VOGONS


What retro activity did you get up to today?

Topic actions

Reply 27820 of 29077, by darry

User metadata
Rank l33t++
Rank
l33t++
PcBytes wrote on 2024-06-19, 18:35:
darry wrote on 2024-06-19, 15:53:
PcBytes wrote on 2024-06-19, 13:54:

Well, that's something. I get 2.25V instead of 3.3v. No ISA slots present on the VP6.

EDIT: Yup, there's something going on with the 3v3 line of the mainboard - probing directly on the PSU gives 2.81V, and testing the PSU alone on the 3v3 rail gives 3.4v (since it has a 3v3 sense, these tend to run a tad higher on that line.)

Maybe something on the board is semi-shorted and bringing down the 3.3V rail, likely generating extraneous heat in the process ?

If so, a thermal imager, IR thermometer or a touch test might help find the culprit .

Touched every component I could for extra heat. Nothing heats up out of the ordinary.

It might be worth measuring the actual current draw on the 3.3V rail. If it's within the PSU's nominal capabilities, then maybe the PSU has trouble with regulation of that rail ?

Reply 27821 of 29077, by Dan386DX

User metadata
Rank Member
Rank
Member

Not something I did, but something I learned when I was researching high end XP era parts.

TIL that there are a few AGP graphics cards that require additional PCIe 6-pin power because they launched during the transition period into PCIe.

Most of you probably already know this, but I just thought it was interesting and wanted to share.

Also, I want an X1950.

90s PC: IBM 6x86 MX 233MHz. TNT2 M64. 256MB RAM, 2GB CompactFlash.
Boring modern PC: i7-12700, RX 7800XT. 32GB/1TB.
Fixer upper project: NEC Powermate 486SX/25. 16MB/400MB.

Reply 27822 of 29077, by PcBytes

User metadata
Rank l33t
Rank
l33t
darry wrote on 2024-06-20, 02:42:
PcBytes wrote on 2024-06-19, 18:35:
darry wrote on 2024-06-19, 15:53:

Maybe something on the board is semi-shorted and bringing down the 3.3V rail, likely generating extraneous heat in the process ?

If so, a thermal imager, IR thermometer or a touch test might help find the culprit .

Touched every component I could for extra heat. Nothing heats up out of the ordinary.

It might be worth measuring the actual current draw on the 3.3V rail. If it's within the PSU's nominal capabilities, then maybe the PSU has trouble with regulation of that rail ?

I went back to the square one with the 3v3 rail - inserting an AGP card brings the 3v3 line back to the standard spec.

"Enter at your own peril, past the bolted door..."
Main PC: i5 3470, GB B75M-D3H, 16GB RAM, 2x1TB
98SE : P3 650, Soyo SY-6BA+IV, 384MB RAM, 80GB

Reply 27823 of 29077, by bakemono

User metadata
Rank Oldbie
Rank
Oldbie

To try and solve the sudden issue of CHKDSK not running because "does not appear to be a Windows 2000 disk", I compared MBRs and boot sectors between the new HDD and the (partial) dump I made of the original HDD. The only difference I found was that bit 0, offset 0x41 in the boot sector had toggled. So I cleared the bit on the new HDD and restarted. Windows did not prompt to run CHKDSK. I right-clicked on the C: drive, went to error-checking and clicked yes to run the check at next boot. Restarted again. Prompted to run CHKDSK but still "does not appear to be a Windows 2000 disk". Haha. So I haven't solved the issue but I did discover that bit 0 offset 0x41 in the boot sector appears to be the flag that tells Windows to run CHKDSK.

GBAJAM 2024 submission on itch: https://90soft90.itch.io/wreckage

Reply 27824 of 29077, by GemCookie

User metadata
Rank Member
Rank
Member

I used Arch btw.

The Arch Linux 32 installation images are meant to be burned to a CD; I don't have any spare optical drives, so I chose to install the OS from a flash drive.
I eventually succeeded, though I had to run isohybrid, specify a custom root= parameter, write the image to an unused hard disk partition and... rebuild the initramfs image with IDE drivers for the ALi chipset. That last step was a load of fun.

