Thank you!

Running:

DEBUG
g=dc00:7

opens the SCSI BIOS configuration utility, the same one as CTRL-A at boot time. However, it looks as if the DMA test routine has been built into this CTRL-A BIOS boot menu already. Perhaps it isn't in the *B variants of the 154X card? Either I didn't notice it previously, or I simply thought it was the common "verify media and remap sectors" feature.

HOWEVER, the benefit of running it from debug is that I can have all the Cyrix CPU registers setup first and thus determine if they are interefering with DMA transfers.

Hmm.... even with a [safe] DMA rate of 5.0 MB/s, if I run the DMA test after I've booted the system into DOS, I get

1BUFFER MISCOMPARE!!! PHYSICAL ADDRESS: 020036h  OFFSET: 00h

Can himem not be loaded?

If I run the test from CTRL-A at boot time, 5.0 MB/s and 6.7 MB/s pass the test. It fill fail at 8.0 MB/s.

feipoa wrote on 2022-02-28, 13:28:

Hmm.... even with a [safe] DMA rate of 5.0 MB/s, if I run the DMA test after I've booted the system into DOS, I get […]
Show full quote

Hmm.... even with a [safe] DMA rate of 5.0 MB/s, if I run the DMA test after I've booted the system into DOS, I get

Copy code to clipboard

1BUFFER MISCOMPARE!!! PHYSICAL ADDRESS: 020036h  OFFSET: 00h

Can himem not be loaded?

If I run the test from CTRL-A at boot time, 5.0 MB/s and 6.7 MB/s pass the test. It fill fail at 8.0 MB/s.

The test needs to be run with nothing loaded. Since you have it built in to the SCSISelect I would suggest just running it in the pre-DOS environment

1542B cards do not have SCSISelect FYI

maxtherabbit wrote on 2022-02-28, 14:17:

feipoa wrote on 2022-02-28, 13:28:

Hmm.... even with a [safe] DMA rate of 5.0 MB/s, if I run the DMA test after I've booted the system into DOS, I get […]
Show full quote

Hmm.... even with a [safe] DMA rate of 5.0 MB/s, if I run the DMA test after I've booted the system into DOS, I get

Copy code to clipboard

1BUFFER MISCOMPARE!!! PHYSICAL ADDRESS: 020036h  OFFSET: 00h

Can himem not be loaded?

If I run the test from CTRL-A at boot time, 5.0 MB/s and 6.7 MB/s pass the test. It fill fail at 8.0 MB/s.

The test needs to be run with nothing loaded. Since you have it built in to the SCSISelect I would suggest just running it in the pre-DOS environment

1542B cards do not have SCSISelect FYI

In that case, 6.7 MB/s is max for SLC2. I'll see what my SXL2/ULSI2 can do, which has the same scsi card. EDIT: Best the SXL2-66 could do is 5.7 MB/s.

Small update - just because the SCSI BIOS diagnostic utility passes a particular DMA speed, doesn't mean it is stable in DOS and Windows.

I was running it at 6.7 MB/s, thinking it was stable, but HDD access will eventually hang up. I've put it back to 5.0 MB/s.

feipoa wrote on 2022-03-09, 07:53:

Small update - just because the SCSI BIOS diagnostic utility passes a particular DMA speed, doesn't mean it is stable in DOS and Windows.

I was running it at 6.7 MB/s, thinking it was stable, but HDD access will eventually hang up. I've put it back to 5.0 MB/s.

Hmmm this does not match my experience. Are you using active or passive termination?

Were your tests done with an SXL, SXLC, SLC, or DLC?

I don't recall off the top of my head; I'd have to pull the case off to check. Probably passive as I've had better luck with passive terms compared to active.

feipoa wrote on 2022-03-09, 19:58:

Were your tests done with an SXL, SXLC, SLC, or DLC?

I don't recall off the top of my head; I'd have to pull the case off to check. Probably passive as I've had better luck with passive terms compared to active.

Negative. I've tested on a real 486 as well as a real 286, no upgrade processors.

