ASUS A7V880 Mainboard Review. First Look at VIA KT880 Chipset

by Ilya Gavrichenkov
04/19/2004 | 09:05 AM

Is Socket A dead? Well, I would say it is a way too early to announce that. Yes, AMD is actively promoting its CPUs with new AMD64 architecture in all market segments, but it in no way means that the manufacturer is about to discontinue the production and technical support of its processors for the good old Socket A. Even the release of the new low-cost Athlon XP processors based on Paris core and designed for Socket754 processor form-factor should not push the Socket A solutions into oblivion. According to AMD’s current plans, the peaceful co-existence of Socket A, Socket754 and Socket939 processors will stay for at least another year or so.

It is true the current Socket A Athlon XP processor family is more than successful. Their production cost is low enough while the demand for these CPUs is still pretty high due to a very well-developed infrastructure. In fact, contemporary Athlon XP processors on Barton core remain among the best solutions for the mainstream and value systems from the price-to-performance point of view. This way, Socket A shouldn’t be disregarded even despite the fact that we keep putting up more and more articles devoted to the fresh new Socket754 products.

The chipset and mainboard manufacturers seem to be guided by the same conclusions as we are. They not only continue manufacturing their old solutions for Socket A systems, but they still announce new solutions for the same platform. Today we are going to pay special attention to two new products of the kind: the new VIA KT880 chipset with a dual-channel memory controller and a new ASUS A7V880 mainboard based on this chipset.

So, as you may have already understood, VIA KT880, which the company promised to release about half a year ago, has finally reached the mass production stage. This is the first (and the last) Socket A chipset from VIA featuring a dual-channel memory controller. The final arrival of the VIA KT880 seems to be putting a full stop in the constant rivalry between VIA and NVIDIA in the Socket A chipset field. Until now VIA didn’t have a chipset as fast as NVIDIA nForce2, although the company did undertake numerous attempts to do it. However, all the company’s previous chipsets, which were intended to compete with NVIDIA nForce2 (and there were three of them: KT400, KT400A and KT600) featured a single-channel memory controller. Keeping in mind that the FSB bus between the CPU and the chipset features limited bandwidth, VIA assumed that Athlon XP based systems will hardly require two 64bit memory channels. Therefore, the company hasn’t yet had a chipset running as fast as nForce2 with its dual-channel memory architecture. That is why now VIA decided to try its luck by introducing a dual-channel memory controller, especially since the controller itself has already been developed for the Pentium 4 PT880 chipset. This way, having combined the memory controller from the PT880 chipset with the Socket A KT600, VIA got a new product called KT880, which on the one hand, looked more advanced than nForce2 from the formal characteristics point of view, and on the other hand, could work not any slower than its major rival due to the new dual-channel memory architecture. Let’s see what kind of chipset VIA got in the long run.

VIA KT880: Quick Glance

I called this section “Quick Glance” because there is not much to talk about regarding the features and specifications of the new chipset. Just look at the flow-chart below:

VIA KT880 includes the already familiar VIA VT8237 South Bridge, which we all know from KT600, PT800 and PT880 that is why most characteristics of the new KT880 is identical to those of the core logic sets mentioned above. This way, the distinguishing features of the new VIA solution are all hidden in the chipset North Bridge. To be more exact, the only distinguishing feature of the newcomer from the KT600 it is actually based one is the support of two memory channels, which ensures 128bit memory bus and allows reaching peak bandwidth of 6.4GB/sec with DDR400 SDRAM.

In order not to make too much of the bla-bla talking, I suggest comparing the formal features of the VIA KT880 with those of its direct competitor – NVIDIA nForce2 Ultra 400: the dual-channel Socket A chipset from NVIDIA. For a better comparison, I will also include the specifications of the VIA KT600, the predecessor of our today’s hero:

The comparison of the nForce2 Ultra 400 and KT880 features is evidently in VIA’s favor. It really allows implementing 4 memory slots instead of three, which is a pretty important thing for dual-channel configurations. Besides, it also ensures Serial ATA support and allows creating RAID arrays, features 8 USB 2.0 ports instead of the six ports supported by nForce2 Ultra 400. To be fair, I have to point out that the more advanced NVIDIA MCP-T South Bridge also features a few unique peculiarities. For instance, it contains a fully-fledged sound APU processor and supports IEEE1394 interface. However, the absence of the Serial ATA controller is a significant drawback, which casts a big shadow over all the previously listed advantages. This way, when we look at the formal characteristics of the reviewed solutions our preferences will move towards VIA KT880.

