Very interesting, I thought AHA-1542 cards were meant to be kind of crappy, but they do okay here. I have a couple kicking around. May have to plan to put one in my potential ISA 486DX50 build...... Might be an idea for my 386DX40 board as well, but I was tending toward keeping that "middle of the road"/normal. Think I maybe had a future domain 16bit ISA SCSI of some sort too. Might only have kinda lame and chonky clonky 300 and 80 MB drives..
Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.
@rasz_pl
The post is big and things get mixed up.
Repeated the test configuration details on top of every chart.
Gotek floppy drives are working.
Bus-mastering SCSI controllers are the only factor that causes the issue with BL3 L1 cache.
If IDE or non bus-mastering SCSI controllers are used all is good.
So far i tested on the 2 fastest 386 motherboards for obvious reasons. There is a chance both of them exhibit this problem.
In the coming days will check how other motherboards handle things.
ECS Panda 386V and Chaintech 340SCB come on top of the list so far.
@feipoa
Yes, missing characters can happen with several VGA and local storage controllers.
Especially if the two controllers are slot-neighbors. Moving them to specific slots away from each other sometimes can fix things.
So far i didn't have data corruption from the write-access issue.
In fact i thought i escaped it all together with "magical" Promise EIDE Pro, but the recent in-depth testing proved me wrong.
The solution is to switch to mechanical HDDs, or if the preference is to stay with CF cards just increase the ISA bus divider and loose a lot of system performance.
Scraped some pictures from the internet of Winbond based IDE controllers that i think i tried before. Cannot be 100%. Going by memory here.
ROM shadowing
None of the caching controllers handle their own ROM shadowing.
So far only some Adaptec, Bustek, Buslogic, Procom models do it.
Keyboard controllers
I can safely say that no stone was left unturned there.
The problem is clearly the local storage controllers.
Advansys ABP-5140/5142
The best SCSI controller indeed.
So far tested with 2 motherboards - see my note to rasz_pl above.
Will check few more soon. As well as SXL2 processors.
But regardless of what the outcome is, it is unlikely i will be moving to other motherboards.
SIIG i540/542
It is impressive indeed, but does not handle overclocking very well.
Starts to flake beyond 16.67MHz ISA bus.
Didn't test anything yet with DLC/SXL2 CPUs.
Only 386DX and BL3 for now. It is not included in the charts with BL3 CPU because it is unstable or not working at all at 25 and 27.5 MHz.
These tests are big churn. Adding more configurations is borderline unfeasible.
At most i will check how 2-3 other mobos/chipsests handle the BL3 L1 cache issue with SCSIs and check few key controllers of interest with SXL2, but thats it.
SIIG i540/542 and Advansys ABP-5140/5142 are basically the same design/implementation/BIOS_microcodes, yet they behave differently.
i540 is faster at 10MHz, ABP-5140/42 handles overclocking just fine.
@BitWrangler
I always thought that the late AHA-1542 models such as CF/CP are some of the best controllers in their time.
Future Domain in the other hand failed to impress. They are definitely more trouble than the AHA CF/CP.
Jesus tits, that's a big pile of disk controllers. Most of the controllers I had back in the day were either winbond or UMC, with an occasional DTK PTI-217. You seem to have a lot of the more obscure ones.
I think I have a PTI-227B, which is also based on a winbond chip. Is it worth testing?
I also have one of these. It's meant for CD-ROMS. I think it's normally configured for a tertiary drive, but it might be possible to use it with a hard drive using the primary address. https://www.amazon.com/CREATIVE-TECHNOLOGY-CT … D/dp/B00450I8U0
I think the notion I had about adaptecs being crappy might have come from early Linux SCSI documentation, where the authors were obviously preferring PCI solutions. However, there was in the mid 90s somewhat of a plague of crappy SCSI cards that came "free" with SCSI scanners and other peripheral types.
Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.
@anonymous coward
Convincing number of test samples is the main factor, otherwise we are only speculating.
PTI-227B. The B variant didint do well. It was my first attempt to find a counterpart to just 227.
@BitWrangler
Fact is that SCSI controllers are more pretencious in general, so i know what you mean.
