First post, by LSS10999
UPDATE (2024-01-31): I'm rearranging the OP to reflect some recent discoveries with a second AIMB-865 (1.0s1.3), that I'm currently using now.
With my tests first on a v1.0s1.2 then on a v1.0s1.3 board of AIMB-865, there are some known problems with these boards, with some having possible workarounds.
- So far any 65nm or 45nm FSB800 CPUs should work without any hardware mods, though for 45nm CPUs you may need to include additional microcodes if your CPUID is not 10676 which the BIOS has.
- As stock BIOS has only microcodes for CPUID 10676h, you may want to add additional microcodes (10677h and 1067Ah) into the BIOS, and preferably updated ones for all Core 2 CPUs, to maximize support of functionality.
- Core 2 CPUs may drop to 100MHz bus speed in some circumstances. The reason behind the reduction of bus speed appears more complicated than I originally assumed, as there are other circumstances in which the CPU does operate at intended bus speed.
- If you're using X6800, be noted that the CPU multiplier setting is flawed that it would actually set the CPU to use the maximum possible value (about 60x) which obviously cannot work, then switches to a fail-safe saying overclocking failed and lets you set the correct/desired multiplier.
- Someone in Modlabs forum actually managed to make an onboard BSEL mod to force FSB1066 CPUs operate at FSB800 which is more compatible while retaining compatibility with actual FSB800 CPUs. A minor caveat would be that this will break FSB533, but AFAIK there are no Core CPUs with that FSB. Not sure if this mod really works. I did such a mod on my 1.0s1.2 board but from what I could test with a FSB1066 CPU, the BIOS is still reporting the original speed, not the one set to FSB800. Maybe it depends on the clock generator also...
- You may need to set memory frequency to 266MHz when using a CPU with 266MHz bus speed (FSB1066).
- For 200MHz (FSB800) CPUs, any memory frequency can be used.
- The board can be very picky when it comes to memory. Some combinations work but not reliable, would end up running at an undesired 3-4-4-12 timing and/or report errors. You may need some efforts to get the right combinations that would work at desired timing as well as getting no errors during tests. (Manually specify DRAM timings instead of SPD may not always work)
- Can't be sure about this board's RAM slot quality. It's entirely possible to have faulty slots which will lead to unstable system as well as the inability to utilize dual-channel configurations.
- When using only onboard GPU, you can get up to 3800MB of RAM. However, it seems video cards including PCI ones may additionally map certain areas in the system arena to their liking, further reducing available RAM. In my case with a half-height PCIe video card (AMD) with a PCI-PCIe bridge, I ended up with only 3166MB of RAM available (with onboard graphics set to disabled, or it will be minus the shared memory specified).
Considering there are different revisions of this board as well as different hardware configurations the BIOS options are a bit messy.
Actually all board variants (1.0s1.2, 1.0s1.3, etc.) share the same BIOS. Some features are only visible on specific board variants.
- In CPU configuration sections some NetBurst-specific options are visible to Core 2 CPUs. I think it's advised to configure the options appropriately depending on the CPU architecture to ensure system stability.
- The Hyper-Threading switch actually controls the availability of the second core for Pentium D CPUs. If set to disabled, the second CPU core would not be available. It has no effect for Core 2 CPUs.
- Onboard GPU cannot be fully disabled with 1.0s1.3 revision (865GV). It remains visible to the system even when you set it to Disabled. After checking the 865GV chapter in the 865G/865GV GMCH datasheet, it seems this is by design, as the bit responsible for enabling the internal graphics (IGDIS) is hardcoded to enabled and read-only for 865GV, whereas the same bit is R/W for 865G. (UPDATE: Setting it to Disabled in this case will still prevent the system from reserving RAM for the onboard GPU, as otherwise the amount of RAM specified would be reserved on top of existing system arena and unavailable.)
- Boot Video card init order will only show AGP for 1.0s1.2 board (865G). For 1.0s1.3 boards (865GV) only PCI options will be provided.
- Enabling "Memory Hole" will actually limit available system memory to 15MB (at least from what BIOS reports). Not desirable for other use cases, but it can be useful for DOS programs that would misbehave when the system has a lot of installed RAM.
