VOGONS


First post, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie

Ok, due to constantly arising (off-)topic for PCChips fake cache motherboards, I'm opening this thread. A list with motherboards models and revisions with (possibly) fake cache chips with categories from worst to best cases according to the possibilities for adding real cache:

DCA sticker, cache upgrade is impossible:
1. Octek HIPPO DCA 1 rev. 1.x (claims DCA(1) support, seems like it can't use special EDRAM modules (fast DRAM with some tiny amount of SRAM cache))
2. Octek HIPPO VL (claims DCA(1) support, seems like it can't use EDRAM modules)
3. Octek HIPPO VL+ (claims DCA(1) support, seems like it can't use EDRAM modules)

Upgradable with EDRAM module:
1. Digicom DIGIS-486EL (probably it can use EDRAM modules)
2. Octek HIPPO DCA II (claims DCA2 support, possibly it can use EDRAM modules)

Fake cache chips, upgradable with proprietary COASt module:
1. PCChips/Hsin Tech M507 (has proprietary COASt module slot)
2. PCChips/Hsin Tech M919 ver. 2, ver. 3.2, ver. 3.3 and ver. 3.4 (ver. 1.x is with real cache, all fake cache chips revisions have proprietary COASt module slot)

Fake cache chips, upgradable with real SRAM chips:
1. Amaquest/Anson/Ansoon Technology Co. AP8548
2. Amptron DX-9300A ver. 1.1
3. BEK-tronic BEK-V429S
4. EPoX/ProNiX GXA486SG rev. 0.0
5. Full Yes SIS 486G 3.3/5V rev. G/H
6. Kaimei SA-486 VL-BUS M.B. rev. V2C
7. PCChips/Hsin Tech M912 ver. 1.7, ver. 1.9, ver. 6.0 and ver. 6.1 (ver. 1.4 is with real cache)
8. PCChips/Hsin Tech M915i
9. PCChips/Hsin Tech M918
10. PCChips/Hsin Tech M921
11. Rectron RT-4s3
12. U-Board ST1A-JK-042E rev. B
13. (Unknown brand) PCI486 V1-HJ2

Unnecessary fake cache chips as L2 cache is integrated into CPU:
00. PCChips/Hsin Tech M666
(non-existing motherboard with deliberately written fake article, shame on TRW)

I'm waiting for your suggestions with links, pictures, descriptions, etc. I'll keep the list in the post updated.

Last edited by analog_programmer on 2025-01-16, 06:53. Edited 22 times in total.

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 2 of 28, by rasz_pl

User metadata
Rank l33t
Rank
l33t
analog_programmer wrote on 2024-10-11, 07:50:

DCA wasnt fraud Re: Octek Hippo-VL Motherboard Repair Help Needed (MR BIOS)
Afaik It was sold bundled with ED-RAM. Then again this particular one has 30pin simm sockets, so either they also shipped custom 30 pin ED-RAM or sold boards before ram was ready.

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Reply 3 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
rasz_pl wrote on 2024-10-11, 09:07:
analog_programmer wrote on 2024-10-11, 07:50:

DCA wasnt fraud Re: Octek Hippo-VL Motherboard Repair Help Needed (MR BIOS)
Afaik It was sold bundled with ED-RAM. Then again this particular one has 30pin simm sockets, so either they also shipped custom 30 pin ED-RAM or sold boards before ram was ready.

Ok, Octek are a bit "smarter" than the others. I'll dedicate a special "questionable" category for them 😀

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 4 of 28, by jtchip

User metadata
Rank Member
Rank
Member
rasz_pl wrote on 2024-10-11, 09:07:
analog_programmer wrote on 2024-10-11, 07:50:

DCA wasnt fraud Re: Octek Hippo-VL Motherboard Repair Help Needed (MR BIOS)
Afaik It was sold bundled with ED-RAM. Then again this particular one has 30pin simm sockets, so either they also shipped custom 30 pin ED-RAM or sold boards before ram was ready.

The post below yours points to the Octek Hippo DCA II, which has the option to use EDRAM. That post points to another thread, one of the posts points out the difference between DCA and DCA II and only the latter supports EDRAM.

DCA uses the Headland HTK340 chipset, which requires an external HT44 cache controller. Another board marked with DCA is the Octek HIPPO DCA 1 486. That and the VL/VL+ boards don't appear to have the cache upgrade socket so can't be upgraded to support L2 cache so Octek misleadingly uses the term cache (the chipset only has a write buffer).

