Xicor wrote on 2026-06-28, 11:59:

[...]

This Shuttle had a corrupted bios, and for some reason the retroweb had the info as if it was a VIA chipset, only that mine has an Intel Bx.... All the bios form retroweb failed. Managed to find a 128GB patched bios that did work from another source.

Something tells me the "V" in the HOT-661V refers to Via 😉

This one, the 661/P is likely to be yours.

rasz_pl wrote on 2026-06-28, 22:30:

Yamaha XG compatibility is just software Synth. They also shipped software library for some level of A3D compatibility.

A software synthesizer from Yamaha and the A3D libraries—and how great it look on stickers!
But anyway, here it’s taken things to a whole new level:
https://www.youtube.com/watch?v=KTHY5YoDN4c

New aquisitions.

nuno14272 wrote on 2026-06-29, 16:00:

New aquisitions.

Nice Gravis Ultrasound clone you got there.

Found a nice X-Fi Fatal1ty, I think the only difference from the Titanium other than the shield is the 64MB X-RAM that like 3 games use 🤣

AaronS wrote on 2026-06-29, 19:34:

Found a nice X-Fi Fatal1ty, I think the only difference from the Titanium other than the shield is the 64MB X-RAM that like 3 games use 🤣

I have a few of those, they're perfect for those ultra high end XP builds and windows 7. Try to find the 3.5 drives if you can, they add a nice touch and are actually practical.

The Fatality ones don't seem to popup very often on ebay but then I haven't really been looking for one particularly, saw it pop up for £20 and seemed like a fair price. I see a ton of the Titaniums though, especially the ones with blue PCBs.

weedeewee wrote on 2026-06-29, 16:33:

nuno14272 wrote on 2026-06-29, 16:00:

New aquisitions.

Nice Gravis Ultrasound clone you got there.

Now that's a brand I haven't heard about in a long time... I knew they made sound cards, but I never got a hold of one. How good are they compared to Sound blaster? All I had from Gravis was a Gameport Joystick, and it was very good.

Shader_BiH wrote on Yesterday, 18:03:

weedeewee wrote on 2026-06-29, 16:33:

nuno14272 wrote on 2026-06-29, 16:00:

New aquisitions.

Nice Gravis Ultrasound clone you got there.

Now that's a brand I haven't heard about in a long time... I knew they made sound cards, but I never got a hold of one. How good are they compared to Sound blaster? All I had from Gravis was a Gameport Joystick, and it was very good.

Like any soundblaster has their quirks, the gravis line also has their own quirks, one of them being no direct hardware soundblaster compatibility. for any more info someone else will have to supply it since I'm no expert even though I have four or so gravis ultrasound cards.
My personal opinion. they're a pain in the ass at times, but they're nice to have and play with, especially when the software supports them directly.
Though... the first generations (maybe all of them, aside from some clones) could have been made better if they had some good analog pcb layout.

Couple of lowend cards. Cut down Riva TNT (8 MB / 64-bit). And below the elusive Savage IX. It was commonly used in noteboks and few of them found their way on AGP card too. Notice the mid-2002 datecode - at the time it was horribly outdated, maybe comparable to VIA integrated solutions that were based on the same technology.

Interesting card that Savage IV, the video memory is located on the back side?

That S3 is awesome, it screams CHEAP, super small PCB, only the chip, without cooling and the BIOS, memory on the back, I assume a single module of cheap and slow RAM.

PD2JK wrote on Yesterday, 21:12:

Interesting card that Savage IV, the video memory is located on the back side?

Nope, the Savage IX has 8MB of embedded SDRAM in the chip.

PD2JK wrote on Yesterday, 21:12:

Interesting card that Savage IV, the video memory is located on the back side?

nothing on the back https://www.vgamuseum.info/index.php/news/ite … 78-s3-savage-ix

dionb wrote on Yesterday, 22:31:

Nope, the Savage IX has 8MB of embedded SDRAM in the chip.

Are you sure? Obviously eSDRAM is out of the question, In 1999 8MB of eSDRAM would be on the cutting edge of technology, super expensive, super failure rate prone, and at 0.18μm take almost 1cmx1cm of die area alone, thats more than the whole 86C298. This leaves us with some esoteric for the time dual die package situation, except there were no single die 64Mbit 64bit (not to mention 128bit) chips available in 1999. The best you could get at the time was 4 MB x 16, and that would be top end expensive die.
Just look at normal 8MB Savage 4/lt https://www.vgamuseum.info/index.php/news/ite … 6-s3-savage4-lt its using 1 MB x 16 bit ram chips and they take more space than the graphic chip 😀

My guess is either a tiny internal on-die cache buffer (tens of kilobytes at max) and marketing claiming 8MB (because AGP) or single 512 KB x 16 Bit x 2 Banks = 2MB (1999 $1) with everything else streaming over AGP sideband.

Edit: Curiously HP advertised S3 Savage/IX in two versions? "Integrated 4MB or 8MB SGRAM" https://macdat.net/laptops/hp/pavilion_n5100.php https://www.macdat.net/laptops/hp/omnibook_xe3.php

So its most likely 2 MB x 16 and 4 MB x 16 embedded ram dies. Not great for BW and obviously a lie when it comes to memory bus width, but not tragic otherwise.

