VOGONS


3 (+3 more) retro battle stations

Topic actions

Reply 2360 of 2371, by H3nrik V!

User metadata
Rank Oldbie
Rank
Oldbie
pshipkov wrote on 2025-05-12, 05:01:
My initial plan was to place the CF-IDE adapter at the back of the case, but i realized that with a single IDE connector only th […]
Show full quote

My initial plan was to place the CF-IDE adapter at the back of the case, but i realized that with a single IDE connector only the cable will not be long enough to reach the 3 contact points - middle of the case the on-board IDE connector, the CF-IDE adapter on the back and the CDRW drive in the front. Additionally i wanted to remove the 3.5" floppy bay as it obscured too much from the motherboard.
case_floppy_drive_bay.jpg
To resolve these issues, I designed a custom 3D-printed mounting bracket that accommodates the optical drive, floppy drive, and CF-IDE adapter in a single consolidated arrangement. This integrated assembly fits neatly into the 5.25-inch drive bay positioned adjacent to the optical drive, reducing the cable clutter.
floppy_drive_and_cf_ide_adapter.jpg
And this is how the computer ended up looking on the inside:
case_top_lo.jpg

Any chance, you'd be willing to share the STL for that floppy/CF mount? And maybe even the source?

If it's dual it's kind of cool ... 😎

--- GA586DX --- P2B-DS --- BP6 ---

Please use the "quote" option if asking questions to what I write - it will really up the chances of me noticing 😀

Reply 2361 of 2371, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++
pshipkov wrote on 2025-05-14, 05:50:

While it brings interesting details about brand, products, business dynamics, it does not mention "Ambra Lightning 100" which i found strange.
There is no record online about it either, other than some magazine where a pre-production prototype was tested, or something like that.
Makes me wonder if this model hit the shelves at all ... or got shelved.

I'm not totally convinced that the 100MHz BL3 made it to market. If it did, it was in small batches. A 100MHz part was definitely planned though. I even remember seeing an article about the 100MHz SLC3 in a 1993 article of PC World.

As for the link about Ambra, I think the materials are not complete. I am pretty sure that they lasted long enough to deliver products with Pentium CPUs. I think at least some of the Ambras were rebadged Acers.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 2362 of 2371, by pshipkov

User metadata
Rank l33t
Rank
l33t

@H3nrik V!

Attached the OBJ file. It is one of the most portable 3D formats. You can work it out from there.

@Anonymous Coward

That Ambra model most likely was never released. Also, there is no record about 100MHz BL3 in commercially available computers. So, i think i agree with you.

As for the Ambra brand - i only heard about them until recently - when started checking more carefully around Alaris brand and products.
There is not a lot of info about them. They apparently had some collaboration with IBM but was that exclusive in any way - i don't know - probably not.
Looks like an OEM partner of sort. Probably one of many. It just happens that we are interested in the vertical they operated.

retro bits and bytes

Reply 2363 of 2371, by H3nrik V!

User metadata
Rank Oldbie
Rank
Oldbie
pshipkov wrote on 2025-05-15, 01:13:

@H3nrik V!

Attached the OBJ file. It is one of the most portable 3D formats. You can work it out from there.

Thanks a lot!

If it's dual it's kind of cool ... 😎

--- GA586DX --- P2B-DS --- BP6 ---

Please use the "quote" option if asking questions to what I write - it will really up the chances of me noticing 😀

Reply 2364 of 2371, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Ambra was IBM's low end clone-fighter brand, structured as an independent business unit with the autonomy to operate outside normal IBM structures and strictures to be flexible in the market. They appeared to rely on direct sales on TV advertising and some big box store deals, was also what you might call a "supermarket PC". Ergo the enthusiast, business and prosumer markets just completely ignored them, for the masses not the classes. So they barely made a blip in the computer press. Possibly they were cropping up in general magazines and regular newspapers in ads and specials from the big boxers. It's US existence was the shortest I think, but it clung on in Canada and maybe some Euro markets a couple of years longer.

The other IBM brand that they kept on a tighter rein to be slightly more prestigious and actually "from IBM" rather than "with IBM technology" as I think the Ambra association was admitted, was Aptiva, and they appear also to have had some BL models in the same time frame.

The standard amount of support for both of these was three years, the machines of interest were 93-94 machines and support was expiring in 1997. This is significant because IBM and IBM Microelectronics the division that made the motherboards, had no support on WWW until 1998, it was all in parts of compuserve, or direct via BBS. Now while the PC/AT/PS/2 stuff had a large techie following, their files got archived, but the poor relation Ambra and Aptiva info never made it to the web or got in the full BBS archives that cut off in 1997. IBM microelectronics appear to have made everybody's BL2/3 boards and it was the only way to get the CPU. Now adding to the difficulty of pulling stuff from the archival black hole is that IBM microelectronics got more "famous" for their following power PC products, which you've note has PC in the name, like every damn thing, so you search IBM microelectronics PC motherboard and you'll get firehosed with the power PC stuff. Archive.org is also very inefficient for early years, massive sites only, then gradually improving to being nearly representative around year 2000 and getting pretty much everything only from the mid noughts.