But also, a byproduct of the examination was identifying 2-3 of them that are kind of just working and also very fast. So, that changes the story to a point.
Gemlight 386DX-3340 based on Symphony 'Haydn II' SL82C362, SL82C461, SL82C465
Several motherboards based on this chipset have been examined already and was determined that they are the fastest clock-to-clock 386 class assemblies.
This one here is no different.
Not far ago i purchased two of these boards for about $50 each. Not a bad price.
The reason i spend the $ and time on them was see if they scale well to 50MHz FSB.
Not all boards based on this chipset can do it, but if these two can - that will be neat.
The type of chips used in them is similar to DTK PEM-0030Y and DTK PEM-0036Y, which can do 50MHz FSB just fine, so that reinforced my decision.
Was not able to find relevant documentation online, but that's fine since the assembly is basically jumperless and the level 2 cache jumper configuration is printed on the PCB.
Supported are IBM BL3, TI 486SXL2 and 386DX processors.
Up to 32Mb FPM RAM.
Default BIOS is rudimentary, but used microcodes from DTK PEM-0036Y - worked fine.
Despite silkscreen prints specify jumper configurations for 64Kb, 128Kb, and 256Kb level 2 cache, only 64Kb is possible. 128/256Kb lead hard to POST not completing or error reported about external cache not recognized.
I don't think there are 8K x 8 cache chips with faster rating than 20ns. I have 3 full sets of such chips. None of them function properly above 40MHz. So, that's the main limiting factor for this motherboard.
50MHz FSB is possible, but without level 2 cache. Obviously performance is not the best.
Decided not to do a full round of testing - it won't reveal anything important.
Basically, at 40MHz everything just works.
Same for 50MHz without level 2 cache.
If overclocking is not an objective, these boards are great.
Despite the smaller level 2 cache buffer they still outdo pretty much everything out there.
And that's all.
Nothing more to say really.
Thanks for donating this board to me some time ago. The comment concerning 256k not functioning should be corrected. 256K works fine with two 64kx4 TAG chips plus eight 32kx8 chips. The main drawback of this board is its glaring colour.
Nice to hear you made it work with 256Kb level 2 cache.
It refused the 64kx4 Micron chips here. I don't have others.
Sound card blocks the crystal oscillator. What's the frequency ?
Now this is getting interesting.
With working 256Kb L2 cache these boards can easily be alternative to the rare DTK PEMs.
In my tests early they appeared solid.
Can you spin them while you still have the setup on the test bench ?
You always have to check the pinouts on tag RAMs. Sometimes you'll have two chips with the same number of pins and the same memory arrangement, but different pinout. Sometimes you get boards that need tags that were only produced by one company and almost impossible to find.
pshipkovwrote on 2024-05-31, 23:28:Nice to hear you made it work with 256Kb level 2 cache.
It refused the 64kx4 Micron chips here. I don't have others.
Sound card […] Show full quote
Nice to hear you made it work with 256Kb level 2 cache.
It refused the 64kx4 Micron chips here. I don't have others.
Sound card blocks the crystal oscillator. What's the frequency ?
Now this is getting interesting.
With working 256Kb L2 cache these boards can easily be alternative to the rare DTK PEMs.
In my tests early they appeared solid.
Can you spin them while you still have the setup on the test bench ?
Could you show me a photo of the two 64kx4 Micron chips you tried? How many did you try? Maybe one was bad. Looking at the pin configurations for a common Micron 64kx4 and Cypress 64kx4, the pinouts look the same.
The installed crystal is the supplied 80 MHz crystal. Everything works fine so far.
I'm running NT 3.51 and Win 3.11 on it now. Wanting to do some mp3 tests when I have time.
You always have to check the pinouts on tag RAMs. Sometimes you'll have two chips with the same number of pins and the same memory arrangement, but different pinout. Sometimes you get boards that need tags that were only produced by one company and almost impossible to find.
That wouldn't surprise me. There was some pretty uncommon configurations in the wild, wild west of the early 90's. However, for the part numbers I looked up, Micron and Cypress 64kx4 had the same pinouts. The only difference being that one manufactorur may call pin 1 A0, while the other may call it A6, etc. But all pins 1-10, and 18-23 are address pins, pins 14-17 are I/O pins, pin 24 is Vcc, etc.
Plan your life wisely, you'll be dead before you know it.
i have 5 of them.
2 are in use.
all of them are solid.
lets see if you can escape some of the sound issues you had with PEM 3000.
Also, try a bit higher than 40mhz.
40 is boring : )
it is true that no brand/model level 2 cache chips are "best". only trial and error gets you a working set. especially true when operating outside specifications.
What's the probability that you thought you tested those 64kx4 modules, but forgot to do it? I'll have to try these chips for myself. If you're sure you tested those, do you recall what symptoms you received after installing them? Did you try at lower frequencies, like 25-33 MHz FSB? 15 ns should be plenty for 25-40 MHz FSB.
Trying the sound in non-single mode DMA is definately on my agenda. Considering that the PEM-4030Y and PEM-4036YB both exhibited this issue, my hopes aren't high for the Gemlight MB386-40-SYM.
To go beyond 40 MHz FSB, I'd need to either run the SXL in 1x mode at 50 MHz, or run the IBM BL3 at 2x50. But since the BL3 is slow on these Symphony based boards, it isn't very motivating. I'll try it anyway. Maybe I can get 2x55 running.
Plan your life wisely, you'll be dead before you know it.
What is the probability of me not testing these chips ?
Zero.
I bought the two boards specifically for the purpose of finding an alternative for the DTK PEMs.
The level of interest was very high.
Cannot remember if tested 33MHz. The way i roll 45 is the base. Below that is no story.
I sold the other board, so cannot verify your result here anymore.
Let's see what you find out. Very curious.
You already have SXL2 in the PGA132 socket. Just run it without clock multiplier.
Couldn't push ISA any faster or system wouldn't boot. How you got 25 MHz on your AHA-1520B is puzzling. Maybe only the BL3 will run with this high of ISA clock when using 1520/1530?
EDIT:
I tried SXL at 1x50 MHz, but had some error while booting DOS (missing command interpreter). I was going to play with RAM and SRAM, then both plastic clips on one of the SIMMs broke. I need to swap the SIMMs before I can do any further testing. Looking back at my notes from the PEM-4030Y, I wasn't able to get 50 MHz FSB running. Didn't you need the PEM-4036Y (not YB) to get 50 MHz operational?
EDIT2:
Swapped all SIMM sockets, then did some more tests. 50 MHz FSB, even with SRAM chips removed, is a hard no go. Can you confirm that you successfully tested an SXL chip on this motherboard at 50 MHz w/out SRAM?
Plan your life wisely, you'll be dead before you know it.
I usually publish videos for the more interesting experiments, but didn't do it for the ISA IDE/SCSI investigation.
Here are two for Adaptec AHA-1520B:
DTK PEM-4036Y, STB Nitro 2Mb (CL GD-5434), LAN, Sound
ISA bus at 25MHz
ZuluSCSI + 8Gb CF card
SCSI BIOS set to 6.67Mb/s transfer speed with TI 486SXL2-50 processor at 50MHz (1x50MHz) with IBM BL3 processor at 100MHz (2x50MHz)
How i do it ?
AHA-1520B and 1530P are excellent controllers. They are the best Adaptec ISA devices if you ask me.
The ones here are nothing special - random picks from Ebay.
They just worked. No special treatment.
The system itself took some effort to get there. Two useful links:
RAM modules here
Level 2 cache chips as shown here, also check the EDIT section below in the same post. It is for the SXL2 CPU.
Btw, this morning i stabbed DTK PEM-4030 (3330) on Ebay.
It hit me in the face, so i took it.
It is the same model you have.
Will be interesting to cross-check it on my side.
The Gemlight boards worked fine at 50MHz, but since i was not able to get 256Kb L2 cache working and the installed chips were too slow for 50 i tested briefly only. But technically the boards worked fine.
Last edited by pshipkov on 2024-06-01, 21:24. Edited 1 time in total.
In that video, you are testing the PEM-4036Y. I recall you saying only the 4036Y would overclock well. Can you confirm that the PEM-4036YB works at 25 Mhz ISA? Can you also see if another graphics card will work? I'm out of GD-5434 cards to test.
The eBay listing was for PEM-3330Y. I'm not sure what the difference is between the PEM-4030Y and the 3330Y labels, but if I had to guess, the "33" prefix implies that it doesn't do 40 MHz well. I know they used to bin the chipsets and C&T CHIPS would actually stamp them with 33 MHz or 40 MHz depending on the binning.
The Gemlight boards worked fine at 50MHz, but since i was not able to get 256Kb L2 cache working and the installed chips were too slow for 50 i tested briefly only. But technically the boards worked fine.
Perhaps it worked with some very select DRAM modules at 50 MHz, but I do not have those. I have a few of what you've highlighted in your thread, but only 4 pieces total.
pshipkovwrote on Yesterday, 18:25:I usually publish videos for the more interesting experiments, but didn't do it for the ISA IDE/SCSI investigation.
Here are two […] Show full quote
I usually publish videos for the more interesting experiments, but didn't do it for the ISA IDE/SCSI investigation.
Here are two for Adaptec AHA-1520B:
DTK PEM-4036Y, STB Nitro 2Mb (CL GD-5434), LAN, Sound
ISA bus at 25MHz
ZuluSCSI + 8Gb CF card
SCSI BIOS set to 6.67Mb/s transfer speed
Was there a zuluscsi variant which had CF card's? I thought they were all SD card based.
Plan your life wisely, you'll be dead before you know it.
PEM-4036YB is an awesome board, but it does not overclock well. 40MHz is the limit. Can do 45 but unstable.
Because of that the highest ISA bus frequency can be 20MHz only. That is not a problem for almost all decent SCSI adapters.
If the 3330 board arrives as one piece will find out what 33 means.
You will probably need more than 4 RAM modules in case yours are not "perfect" and binning is needed. But give them a try first.
Yes. ZuluSCSI is with SD cards. Typo in my previous message.
I can confirm that most decent+ video cards work just fine at 50MHz on the DTK PEM-4036Y motherboard.
For example: ET4000/W32i, Trident 8900C#, CL-5426/28/29 and so on.
Last edited by pshipkov on 2024-06-02, 01:17. Edited 1 time in total.
I can confirm that most decent+ video cards work just fine at 50MHz on the DTK PEM-4030Y motherboard.
For example: ET4000/W32i, Trident 8900C#, CL-5426/28/29 and so on.
I think you meant to write PEM-4036Y. You don't have a 4030Y (yet).
PEM-4036YB is an awesome board, but it does not overclock well. 40MHz is the limit. Can do 45 but unstable.
Because of that the highest ISA bus frequency can be 20MHz only. That is not a problem for almost all decent SCSI adapters.
The 4036YB cannot even do 45 MHz FSB? That renders it pretty useless. I remember having issues getting the 4036YB to boot with my SCSI card, a symptom I plan to investigate more in the next few days. On the other hand, my 4030Y, which contains less 7400 chips, can do 45 MHz FSB just fine.
Curiously, the only difference between the 4036Y and 4036YB are the use of QFP 7400 series chips on the *YB, whereas the *Y uses all DIP 7400 chips. Previously, I sent you a spreadsheet tabulating the difference between the three boards - I thought there were a few different components, but once I had the board in my hands, I realised the missing QFP 7400 chips from the YB were under the KBC. Thus, both the *Y and *YB are identical, except for QFP vs. DIP.
Thus, it is surprising that the 4036Y can handle up to 50 MHz stable, while the 4036YB can only go up to 40 MHz. Can someone explain how DIP over QFP chips can allow for 50 MHz? Simple case of PCB design differences, or is there some magic in the DIP 7400 chips?
If the 4036YB can only do 40 MHz, and it is a tightly integrated baby AT board, I'd be surprised if the Gemlight 386DX-3340 (aka MB386-40-SYM), which is much larger motherboard, can do 50 MHz with SRAM installed. I'll try to get 50 MHz, no SRAM, working again, but it's not looking hopeful. I even tried 60 ns, 1 MB chips.
Yes. ZuluSCSI is with SD cards. Typo in my previous message.
What did you use as a benchmark app for your massive SCSI bench thread? I used coretest. Do you have an explanation why, at only 40 MHz FSB, and SXL2-80 MHz, ISA at 13.3 MHz, why my 1522B system benched faster than yours at 25 MHz ISA? Simple difference between BL3-100 and SXL2-80? I didn't think the perf. difference between SCSI2SD and ZuluSCSI would be so different. I thought SCSI2SD v6 was only supposed to be a hair faster than ZuluSCSI.
Plan your life wisely, you'll be dead before you know it.
You and i seem to be the only ones who are into 386 overclocking and stuff.
The wide masses, you know, those five or so other guys out there, seem to be satisfied with 40MHz.
So, i wouldn't call DTK PEM-4036YB useless.
It is in the family of the fastest clock-to-clock 386 motherboards, and can be a great fit for builds within spec.
At some point i was thinking to replace the QFP chips with DIP ones, but common sense prevailed.
For now there is no strong enough reason for me to risk ruin the YB board.
Also, i am not sure if it is the QFP chips really. Can be some subtle difference in the wiring, or who knows what else.
The Gemlight boards did well at 50MHz, but without L2 cache. That's for sure. Maybe start from there again. Hope you figure it out.
You have the AHA-1520B setup in 10Mb/s transfer speed, while i have it in 6.67Mb/s only (otherwise boot to DOS does not happen at 25MHz ISA). This setting is the dominant factor for SCSI.
ISA bus speed is secondary.
CPU type and multiplier, presence/absence of level 2 cache, CF/SD cards are no factors.
So, 13MHz ISA + 10Mb/s == 25MHz ISA + 6.67Mb/s, give or take couple of Kb/s.
How i tested ?
Coretest - easy, but often misleading.
SpeedSys - slightly harder, somewhat less misleading.
Copy directory with files. Around 50Mb in size. Few big and lots of small ones. Stop-watched manually. Results confirm that Speedsys linear read speed is the more accurate number to look at from the synthetic tests.
Open a text file in NC or DN text editors and hold F2 (save) to spam the IDE. This is surprisingly flaky test when overclocking the old hardware. With high ISA bus frequencies some SCSI controllers can corrupt the file or completely mess-up the file system. Some IDE controllers may hit the write access issue.
Run DOS and Windows games/apps. Some controllers can cause auto-progression when overclocked. Like i hold down the ENTER key. In Windows the results are funny - anything the cursor points at gets executed automatically - dialogs fly around. From the DOS apps/games, Wolf3D seems to be the most sensitive to this issue.
I missed a few descriptors in my last message. I should say, PEM-4036YB is useless if you want to run higher than 40 MhHz. Now that 90 MHz is achievable on an SXL2 chip, I see little desire for 80 MHz. BL3 at 2x50 is also a nice option, thus the YB isn't so interesting for me. If I need to run the BL3 at 33 or 40 MHz FSB, there's faster boards than the YB.
Did you try YB with BL3 at 2x45? Or it failed even with 1x45?
The YB board doesn't contain the DIP real-estate anywhere on the PCB, so trying to patch in DIP chips onto QFP pads will be a wiring mess. While the *Y board used exclusively DIP, *YB is a mix of DIP and QFP, without any DIP "pads" adjacent to the QFP chips. I suspect the speed issue with *YB is related to how the traces are routed, thus limiting FSB stability.
I played again with the Gemlight at 50 MHz. Even with DRAM at 2 ws, it will not POST successfully. Perhaps the key is your 100 MHz crytals? I remember you saying that this could matter when on the edge. What is the branding on yours? Mine have ECLIPTEK ECH11, 100.00M.
I already tried all sorts of SIMMs, 4 modules and 8 modules. 50 MHz isn't happening for me, SRAM or otherwise. Also tried 6.67 MHz. Is it possible that you tested up to 50 MHz on just one of your Gemlight boards (the one that was sold online) and the secondary board (the one I received) wasn't confirmed up to 50 MHz? I'm thinking that certain chipsets cannot cope with 50 MHz. I did try a heatsink/fan on the chipset though.
