The chip has package features that definitely make it a real NEC part and some pinout details suggests it is probably based on one of NEC's gate array products. It may not be something NEC themselves sold, but they definitely produced it.

Tiido wrote on 2025-09-05, 17:15:

The chip has package features that definitely make it a real NEC part and some pinout details suggests it is probably based on one of NEC's gate array products. It may not be something NEC themselves sold, but they definitely produced it.

Tnx!
Some other chipsets from ADC with "NEC" prints

You can see the same type of date+lot code, font and same type of package features. These are consistent with all other actual NEC made chips since mid-late 80s or so.

Several observations of this rather unusual Eagles NEC Power TX "chipset" (merely calling it "chip" or "core logic" might be more appropriate).
About a year or so ago, I could have bought this mainboard on eBay from someone in Kazakhstan for about $75 with shipping, but ended up passing it up, and did not see it again.
Former eastern bloc countries often got hardware products that were too low performance for wealthier countries like US / Canada, western / northern Europe, and Japan, and over the years, I have purchased several of such hardware products.
One should probably run lspci command to figure out the PCI vendor ID to see which vendor might have developed the item.
Based on my own "investigation" (not that it is that sophisticated), it appears that the chip was developed by Taiwan based RDC Microelectronics.
It is a rather low key firm probably for legal reasons (i.e., avoid getting sued by Intel's notorious legal department for patent infringements), but they seem have developed a 486 compatible microprocessor core some years ago, and this Power TX chip is part of their x86 eco system.
It appears that DM&P Vortex86 DX and EX SoC were developed by this same RDC Microelectronics, it is logical to think that this 486 compatible microprocessor and the Power TX chip are their design ancestors.
Personally, I am surprised that all the functionality of a P54C bus + 64-bit FPM DRAM / EDO DRAM / SDRAM controller + 32-bit PCI + ISA bus all fit in a single PQFP package (Is it a 240 pin part or 208 pin part? Sorry, I am too lazy to count the pin count.).
I have two copies of Utron (HiNT) 85C501 / 502 based mainboards from PC Chips, but that one still used 2 pieces of a 208 pin PQFP package to form most of the functionality of the mainboard.
That's my 2 cents.

The Retro Web does not have it catalogued, but there appears to be a 486 version of this mysterious "NEC" chip set from the same Eagles brand.
At least I have seen the photo of the mainboard.

Just to inject some nostalgia about the Japanese semiconductor industry.
A few months ago, I purchased an FIC mainboard with VIA Technologies Apollo Master chipset.
At least 1 out of the 2 208 pin PQFP part was definitely made by Toshiba.
It was manufactured sometime in 1995, but by around 1997, VIA completely switches to TSMC.
In rest of Asia outside of Japan (Japanese sometimes act like they are not Asians, and often compare themselves to Westerners, but Japan is, by definition, will always be an Asian nation due to its culture and geography.), being associated with Japanese technology / brand used to have some currency back then, and I strongly believe why the firm that designed this Power TX ASIC wanted to emphasize the fact that NEC was the foundry of this mysterious chip.
Japan has had 0% average economic growth since 1991, and one of the reasons can be traced to losing the DRAM industry to Samsung, Micron Technology, and Hynix (now SK hynix).
DRAM immensely helped depreciate their fabs, and once Japan Inc. lost their DRAM industry around 2001, that meant that they could not easily recoup their investment in the fabs.
What is likely an under appreciated event that happened in the second half of 1990s is, Taiwanese foundries like TSMC and UMC quietly stole NEC / Toshiba / Fujitsu / Hitachi's foundry business between 1995 to 2000.
Particularly, NEC and Toshiba did brisk business in the foundry business until late 1990s.
With DRAM and foundry business gone by early 2000s, it became increasingly difficult to pay for their fabs, and by 2007, these 4 firms give up on developing semiconductor manufacturing technology beyond 40 nm process.
Perhaps, this is one of the major reason why Japan has had 0% average economic growth since 1991.