DCA II uses the OPTi 82C802G chipset, as can be seen on the silkscreen, so can be upgraded with EDRAM modules.

Reply 5 of 28, by Repo Man11

User metadata
Rank l33t
Rank
l33t

The PCChips M520 has fake cache. The one I have was set up with a K5 PR133 (100 MHz) and because it had fake cache chips and an empty cache slot, it had no L2 at all! The case sticker indicates it was upgraded from a 486, so I suppose it was faster than what had preceded it, but maybe not by much.

"We do these things not because they are easy, but because we thought they would be easy."

Reply 6 of 28, by kmeaw

User metadata
Rank Member
Rank
Member

Let's attempt to "defuse" those BIOS hacks on those boards.

I will start with 08/15/95-SIS-85C471B/E/G-2C4I9PA0-00 for EPoX/ProNiX GXA486SG.
Cache memory is printed out at F000:23F9, the actual call to fetch the amount of cache is at F000:2409 -> F000:A164. The target function is overwritten with "mov al,5; ret" and some garbage bytes follow (21 48 C3 87 DB 87 DB ...). By guessing the overwritten bytes, the function should read as follows: mov al,0e0h (B0 E0) ; call 0xE98A (E8 21 48) ; ret (C3). Luckily the 8-bit checksum stays the same.
So the undo patch is: A164: 05 C3 -> E0 E8

BEK-tronic BEK-V429S 2C4UKB11 2.01r has no overwritten bytes, they used free space and diverted the same call; the undo patch is: 24FE: 00BA -> 2379, FFFF : 12 -> 30.

Reply 7 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
jtchip wrote on 2024-10-12, 00:31:

The post below yours points to the Octek Hippo DCA II, which has the option to use EDRAM. That post points to another thread, one of the posts points out the difference between DCA and DCA II and only the latter supports EDRAM.

DCA uses the Headland HTK340 chipset, which requires an external HT44 cache controller. Another board marked with DCA is the Octek HIPPO DCA 1 486. That and the VL/VL+ boards don't appear to have the cache upgrade socket so can't be upgraded to support L2 cache so Octek misleadingly uses the term cache (the chipset only has a write buffer).

DCA II uses the OPTi 82C802G chipset, as can be seen on the silkscreen, so can be upgraded with EDRAM modules.

Thanks for this detailed info. I'll add Octek HIPPO DCA II mobo in "questionable" category as obviously Octek's "cache" option is not based on true L2 caching scheme.

Repo Man11 wrote on 2024-10-12, 01:33:

The PCChips M520 has fake cache. The one I have was set up with a K5 PR133 (100 MHz) and because it had fake cache chips and an empty cache slot, it had no L2 at all! The case sticker indicates it was upgraded from a 486, so I suppose it was faster than what had preceded it, but maybe not by much.

I can't find anything about M520 model's cache...

EDIT: I've just looked at some detailed pictures of the M520 board and I don't see any cache "chips" on board. In the mobo's manual there's nothing mentioned about onboard cache, the only info is for COASt module. It seems like this model is another cacheless board with slot for proprietary COASt module. I have a similar board PCChips M519 (with OPTi chipset), so I don't see any reason to add this M520 board in "fake cache" category.

Last edited by analog_programmer on 2024-10-12, 09:48. Edited 1 time in total.

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 8 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
kmeaw wrote on 2024-10-12, 02:16:
Let's attempt to "defuse" those BIOS hacks on those boards. […]
Show full quote

Let's attempt to "defuse" those BIOS hacks on those boards.

I will start with 08/15/95-SIS-85C471B/E/G-2C4I9PA0-00 for EPoX/ProNiX GXA486SG.
Cache memory is printed out at F000:23F9, the actual call to fetch the amount of cache is at F000:2409 -> F000:A164. The target function is overwritten with "mov al,5; ret" and some garbage bytes follow (21 48 C3 87 DB 87 DB ...). By guessing the overwritten bytes, the function should read as follows: mov al,0e0h (B0 E0) ; call 0xE98A (E8 21 48) ; ret (C3). Luckily the 8-bit checksum stays the same.
So the undo patch is: A164: 05 C3 -> E0 E8

BEK-tronic BEK-V429S 2C4UKB11 2.01r has no overwritten bytes, they used free space and diverted the same call; the undo patch is: 24FE: 00BA -> 2379, FFFF : 12 -> 30.

