Gigabyte X38-DQ6 (Intel X38)
||LGA775 Pentium 4 FSB 1066/800/533 MHz
Celeron Conroe/Prescott FSB 800/533 MHz
Dual-Core Pentium4 Smithfield/Presler
Yorkfield, Wolfdale FSB 1333/1066/800 MHz
Quad-Core Kentsfield, Dual-Core Conroe/Allendale
Audio Intel HDA
12 USB 2.0
2x Gigabit Ethernet
We'd like to start this review with the question to the reader: which modern chipset for the Intel platform is the top-end? I am sure you answered this question with some delay - this is NVIDIA nForce 680i SLI.
And what about the AMD platform? Is it nForce 590 SLI? Wrong you are! After the release of the Core architecture, AMD left the top-end sector.
We've been looking forward to the release of the new top-end chipset Intel X38, and with some skepticism as well. Intel's latest high-end chipset was released quite a long time ago (I mean 975X), and motherboards on its base are no longer treated by users as a goal of purchase (therefore, these boards almost vanished off the store shelves). The reason for that is straightforward: the chipset does not support Intel's new processors with the 1333 MHz bus, and the used south bridge (ICH7(R) is outdated both morally and technically. We also note that at overclocking capabilities the 975? lost to both P965 and P35. In the end, motherboard manufacturers had to wringer P35, and we must admit they've done a good job of that. Motherboards like ASUS Blitz or Gigabyte P35-DQ6 can be rightfully regarded as ultimate high-end products, and novelties based on X38 are going to have a really hard time competing against them.
As regards the pessimism with respect to X38, its cause is just behind that. It suffices to look through the major traits of the new north bridge. First, X38 supports all modern processors as well as processors of the nearest future. I mean the products manufactured following the 45-nm process technology with the 1333 MHz bus. In fact, P35 offers the same functionality. On the other hand, Intel is planning to migrate its processors to the 400 MHz bus (1600 QPB) in the nearest future, and some manufacturers have already "enabled" support for such processors on motherboards based on the X38 chipset. But most likely (the X38 is not yet announced), the X38 will not offer official support for this bus because in 2008 Intel is releasing another high-end chipset - X48 which will definitely offer support for this increased frequency. As our tests showed, a well-developed motherboard based on P35 is already now capable of running at 500 MHz (i.e. 2000 MHz QPB), which almost levels the "official" advantage of X38 at overclocking.
Secondly, Intel accentuates the presence of a modified memory controller in X38 which should provide a high performance level. But such statements are normal to the release of each Intel's high-end chipset starting with 875X. But in fact there were either no difference between high- and middle-end chipsets or, it was extremely minor.
Quite probably, the advanced memory controller on X38 will provide a speed gain with only DDR3 memory (I mean the Extreme Memory technology). But by that time when the chipset turns to be relatively affordable (or, in other words, becomes equal at price to DDR2), at least one more generation of Intel chipsets will have been released. But it is not quite correct to treat the Extreme Memory as a solely marketing trick. The thing is, this is some sort of an analog to the NVIDIA EPP (Enhanced Performance Profiles) technology which allows optimizing operation of the chipset for some specific memory modules in terms of setting secondary latency timings.
What will that give in practice? I don't think this will give much - since motherboard manufacturers usually test and optimize the BIOS thoroughly to operate a limited number of "overclocker" modules. Therefore, the Extreme Memory technology is more a plus to the developers rather than for the end users. We also note that this technology is really needed for Intel to compete against NVIDIA.
Thirdly, Intel X38 is the first product that offers support for the PCI Express 2.0. The number of bus lanes is quite enough to arrange two slots having 16 lanes in each. At the same time, there will be some "vacant" lanes for PCI Express x1 slots, for Gigabit LAN controllers, for communication between the north and south bridges over the DMI bus, and quite probably for another PCI Express x8 slot.