Babasha wrote on 2025-09-05, 17:20:

Tiido wrote on 2025-09-05, 17:15:

The chip has package features that definitely make it a real NEC part and some pinout details suggests it is probably based on one of NEC's gate array products. It may not be something NEC themselves sold, but they definitely produced it.

Tnx!
Some other chipsets from ADC with "NEC" prints

I will second your comment.
The silk print looks like an NEC part based on their fonts, so I think the chip was manufactured by NEC.
NEC also did in-house test and assembly (packaging) like many other Japanese semiconductor firms did.
I wonder how "real" NEC employees at their test and assembly site felt when they saw their company name being imprinted on a chip that is not really their product.

Tiido wrote on 2025-09-05, 17:31:

You can see the same type of date+lot code, font and same type of package features. These are consistent with all other actual NEC made chips since mid-late 80s or so.

Perhaps, one might be able to barely run the latest Linux kernel as long as a customized 586 instruction set level kernel can be compiled on Gentoo Linux.
I have done this, so it is possible.
If somehow 128 MB of RAM can get recognized by this mainboard, LXDE can be launched, although it will be pretty useless due to lack of RAM.

"Eagle" sounds suspiciously like a brand name PC Chips might have used and they are notorious for re-labeling chipsets...

That said, this chipset is pretty unusual - a highly integrated single-chip design, but in QFP package. Add to that its late date and support for SDRAM and bus speeds up to 90MHz (if it can actually go as fast as the silkscreen promises) and it doesn't look like any other chipset; the featureset is similar to the SiS 5571 or 5582, but that used a BGA form factor and the chip was bigger than this one. Even if it weren't an actual NEC chip, it doesn't look like any other chip either.

Oh wow, the Eagles mainboard has an ITE IT8661F!
I recently purchased 10 unused IT8661F from a seller on AliExpress for my future retro mainboard project I am working on.
Several other copies of this Eagles mainboard had LGS Super I/O chip for serial/ parallel/ floppy, if I recall.

While IDE can be slow, PIO4 mode should get 16.6 MB / s, at least in theory.
I believe my Utron / HiNT UT85C501 / 502 also "only" supports PIO4 mode as well.

If you try a 64M bit x8 based 128 MB module (total of 16 devices), will it recognize as a 128 MB module?

Perhaps, its power circuit not being able to handle the large current draw of faster AMD K6-2s like ones above 500 MHz.
Interesting to see if AMD K6-III / III+ can run on this board.

Babasha wrote on 2025-09-05, 17:09:

. . . […]
Show full quote

. . .

CPUs: The board supports most Socket 5/7 processors, including the AMD K6-2/350. I personally tested it with an Intel Pentium 100 MHz and an AMD K6-2/K6-2+ (233 to 500 MHz). One quirky feature: if you set the multipliers to 5.5x or 6x in the BIOS, the system just halts during the POST.

CPU Speed: 90 - 350 MHz or above. It supports a wide range of core frequencies and voltages (from 1.8 to 3.5 V) and bus speeds (from 60 to 100 MHz). I noticed the clock generator can unpredictably switch from 100 MHz to 90 MHz.
. . .

bracecomputerlab wrote on 2025-09-09, 12:24:

If you try a 64M bit x8 based 128 MB module (total of 16 devices), will it recognize as a 128 MB module?

All my 128MB modules in any organization count on first stage of BIOS POST up ti 128MB but shows only 64MB in System summary screen and in DOS/WIN9X

bracecomputerlab wrote on 2025-09-09, 12:34:

Perhaps, its power circuit not being able to handle the large current draw of faster AMD K6-2s like ones above 500 MHz.
Interesting to see if AMD K6-III / III+ can run on this board.

Babasha wrote on 2025-09-05, 17:09:

. . . […]
Show full quote

. . .

CPUs: The board supports most Socket 5/7 processors, including the AMD K6-2/350. I personally tested it with an Intel Pentium 100 MHz and an AMD K6-2/K6-2+ (233 to 500 MHz). One quirky feature: if you set the multipliers to 5.5x or 6x in the BIOS, the system just halts during the POST.

