Mau1wurf1977 wrote:

No Ram amount shouldn't make an impact.

Check the turbo button and also run a CPU benchmark to see if the CPU speed is ok.

Speedsys gives me a processor benchmark of 5.08 which looks about right accordimg to this thread:
386DX40 - Not as fast, as it used to be

where a 386dx40 gives 5.98.

I don't have a turbo button as it's in an old atx case, I'll check the manual for jumpers but speedsys etc all say its a dx33 running at 33 or 34mhz.

I'll swap the graphics card for another and see what happens. I thought this particular speedstar was a tseng et4000 model but maybe it's not.

What are your RAM Waitstate settings?

Maybe try cachechk.exe. My 386DX-40 gives
33 for cached read
64 for cache miss read
54 for uncached read
(the last 16 MB are uncached)
Ratios also tell you the relative access speed for the different scenarios.
My Waitstate setting is 2R / 0W.

This should give you some impression how fast your memory is.

I wonder what Mau1wurf1977 gets with his insane 386 machine...

cachechk tells me I have no cache 😳 but it says cache is enabled. On booting up I get the message "64kb cache memory".

The waitstate settings available are 0/1 for each and it's currently on 0/0.

Cachechk was made for 486, so it may report wrong assumptions since the 386 access is slower.
The important thing is, what numbers does it report?
Then switch off the Cache in BIOS and rerun, to see the difference.
It should be compareable to 33 / 54.

Next time I use my 386 I will save a report (also for a 4R / 2W setting) and post it here for comparison.

Hmm...whether I enable or disable external cache in the bios seems to make no difference to either the "64kb cache" message on bootup, or to cachechk results. But the manual seems to suggest that the setting is always enabled for a 386.

Cachechk results are these [copied by hand so some abbrevaitions here]:

Megabyte # ------ Memory Access Block Size (kb) ------
1 2 4 8 <etc>
0: 73 <repeated>
1: 73 <repeated>
2 <----- same as above

Extra tests -----
Wrt 40 <repeated> 41 40 <repeated>
This machine does not seem to have any cache.
Main memory speed - 15.1 MB/s 69.7 ns/byte (100%) [read] 8.7 clks
Effective RAM access time is [read] 139 ns....
Effective RAM access time is [write] 76 ns....
Clocked at 386 32.7mhz

Maybe it has socketed cache chips and some previous owner pulled them out, and the bios really isn't doing anything regarding the cache enable/disable but simply assumes 64kb because that's what it should be.

I'll open the case later and see if anything looks like it's missing.

This really looks like there is no caching. But this would also explain your low score.
Why the BIOS reports 64 kB is still a question. You may also have to check if the Cache config is jumpered correctly (if there are chips).
Maybe you even have one of those fake cache boards.

elianda wrote:

This really looks like there is no caching. But this would also explain your low score.
Why the BIOS reports 64 kB is still a question. You may also have to check if the Cache config is jumpered correctly (if there are chips).
Maybe you even have one of those fake cache boards.

Could be.

I "think" the BIOS will report the amount of cache that the board is jumpered to, regardless of if, and how much, cache is actually installed on the board.
If these are not fake cache chips, then something strange is happening here. Perhaps defective, incompatible or incorrectly installed cache? Maybe one of the chips (1st chip) isn't seated properly so the board thinks the 1st socket is empty and doesn't check the other sockets for the presence of cache chips??

I ran cachechk on my 386. It's not set to "insane performance" at the moment. The cache is 128Kb, and they are the slow 20ns variety. The mobo's osci is 100 MHz, and the DX-40 is running at 50 MHz. The mobo's bus speed is 12.5 MHz.

I made some adjustments to the 386 mobo:

Removed 100 MHz osci. Replaced with 80 MHz osci.
Tightened BIOS memory timings. Previously, they were set to "ignore manually set timings".
Increased bus speed to 20 MHz.

Last one, I promise. 😉 I made some more adjustments. I put the 100 MHz osci back in to the board, and set the bus speed to 16.6 MHz. The BIOS memory timings are still set to "auto" rather than "manual". For all of these 3 cachechk tests, the cache is the "slow" 20ns stuff.

Looks like you are sitting in front of HAL-9000 😉.
Well but it seems this tool is very nice for checking memory / cache subsystem speed.
It also reveals ratfinks 64 kB 'where is it?' cache.

elianda wrote:

Looks like you are sitting in front of HAL-9000 😉.

I am a HAL-9000 unit. That's my reflection. 😉

Maybe we should split this a bit up in a different thread.