Would be cool if someone had a broken card and decapped Savage/IX chip. At $15 they are normally selling working its a waste of money just to satisfy curiosity 😀

H3nrik V! wrote on 2026-06-26, 12:17:

myne wrote on 2026-06-26, 07:31:

Right you are. My bad. I guess they did the whole "move the northbridge on-die" in steps. […]
Show full quote

H3nrik V! wrote on 2026-06-24, 15:36:

Did a bit of reading. Looks like the i7-9xx (Nehalem based) doesn't have PCIe on board, but that it is on the chipset via QPI

Right you are.
My bad.
I guess they did the whole "move the northbridge on-die" in steps.

https://www.intel.com/content/dam/www/public/ … sheet-vol-1.pdf

Interesting document. Thanks!

You know what's confusing?
1156 has PCIE on-chip. 1366 doesn't.

I do know that QPI is basically a variant of PCIE, so perhaps they figured a MUX on the chipset would be 'better'

Just got this Hercules Dynamite 128/Video (ET6000). These were blazing fast cards for a while back in the day, I think it was just later in 1997 or so, when new cards actually started to beat these MRAM ET6000s in all scenarios.

Seeing this card also got me thinking that these Dynamite cards must be one of the first cards that use silk screen this heavily for branding and especially on the "upper" side. Pretty neat.

myne wrote on Today, 03:19:

You know what's confusing? 1156 has PCIE on-chip. 1366 doesn't. […]
Show full quote

H3nrik V! wrote on 2026-06-26, 12:17:

myne wrote on 2026-06-26, 07:31:

Right you are. My bad. I guess they did the whole "move the northbridge on-die" in steps. […]
Show full quote

Right you are.
My bad.
I guess they did the whole "move the northbridge on-die" in steps.

https://www.intel.com/content/dam/www/public/ … sheet-vol-1.pdf

Interesting document. Thanks!

You know what's confusing?
1156 has PCIE on-chip. 1366 doesn't.

I do know that QPI is basically a variant of PCIE, so perhaps they figured a MUX on the chipset would be 'better'

Yes, that's a funny note - but you wrote it yourself: "I guess they did the whole "move the northbridge on-die" in steps."

If I look at it, it seems like the 1156 launched close to a year later than 1366 did. I guess the 1366 was the first "flagship" on-die memory controller chip, where they only put the memory controller on-board, rather than everything all at once. The thing is that the on-board PCIe on 1156 was on the expense of "only" dual channel memory. (not that it really made any practical difference - but data sheets are data sheets 🤣 )

I guess the fact that the QPI link was scalable (4.8GT/s to 6.4 GT/s) made it possible to also scale the number of PCIe channels from the northbridge side, whereas the 1156 had one x16 interface. That was probably not enough for the HEDT platform in Intel's mind back then. Also, if I understand correctly, the QPI is also used on Xeons for inter-socket communications?

rasz_pl wrote on Yesterday, 23:47:

nothing on the back https://www.vgamuseum.info/index.php/news/ite … 78-s3-savage-ix […]
Show full quote

PD2JK wrote on Yesterday, 21:12:

Interesting card that Savage IV, the video memory is located on the back side?

nothing on the back https://www.vgamuseum.info/index.php/news/ite … 78-s3-savage-ix

dionb wrote on Yesterday, 22:31:

Nope, the Savage IX has 8MB of embedded SDRAM in the chip.

Are you sure? Obviously eSDRAM is out of the question, In 1999 8MB of eSDRAM would be on the cutting edge of technology, super expensive, super failure rate prone, and at 0.18μm take almost 1cmx1cm of die area alone, thats more than the whole 86C298. This leaves us with some esoteric for the time dual die package situation, except there were no single die 64Mbit 64bit (not to mention 128bit) chips available in 1999. The best you could get at the time was 4 MB x 16, and that would be top end expensive die.
Just look at normal 8MB Savage 4/lt https://www.vgamuseum.info/index.php/news/ite … 6-s3-savage4-lt its using 1 MB x 16 bit ram chips and they take more space than the graphic chip 😀

My guess is either a tiny internal on-die cache buffer (tens of kilobytes at max) and marketing claiming 8MB (because AGP) or single 512 KB x 16 Bit x 2 Banks = 2MB (1999 $1) with everything else streaming over AGP sideband.

Edit: Curiously HP advertised S3 Savage/IX in two versions? "Integrated 4MB or 8MB SGRAM" https://macdat.net/laptops/hp/pavilion_n5100.php https://www.macdat.net/laptops/hp/omnibook_xe3.php

So its most likely 2 MB x 16 and 4 MB x 16 embedded ram dies. Not great for BW and obviously a lie when it comes to memory bus width, but not tragic otherwise.

Would be cool if someone had a broken card and decapped Savage/IX chip. At $15 they are normally selling working its a waste of money just to satisfy curiosity 😀

Supposedly there is 8MB / 64-bit integrated in the package. Savage IX should be similar to Savage 4 or some hybrid between Savage 3D and S4. Both core and memory running at 100 MHz. I want to make some benchmarks of nearly all early 3D cards, the IX included... but it will take time 😁

There are new IX chips available on ebay. So if anyone wants to cut one for science, it is possible without killing working card 😁

nuno14272 wrote on 2026-06-29, 16:00:

New aquisitions.

brutal awesome acquisition !!

Yes a GUS clone!

X1950 Pro 256MB came in yesterday. Managed to undo its unnecessary overkill Rialto cooling.

We went from this absolute UNIT of a ghetto Rialto cooling (which would be enough for a Pentium MMX!!!!!!)

To something more within normal ranges, recycled from a broken RS481 motherboard's southbridge, although I had to clip off the old pushpin holes (they wouldn't reach the screw next to the Rialto anyways) before pasting it on.

For the concerned, the heatsink is NOT touching anything with its corners, and I've filed the corners to the fullest extent possible.