- USB boot is supported, but you have to modify BIOS hard disk order in order to do so. Should be noted that you must make sure the BIOS_WP1 jumper (near the BIOS chip) is not shorted (which write-protects the BIOS. The manual I could find online was wrong and suggested the opposite) to be able to actually make changes to boot priority as these parts are stored in NVRAM (an area inside BIOS chip).
- There's an option called "C.S.A Gigabit Ethernet" that I'm not sure about the actual function. Most likely it's a toggle to enable the 1Gbe interface on board variants that actually equipped it. I don't think my boards have it...
No, AGP cards cannot be used on 1.0s1.3 boards even if you solder an AGP slot on it, as they use 865GV. If you do have such a slot it may be possible to use the so-called ADD card which enables you to use DVI with the onboard graphics, but I haven't seen such a card myself, so I cannot be entirely sure about this.
My 1.0s1.3 board comes with such a slot but as far as I've tested, it's not usable. You can only use AGP cards with 1.0s1.2 boards, but only half-length ones, since full-length cards will get in the way with the memory slots.
Additionally, since the onboard graphics cannot be disabled on 865GV, if you're using a PCI video card, be sure to disable Intel graphics driver whenever possible (for Windows, Linux, etc.) to prevent the onboard graphics from interfering.
The BIOS incorrectly sets all ISA DMA channels to the LPC interface, with no convenient way to toggle the ones used by ISA cards to PC/PCI. As such, while the card can be detected and configured (including via ISA PnP), DMA operations will not work unless you manually change the respective DMA channels to PC/PCI.
You can refer to ICH5 datasheet and modify the bits in registers 90-91h of the ISA interface (Device 1Fh, Function 0) to make the channels you need accessible to PC/PCI, but do not change all of them. You need to keep the DMA channels used by SuperIO on LPC, namely FDC and ECP Parallel Port, or you'll break the functionality.
This is an interesting board, using a soldered ER14250 battery (3.6V) instead of the usual replaceable CR2032 coin cells. Since it's soldered to the board, it is difficult to replace should it have been drained, although 14250 battery is supposed to hold much more capacity than coin cells thus last longer.
If you're skilled enough, it is possible to solder a battery holder in its place for easier access/replacement of the battery, but keep in mind that the battery holder will be a bit larger in size than the actual battery and you may get in the way of other components, namely the IDE1 connector as well as the BIOS chip socket.
On the board I'm currently using, I experimented with a USB-rechargeable 14250 battery after soldering a battery holder, and it turned out to be working fine, despite the battery, rated 3.7V, emits ~4.1V when fully charged. Not sure whether my board's VBAT sensor was broken (as it showed very low values like +0.032V), but CMOS data is indeed being preserved.
* Theoretically you don't really have to use a 14250 battery for the CMOS/RTC part, considering the size of a 14250 battery holder. You can pretty much install anything that could provide a reasonable +3V there.
** If you're using a USB-rechargeable battery (that has a USB connector for charging input), you can pretty much "plug" the battery into one of the board's USB front panel header so you don't ever have to worry anymore, as it'll automatically gets charged so long as you keep the system plugged in. Just that it may have an impact on the battery's lifespan to some extent.
CF card by default assumes Secondary Master and unlike other motherboards such as IMB200 there's no way to toggle it. Should note that like the BIOS_WP function, the info in the manual was wrong and in reality what the board does is actually the opposite!
As such, if you use a IDE44-SATA adapter to connect SATA drives to the IDE-SLIM port you need to do some hardware mods, and how it is done depending on the adapter PCB you use (mSATA-IDE44 also counts):
- Force the disk to IDE Slave. Some PCBs have jumper or footprint for such a toggle that you can just use it. In JM20330's case, if unsure, you can use a multimeter to inspect where you can access pin 33 (MSSEL). If this is not properly configured and the disk operates as Master, it will prevent installed CF card from working.
- "Cheat" 80-wire detection. In order to allow higher UDMA levels the pin 34 (PDIAG) must be grounded. A simple way to do is connect pin 30 and 34 with a wire. This is very important, as if 80-wire detection is not correctly configured, even the CF card will be forced to UDMA2. Additionally, 80-wire detection must be set to "Host" in BIOS to ensure everything works correctly.