Again, what will that bring in practice? To start with, the video cards will not run faster because their performance depends on the core architecture, on the core frequency and the graphic memory. As regards the graphic subsystem in general, the data transfer rate via the PCI Express bus is not a bottleneck. Also, increase in the bandwidth (up to 16 Gb/s) by no means affects the graphic performance. Clearly, support for PCI-E 2.0 by the motherboard will result in the emergence of respective video cards for which the capabilities of PCI-E 1.0 will no longer be sufficient. But today the user of X38-based motherboards will not get any advantage or harm from PCI-E 2.0 since PCI-E 2.0 is backward-compatible to PCI-E 1.0.
We should also note that two PCI-E slots with 16 lanes each will make it possible to use the AMD Crossfire technology in the maximum performance mode. By analogy with NVIDIA SLI, we can assume that the difference in speed for the 16+16 mode will (on the average) be about 5% higher than the sped in the 8+8 mode, and much higher than the speed in the 16+4 mode. In other words, users of AMD Crossfire will get a chipset that most fully reveals the potentials of this technology. We have to admit that before the release of X38 the users of Crossfire had a very scarce choice: symmetrical division of lanes was officially supported by only the i975X chipset (following the 8+8 scheme) which, as we stated above, has irretrievably outdated. But the modern chipset P35 supports division of lanes only following the 16+4 scheme only, which is not suitable for the Crossfire. In fact, ASUS implemented the 8+8 scheme on the Blitz board based on P35, but for that an additional chip was needed. That is, this is an exception which confirms the rule.
By the way, we are bringing in the diagram of the Intel X38 chipset as applied to Gigabyte X38-DQ6:
As regards the expansion options of X38, they are absolutely the same as in P35, since in both cases the same south bridge - ICH9(R) - is used. Therefore, we can formulate the conclusion regarding the X38 chipset in the following form: "Motherboards based on X38 are aimed at overclockers and computer enthusiasts who use the AMD Crossfire technology". But if you don't use the Crossfire, and you don't need the bus speed higher than 500 MHz, then it turns out that a X38-based motherboard will nor of any need to you.
As regards the various rumors around X38, we should note two most important points. The first one regards the developments of both the chipset itself and motherboards on its base. You should not take these rumors for granted: only motherboards of 100% proven design and correct stepping of the chipset will be shipped for sales. The root of these rumors is the same as usual - testing of engineering samples of boards with debug versions of the BIOS and at the early stages of the chipset steppings. Actually, we'll be testing an engineering sample of Gigabyte X38-DQ6, so all the found shortcomings (if found) will be respectively commented.
Running ahead, we say that we have not found any technical shortcomings.
The second rumor was more interesting: it touched upon the support for SLI by X38-based motherboards. We should note it first that X38 does not officially support the SLI technology. The reasons for these rumors are the following. This is above all the info leakage from Foxconn disclosing that their motherboard based on X38 will be equipped with NVIDIA's drivers. That is feasible both theoretically and practically: the NVIDIA SLI technology works on any Intel motherboard with symmetrical distribution of PCI Express lanes. But for that only specially modified drivers are needed. And such drivers are developed by hardcore computer enthusiasts, which is slow and the results are far from "good". NVIDIA itself is not going to make the life of Intel easier since it is competing against the latter on the market of system logic, and in the nearest future – on the market of graphics.
The second reason for rumors regarding X38 and SLI is fabulous and elegant at the same time: one of the manufacturers has allegedly installed a single-chip chipset by NVIDIA with support for SLI on the X38-based motherboard. From the technical viewpoint, that looks fabulous, but if the plan is embodied into reality, this product would enjoy a success. The reason is simple - top-end motherboards based on X38 will cost about $300, and the cost of additional chip priced at $40-50 will not radically change the final price.
