PHOTOS CONTINUED FROM PREVIOUS PAGE

Showing the PS/2 adaption to the keyboard controller; how I mounted an internal CF-to-SCSI adapter; the BIOS memory count without the expansion card, but with shadowing enabled (3712 KB); the BIOS memory count with the expansion card, and with shadowing enabled (6784 KB) - note that some ISA I/O cards make it so that BIOS cannot see the expansion card. I didn't investigate further, but perhaps some port conflicts. Since I'm using the floppy port on the SCSI card, an general purpose I/O card wasn't of interest in the final design.

Image of the POST screen followed by Speedsys and cachechk screenshots showing the memory slowdown in the past 4096 KB address range. Doom also shows some slow down with the memory expansion card installed (from 7.19 fps to 6.79 fps). I'm guessing this slow down would be larger the more the program uses memory in the address range of the memory expansion card.

pshipkov wrote on 2022-02-07, 18:58:

@maxtherabbit
So what happens after set that flag ?
Can you explain ?

Sorry I just saw this

Running that command in DEBUG begins execution of a built-in test routine in the Adaptec ROM. It just loops dummy DMA transfers to/from system memory in different segments to verify data integrity. It allows you to safely test a given DMA speed (selected via jumpers) without risk of losing actual data on disk.

Hi, the memory card does really provide XMS?
I have similar one but it serves as EMS and standard p-mode games like Doom cannot use EMS.
The memory bandwidth is limited by ISA bus...

Thanks Max.
Makes sense. Never tried that before.

@Feipoa
Good photo report.
The rig turned out great - packed and tight, including the semi-messy cables in the box - for extra retro flavor.
Looking at your screenshots - it occurred to me that i never ran SpeedSys and Doom with that 286 CPU upgrade here (i think). Going to check that soon.
Your PS/2 mouse insistence is admiring 😀

RayeR wrote on 2022-02-25, 18:26:

Hi, the memory card does really provide XMS?
I have similar one but it serves as EMS and standard p-mode games like Doom cannot use EMS.
The memory bandwidth is limited by ISA bus...

Yes, the memory card has settings (dip switches) for various conditions of memory allocation. You can set all the onboard memory up as XMS, as I have done, or you can setup part of the memory as EMS with two address control ports. This particular card should support LIM 4.0 EMS, which I've been told that the cheaper ones do not.

maxtherabbit wrote on 2022-02-07, 14:57:

Adaptec has a DMA test in ROM you can use to safely test the higher speeds without risking data corruption. Open DOS DEBUG and g […]
Show full quote

feipoa wrote on 2022-02-07, 11:38:

I'm a bit nervous to leave the DMA speed at 6.7 MB/s, given that 8.0 MB/s didn't work at all.

Adaptec has a DMA test in ROM you can use to safely test the higher speeds without risking data corruption. Open DOS DEBUG and go:

g=c800:9

replace c800 with the base address of your ROM if different

I will test this out shortly, since I currently have the system out for troubleshooting HWiNFO and CHKCPU, both of which incorrectly report the CPU. Only Check-It Pro and AGSI v1.23 are reporting the CPU speed correctly.

pshipkov wrote on 2022-02-25, 20:11:

@Feipoa Good photo report. The rig turned out great - packed and tight, including the semi-messy cables in the box - for extra r […]
Show full quote

@Feipoa
Good photo report.
The rig turned out great - packed and tight, including the semi-messy cables in the box - for extra retro flavor.
Looking at your screenshots - it occurred to me that i never ran SpeedSys and Doom with that 286 CPU upgrade here (i think). Going to check that soon.
Your PS/2 mouse insistence is admiring :)

The cables couldn't have been neater because the ACARD was VERY particular about not being the last device on the SCSI bus. I tried various cables, terminators, and prayers. Sometimes it seemed like it would work as the final device, but next boot it would not. Some photos may show the ACARD as the final device, but that wasn't the final configuration. I also really wanted the CF card slot bracket to be on the outside of the case, but the length of the ACARD PCB prevented this. I'd have had to dangle the ACARD floating somewhere between the expansion cards, or create an extra long adapter cable to facilitate this. I did not want to bother. I also wanted to keep one slot bay open for the possible expansion of a second Everex memory card.