The original OP has been kept in a spoiler block below just in case.
A while ago I got myself an ANOVO AIMB-865 (V1.0S1.2, 865G with 1 ethernet controller placed around the lower two USB ports) intended to replace an Axiomtek IMB200 mainly for better CPU support (that board supports only up to Pentium D 960 as the official manual stated, and any attempt to extend the CPU support past it, mainly by adding microcodes and such, was not successful), yet I've found quite a few problems with the new board.
1. The board officially claims to support Core 2 Duo up to 2.66GHz, and as tested, up to Core 2 Extreme X6800 can boot. While the BIOS shows correct clock values (so does Windows XP's System Information, which doesn't seem to reflect actual values, either), the actual system bus speed is only 100MHz, regardless of whether it is FSB 800 (200MHz) or FSB 1066 (266MHz). Also, when using a FSB 1066 Core 2 Duo, the board can only boot when memory speed is set to 266MHz (while a FSB 800 Core 2 Duo can boot with any option, 266/333/400 MHz, despite in both scenarios, the CPUs will have an actual system bus speed of 100MHz).
2. The board seems to have issues handling the multipliers when booting with an extreme edition CPU, as it initially sets the multi to the maximum value (60x I think) then boots with an overclock failed message and has to manually set the multi back to the intended value in order to be able to boot properly during subsequent boots.
3. Somehow the AIMB-865 board I have fails to boot on any valid dual-channel memory configuration regardless of installed CPU. Using single-channel memory configurations (by leaving one of the slots empty, or using memory sticks of different sizes), however, will boot properly.
4. Tried a Pentium D 965 and it could not boot at all. Perhaps a BSEL mod that sets to 200MHz bus speed (which somehow had no effect on IMB200) is needed, but I haven't had time to try it there.
5. The ISA DMA does not work out of box. As tested, a chipset register hack similar to the one in this video is required for ISA DMA to function, and it only lasts for this boot instance only (PS: Setting the register values involved to exactly the ones shown in the video actually caused a hard freeze during my initial tests after playing some ISA sounds, as I might have flipped some other register bits that I shouldn't have. Doing it by figuring out which bits the actions actually flipped according to the video, however, I did not encounter freezes anymore).
The only detailed information about the board I could find is this one (from Modlabs, in Russian), and it appears to have mentioned some of the problems I'm having (namely the ISA DMA and the actual system bus speed issues). Besides, there is an official BIOS image named 865GV008.ROM which bears the same BIOS version number of the board I currently own, but I cannot confirm whichever variant the BIOS was intended to (or is not specific to variants), as there is also another major variant called V1.0S1.3 which is 865GV-based and lacked the AGP slot.
Apparently regarding the 100MHz system bus issue when running Core 2 CPUs on 865-based mainboards, there's already a post talking about it in an attempt to get it working on a different board than this one. And I'm not certain if there are any methods other than BSEL mod, to force 200MHz bus speed for Pentium D 965, if the motherboard itself is not (or not entirely) using BSEL to determine CPU clocks (like my IMB200, the BIOS always gives it a 33MHz system bus for Pentium D 965 then hangs, even after doing a BSEL mod that has been confirmed working using another 865GV-based board I had for reference, MSI 865GVM3-V).
Besides, from the Modlabs post I did not see it mentioning the inability to boot when using dual-channel memory configurations, so whether this is due to the board's BIOS or my board itself (hardware issue) could not be determined.
So again, there are still a lot of things needed in order to make AIMB-865 useful, so I'll have to stay with IMB200. Compared to AIMB-865, while it lacks the potential to extend CPU support (for better forward compatibility), and it doesn't have an easy way to manage USB boots (so I have to install OSes using optical discs instead), it has out-of-box ISA support (including ISA PnP like AWE64 Gold).
LATE EDIT (2023-06-05): The video I referenced is long gone so I removed the dead link. I don't know if anyone has ever re-upped the video, or has any documentation that could be made public so it could be put there for reference.