Gigabyte X38-DQ6 Specifications
||- Intel Pentium 4 (Prescott (2M)/Gallatin/CedarMill) with the bus speeds 1066/800/533 MHz;
- Dual-core Intel Pentium D/EE (Smithfield/Presler) with the bus speeds 800/1066 MHz;
- Intel Celeron-D (Conroe-L, Prescott) with the bus speeds 800/533 MHz;
- Support for Intel Core 2 Duo/Quad (Kentsfield (4 cores), Conroe/Allendale (2 cores)) with the bus speed 1066 MHz;
- Support for Intel Yorkfield, Wolfdale with the bus speeds 1600*/1333/1066/800 MHz;
- Socket LGA775;
- Support for HyperThreading processors;
||- North bridge Intel X38 (MCH);
- South bridge Intel ICH9R;
- Interbridge communication: DMI;
||- Four 240-pin slots for DDR2 SDRAM DIMM;
- Maximum memory capacity – 8 GB;
- Supported memory DDR2 533/667/800/1066;
- Dual-channel memory access;
||- Two PCI Express x16 slots;
||- Two 32-bit PCI Bus Master slots;
- Three PCI Express x1 slots;
- Twelve USB 2.0 ports (8 integrated + 4 additional);
- Three IEEE1394 (Firewire; 2 integrated + 1 additional);
- Integrated High Definition Audio 7.1;
- Two Gigabit Ethernet LAN controllers;
||- FSB adjustable within 100 to 700 MHz in 1 MHz increments; multiplier adjustable;
- Adjustable voltage on the CPU, memory, FSB, PCI-E, and the chipset (MCH);
- Gigabyte C.I.A 2;
- EasyTune 5;
- 1 line for UltraDMA133/100/66/33 Bus Master IDE (Gigabyte SATA2; with support for up to 2 ATAPI devices);
- Support for SerialATA II (6 lines - ICH9R, with support for RAID 0,1,0+1, 5);
- Support for SerialATA II (2 lines - Gigabyte SATA2, with support for RAID 0, 1,JBOD) ;
- Support for LS-120 / ZIP / ATAPI CD-ROM;
||- 2x8 Mbit Flash ROM;
- Award BIOS Phoenix with support for Enhanced ACPI, DMI, Green, PnP Features and Trend Chip Away Virus;
- Gigabyte Q-Flash;
- @BIOS, and Face-Wizard;
- Gigabyte Virtual Dual BIOS;
||- One port for FDD, one serial and one parallel port, ports for PS/2 mouse and keyboard;
- STR (Suspend to RAM);
- SPDIF In/Out;
||- Wake-up on modem, mouse, keyboard, LAN, timer, and USB;
- Main 24-pin ATX power connector;
- Additional 8-pin power connector;
- Additional 4-pin Molex power connector;
||- Monitoring the temperature of the CPU, system, voltages,rotational speeds of the four fans; |
||- ATX form factor, 245 mm x 305 mm (9.63" x 12");
The motherboard Gigabyte X38-DQ6 is packed within a huge box having holographic coating, equipped with a handle to carry it around.
One page tips up, and the user can get a brief idea of all Gigabyte's proprietary technologies, as well as watch the board through a transparent window.
- User's Manual in English + Quick Setup Manual;
- 1x software & drivers DVD;
- One ATA133 cable, FDD cable;
- Four SerialATA cables;
- 1x External SerialATA cable + power supply adapters;
- An external module to plug in SerialATA devices;
- A cap to the rear panel of the case.
We note it straight off that the number of USB2.0 ports on the rear panel is unusually great - as many as 8(!). So, at Gigabyte they decided not to include brackets with additional ports into the package bundle. The package bundle also lacks a bracket for the outdated LPT and COM ports. Finally, the board supports 8 SerialATA lines, but the package bundle contains only 4 matching cables. However, that is quite enough since there is a plate in the bundle that facilitates plugging in two external SerialATA devices, as well as an eSATA cable with the respective power adapter.
But there aren't other power adapters in the package bundle, so we give a lower mark for the bundling: a top-end product is supposed to maximally satisfy the user needs and not to make the user come back to the shop to buy a penny-worth item!
As regards the master and brief user manuals, they have nothing to complain about.
Nor there is anything to complain about the DVD disk which, along with the required drivers and Gigabyte's proprietary utilities offers third-party software: Norton and Kaspersky anti-viruses.
||CPU & Memory: