SB-LINK cable. First time I've ever seen one pop up on ebay:

Lol, take one ordinary ATA cable and cut it into few smaller PC-LINK cables 😀

idspispopd wrote:

Voodoo3 3000 and Voodoo4 4500 have the same memory bandwidth, fill rate and triangle throughput.
My guess as to why the 4500 is faster at higher resolutions is that it has more RAM so less texture swapping is going on - the 3000 only has 16 MB and it doesn't really use any AGP features so texture uploads are slower on a TNT or TNT2.
See also this (possibly CPU limited with a PIII 850?) and this (not CPU limited for several benchmarks) test.

Don't forget, though, that even though the paper specs may seem the same, the V4 is doing it at a greater color depth than the V3. The V4 has to have more computing power than the V3 to pull that off. Memory alone isn't going to account for that.

Oh so the V4 DOES do 32bit colour?

Bagged myself a second Toshiba T3200, in order to replace the failed panel in my current one. I got the seller to remove and send just the panel to save on postage.

Also picked up a Texas Instruments CGA card, or at least that's what I'm 80% sure it is.

sliderider wrote:

Don't forget, though, that even though the paper specs may seem the same, the V4 is doing it at a greater color depth than the V3. The V4 has to have more computing power than the V3 to pull that off. Memory alone isn't going to account for that.

V4 can do 32bit color, that's right. But it doesn't do it at the same speed as 16bit color.
The benchmarks I linked to all mention if they were done at 16bit or 32bit color.
From the second set of benchmarks I linked to you can see that the V4 at 32bit color is running slower that a V3 143MHz at 16bit color.

Based on my benchmarks:
http://www.vogonswiki.com/index.php/3dfx_Benchmarks

A V4 4500 32bit colour in Quake 3 1024x768 is almost as quick as a V3 3000 in 16bit Colour.

All my testing has shown that the V4 is quicker than the V3 clock for clock.

Those tests were done years ago so I am planning to do them again and I am interested, but again Q3 seems to work well on the V4.

In bandwidth limited benchmarks (e.g 32bit) the V4 does come close to a Geforce 2 MX in certain tests (not a surprise as both have 166MHz ram).

Maybe it is just having 32mb of ram vs the 16mb on the V3

internally also the V3 does 32bit color, only the output is dithered 22bit color.
I am quite sure nobody will see a difference between 22bit and 32bit...

schlang wrote:

internally also the V3 does 32bit color, only the output is dithered 22bit color.
I am quite sure nobody will see a difference between 22bit and 32bit...

That doesn't make sense if you think about it. The reason Voodoo chip is built to process with a lower bitrate seems to be to improve performance and have a simpler chip.

You can actually adjust how high you want to take the color depth in the display properties of the V3, 16- or 22bit.

The V3 uses a hybrid 32bit color raster, 16bit z-buffer and 8bit stencil buffer. This means the card renders the picture in full 32bit color internally, but when it displays the image it must cull the colors to fit the z-buffer.

Even the GeForce culls the colors from 32- to 24bit colors.
TRUE 32bit would be 32bit color raster, 32bit z-buffer and 8- or 16bit stencil buffer.

not anything "special" but finally after an army of MK1s I just received a MK2 SC-55 , and it's excellent looking/working 😀

keropi wrote:

just received a MK2 SC-55 , and it's excellent looking/working :)

And nice and clean amiga 1200 :)

Is there real difference between sc-55 revisions?
As far I know, rev1 has a bug, that used by some dos games ost, so it's better to have old one?

bug? no MK1 does not have a bug, just the "capital tone fallback" feature that is absent in MK2s ... but MK2s have more polyphony/tones and sound a little different AFAIK .
If I was to have only one SC-55 I would chose a MK1 personally. But it's better to have both 🤣 🤣 🤣

Yes, i found post http://nerdlypleasures.blogspot.ru/2012/04/un … me-support.html - this is about "Capital Tone Fallback"

That's Roland for you 😀 Every unit is a little bit different 🤣

^ even MK1s aren't exactly the same, anyone remembers the DN3D breath/fl. key click difference? Interplay - sound driver hacking

Better have 3x SC-55s now 🤣 🤣 🤣

Got these yesterday:

Western Digital 43MB AT
Compaq SystemPro Processor board with 80386 @ 33MHz
Western Digital Winchester interface HDD Controller

Scored my 3th Asus P5A today (the first revision 1.06 for me, so no mobile k6+ support) and two voodoo3 2000 pci.

The collection keeps growing and growing 😎

A pair of JBL L20T bookshelf speakers. The seller said "consecutive serial number" or such, but I don't really care about such thing (or do I need to?). What's important is that a pair of JBL L20Ts - with JBL 035Ti titanium dome tweeters - would make a nice surround speakers for my JBL 120Ti mains. Titanium Dome (Gaming) Theater, anyone? 😀

One may argue that JBL 18Tis - with JBL 044Ti titanium dome tweeters - will make a better match for the 120Ti mains. However, JBL 044Ti tweeters are as rare as hen's teeth.I would be lucky if I could find a pair of spare 044Ti tweeters for my 120Tis. 035Tis, on the other hand, are easier to find. And should the worst happen, the newer 052Tis can perfectly replace the 035Tis. So for the sake of easier maintenance, I went for L20Ts instead of 18Tis.

Anyway, if anyone just happen to find a pair of JBL 044Ti tweeters, please let me know. 😀

Received a Diamond Stealth SIII S540 for my '99-00 Pentium III build. Not the quickest card I know and the drivers are poor - but I was interested in the S3 Metal drivers for Unreal.

Its boxed, although not in the best condition the card was brand new - never been used.

I've just had a play and although its not quick, I'm impressed. The textures looks fantastic in Unreal using the S3 Metal Driver and Unreal Tournament looks very clear, clearer/crisper than the Geforce 256 Pro I had in there.