However, even though this topic is not directly connected with the major goal of our today’s review, we have to stress that NVIDIA is about to introduce new South Bridges in the near future, which will allow nForce2 Ultra 400 to offer more attractive specifications than those of the today’s KT880. Within the next few weeks, they are planning to announce RAID MCP and Gigabit MCP, with the integrated SerialATA RAID controller and a Gigabit network controller with a hardware Firewall:

The new South Bridges from NVIDIA are due in June-May timeframe.

But even if NVIDIA keeps its promise and will not delay the launching of its new South Bridges, VIA KT880 will still have one very important advantage over the rival. It is the price. VIA chipsets for Socket A platform cost considerably less than the competing solutions from NVIDIA, which makes the VIA stuff much more popular. For instance, VIA KT880 mainboards will cost about the same as VIA KT600, which is about $10-20 less than NVIDIA nForce2 Ultra 400 solutions cost.

In conclusion I would like to point out one more thing, which will be of interest to overclockers in the first place. Speaking about VIA KT880 as of VIA KT600 + dual-channel memory controller I left out one very important fact. Namely, I have to stress that VIA KT880 has finally acquired the so long-awaited option to clock the AGP/PCI buses asynchronously, which is also known as AGP lock. In other words, now when you want to increase the FSB frequency beyond the nominal values, such as 100/133/166/200MHz, you will also be able to avoid synchronous growth of the AGP and PCI frequencies together with the FSB one, so that they exceeded the nominal 33/66MHz. This way, now it will be much easier to overclock the processors on VIA KT880 based mainboards, because you will not need to keep an eye on the proper functioning of the AGP and PCI devices during overclocking. As a result, VIA KT880 will be at least no worse than NVIDIA nForce2 from the overclocking point of view, because the latter has had the option of asynchronous FSB and AGP/PCI buses clocking from the very beginning.

ASUS A7V880 – First VIA KT880 Based Mainboard

Just like in case of VIA KT600, the first mainboard on the new KT880 was also released by ASUS. The newcomer got a very predictable name: ASUS A7V880. You can see it on the photo below:

And here are the product specifications:

Since we got a preproduction sample for our tests, it is a way too early to discuss the package contents and the bundles, which will be supplied together with the board. Therefore, let’s pass over to the features of the new ASUS A7V880.

First of all, I would like to stress that the features list for this product doesn’t look like that of a High-End solution. In fact, there is only one Gigabit network controller from Marvell onboard, which should enrich the list of product features. Nevertheless, due to highly functional VIA VT8237 South Bridge, ASUS A7V880 supports all necessary protocols and interfaces. I should say that this ASUS’ approach is quite logical. Socket A processors little by little move to the low-cost category that is why the mainboard for these CPUs also shouldn’t be very expensive.

In fact, the absence of any additional onboard controllers will make the off-the-shelf price of ASUS A7V880 almost equal to that of the VIA KT600 based ASUS A7V600. In other words, you will be able to buy a modern Socket A mainboard with a dual-channel memory controller for about $80.

This way, ASUS A7V880 will represent an excellent combination of low price, rich features and high performance, which will undoubtedly make it a very attractive product for the users looking for inexpensive mainstream and value PCs. Due to pretty low price the life cycle of ASUS A7V880 mainboards, that is the time when the production of these mainboards will be in progress, will make at least one year, until AMD discontinues the top Socket A processors completely.

Of course, ASUS A7V880 will be able to work with any Socket A CPUs with 200/266/333/400MHz bus frequency without any problems. Also this mainboard will have no issues supporting DDR266/DDR333/DDR400 SDRAM. The mainboard can work with the memory in synchronous and asynchronous modes, and allows using single-channel 64bit and dual-channel 128bit configurations. The BIOS Setup of ASUS A7V880 is based on AMI microcode, boasts sufficient flexibility of the memory configuring options and allows adjusting all the key timings. Note that just like with the NVIDIA nForce2 mainboards the synchronous mode is the highest performance mode for the VIA KT880, that is when the memory working in the dual-channel configuration supports the same frequency as the FSB.

By the way, note that unlike NVIDIA nForce2 based mainboards, ASUS A7V880 features 4 memory slots, 2 for each channel, which makes its much easier to upgrade the memory subsystem towards larger overall memory capacity.

