VOGONS


Problems regarding ANOVO AIMB-865

Topic actions

Reply 20 of 38, by JTK68

User metadata
Rank Newbie
Rank
Newbie
White wrote:

I will NOT send you this BIOS via email, sorry.

Folder is 0% full (17 from 9999 messages stored)

I do not know why you cant't send me PMs/

You send PM work but I can´t send PM back.

Reply 21 of 38, by LSS10999

User metadata
Rank Oldbie
Rank
Oldbie
JTK68 wrote:
White wrote:

I will NOT send you this BIOS via email, sorry.

Folder is 0% full (17 from 9999 messages stored)

I do not know why you cant't send me PMs/

You send PM work but I can´t send PM back.

Maybe you need to ask an administrator... it's very likely that you need a minimum amount of posts before you can send PM (for spam security concerns).

Reply 22 of 38, by Scraphoarder

User metadata
Rank Member
Rank
Member
White wrote:

This BIOS is exists. I have it. But for now i do not provide it to public and it is possible i will not.
You can send me PM if you really want it.

PM sent. This bios is highly wanted.

Reply 26 of 38, by ejipcikarang

User metadata
Rank Newbie
Rank
Newbie
White wrote on 2020-03-12, 07:36:

Later i will give you exact information about pinout, but for now - Power Switch is two vertical pins near right USB header

Thank you for the reply, but i try and doesn't work.

Attachments

  • AIMB-865.jpg
    Filename
    AIMB-865.jpg
    File size
    26.87 KiB
    Views
    1837 views
    File license
    Public domain

Reply 28 of 38, by ejipcikarang

User metadata
Rank Newbie
Rank
Newbie
White wrote on 2020-03-13, 04:54:

This jumper ahould work. And whats happenning when you short it?

I'm sorry, its work.. i completely forgot about jumper clear cmos.
still look forward with complete front panel pin.

Reply 31 of 38, by popfuture

User metadata
Rank Newbie
Rank
Newbie

There seem to be two revisions of this board available on eBay. There’s a V1.0S1.2 and V1.0S1.3 but I can’t seem to find any info on what the differences are. Anyone know? Do they use the same firmware images?

Reply 32 of 38, by LSS10999

User metadata
Rank Oldbie
Rank
Oldbie
popfuture wrote on 2020-03-25, 15:34:

There seem to be two revisions of this board available on eBay. There’s a V1.0S1.2 and V1.0S1.3 but I can’t seem to find any info on what the differences are. Anyone know? Do they use the same firmware images?

V1.0S1.2 has AGP slot (usually black PCB, probably 865G). V1.0S1.3 doesn't have AGP slot (usually green PCB, probably 865GV).

However, the design is flawed that you can only use a half-length AGP card, as a full-length video card will get in the way of the RAM slots (PS: RUBY-9719VG2AR also has the same flaw).

However, for both revisions, there seem to be different variants with either one or two network adapters.

Reply 35 of 38, by ackmangogo

User metadata
Rank Newbie
Rank
Newbie
White wrote on 2019-07-05, 07:11:

This BIOS is exists. I have it. But for now i do not provide it to public and it is possible i will not.
You can send me PM if you really want it.

Hi White, I am looking for build a powerful machine with isa sound card, after googling around found this thread, plan order ANOVO AIMB-865 board, would you send me your mod bios ?

Reply 37 of 38, by LSS10999

User metadata
Rank Oldbie
Rank
Oldbie

Necroing this old thread of mine...

I've recently acquired a second AIMB-865 board (1.0s1.3). I've kept my previous 1.0s1.2 board in a stash after my initial experiments, that I currently don't have time to bring it out, but my experiments with the second board offered some more insights.

- The second board has a BIOS whose version string is 865GV009, which is a bit newer than the 865GV008 I had on the first board as well as the one that was once available on Anovo's website, but from all my tests it doesn't really differ much compared to my first one. All the issues I'm having is still there.
- Dual-channel DDR works fine on this one. I wonder if I'm really having some faulty components on my first board that prevented dual-channel from working back then. I need to use some specific combinations of sticks, however, as some sticks don't play well with each other and would end up making one stick completely absent, leaving the system running in single channel mode as a result.
- This second board is kind of special as it actually has an AGP slot while being 1.0s1.3, but after some testing, it doesn't really work. I can't get output from the video card and afterwards the onboard VGA port output some garbled contents. Later on I confirmed in HWINFO that the northbridge was indeed 865GV, not a 865G in disguise, so no real AGP at all. Perhaps one could use an ADD card to get DVI using the integrated GPU, but I'm not sure as I don't have such cards myself. The AGP port is not important, as I'm considering using a half-height PCIe video card with the help of a PEX8112-based bridge adapter to plug it into the first PCI slot, though I do worry about power supply as the neither the adapter nor the card has any external power input and the card I'm using is rated about 50W TDP while PCI slot normally only supplies ~25W as far as I could find info about.
- During my tests I did see my CPU (E4700) operating at intended (200MHz) bus speed while checking with HWINFO, though only briefly. After some further tests and boot cycles, the CPU no longer runs on such clock anymore, stuck at 100MHz bus speed, and I could not reproduce 200MHz speed anymore (HWINFO can report actual CPU speed). In other places (MemTest86+, Linux, etc.) the CPU always runs at 100MHz bus speed just like my first board.
- The onboard GPU cannot be fully disabled with 865GV, even though I did set it to disabled in BIOS (as I'm to use a PCI video card). I'm not sure if this will somehow interfere with operating systems when it comes to determining the default display output device. Sadly according to the 865GV section of the 865G/865GV datasheet, this is by design.

