Quality and Setup of the Abit BH7
| Lean, mean, orange machine
The first thing that strikes you about the BH7 is the color of the PCB. This would have to be the first bright orange board we have seen. The layout of the board is well thought out with ample space between the various components and connections. The "busiest" section of the board may well be the power array. The capacitors that line the CPU socket are all made by Rubycon, which are some of the most reliable out there. This should prevent problems that have plagued some other boards in the past. We actually ran into this problem with an Abit BE6-II ourselves recently. In fact, all of the capacitors on the board are Rubycon, and coupled with the 4-phase power that Abit uses on the BH7 should increase stability, even when overclocking. Abit also claims that this 4-phase power circuitry will allow the BH7 to work with forthcoming 800MHz FSB Pentium 4's as well, perhaps prolonging the lifespan of the board in the process.
The device connections consist of two ATA100 ports, a floppy drive port, and a Serial ATA connection. The most noticeable feature about these connections would be the way the IDE ports are positioned, mounted on their sides and placed on the edge of the board. Rarely used, this mounting of the IDE ports helps prevent restriction of airflow around the case with some neat folding of the IDE cables. The Serial ATA connector is controlled by the Marvell 88i8030 bridge, which replaces the one by Silicon Image listed on Abit's website. This bridge is akin to using Abit's SERILLEL adapter in the second IDE port, and in effect cancels it out. That is, if a hard drive is connected to the Serial ATA connector, the 2nd IDE ports is rendered inoperable. While we applaud Abit in tinkering with new technology, we find the application of the SERILLEL bridge to be a shortcoming of the board. For this review, at least, we did not suffer from this since we had no SATA drives for testing.
In what seems to be a rarity these days, the DIMM slots are finally placed far enough away from the AGP slot that no problems arise when swapping out memory sticks when a video card is inserted into the slot. We feel this is partially due to the relative lack of features on the board, but is a nice move all the same. The location of the 20 pin ATX connection is also satisfactory, placed along the edge by the floppy port, and away from the other components. The on-board components all come by way of RealTek. These consisted of 6-channel audio supplied by the popular ALC650 chip, 10/100M Ethernet connections by the RTL8101L chip, and the RealTek RTM360-11R clock generator, which Abit often uses for overclocking purposes. The reason being this chip supports up to 250MHz FSB while allowing fixed frequencies for the AGP and PCI buses.
The external connections
included the standard PS/2, parallel, and serial
ports. For full support of the 6-channel audio, Abit included the three usual audio jacks for line-in, line-out, and mic-in, but also had an additional two jacks for the center/subwoofer channel and the surround speakers. In addition, there were also two S/PDIF jacks, one each for input to or output from digital devices. For external connections to other devices, there are four USB 2.0 ports and a RJ-45 LAN jack standard. If the four USB ports are not enough, we should mention that there is another USB 2.0 header on the board for connecting another 2 USB ports.
THE BIOS AND SETUP:
As with all current Abit boards, the Phoenix/Award BIOS incorporates Abit's SOFTMENU technology. In this section we found all of the settings necessary for controlling the front side bus, CPU, and AGP/PCI bus speeds. The FSB can be raised in 1MHz steps from 100MHz all the way up to 250MHz. With this wide range of speeds, Abit has expanded on the usual voltage options. Normally with Abit boards, we have seen options for raising the VCore in presets of 5%,10%, and 15% over the default 1.5V. The BH7 has double the number of options, allowing for 20%, 25%, and even 30%, bringing the VCore up close to 2V. While we briefly went up as high as 30% during testing, it is not recommended for prolonged use. The AGP and PCI bus speeds can be locked in at 66/33 MHz, which prevents devices using these buses from running too far out of spec, and hindering overclocking of the system. We also found options for raising the DIMM voltages in .1V steps up from 2.55V to 3.05V, but the AGP voltage only had two options, 1.5V or 1.6V. The memory speed is also determined in the SoftMenu, by selecting 1:1, 3:4, or 4:5 dividers depending on the CPU/RAM speed ration desired.
Further tweaking of the memory could be found in the Advanced Chipset Features. There we could leave the memory set "By SPD", or manually configure the CAS Latency, Tras, Trp, and Trcd, but further optimizations such as the Command Rate were not present. The PC Health Status was helpful in that it can either alert the user to a potentially harmful situation, or shut down the system entirely. This actually came into play while we were overclocking the BH7. While trying to achieve a stable overclock at a FSB of 165MHz, we had the VCore set at 25% while using the stock Intel cooler. An alert rang out soon after benchmarking the system, alerting us that the temperature threshold had been met.
| || |
| Overclocking the Abit BH7 |
| A little tweak here, a little voltage there, and Voila! |
Overclocking the BH7 was as easy as entering in a value for the front side bus in the SoftMenu and saving our settings. We locked the AGP and PCI bus speeds to prevent any problems from arising there. We were able to go directly up to 152MHz before we hit our first snag. Starting up Windows XP immediately caused Fatal System Errors. At this point, we raised the CPU voltage to 10% over the default and continued on up to 160MHz. After a few more errors or failed starts, we lowered the memory timings from their optimized numbers to 'By SPD'. This allowed us to continue, but we were alerted at 165MHz that the CPU was running a bit too hot. Definitely an issue where a better cooling solution was needed. We swapped out the stock Intel cooler and installed a Zalman fan cooler that can be seen in one of the board shots above. Even though this lowered the CPU temperature by a good 10 degrees, we found that benchmarking the system was near impossible. We did not completely crash Windows, but we could not complete the longer, more involved tests. We went as high as the full 30% that the BIOS allowed for raising the CPU voltage, but to no avail. We backed off a few MHz on the FSB to obtain a completely stable overclock at 163MHz, and then lowered the voltage to run the CPU a bit cooler. For some quick calculations, by raising the FSB to 163MHz, we were effectively using the CPU at 3.09GHz, while the memory was running at 202MHz.
System Comparisons and Sandra Scores