As for other expansion options, ASUS A7V880 doesn’t prepare any surprises for us. The mainboard is equipped with 5 PCI and 1 AGP 8x slot, which will remain demanded even despite the fact that PCI Express x16 is coming out very soon now. The owners of Socket A systems will hardly hunt for the top VGA cards preferring more attractive solutions from the price-to-performance point of view. Just like many other ASUS mainboards, the new A7V880 features the company’s unique brand name WiFi slot intended for wireless network cards from ASUS. This way, ASUS A7V880 owners will have an easy option to expand the functionality of their mainboards in terms of 802.11a/b and later 802.11g standards support at a very low expense.

Due to the VIA VT8237 South Bridge, ASUS A7V880 supports two Serial ATA-150 channels and allows creating RAID 0 or RAID 1 arrays. Also the mainboard features 8 USB 2.0 ports, with four ports laid out on the mainboard rear panel, and four represented as ready to work PCB connectors.

As for the integrated sound, ASUS A7V880 features a six-channel ADI AD1888 AC’97 sound codec with SPDIF Out support (you can see its coaxial connector as well as three regular audio jacks on the rear panel of the board). Although the manufacturer considers this codec a not very expensive one, it still supports the brand new AC97 2.3 standard, which makes it much easier to connect all external audio devices to the system. Also due to the thoroughly laid out analog sound tract, we revealed no issues with the performance of the onboard sound solution by ASUS A7V880.

Just like the majority of contemporary mainboards, ASUS A7V880 is equipped with an Ethernet controller from Marvell, which also supports the technology called Virtual Cable Tester. You can read more about this technology on Marvell’s official web-site. In our turn, we will just point out the major idea behind this technology. Virtual Cable Tester allows tracking all most common problems with the Ethernet cable connected to the chip. Among these problems we could list the following: torn spots or dead shorts (the technology detects the problem with 1 meter precision), pair swaps, pair polarity problems and a few others.

The PCB used as a basis for ASUS A7V880 looks quite original and has nothing in common with any other mainboards. In fact, this is not at all surprising, because VIA KT880 chipset is not pin-compatible with the previous VIA’s North Bridges. Since there are no additional onboard controllers the PCB design is pretty simple is hardly has any significant drawbacks or bottlenecks. The only thing we were quite upset with was the location of the FDD connector, which was placed in front of the PCI slots, so that the FDD cable will always be in the way either for the PCI or the AGP cards.

Note also that the mainboard doesn’t require any PSUs with an additional 12V connector. There is simply no four-pin 12V ATX power supply connector on ASUS A7V880 mainboard at all. The chipset North Bridge is covered with a massive passive heatsink, which is more than enough to ensure proper cooling of the product, since VIA KT880 doesn’t generate too much heat.

The processor voltage regulator on ASUS A7V880 is designed according to a two-channel scheme. By the way, since we came to speak about the CPU voltage regulator, I would like to mention also that there are quite a few empty spots right around the area where the CPU power supply circuitry is located. We managed to find four empty spots intended for MOSFETs, three spots for large capacitors, and more than a dozen of spots for smaller electronic components, such as resistors and small capacitors. Frankly speaking, I got the impression that ASUS engineers suddenly decided to act economically ad removed the “unnecessary” elements from the PCB at all. However, this hasn’t affected the mainboard stability at all, I should say. At least I haven’t revealed any problems with the mainboard performance during our test session.

I would like to devote special attention to the connectors laid out on the back panel of the mainboard. There is one parallel and one serial port (the second serial port is also laid out on the PCB, but the pin-connector is not available in the corresponding spot), four USB 2.0 ports (the other four ports are laid out as pin-connectors on the PCB), RJ45 network connector, three audio jacks and a coaxial SPDIF Out.

The overclocking friendly features supported by this mainboard are also of interest to us. The thing is that the peculiarities of the VIA KT880 chipset, namely the possibility to clock the AGP and PCI buses asynchronously, ASUS A7V880 may turn into an overclockers’ favorite product, just like any other mainboards based on the same chipset, actually. Especially, since overclocking with an nForce2 mainboard always prepares some unpleasant surprises, such as BIOS resetting, floating frequency values, etc.

So, the BIOS Setup of ASUS A7V880 offers overclockers the following tool kit:

• Allows changing the FSB frequency within 100-227MHz interval with 1MHz increment;
• Allows changing the CPU clock frequency multiplier within the interval from 5x to 22.5x;
• Allows clocking the AGP/PCI buses asynchronously, that is when the frequency of these buses remains at the nominal value, i.e. 66/33Mhz independent of the FSB frequency at the given moment. Otherwise, AGP/PCI frequency is connected with the FSB frequency linearly and increases every time the FSB frequency goes beyond the nominal values;
• Allows adjusting processor Vcore. The voltage can be set from 1.65V to 1.85V with 0.05V increment;
• Allows changing the Vdimm voltage. The available range starts at 2.55V and finishes at 2.85V with 0.1V increment.
• Allows adjusting the AGP bus frequency within the interval from 1.5V to 1.8V with 0.1V increment.