Are you suggesting some ways to fix those factory broken BIOSes and restore their functions to detect and report the correct amount of installed cache is someone replace fake "chips" with real ones?

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 9 of 28, by rasz_pl

User metadata
Rank l33t
Rank
l33t
jtchip wrote on 2024-10-12, 00:31:

DCA uses the Headland HTK340 chipset, which requires an external HT44 cache controller. Another board marked with DCA is the Octek HIPPO DCA 1 486. That and the VL/VL+ boards don't appear to have the cache upgrade socket so can't be upgraded to support L2 cache so Octek misleadingly uses the term cache (the chipset only has a write buffer).

Re: What chipsets/devices took advantage of Enhanced EDRAM?
appiah4 : "I own a Hippo VL+ with that chipset and it uses EDRAM"
Would be cool to find someone with DCA and DCA2 EDRAM setups to run some tests.

https://github.com/raszpl/FIC-486-GAC-2-Cache-Module for AT&T Globalyst
https://github.com/raszpl/386RC-16 memory board
https://github.com/raszpl/440BX Reference Design adapted to Kicad
https://github.com/raszpl/Zenith_ZBIOS MFM-300 Monitor

Reply 10 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
rasz_pl wrote on 2024-10-12, 10:06:

Would be cool to find someone with DCA and DCA2 EDRAM setups to run some tests.

One question to clarify something that I still can't get, 'cause I've never seen such a 486 motherboards: Do these Octek boards really have onboard chips with DRAM in combination with some amount of SRAM cache or they can also take same special EDRAM (DRAM+SRAM) modules instead of standard EDO/FP-RAM modules? If the second case is valid (cache is only on the RAM modules), then they're just cacheless motherboards and have no place in this fake cache motherboards thread.

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 11 of 28, by Repo Man11

User metadata
Rank l33t
Rank
l33t
analog_programmer wrote on 2024-10-12, 09:03:
Thanks for this detailed info. I'll add Octek HIPPO DCA II mobo in "questionable" category as obviously Octek's "cache" option i […]
Show full quote
jtchip wrote on 2024-10-12, 00:31:

The post below yours points to the Octek Hippo DCA II, which has the option to use EDRAM. That post points to another thread, one of the posts points out the difference between DCA and DCA II and only the latter supports EDRAM.

DCA uses the Headland HTK340 chipset, which requires an external HT44 cache controller. Another board marked with DCA is the Octek HIPPO DCA 1 486. That and the VL/VL+ boards don't appear to have the cache upgrade socket so can't be upgraded to support L2 cache so Octek misleadingly uses the term cache (the chipset only has a write buffer).

DCA II uses the OPTi 82C802G chipset, as can be seen on the silkscreen, so can be upgraded with EDRAM modules.

Thanks for this detailed info. I'll add Octek HIPPO DCA II mobo in "questionable" category as obviously Octek's "cache" option is not based on true L2 caching scheme.

Repo Man11 wrote on 2024-10-12, 01:33:

The PCChips M520 has fake cache. The one I have was set up with a K5 PR133 (100 MHz) and because it had fake cache chips and an empty cache slot, it had no L2 at all! The case sticker indicates it was upgraded from a 486, so I suppose it was faster than what had preceded it, but maybe not by much.

I can't find anything about M520 model's cache...

EDIT: I've just looked at some detailed pictures of the M520 board and I don't see any cache "chips" on board. In the mobo's manual there's nothing mentioned about onboard cache, the only info is for COASt module. It seems like this model is another cacheless board with slot for proprietary COASt module. I have a similar board PCChips M519 (with OPTi chipset), so I don't see any reason to add this M520 board in "fake cache" category.

I stand corrected. I never checked it that carefully so when I realized it had no L2 cache I assumed the pair of chips above and to the right of the cache slot were fake cache chips the same as my M507. It never occurred to me that someone would set a computer up knowing it had no L2 cache.

"We do these things not because they are easy, but because we thought they would be easy."

Reply 12 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
Repo Man11 wrote on 2024-10-12, 13:47:

I stand corrected. I never checked it that carefully so when I realized it had no L2 cache I assumed the pair of chips above and to the right of the cache slot were fake cache chips the same as my M507. It never occurred to me that someone would set a computer up knowing it had no L2 cache.

The two smaller rectangular Intel chips in the upper right corner (from tower-case perspective) of your PCChips M520 board are parts of 430VX chipset 😀 There's no cache fraud with M520 board - the only option for L2 cache is a proprietary COASt module and it's described in the user's manual.