Here my results:
This is 2 WS Read, 0 WS Write:

1 CACHECHK V5 10/6/96  Copyright (c) 1995-96 by Ray Van Tassle. (-h for help)
2 CMOS reports: conv_mem= 640K, ext_mem= 31,744K, Total RAM= 32,384K
3 386 Clocked at 39.0 MHz
4 Reading from memory.  
5 MegaByte#:      --------- Memory Access Block sizes (KB)-----
6       1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
7 0:   39   39   39   39   39   39   39   64   64   64   --   --   --    us/KB
8 1:   39   39   39   39   39   39   39   64   64   64   64   64   64    us/KB
9 2  3  4  5  6  7  8  9 10 11 12  <--- same as above.
1013:   39   39   39   39   39   39   39   64   64   64   64   64   61    us/KB
1114:   39   39   39   39   39   39   39   64   64   64   64   64   59    us/KB
1215:   39   39   39   39   39   39   39   64   64   64   64   59   56    us/KB
1316:   54   54   54   54   54   54   54   54   54   54   54   54   54    us/KB
1417 18 19 20 21 22 23 24 25 26 27 28 29 30 31  <--- same as above.
15
16 Extra tests----
17 Wrt  33   33   33   33   33   33   33   33   33   33   33   33   33<-Writing
18 This machine seems to have one cache!? [reading]
19 >>>> If you think you do have L2 cache, you might have FAKE CACHE chips! <<<<
20    !! cache is  64KB --  27.9 MB/s  37.6 ns/byte (162%)             5.6 clks
21  Hmmm. It looks like megabyte #16 isn't being cached!
22  Hmmm. It looks like megabyte #17 isn't being cached!
23  Hmmm. It looks like megabyte #18 isn't being cached!
24  Hmmm. It looks like megabyte #19 isn't being cached!
25  Hmmm. It looks like megabyte #20 isn't being cached!
26  Hmmm. It looks like megabyte #21 isn't being cached!
27  Hmmm. It looks like megabyte #22 isn't being cached!
28  Hmmm. It looks like megabyte #23 isn't being cached!
29  Hmmm. It looks like megabyte #24 isn't being cached!
30  Hmmm. It looks like megabyte #25 isn't being cached!
31  Hmmm. It looks like megabyte #26 isn't being cached!
32  Hmmm. It looks like megabyte #27 isn't being cached!
33  Hmmm. It looks like megabyte #28 isn't being cached!
34  Hmmm. It looks like megabyte #29 isn't being cached!
35  Hmmm. It looks like megabyte #30 isn't being cached!
36  Hmmm. It looks like megabyte #31 isn't being cached!
37  Hmmm. It looks like RAM at megabyte #1 is REALLY slow!
38  Hmmm. It looks like RAM at megabyte #2 is REALLY slow!
39  Hmmm. It looks like RAM at megabyte #3 is REALLY slow!
40  Hmmm. It looks like RAM at megabyte #4 is REALLY slow!
41  Hmmm. It looks like RAM at megabyte #5 is REALLY slow!
42  Hmmm. It looks like RAM at megabyte #6 is REALLY slow!
43  Hmmm. It looks like RAM at megabyte #7 is REALLY slow!
44  Hmmm. It looks like RAM at megabyte #8 is REALLY slow!
45  Hmmm. It looks like RAM at megabyte #9 is REALLY slow!
46  Hmmm. It looks like RAM at megabyte #10 is REALLY slow!
47  Hmmm. It looks like RAM at megabyte #11 is REALLY slow!
48  Hmmm. It looks like RAM at megabyte #12 is REALLY slow!
49  Hmmm. It looks like RAM at megabyte #13 is REALLY slow!
50  Hmmm. It looks like RAM at megabyte #14 is REALLY slow!
51    Main memory speed --  17.2 MB/s  60.9 ns/byte (100%) [reading]   9.1 clks
52    Effective RAM access time (read ) is 121ns (a RAM bank is 2 bytes wide). 
53    Effective RAM access time (write) is  63ns (a RAM bank is 2 bytes wide). 
54    386 Clocked at 39.0 MHz.  Cache ENABLED.
55 Options: -t0

And 4 WS read / 2 WS write:

1 CACHECHK V5 10/6/96  Copyright (c) 1995-96 by Ray Van Tassle. (-h for help)
2 CMOS reports: conv_mem= 640K, ext_mem= 31,744K, Total RAM= 32,384K
3 386 Clocked at 39.0 MHz
4 Reading from memory.  
5 MegaByte#:      --------- Memory Access Block sizes (KB)-----
6       1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
7 0:   39   39   39   39   39   39   39   76   76   76   --   --   --    us/KB
8 1:   39   39   39   39   39   39   39   76   76   76   76   76   76    us/KB
9 2  3  4  5  6  7  8  9 10 11 12  <--- same as above.
1013:   39   39   39   39   39   39   39   76   76   76   76   76   74    us/KB
1114:   39   39   39   39   39   39   39   76   76   76   76   76   72    us/KB
1215:   39   39   39   39   39   39   39   76   76   76   76   72   69    us/KB
1316:   67   67   67   67   67   67   67   67   67   67   67   67   67    us/KB
1417 18 19 20 21 22 23 24 25 26 27 28 29 30 31  <--- same as above.
15
16 Extra tests----
17 Wrt  33   33   33   33   33   33   33   33   33   33   33   33   33<-Writing
18 This machine seems to have one cache!? [reading]
19 >>>> If you think you do have L2 cache, you might have FAKE CACHE chips! <<<<
20    !! cache is  64KB --  27.9 MB/s  37.6 ns/byte (193%)             5.6 clks
21  Hmmm. It looks like megabyte #16 isn't being cached!
22  Hmmm. It looks like megabyte #17 isn't being cached!
23  Hmmm. It looks like megabyte #18 isn't being cached!
24  Hmmm. It looks like megabyte #19 isn't being cached!
25  Hmmm. It looks like megabyte #20 isn't being cached!
26  Hmmm. It looks like megabyte #21 isn't being cached!
27  Hmmm. It looks like megabyte #22 isn't being cached!
28  Hmmm. It looks like megabyte #23 isn't being cached!
29  Hmmm. It looks like megabyte #24 isn't being cached!
30  Hmmm. It looks like megabyte #25 isn't being cached!
31  Hmmm. It looks like megabyte #26 isn't being cached!
32  Hmmm. It looks like megabyte #27 isn't being cached!
33  Hmmm. It looks like megabyte #28 isn't being cached!
34  Hmmm. It looks like megabyte #29 isn't being cached!
35  Hmmm. It looks like megabyte #30 isn't being cached!
36  Hmmm. It looks like megabyte #31 isn't being cached!
37  Hmmm. It looks like RAM at megabyte #1 is REALLY slow!
38  Hmmm. It looks like RAM at megabyte #2 is REALLY slow!
39  Hmmm. It looks like RAM at megabyte #3 is REALLY slow!
40  Hmmm. It looks like RAM at megabyte #4 is REALLY slow!
41  Hmmm. It looks like RAM at megabyte #5 is REALLY slow!
42  Hmmm. It looks like RAM at megabyte #6 is REALLY slow!
43  Hmmm. It looks like RAM at megabyte #7 is REALLY slow!
44  Hmmm. It looks like RAM at megabyte #8 is REALLY slow!
45  Hmmm. It looks like RAM at megabyte #9 is REALLY slow!
46  Hmmm. It looks like RAM at megabyte #10 is REALLY slow!
47  Hmmm. It looks like RAM at megabyte #11 is REALLY slow!
48  Hmmm. It looks like RAM at megabyte #12 is REALLY slow!
49  Hmmm. It looks like RAM at megabyte #13 is REALLY slow!
50  Hmmm. It looks like RAM at megabyte #14 is REALLY slow!
51    Main memory speed --  14.4 MB/s  72.8 ns/byte (100%) [reading]  10.8 clks
52    Effective RAM access time (read ) is 145ns (a RAM bank is 2 bytes wide). 
53    Effective RAM access time (write) is  63ns (a RAM bank is 2 bytes wide). 
54    386 Clocked at 39.0 MHz.  Cache ENABLED.
55 Options: -t0

So just the WS settings increase 20% memory throughput speed !
(17.2 vs 14.4 MB/s)

Cache seems to cover slow timings quite well.

What I do not really understand with my settings:
Write Speed does not decrease when set to 2 WS.

elianda wrote:

What I do not really understand with my settings:
Write Speed does not decrease when set to 2 WS.

Maybe something else is the bottleneck?

I've 3 old retro machine, here's the results (the 486's are 1.0, K6 is 1.0c):

1.QDI v4p895p3 smt v5.0 mobo (opti895 chipset), am5x86-P75 at 150 (3x50), 256kb L2 cache, 16 mb ram (2x8 edo at fpm mode), Diamond Stealth 64 Vram (S3 964 Chipset) 2Mb Vlb, Ms-Dos 7.1 (win95osr2); 90.9 fps

2. Dataexpert Exp4044 256kb l2 cache(opti895 chipset too, but doesn't support write back cache and L2 cache instable 40 and 50 Mhz fsb, so i stuck at 33Mhz), am486dx2-66@100 (8kb write trough l1 cache, 3x33 strange dx2 doesn't work 2x multi, maybe the mobo's bios bug), 16 mb fpm ram, S3 805p 1 Mb vlb, Ms-Dos 6.22; 66.6 fps

3.MSI Ms-5129 mobo (Intel 430VX Chipset, 256KB L2 Cache), 80 MB 60ns Edo Ram (64 Mb cached), AMD K6 233 MMX, Tseng ET6000 2Mb + Diamond V2 8 Mb, Ms-Dos 7.1 (win95osr2); 217.0 fps

Fascinating.

You can really see the differences among cards and how the cpus contribute with this data.

Have updated some results!

I think we have enough data now. At least plenty so people can compare their own machines and check if the performance is ok...

Mau1wurf1977 wrote:

Have updated some results!

I think we have enough data now. At least plenty so people can compare their own machines and check if the performance is ok...

But we still have very few post P1's in the list, and many processors are still missing.

Also, in the Pentium class motherboard section, the Pentium Overdrive 83Mhz result is on a 486 class motherboard 😉

Yes a lot of CPUs are missing, but everyone is using a different system anyway, so the results are always a bit over the place.

Simply, It will have to do 🤣

Lol k. Maybe I'll create my own private database using the same info.

What I like about this system is the sheer ease of adding new info into the table.