Formally, one of the biggest advantages of ASUS A7V880 from the overclocking point of view is the possibility to change the CPU clock frequency multiplier within a very wide range and the opportunity to lock AGP/PCI bus frequencies during overclocking. However, ASUS A7V880 is not ideal and also has some drawbacks. In particular, the ranges for voltages (especially the processor Vcore) are very narrow, and the top FSB value is too low for serious overclocking attempts, especially since the BIOS Setup allows asynchronous clocking of the AGP/PCI buses.

To check out how ASUS A7V880 behaves during overclocking, we decided to check how the things will work if the FSB frequency exceeds 200MHz. for this experiment we took an Athlon XP 3200+ processor, tried to increase the FSB frequency to the maximum at the same time having reduced the CPU clock frequency multiplier to 9.5x. In this test all the voltages were set to their nominal values, and asynchronous AGP/PCI clocking was enabled, of course. The memory frequency was set synchronous with FSB, i.e. we selected the DDR400 SDRAM setting for this CPU. Overclocking appeared absolutely no problem for PC4200 Enhanced Latency SDRAM from OCZ Technologies.

I should say that these experiments didn’t take us too long. When we set the FSB to the maximum value supported by ASUS A7V880 – 227MHz, the system took off perfectly smoothly and didn’t reveal any stability problems throughout the entire test session.

In fact it means that ASUS A7V880 features a much higher overclocking potential. However, the user cannot take full advantage of it because of the limited opportunities offered by the BIOS Setup. At the same time, I have to stress that this problem can be completely eliminated after the new BIOS versions come out. The clock frequency generator used on this mainboard, ICS94228, is truly capable of generating much higher FSB frequencies than the notorious 227MHz.

This way, the VIA KT880 chipset appeared a very good basis for overclocker’s mainboards. No secret that most NVIDIA nForce2 Ultra 400 based mainboards have the maximum FSB frequency lying between 220-230MHz. Our today’s hero, ASUS A7V880 on VIA KT880 managed to easily reach this top and failed overcome it only because of the non-finalized BIOS.

In conclusion I would like to say a few words about the hardware monitoring implemented on ASUS A7V880. Firstly, I have to point out a few things that disappointed us to a certain extent. Unfortunately, there are only two cooler connectors. Other than that there are no more reproaches. Only one more thing appeared not very pleasing: the CPU die temperature is measured with the thermal resistor located under the processor socket. Although the overheating protection system works using the thermal diode built into the processor core.

Theoretically, this mainboard should also feature Q-Fan technology, which allows adjusting the fan rotation speed depending on the CPU temperature, in order to reduce the noise level. Moreover, ASUS A7V880 requires several parameters to configure this technology: the CPU temperature when the fan starts at its minimal rotation speed, the power required to start the fan at its minimal speed, and the CPU temperature when Q0Fan is disabled, so that the fan starts rotating at its maximum speed. This way, Q-Fan of ASUS A7V880 is a potentially very flexible technology, which allows reducing the noise level of the system based around this mainboard. However, the current BIOS version doesn’t have this technology implemented in full yet. So, let’s wait for the new BIOS versions.

Testbed and Methods

Well, now it’s high time we got to one of the most important and interesting parts of our today’s test session. The major intrigue implies: will the new dual-channel VIA KT880 chipset manage to catch up with NVIDIA nForce2 Ultra 400? Or the additional memory channel appeared of no help to VIA? Let’s find this out now…

To answer the questions set above, we assembled two testbeds: one on NVIDIA nForce2 Ultra 400 and one on VIA KT880. in both cases we took solutions from one and the same mainboard maker – ASUS. So, the overall configuration of the test systems looked as follows:

• AMD Athlon 3200+ CPU (2.2GHz, 400MHz FSB);
• Mainboards: ASUS A7V880 (VIA KT880, BIOS 1001.018) and ASUS A7N8X-E Deluxe (NVIDIA nForce2 Ultra 400, BIOS 1010);
• 2x256MB DDR400 SDRAM from Corsair with 2-2-2-5 timings. The memory worked in dual-channel mode;
• ASUS RADEON 9800 XT graphics card;
• Western Digital WD400JB HDD.