But M507 model... it's practically the same story - cacheless board with COASt slot, except that there are those fake write-back advertising plastic "chips" connected to nowhere. So far this is the only model with fake "chips" which can't accept real ones. The prize for the fakest one amongst fake onboard cache "chips" mobos goes to PCChips M507 model 😁

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 13 of 28, by jtchip

User metadata
Rank Member
Rank
Member
rasz_pl wrote on 2024-10-12, 10:06:

Re: What chipsets/devices took advantage of Enhanced EDRAM?
appiah4 : "I own a Hippo VL+ with that chipset and it uses EDRAM"
Would be cool to find someone with DCA and DCA2 EDRAM setups to run some tests.

Hmm... that thread comes before the one where he says the board would not POST and it's not clear whether appiah4 got the board working at all. Also, the DCA2 manual specifies the first 2 SIMM slots as DynamiCache (EDRAM) slots whereas the DCA1 manual makes no such mention. There are no manuals for the VL/VL+ and they have 30-pin SIMM slots anyway.

What's odd is the datasheet for the 82C802G chipset (DCA II) mentions support for page-mode DRAM and it has an integrated L2 cache controller so Octek could have just added SRAM chips on the motherboard instead of using EDRAM. I agree we need someone to test DCA and DCA2 boards.

analog_programmer wrote on 2024-10-12, 11:58:

One question to clarify something that I still can't get, 'cause I've never seen such a 486 motherboards: Do these Octek boards really have onboard chips with DRAM in combination with some amount of SRAM cache or they can also take same special EDRAM (DRAM+SRAM) modules instead of standard EDO/FP-RAM modules? If the second case is valid (cache is only on the RAM modules), then they're just cacheless motherboards and have no place in this fake cache motherboards thread.

No, they don't have on-board SRAM or DRAM, only support for EDRAM (at least on DCA II, jury's still out on DCA). TRW seem to have plastered a fake cache label on at least some of these Octek boards, which is misleading because they don't have fake cache, just a misleading sticker that includes the word "Cache" on the chipset. In a sense they're worse than PC Chips' fake cache since you have to rely on rare EDRAM modules to upgrade whereas SRAM chips are more easily available.

It's your list but perhaps it can have multiple categories:

  1. Fake cache chips, upgradable with SRAM chips (most PC Chips boards)
  2. Fake cache chips, upgradable with COASt module (M507)
  3. Fake cache chips, unnecessary as L2 cache integrated into CPU (M666)
  4. DCA sticker, no cache upgrade possible (?, HTK340 supports external HT44 cache controller but the Octek boards don't include a socket for this)
  5. DCA2 sticker, upgradable with EDRAM module

Reply 14 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
jtchip wrote on 2024-10-13, 00:12:

No, they don't have on-board SRAM or DRAM, only support for EDRAM (at least on DCA II, jury's still out on DCA). TRW seem to have plastered a fake cache label on at least some of these Octek boards, which is misleading because they don't have fake cache, just a misleading sticker that includes the word "Cache" on the chipset. In a sense they're worse than PC Chips' fake cache since you have to rely on rare EDRAM modules to upgrade whereas SRAM chips are more easily available.

So, Octek motherboards need some special EDRAM modules for their dynamic cache architecture (DCA) to work. Is this the shortest true statement?

And what about the missing external HT44 cache controller? Which motherboard model has it and which one does not?

I found another statement here, that only DCA2 boards can use EDRAM modules. So, if this is true, thеn DCA(1) sticker is fraud, but DCA2 is not.

I'm totally lost with the Octek's "cache" thing 😁 It would be best if someone can give clear information on which board models can use EDRAM modules and which can't, but have an advertising sticker for missing DCA support.

jtchip wrote on 2024-10-13, 00:12:
It's your list but perhaps it can have multiple categories: […]
Show full quote

It's your list but perhaps it can have multiple categories:

  1. Fake cache chips, upgradable with SRAM chips (most PC Chips boards)
  2. Fake cache chips, upgradable with COASt module (M507)
  3. Fake cache chips, unnecessary as L2 cache integrated into CPU (M666)
  4. DCA sticker, no cache upgrade possible (?, HTK340 supports external HT44 cache controller but the Octek boards don't include a socket for this)
  5. DCA2 sticker, upgradable with EDRAM module

Your idea for more categorized list seems good. I'll think about it.