The new board had its ER14250 battery completely drained when I got it, but unlike my first board where I simply replaced the battery, I tried something new this time.

- I replaced the drained battery with a 14250 battery holder and put a USB-rechargeable 14250 battery in it.
- The battery holder I used was a bit too large and the soldering process was rather tricky, but I was able to solder it in place without causing too much interference with nearby components.
- A minor problem, however, is that the VBAT sensor looks broken, showing very low values like ~0.032V. The battery is rated 3.7V but actually outputs ~4.1V when fully charged. Not sure if this relatively high voltage kind of broke the sensor, but everything is working fine -- CMOS RAM is indeed being preserved.

Maybe I can try doing a similar hack on my first board when I have time to take it out.

On the other hand, I've figured out how to properly enable ISA DMA on this board. According to outputs from HWINFO it appears the BIOS incorrectly set all ISA DMA channels to LPC interface so the board's PCI-ISA bridge (which uses PC/PCI) cannot use them by default. According to the ICH5 datasheet, by modifying the register 90h-91h on the Intel ISA bridge (D31:F0), changing the respective bits for the DMA channel you want to use from 11b (LPC I/F) to 01b (PC/PCI) will allow the PCI-ISA bridge to access it. Tested with a Sound Blaster 16 and everything works after making this change.

Despite known issues, so far I think this board is good enough for me to actually use it to supersede the IMB200 I'm currently using, as that board is suffering from some annoying issues preventing it from POSTing properly every time.

EDIT: Some updates after running this board for a while:
- The issue with the actual clock the CPU is running seems more complicated than I thought. Before I actually installed the board into the case, I was booting to Linux distros from USB sticks and the boot process was very slow (taking a while to even show GRUB), with all places showing the system running at 100MHz bus speed. When I booted the same Linux distro from the DVD drive things were a bit faster (GRUB showed up within expected time threshold), and /proc/cpuinfo output suggested it was indeed running at expected bus speed (200MHz). I wonder if USB boot is somehow related to the CPU getting stuck at 100MHz bus speed...
- Not sure whether the cannot-be-disabled integrated graphics would interfere with Linux startup, but so far all my successful boot attempts into the Linux live environment has had the modesetting of non-radeon modules set to 0 (as I'm using an AMD card). Will need some further testing, however.
- I've dumped my BIOS (865GV009) and updated whatever microcode files I could. Actually the BIOS contains a few 45nm entries (10676h), but I wonder if anyone has ever managed to boot a 45nm CPU on this board using stock BIOS? (UPDATE: 45nm CPUs do work.)
- The part regarding BIOS write protect jumper in the manual I could find is wrong and misleading. The BIOS would be write-protected if the jumper is shorted, while the manual suggested the opposite. Also, when write-protected, some BIOS settings can no longer be saved (namely disk boot priority), as "Checking NVRAM" would report "update failed".

EDIT 2: I decided to conduct a memory test using MemTest86+ (from my Linux distro) on my running board and discovered something.
- Booting my Linux distro's DVD (not USB stick) and use MemTest86+, and this time MemTest86+ reports the CPU running at 200MHz bus speed.
- Don't know whether my (quite dated) memory stick is faulty, or some kind of training issues, I'm seeing random bit flip errors during the beginning of the test process. Will be testing the sticks for a bit longer to see what's really going on...

* Looks like the board is extremely picky about RAM. Actually there were incompatibilities with the sticks leading to it running at an unexpected 3-4-4-12 timing (manually specifying the timing from BIOS has no effect). I replaced the sticks with a few others and it took me a while to find a few sticks that work properly. Tests are reporting no errors and it's running at 3-3-3-8 which is expected.

Last edited by LSS10999 on 2024-02-07, 02:22. Edited 2 times in total.

Reply 38 of 38, by LSS10999

User metadata
Rank Oldbie
Rank
Oldbie

I took out my old 1.0s1.2 board (which had trouble with RAM) and continued my experiments.