Open a text file in NC or DN text editors and hold F2 (save) to spam the IDE. This is surprisingly flaky test when overclocking the old hardware. With high ISA bus frequencies some SCSI controllers can corrupt the file or completely mess-up the file system. Some IDE controllers may hit the write access issue.
Run DOS and Windows games/apps. Some controllers can cause auto-progression when overclocked. Like i hold down the ENTER key. In Windows the results are funny - anything the cursor points at gets executed automatically - dialogs fly around. From the DOS apps/games, Wolf3D seems to be the most sensitive to this issue.
I'll have to try this. Are you holding down F2 while opening Notepad with the mouse left-click? Or do you already have Notepad open with some text on it, and only then start to hold down F2 and wait? Wait how long? Are you still holding down F2 while trying to close Notepad with the mouse? Must you open the saved file to witness the corruption? Or is there a corrupt file on the HDD that Scandisk picks out?
For your other test, you just hold down ENTER while on the Windows 3.1 desktop and move the mouse around? The test passes if apps that the mouse pointer hoovers over aren't opened, and also nothing else flying around?
Plan your life wisely, you'll be dead before you know it.
The YB board doesn't contain the DIP real-estate anywhere on the PCB, so trying to patch in DIP chips onto QFP pads will be a wiring mess. While the *Y board used exclusively DIP, *YB is a mix of DIP and QFP, without any DIP "pads" adjacent to the QFP chips. I suspect the speed issue with *YB is related to how the traces are routed, thus limiting FSB stability.
I played again with the Gemlight at 50 MHz. Even with DRAM at 2 ws, it will not POST successfully. Perhaps the key is your 100 MHz crytals? I remember you saying that this could matter when on the edge. What is the branding on yours? Mine have ECLIPTEK ECH11, 100.00M.
So far all crystals worked fine at up to 100MHz. At least i didn't notice difference between them.
Story changes at 110MHz.
I already tried all sorts of SIMMs, 4 modules and 8 modules. 50 MHz isn't happening for me, SRAM or otherwise. Also tried 6.67 MHz. Is it possible that you tested up to 50 MHz on just one of your Gemlight boards (the one that was sold online) and the secondary board (the one I received) wasn't confirmed up to 50 MHz? I'm thinking that certain chipsets cannot cope with 50 MHz. I did try a heatsink/fan on the chipset though.
Tested both. Wanted to see if they behave differently to determine which one to keep. They functioned the same. One of them went to Ebay, the other one is with you.
I am really very puzzled that you are struggling to get it going at 50MHz. What can be the problem ? I assume FPU is not inserted, right ? Most of them don't like 50MHz very much.
So at 50MHz no lights at all ? Or POST hangs ? Or boot hangs ? Where exactly it stops ?
Maybe you really collected all the bad memory modules on the planet, but i doubt it. : )
I'll have to try this. Are you holding down F2 while opening Notepad with the mouse left-click? Or do you already have Notepad open with some text on it, and only then start to hold down F2 and wait? Wait how long? Are you still holding down F2 while trying to close Notepad with the mouse? Must you open the saved file to witness the corruption? Or is there a corrupt file on the HDD that Scandisk picks out?
You can try it in Windows Commander, Norton Commander, or Dos Navigator.
F4 to open file for edit. Then hold F2 for i don't know - 30 seconds to a minute and see what happens.
For your other test, you just hold down ENTER while on the Windows 3.1 desktop and move the mouse around? The test passes if apps that the mouse pointer hoovers over aren't opened, and also nothing else flying around?
What i meant is that when overclocked some SCSI adapters seem to freak-out the keyboard controller or something like that.
In turn the keyboard controller keeps spamming the system with keys presses, including the Enter one.
As a result any context sensitive actions get triggered.
When in Windows if i select an icon it immediately invokes its function. If i move the mouse over buttons or other active elements they get triggered.
It is a zoo. Very dangerous one because you can easily mess-up windows configs and the content on disk. Quickly learned that Wolf3D is 100% indicative of the problem, so stopped double checking in Windows. Anyway.