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 15 of 28, by jtchip

User metadata
Rank Member
Rank
Member
analog_programmer wrote on 2024-10-13, 06:56:

So, Octek motherboards need some special EDRAM modules for their dynamic cache architecture (DCA) to work. Is this the shortest true statement?

And what about the missing external HT44 cache controller? Which motherboard model has it and which one does not?

I found another statement here, that only DCA2 boards can use EDRAM modules. So, if this is true, thеn DCA(1) sticker is fraud, but DCA2 is not.

That's the same post I pointed to earlier 😀

To answer your questions, yes (though possibly only applies to DCA II), and the HT44 cache controller comes on a separate module that is supported on at least the Dell 486/L motherboard. The Octek DCA1 makes no mention of this special slot and the VL/VL+ don't appear to have this slot either, from visual inspection.

Reply 16 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
jtchip wrote on 2024-10-14, 00:02:

That's the same post I pointed to earlier 😀

To answer your questions, yes (though possibly only applies to DCA II), and the HT44 cache controller comes on a separate module that is supported on at least the Dell 486/L motherboard. The Octek DCA1 makes no mention of this special slot and the VL/VL+ don't appear to have this slot either, from visual inspection.

Sorry, I didn't open the links from your post with the explanation on those DCA and DCA II EDRAM cache tech.

Octek's HIPPO DCA, HIPPO VL and HIPPO VL+ mobos lack any HT44 cache controller and special DCA cache+memory card slot, so there's no chance for them to be upgraded/modified with real cache. You're right, these Octek boards are worse than any PCChips (or other brand) fake cache boards 😀

Hey, rasz_pl, I think we have a new winner in category "fake cache top scammer" and surprise... it's not PCSH*Ts 😁 😜 😜

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 17 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
jtchip wrote on 2024-10-13, 00:12:
It's your list but perhaps it can have multiple categories: […]
Show full quote

It's your list but perhaps it can have multiple categories:

  1. Fake cache chips, upgradable with SRAM chips (most PC Chips boards)
  2. Fake cache chips, upgradable with COASt module (M507)
  3. Fake cache chips, unnecessary as L2 cache integrated into CPU (M666)
  4. DCA sticker, no cache upgrade possible (?, HTK340 supports external HT44 cache controller but the Octek boards don't include a socket for this)
  5. DCA2 sticker, upgradable with EDRAM module

Ok, finally I updated the list adding another sub-category - Fake cache chips, upgradable with real SRAM chips or COASt module.

EDIT: PCChips M919 boards versions 2.x and 3.x can only be upgraded with proprietary COASt module, so for now there's no candidates for "Fake cache chips, upgradable with real SRAM chips or COASt module" category.

Last edited by analog_programmer on 2024-11-06, 07:55. Edited 1 time in total.

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"

Reply 18 of 28, by paradigital

User metadata
Rank Oldbie
Rank
Oldbie
analog_programmer wrote on 2024-10-11, 07:50:

Fake cache chips, upgradable with real SRAM chips or COASt module:
1. PCChips/Hsin Tech M919 ver. 2, ver. 3.2, ver. 3.3 and ver. 3.4 (ver. 1.x is with real cache, all fake cache chips revisions have proprietary COASt module slot)

The M919 definitely cannot be upgraded with real SRAM chips, only the proprietary COASt-like module.

Check out the traces around the “cache” to understand why!

pc-chips-m919-2.jpg

pc-chips-m919-4.jpg

Reply 19 of 28, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie
paradigital wrote on 2024-11-06, 07:10:
The M919 definitely cannot be upgraded with real SRAM chips, only the proprietary COASt-like module. […]
Show full quote

The M919 definitely cannot be upgraded with real SRAM chips, only the proprietary COASt-like module.

Check out the traces around the “cache” to understand why!

pc-chips-m919-2.jpg

pc-chips-m919-4.jpg

Tanks for the pictures and the notice! Now I see my mistake - PCChips/Hsin Tech M919 boards versions 1.x use real DIP SRAM chips, while versions 2.x and 3.x use PQFP chip imitations connected to nowhere (as in PCChips/Hsin Tech M507 mobo). So, M919 boards versions 2.x and 3.x go to "Fake cache chips, upgradable with COASt module" category.

from СМ630 to Ryzen gen. 3
engineer's five pennies: this world goes south since everything's run by financiers and economists
this isn't voice chat, yet some people, overusing online communications, "talk" and "hear voices"