery System - automatic restoring of BIOS settings to default values in case of excessive overclocking;
- A utility for rewriting BIOS from a floppy disk that is launched through CMOS Setup.
|Based on jumpers and switches||CMOS reset jumper|| |
|2 buttons (power and reset)
||Allow to switch on and reboot the computer without having to connect corresponding wires from the chassis' front panel to the board|
|Through BIOS based on Award BIOS v6.00PG
||Ability to turn special CPU functions off
||K8 Cool'n'Quiet, SSE/SSE2 Instructions
|Memory timing settings
||Write Recovery Time, Precharge Time, Row Cycle Time, RAS to CAS Delay, RAS to RAS Delay, Row Precharge Time, Min RAS Active Time
|Memory clock rate selection
||Auto, 400, 533, 667, 800 MHz (actually sets multiplier relative to HTT frequency)
|HT bus operational settings
||Frequency (sets multiplier x1-x5 in integer increments) and width (8 or 16 bits)
|Ability to set frequencies of peripheral buses
|Manual distribution of interrupts by slots
|Adjusting FSB frequency
||200-450 MHz in 1 MHz increments
|Adjusting CPU multiplier
||from x4 in integer increments
|Adjusting CPU core voltage
||+0.012-+0.787 in 0.012 V increments
|Adjusting memory voltage
||1.95-2.50 V in 0.05 V increments
|Adjusting chipset voltage
||1.250-1.325 V in 0.025 V increments
For testing we used BIOS N56BA611 06/21/07 version provided by the manufacturer. The aforementioned BIOS capabilities are available in the specified version of the BIOS. Non-standard settings were not tested for operability. The board has a Boot Menu function that allows to express-select the boot disk without having to change settings in the main BIOS menu. A built-in BIOS Setup utility allows rewriting BIOS from a floppy disk without having to load the OS.
For the users that don't want to bother with specifics of the overclocking settings there are three preset overclocking modes named similarly to car engine types (V6, V8 and V12). An acceleration of "Extra, Extreme and Extraordinary" is promised for each correspondingly. Only the CPU clock rate is increased and in our tests we used 2.3, 2.4 and 2.5 GHz.
In order to evaluate the overclocking capabilities of the board and its BIOS, we overclock our testbed CPU to the highest frequency possible that also allows for stable operation. Applying this test procedure, we are able to effectively use all of the test board's supported abilities, including increasing processor core voltage, and if necessary, correcting multipliers and adjusting system and peripheral bus frequencies. However, if lowering Hyper-Transport frequency, for example, doesn't improve overclocking performance, the default multiplier is used instead. RAM is set (by using multiplier correction) to the standard frequency for the modules being used, unless the manufacturer specifies methods for improving memory overclocking, in which case their effectiveness is also explored. In order to evaluate the overclocked system's stability, we load Windows XP and run performance tests built into WinRAR (Tools menu - Benchmark and hardware test) for 10 minutes. It is important to realize that overclocking performance varies by motherboard and is, to some extent, an individual characteristic of each specific unit. For this reason, it is impossible for us, and any other review, to determine the overclocking potential of any board with megahertz precision. The practical goal of our test is to find out if the CPU's high overclocking potential is hindered by the board and to evaluate the board's behavior in non-standard BIOS modes. This test also assesses the board's ability to automatically revert to correct frequencies in the case of system hang-ups, excessive overclocking, etc.
||Clock rate, MHz
||FSB frequency, MHz
||Core voltage (according to BIOS system monitoring), V
||HT bus frequency (multiplier), MHz
|Athlon 64 X2 4000+ (Windsor, 2.0 GHz)
This acceleration is, of course, not record breaking. Increasing the clock rate any further caused errors during Windows start-up and it is hard to even guess where the bottleneck was. The margin for adjusting voltage is, indeed, not wide, but in similar conditions our testbed CPU has been know to reach a value of at least 2800 MHz on many boards.
- Processor: AMD Athlon 64 X2 4000+
- RAM: 2 Kingston KHX7200D2K2/1G (DDR2-800, 5-5-5-15-2T) modules of 1 GB each
- Hard drive: Seagate Barracuda 7200.10 (SATA, 7200 rpm)
- Video adapter: ATI Radeon X1900XTX, 512 MB GDDR3
- PSU: Chieftec CFT-560-A12C
- OS: Windows XP SP2
For comparison we have chosen Biostar TForce550, because in the hierarchy of Biostar products, more specifically in the TForce series, this board is one grade below the one being considered.
||Biostar TF560 A2+
|Data compression using 7-Zip, min:sec
|MPEG4 (XviD) encoding, min:sec
|Unreal Tournament 2004 (Low@640x480), fps
|Unreal Tournament 2004 (Highest@1600x1200), fps
|Doom3 (Low@640x480), fps
|Doom3 (Highest@1600x1200), fps
The differences in performance are minimal.
System boards from Biostar, especially the ones that belong to T-Series, for quite some time now have been proving to be products of high implementation quality. Only the integrated audio controller has been a target of objective criticism. Consequently, today we find no reasons for criticism anymore. Pioneering in the Socket AM2+ support clearly also deserves praise.
However, in our opinion, this specific model has two problems that are inherited from earlier products and definitely need to be addressed in the future. Surely, it is time to introduce FireWire support for full-size boards and root out the PCI versions of network controllers. Switching to PCI Express isn't all that expensive. This situation seems even more strange considering that the nForce 560 chipset already has a gigabit MAC adapter and supports (unlike the nForce 550) the curious First Packet technology (traffic prioritization). Therefore, in this case it would be most logical to simply enhance chipset support by adding a PHY controller.
Write a comment below. No registration needed!