I have had cases of SCSI bus hangs in the past that were resolved by switching to active termination

I have updated the original post with the DIP switch settings from the official Evergreen 486 SuperChip manual. I've also attached them here, but will bring out the text as well.

1SW1
2ON = Double CPU speed using a 2x multiplier on the Evergreen interposer (not on the CPU)
3OFF = Do not use Evergreen's 2x multiplier, meaning that it uses 1x
4
5SW2
6ON = Set for AST, Sperry, Unisys, etc. Used to obtain compatibility in systems having a slightly proprietary architecture
7OFF = Set for IBM, Compaq, etc.
8
9SW3
10ON = Set SuperChip to match a system with a 16 MHz system clock
11OFF = Not 16 MHz system
12
13SW4
14ON = Set SuperChip to match a system with a 12 MHz system clock
15OFF = Not 12 MHz system
16
17SW5
18ON = Set SuperChip to match a system with a 10 MHz system clock
19Off = Not 10 MHz system
20
21SW6
22ON = Set SuperChip to match a system with a 8 MHz system clock
23OFF = Not 8 MHz system
24
25SW7 
26ON = Set SuperChip to match a system with a 6 MHz system clock
27OFF = Not a 6 MHz system
28
29SW8
30Always ON factory setting.

Turns out that my suspicion that one of the switches were to enable/disable the onboard FPU was incorrect. The good news is that I can disable Evergreen's 2x multiplier. So instead of running my SLC2 at 16.6 x 2 x 2 = 66.6 MHz, I can try running it at perhaps 25x1x2 = 50 MHz or 30x1x2 = 60 MHz. This should substantially increase the memory response, however I'm not sure if all my ISA expansion cards can cope with these high FSBs. It would also be interesting to see where the two configurations meet up with something like Wolf3D. That is to say, does the Wolf3D score of the SLC2 at 66 MHz (16x4) equal the Wolf3D score of the SLC2 at 50 MHz (25x2)?

Where did you find the documentation?

Hi all,

can you tell me where I can find the pshipkov's util pack?

Thx

I am not clear which utils pack you mean.
I use Phil's pack + couple of additional things on the side.

feipoa wrote on 2022-02-07, 11:28:

I ran a few tests from pshipkov's util pack. Note that SCSI DMA left at 5.0 MB/s. […]
Show full quote

I ran a few tests from pshipkov's util pack. Note that SCSI DMA left at 5.0 MB/s.

Wolf_286.exe TIMEDEMO = between 22.1 - 22.3 fps (no sound), or between 20.9 - 21.0 fps (with sound)

Speedsys score = 3.36. I need more RAM to run more tests.

3DSpace = 42%

DMA Speed = 1040K/s

CABT = 4.67 seconds

ATFLOAT = 5.5

Memspeed, 512KB, =8000 KB/s

Waitstat, clock=72.5 MHz, 11 waitstates, perf=50.2

AGSI reports 66.5 MHz. So far the only program to get the speed right on.

CoreTest 2.7, block size 64KB, ===================> 2965.3 KB/s
(what's the fastest CoreTest speed you've recorded on the ISA bus?)

DiskMeter 1.3, block size 64KB
Random =========> 3333.3 KB/s
Sequential ========> 3375.5 KB/s
Buffered =========> 3382.6 KB/s
Max seek =========> 0.3 ms
Perf index ========> 1249
(what's the fastest DiskMeter speed you've recorded on the ISA bus?)

Maxspeed (not in pshipkov's pack), which uses some fancy timing algorithm to measure the speed, reports between 64-69 MHz

Dear all
I was referring to this message were exactly that title for a bench pack have been mentioned.

Does this help?

Gotcha.

benchmarks.zip

Contents:
DOS benchmarks, including Phil's pack.
Portable installations of 3D Rendering applications, Windows 3.11 and WinTune 2.
Everything ready to go (excluding Windows graphics driver).
Let me know if you want compilers + test projects in the same portable fashion.

cool. thanks. just downloading.
BR