CPU Speed: 90 - 350 MHz or above. It supports a wide range of core frequencies and voltages (from 1.8 to 3.5 V) and bus speeds (from 60 to 100 MHz). I noticed the clock generator can unpredictably switch from 100 MHz to 90 MHz.
. . .

I dont think its something about power circuit - it looks like BIOS limitation with high 5.5x and 6x koefs inside BIOS table
PS. BIOS really ugly inside

Okay, you can see if 5.5x or 6x multiplier is working by underclocking the FSB down to 66 MHz.
Another thing is, the clock synthesizer used in this system is the same one used by SiS5591 chip set called Winbond W83194R-17A.
You can get the datasheet from here.

https://www.alldatasheet.com/datasheet-pdf/pd … 83194R-17A.html

The silk print on the mainboard appears to be for W83194R-17, but you should use the settings for -17A by following the datasheet.
This means 90 MHz FSB is not supported, but 95 MHz is instead.

Babasha wrote on 2025-09-09, 12:41:

bracecomputerlab wrote on 2025-09-09, 12:34:

Perhaps, its power circuit not being able to handle the large current draw of faster AMD K6-2s like ones above 500 MHz.
Interesting to see if AMD K6-III / III+ can run on this board.

Babasha wrote on 2025-09-05, 17:09:

. . . […]
Show full quote

. . .

CPUs: The board supports most Socket 5/7 processors, including the AMD K6-2/350. I personally tested it with an Intel Pentium 100 MHz and an AMD K6-2/K6-2+ (233 to 500 MHz). One quirky feature: if you set the multipliers to 5.5x or 6x in the BIOS, the system just halts during the POST.

CPU Speed: 90 - 350 MHz or above. It supports a wide range of core frequencies and voltages (from 1.8 to 3.5 V) and bus speeds (from 60 to 100 MHz). I noticed the clock generator can unpredictably switch from 100 MHz to 90 MHz.
. . .

I dont think its something about power circuit - it looks like BIOS limitation with high 5.5x and 6x koefs inside BIOS table
PS. BIOS really ugly inside

bracecomputerlab wrote on 2025-09-09, 13:28:

Okay, you can see if 5.5x or 6x multiplier is working by underclocking the FSB down to 66 MHz. Another thing is, the clock synth […]
Show full quote

Okay, you can see if 5.5x or 6x multiplier is working by underclocking the FSB down to 66 MHz.
Another thing is, the clock synthesizer used in this system is the same one used by SiS5591 chip set called Winbond W83194R-17A.
You can get the datasheet from here.

https://www.alldatasheet.com/datasheet-pdf/pd … 83194R-17A.html

The silk print on the mainboard appears to be for W83194R-17, but you should use the settings for -17A by following the datasheet.
This means 90 MHz FSB is not supported, but 95 MHz is instead.

Babasha wrote on 2025-09-09, 12:41:

bracecomputerlab wrote on 2025-09-09, 12:34:

Perhaps, its power circuit not being able to handle the large current draw of faster AMD K6-2s like ones above 500 MHz.
Interesting to see if AMD K6-III / III+ can run on this board.

I dont think its something about power circuit - it looks like BIOS limitation with high 5.5x and 6x koefs inside BIOS table
PS. BIOS really ugly inside

Tnx! I will investigate mobo, BIOS, chips in next posts.

Motherboard layout and jumpers (front-panel) settings in attach.

Babasha wrote on 2025-09-05, 17:09:

While the BIOS has a setting for L2 cache, there is no cache on the board. The setting just shows a message that there's 512KB of cache, even though there isn't. The spots for L2 cache are probably just "dummies" to allow for soldering on fake chips.

Looking at the traces where the cache chips would be, I'd guess that if a real cache + tag was soldered it would work (BIOS nonsense aside).
The routing seems too much effort just for nothing, especially given how cramped the PCB is.

The only board with fake QFP cache and fake traces that i know of is the M919 and the routing they did there (just for show) is much more simpler.