The tests were run in MS Windows XP SP1, the BIOS of both mainboards was set for the maximum performance. Moreover, I would like to point out that the actual FSB frequency on both boards was a little beyond the nominal value and equaled 202MHz instead of the nominal 200MHz. This is a peculiarity of all ASUS’ mainboards.

Performance

First of all, let’s check the memory subsystem performance of the nForce2 and KT880 in synthetic benchmarks. This will allow us to draw conclusions later, when we run a few real applications.

So, according to SiSoft Sandra 2004 results, as well as to the data obtained in ScienceMark 2.0, the memory controller of VIA KT880 is faster than the memory controller of NVIDIA nForce2 Ultra 400. Moreover, the VIA KT880 provides lower latency for CPU addressing the memory, as well as higher bandwidth. Is this enough to congratulate VIA on the victory? Not yet! The matter is that the performance of the processor-memory bus, which we are trying to measure with the help of these synthetic benchmarks is not limited by the memory subsystem performance in this case. Here the limitations are imposed by the processor bus, which provides twice as low bandwidth as the dual-channel memory subsystem. That is why the results obtained in these synthetic tests may not reflect the actual memory subsystem performance of the tested platforms.

Moreover, as we have already said a few times, NVIDIA’s major trump is not their dual-channel support. The strength of NVIDIA chipsets hides in the DASP unit (Dynamic Adaptive Speculative Pre-Processor). Its second version is implemented in NVIDIA nForce2 Ultra 400. Theoretically, this unit is just an additional buffer (or cache, if you want), built into the chipset North Bridge. Then this buffer is loaded with the data, which should be required by the CPU later, as the logical part of the DASP unit assumes. So, if the predictions and data sampling from the memory have been successful, the latencies during work with real applications can be significantly reduced due to more efficient use of the available memory bandwidth. It is exactly due to DASP that NVIDIA’s chipset gets some chances to outperform VIA KT880 in real applications, even though its results in the synthetic benchmarks are far from being impressive.

Well, let’s see.

Well, this is exactly what we have been expecting. Due to the DASP unit, NVIDIA nForce2 Ultra 400 chipset didn’t let the new VIA KT880 take the lead in real applications. In 19 benchmarks out of 25, the leading position belongs to the dual-channel solution from NVIDIA. In order to evaluate how much slower VIA KT880 appeared than its rival, take a look at the graph below:

So, the average performance of VIA KT880 is about 0-3% below the leader. However, there are a few more interesting things. Take WinRAR, where NVIDIA chipset with the DASP unit outperforms VIA KT880 by the good 6%. At the same time, in Aquamark3 VIA KT880 is on the contrary, a little bit faster than nForce2 Ultra 400. However, let me repeat once again: the NVIDIA chipset demonstrates evidently higher performance than the newcomer from VIA.

Conclusion

Well, no sensation happened. Another VIA’s attempt to create a chipset outperforming NVIDIA nForce2 didn’t succeed. Due to NVIDIA’s intellectual DASP unit, nForce2 Ultra 400 remains the fastest Socket A chipset today.

However, this doesn’t at all mean that VIA KT880 is a failure as a product. In fact, summing up all the factors you take into account when making your decision about a certain product, I have to admit that VIA KT880 will be an extremely successful chipset. Look here. There is only one major drawback about VIA KT880: not the fastest performance. Is this a dramatic drawback? I don’t think so, because the average lag between VIA KT880 and NVIDIA nForce2 Ultra 400 is only 1.3%. But unlike its rival, VIA KT880 boasts the whole lot of advantages. Firstly, KT880 is less expensive. Secondly, it boasts more advanced features, such as SerialATA and 8 USB 2.0 ports support. Thirdly, its overclocking potential is not any worse than that of nForce2, while the VIA chipset doesn’t suffer from the problems of nForce2 during overclocking (when the BIOS gets corrupt and the frequencies “float” around). I believe that these arguments are more than enough to state that VIA KT880 will definitely turn into a more attractive purchase in the nearest future than an NVIDIA nForce2 Ultra 400 based mainboard.

And in conclusion I would like to mention that the ASUS A7V880 mainboard we have just reviewed is a worthy product on the new core logic set from VIA. It is inexpensive, rather fast and stable. It has all the necessary features, boasts successful design and suits greatly for overclocking experiments. In fact, if ASUS solves all the above pointed issues with the BIOS, this mainboard would become the No.1 I would recommend to all Socket A platform owners. And in the meanwhile we are a little bit concerned with the slightly limited opportunities in terms of CPU overclocking.