I did the same battery mod on this older board. Since the rechargeable 14250 battery I used has a USB (Type-C in my case) connector for charging, I experimented with connecting the battery into one of the front panel USB headers and turned out it works nicely. A caveat would be I have to use a cable tie around the holder to keep the battery in place, as the battery holder I installed is rather loose that I could easily pull out the battery by accident when moving the charging cable. But in any case, this means I don't have to worry about replacing CMOS battery on such board anymore, as it'll be recharged as long as the system is plugged in. (NOTE: USB jumpers are by default set to +5VSB which means they're getting power even when standby/powered off)

I also did the onboard BSEL mod following the photo on ModLabs. I haven't tested actually booting a 1066FSB CPU, though I wonder if it'll also help with ensuring the CPU running at FSB800. I'm not sure if it's possible to add another wire to pull BSEL2 low in a similar manner on top of the current BSEL mod that pulls BSEL1 high, to allow even FSB1333/1600 CPUs to work, as I've no idea on the exact location of all three BSEL bits on the board.

On the other hand, turned out that 45nm (Wolfdale) CPUs do work on this board. I managed to get E5800 (3.2GHz, FSB800) booted. It should be noted that the stock BIOS has only microcodes for CPUID 10676h, while my E5800 is 1067Ah. Some CPUs like E7200 which has 10676h CPUID could probably work out-of-box, while others may require you adding necessary microcode updates into BIOS in order to work reliably.

Unfortunately, neither my 1.0s1.2 board nor my 1.0s1.3 board are perfect.
- Regarding the dual-channel memory issue on my 1.0s1.2 board -- turned out slots 3 and 4 (the two slots farthest from CPU) are faulty. All my working sticks report identical error patterns when plugged into those slots. Using only the first two slots (closest to CPU) will give me a single-channel 2GB in total and without errors in MemTest86+, so these two slots are likely okay. Will probably have to inspect that board thoroughly, though I don't know if these issues are repairable.
- I don't know why, but actually my 1.0s1.3 board's onboard audio is broken -- it does not show up in the system devices list regardless of BIOS options. It's not important, however, as the system would mainly use ISA sound card for old stuffs and PCI sound card for new stuffs.

I don't think I could test other stuffs on my 1.0s1.2 board for the time being. In case someone wants to try 45nm CPUs on this board, I'm providing my own BIOS dumps with updated 65nm and 45nm microcodes here.

Note that the two BIOS versions (865GV008 and 865GV009) differ a lot (especially CMOS structure), that AFUDOS would suggest adding /B /C options when flashing to the other version, which will program the boot block as well as clear CMOS. You can use either version on either board variant but AFAIK they have the same issues with respect to ISA DMA and others, so it's more recommended to stay on the version your board is currently running, though the 865GV008 BIOS has some additional settings in the ACPI section regarding suspend and S3 resume.

PS: I'm currently inspecting the board's ACPI tables. It seems to have entries for only two CPUs. On the other hand, the ACPI table from 775i65G BIOS has entries for 4 CPUs, and that board does have support for Kentsfield (65nm C2Q). I think Yorkfield (45nm C2Q) should theoretically be usable considering Wolfdale is already working, but one needs to somehow force it to run at 800 instead of 1333.

EDIT: I'm not sure whether or not the BSEL mod works so I eventually decided not to actually do the mod on the board I'm currently using, though I did move my E5800 there. Everything works except for some reasons the board can no longer properly ACPI shutdown. Triggering ACPI shutdown would only freeze the system requiring a manual shutdown (holding POWER for 5s). The CPU config section in BIOS additionally contains some switches for C1E and CPU TM function, but hidden. Perhaps I need to unhide it and see if they're responsible for the shutdown issue...

On the other hand... I think I'll have to look deeper into how to correctly mod the ACPI part, as apparently my 1.0s1.2 board was kind of borked when I flashed a modified BIOS using a recompiled DSDT (the only things I added was copy-pasted CPU blocks for a 3rd and 4th instance in hope to allow those cores configured), stuck at POST code 008D (which means building ACPI table from what I found about), and attempts to recover the BIOS was not successful -- the board now no longer boots, giving out two quick beeps followed by 6 short beeps with a longer delay in between...

Attachments

  • Filename
    865GV009.zip
    File size
    350.75 KiB
    Downloads
    6 downloads
    File comment
    AIMB-865 BIOS 865GV009 dumped from 1.0s1.3 board, updated microcode for 65nm and 45nm.
    File license
    Fair use/fair dealing exception
  • Filename
    865GV008.zip
    File size
    352.97 KiB
    Downloads
    6 downloads
    File comment
    AIMB-865 BIOS 865GV008 dumped from 1.0s1.2 board, updated microcode for 65nm and 45nm.
    File license
    Fair use/fair dealing exception