Now despite the interest they garner now, and a very small minority back then, these machines were not owned by computer enthusiasts, or people who knew much about computers, they would get skipped over as trash. These people were also not exactly in the forefront of getting on the internet either, so very low chatter from them on web in the 90s. They might have stuck in an AOL CD and stayed stuck on AOL, not even "dodging around the Interface" to get on web, as AOL didn't seem to make their web connectivity very obvious until near the millennium. Many of them were misinformed about computers and even misinformed about the ones they had, the original owners have been known to think they had a DX2-66 or DX4-100.

So 1998 or so you've got BL2/3 in a load of low end machines, owned in the main by not very computer savvy people, they're probably having performance problems, particularly if they downloaded Bonzai Buddy or something, no web support is available their main support options expired last year, they run the millennium bug check from a magazine cover disk which tells them that the machine is going to explode and kill everyone in the house at midnight on the rollover.... well not quite but will say the SurePath BIOS from IBM microelectronics has the millennium rollover problem. So what do they do? Junk it and buy a Dell/Packard Bell etc. These things had the perfect storm combination for being scrapped early, not very computer literate users, absent support, millennium bug, non-standard for upgrades, limited form factors that you couldn't get another optical, ie CD burner in, meh, scrap it... so I suspect a lot of these were out of circulation before 2000.

Anyway, I am certain that many of the Ambra and Aptiva machines with 4100 in the name had a BL3-100 as does my Patriot SL4100 so that numbering gets into 3rd party stuff with IBM microelectronics boards in. The surepath bios with CMOS wiped though, default boots them at 2x, so maybe 100s are missed, since first boot shows them at 66. Unless it's still in a machine with 100 badging you think you got a 66 that does 100 rather than a sold as 100. Anyway, while we might want to think of it as the best of the 386es, it was actually in the market as the worst of the 486es and sold to people who would not notice, at a time where none of them were very internet active and at a time where even if they were archive.org would preserve it poorly, and at a time where the support would blackhole completely due to not being WWW, and at a time where they would get scrapped if they had a millennium bug. In terms of survival traits, they had none, like an airbreathing fish that liked to eat fishing sinkers (weights).

There is also the probability that the 4100 models had a very short viable product lifespan, as a retail price SKU, 2 months, Patriot was dumping theirs through specials at office stores in late 94. The 466 came out a few months earlier and had that time to sell at retail. Everyone who knew anything by late 94 wanted a real Intel DX2-66 as bare minimum, a real DX4-100 if they could afford it, and with money to burn a Pentium, so these were on none of those people's radars. So when they didn't sell in 1994 due to intense competition and possibly better products being available at like, drug stores, they quite likely got debadged and shifted to discount warehouses. So what you're looking for in magazine ads are in the back cheaper listing pages "Famous name desktop 100mhz $299, free printer". Anyway, not unreleased, but released at such an inauspicious time, they made no ripples. Maybe there's a landfill full of them boxed and new, like that Atari game.

Edit: "Peak Ambra" or Ambra's last blast, ironically a September issue magazine when IBM killed Ambra in August of that year, illustrating the 2 month lead time lag common in magazines at the time. This is the "most ambra" mainstream PC publication I have ever found with several full page ads and reviews page 6 and 7 for instance for the BL systems. https://vintageapple.org/pcworld/pdf/PC_World … tember_1994.pdf Note that I should have said 4100BL as the 100-mhz BL3 indicator the convention 4100i in the model appears to indicate intel dx4. Nevertheless through carefully picking 16 bit only benchmarks Ambra tries to show the BL3 as keeping up with or better than intel. A different notebook review with a SLC2-66 noted it was beating a DX50 on some benches, so 16bit integer combined with the cache is keeping it "up there" on 16 bit DOS stuff.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 2365 of 2371, by Anonymous Coward

User metadata
Rank l33t++
Rank
l33t++

"Buying a system? Sure, you want price and performance. Support? Too often it's an afterthought.
Problem is, too many direct order companies treat support the same way. They make the sale, then disappear.
At AMBRA...."

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 2366 of 2371, by pshipkov

User metadata
Rank l33t
Rank
l33t

@BitWrangler
Good notes and good link to that PC-Mag issue.

As i mentioned in my previous post about the 100/50MHz Alaris Cougar PC, there was a clear business design behind these BL3 based computers.
Offer decent base performance and provide upgrade options for 486 CPUs, Pentium Overdrives, VLB and even PCI extension cards. The budget desktop approach.
I spent a moment to compare the listed in the magazine prices for the 66 / 100 MHz BL3 offerings to other adds from Zeus, HP, and so on.
Here is what i see

Ambra
BL3 66MHZ, 4Mb RAM - $1400
BL3 100MHz, 4Mb RAM - $1700
Bl3 100MHz, 8Mb RAM - $2100

Zeos
486SX 33MHz, 4Mb RAM - $1500
486DX2 66MHz, 4Mb RAM - $1750
486DX4 100MHz, 8Mb RAM - $2000
Pentium 90MHz, 8Mb RAM - $2700

Micron
486SX 33MHz, 4Mb RAM - $1200
486DX2 66MHz, 8Mb RAM - $1900
Pentium 66MHz, 8Mb RAM - $2500

HP
486SX 33MHz, 4Mb RAM - $1100
486DX2 66MHz, 8Mb RAM - $1700
Pentium 60MHz, 8Mb RAM - $3600

No name assembly
486DX2 66MHz, 4Mb RAM - $1500
Pentium 66MHz, 8Mb RAM - $2150

Gateway 2000
486SX 33MHz, 4Mb RAM - $1500
486DX2 66MHz, 8Mb RAM - $2000
Pentium 60MHz, 8Mb RAM - $2500

There is huge amount of other listings that i am not going to include here.

Page 144 - comparison of power desktops. Ambra is there.
Page 152 - comparison of budget desktop. Again, Ambra is there.

It is difficult to do proper comparison between the different systems as everybody is trying to make their product shine in some unique way, so limiting to CPU and RAM for simplicity.
Ambra BL3 66 is clearly better than 486SX based PCs.
It is a hit or miss with 486DX2-66. Depends on the DX2 configuration/price. Some are above some are below the Ambra.
bl3 100mhz is not a match to 486DX4-100 in any way, cannot think of a reason why Ambra will be asking similar price (to DX4 machine) for their BL3-100 system.

But i think you are right - these machines didn't last long, despite the focus on upgradability, something about them didn't click on the market it seems.

Btw, It is amusing to skim through these old magazines, full of blunt adds bordering propaganda, clumsily framed as articles.

retro bits and bytes

Reply 2367 of 2371, by pshipkov

User metadata
Rank l33t
Rank
l33t

I ended up getting a bit carried away the other day while running some performance tests on that Alaris Cougar system I wrote about in the previous page.
Couple of custom tests targeted at specific CPU/FPU/VGA workloads escalated on the spot into a 2D fluid solver thing.
I thought I'd share as it is a bit comical - CFDs are one of the worst possible applications for vintage computers - intense floating point calculations, memory hungry algorithms, and in-memory data sets that grow rapidly - basically everything these old machines struggle with most.
I find this impractical aspect so amusing. Anyhow.

Navier-Stokes 2D fluid solver based on Jos Stam's "Stable Fluids" paper.
Linear solve, velocity and density grids only with handling of boundary conditions.
Mouse-controlled emitter - inject density into the grid at location.
Turbo-C 3.x compatible. No dependencies. Single C script with ~400 lines (can be ~300 if compacted).
16-bit (real) DOS mode.
Double buffered rendering in VGA 13h mode.
Command line arguments for setting various parameters (explained in the included source).
Average framerate is reported upon exit.
cfd.jpg

A run captured on video. Sped it up 4 times as the poor PC is on its knees and everything takes a while despite the very humble 40x25 (first solve) and 80×50 (second solve) grid resolution.

Archive includes source code and compiled executable.

The attachment cfd.zip is no longer available

retro bits and bytes

Reply 2368 of 2371, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Maybe you could also run the code on a system like an Am5x86-180 to offer more perspective. From slowest 486 to fastest 486, here's how much you gain...

Plan your life wisely, you'll be dead before you know it.

Reply 2369 of 2371, by pshipkov

User metadata
Rank l33t
Rank
l33t

Good idea.
Here:
486DX5 200MHz
Pentium 3 Tualatin-S @1575MHz
The P3 machine is a dual CPUs rig. The second CPU can make a big impact on performance as the code is very multithreading friendly. Simple omp parallel for would fit right in but this kind of stuff is a different toolchain entirely.

retro bits and bytes

Reply 2370 of 2371, by feipoa

User metadata
Rank l33t++
Rank
l33t++

It is hard to visualise as the BL3 is sped up 4x and, I think, the Am5x86 is not sped up? So it just boils down to the numbers, assuming both are at the same resolution, then:

BL3-100/2x : 1.36 fps
Am5x86-200 : 5.76 fps
dual Tualatin 1.6 GHz: 70 fps

What do that '2' and '3' signify in the command line when running this simulation?

Plan your life wisely, you'll be dead before you know it.

Reply 2371 of 2371, by pshipkov

User metadata
Rank l33t
Rank
l33t

Yes, you gathered the results correctly.
The first argument sets the grid resolution.
0 is 10x10=100 voxels.
1 (default, if no argument) is 40x25=1,000
2 is 80x50=4,000
3 is 160x100=16,000

The 200mhz 486 cpu is 4.235 times faster than the 100mhz bl3 one, or on a clock to clock basis 486 is 2 times more efficient than 386/387.

The 1575mhz p3 is 51.4 times faster than the 100mhz bl3, or in other words p3 is 1.5 times more efficient than 386/387.
I guess with simd and 2 threads the dual p3 system will outperform both 386 and 486 by large extent. Current code is dos/386 centric and does not take advantage of newer tech.

retro bits and bytes