Because of my KVM's on both test stations, I really want the PS/2. Shuffling around mice is not for me.

EDIT: Note that there are more photos of the build on page 7 of this thread. Looks like the photos got split between pages, causing the first set of photos to be missed, as determined by the views differential.

Did you determine whether or not the EV-159 supports zero wait state operation?

Anonymous Coward wrote on 2022-02-25, 23:54:

Did you determine whether or not the EV-159 supports zero wait state operation?

There's only a jumper for 120 ns and faster memory, or 150 ns memory. I ran cachechk with that 120 ns vs. 150 ns memory jumper, but the results were the same in both jumper positions. Disabling parity didn't help the speed either. I also played with other DIP switch settings that selects between 8-bit and 16-bit EMS transfer modes, but it did nothing, maybe because I'm using XMS. Perhaps if I start using the various EMS modes I might see some changes in memory speed?

8-bit/16-bit EMS transfer mode will only affect EMS memory.
Forget about EMS. For the most part, you don't need it unless specific software requires it.
Software that isn't designed to use EMS won't care.

Also, there's not much point to use the EMS functions on the card with the Superchip installed, because 386 and greater CPUs can create their own EMS using EMM386.

maxtherabbit wrote on 2022-02-07, 14:57:

Adaptec has a DMA test in ROM you can use to safely test the higher speeds without risking data corruption. Open DOS DEBUG and go:

g=c800:9

replace c800 with the base address of your ROM if different

It has been sitting like this for 15 minutes. Is this the expected behaviour? There's no HDD LED activity.

feipoa wrote on 2022-02-26, 13:07:

maxtherabbit wrote on 2022-02-07, 14:57:

Adaptec has a DMA test in ROM you can use to safely test the higher speeds without risking data corruption. Open DOS DEBUG and go:

g=c800:9

replace c800 with the base address of your ROM if different

It has been sitting like this for 15 minutes. Is this the expected behaviour? There's no HDD LED activity.
Adaptec_DEBUG.JPG

No, it should print some text about adaptec DMA test etc. and the activity light should be on

https://retrocmp.de/ctrl/llf/llf.htm
Bottom of this page shows what it should look like. It's possible the ISA cards with SCSISelect omitted this test routine from ROM. My only 154x SCSI adapters are the B revisions

I tried the debug command on my SXL2/ULSI-66 system which also contains the AHA-154xCP, but the result was the same. Perhaps the C adapters use different commands for this.

You sure about that base address? I see dc800 on the screen not c800

Maybe I misunderstood. Is this not the correct procedure:

Oh right, I thought you typo-ed the example. Does it not like going to the top half of a segment or something and you've gotta supply base address of segment and it will look upwards, like D800 or something. IDK, I have to have a double espresso before I do memory math.

Just use debug with command
u dc00:0
to see if ROM is really there - it must start with bytes 55 AA
then disassemble again
u dc00:3
and you should see some JMP instruction that points to entry point of BIOS init code and
u dc00:9
should show JMP to diag code, if the offset is correct...

feipoa wrote on 2022-02-28, 02:39:

Maybe I misunderstood. Is this not the correct procedure:

It is correct. Your card simply must not have the test in ROM

RayeR wrote on 2022-02-28, 03:34:

Just use debug with command u dc00:0 to see if ROM is really there - it must start with bytes 55 AA then disassemble again u dc0 […]
Show full quote

Just use debug with command
u dc00:0
to see if ROM is really there - it must start with bytes 55 AA
then disassemble again
u dc00:3
and you should see some JMP instruction that points to entry point of BIOS init code and
u dc00:9
should show JMP to diag code, if the offset is correct...

Does this look right to you?

.del.dupl.

RayeR wrote on 2022-02-28, 12:48:

ROM is there 😀
It looks that the correct entrypoint is DC00:7 not DC00:9 so try
-g dc00:7