I have yet to set up X due to an outage at my ISP. I'll see if I can play downloaded YouTube videos on this machine; I actually pulled this off on Windows XP with mplayer.

Gigabyte GA-8I915P Duo Pro | P4 520 | GF6600 | 2GiB | 256G SSD | DRDOS/XP/Vista/Arch/OBSD
MSI MS-5169 | K6-2/350 | TNT2M64 | 384MiB | 120G HDD | DR-DOS/MS-DOS/NT/2k/XP/OBSD
Dell Precision M6400 | C2D T9600 | FX2700M | 16GiB | 128G SSD | 2k/Vista/Arch/OBSD

Reply 27825 of 29077, by Thermalwrong

User metadata
Rank Oldbie
Rank
Oldbie

Today has been fun, I got a dead Toshiba Satellite 460CDX for less than a tenner which I'm not even sure covered postage. The seller knew it didn't work, they had stripped out the hard drive and half the screws were gone.
Through the course of today I've taken it from a dead Toshiba Satellite 460CDX that gaves codes 05 & 4C - which according to the Tecra 730CDT maintenance manual, 05H / 05Hex is "SM-RAM Check" and 4C is meaningless i.e. it stopped running meaningful code after reaching post code 05, so a RAM fault...

The attachment 460CDX-initial-POST-codes.JPG is no longer available

Back to a working Toshiba Satellite 460CDX: Re: Toshiba 460CDT - POST/Flash codes?
And hopefully that will help mattlacey and anyone else with a Toshiba Satellite 440CDX / 445CDX / 440CDT / 445CDT / 460CDX / 465CDX / 460CDT / 465CDT laptop, four little traces completely broke the computer.

Once it was working, I found out the original screen is not only garbage (DSTN) but also broken:

The attachment 460CDX-original-screen.JPG is no longer available

So for fun and because I've wanted to do this for a while now, I converted it from a 460CDX motherboard that drives that DSTN screen, into one that drives an active matrix TFT - it's a Toshiba Satellite 460CDT now

The attachment 460CDX-TFT.JPG is no longer available

That line? This is a faulty screen, I'm not gonna do weird electronic tests on a good screen 😁
The mod only requires moving 20-ish 0603 resistors from one spot to another, there's good clearance around them so it's pretty easy to do, gonna detail it in a new thread in a bit. It's much easier and less involved than this previous hackup: Upgrading a Toshiba laptop from DSTN to TFT panel - The Satellite 400CS becomes a 400CDT
(which I made look way worse than it is, you can easily make a 430CDS into a 430CDT by putting the video card + LCD from the 430CDT onto the 430CDS motherboard. Good if you've got a dead 430CDT but a working 430CDS)

Reply 27826 of 29077, by pjturpeau

User metadata
Rank Newbie
Rank
Newbie

Today I replaced the original P133 CPU of my Dell OptiPlex GS-5133L by a P200.

I think I'll replace the passive heatsink by one with a fan...

Reply 27827 of 29077, by Veeb0rg

User metadata
Rank Member
Rank
Member

Picked up this lovely little desktop from the local makerspace.

486dx33, 4meg ram, 165mb hdd. Had to tinker with the video card but it booted right up after that.

Reply 27828 of 29077, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t

Does thinking about a weird retro topic count as retro activity?

I was just reading about some different CPUs (because reasons) and I read that the A8-3870K was the last AMD CPU to support 3DNow! instructions. I had never considered that they stopped including these instructions at some point. That got me wondering if anyone has ever built an overly specific build focused on 3DNow! performance. A higher end Phenom II X4 or X6 would likely be decently faster than a 3870K, so if there were some old versions of games or programs that heavily made use of 3DNow! it'd be interesting to see how they ran on the highest performing 3DNow! processors versus a Sandy Bridge, Ivy Bridge or Haswell Core i5, since those are generally seen as the top chips for an overpowered Windows XP system.

Anyone have all the requisite hardware and software to do a test like this? The closest thing I have on hand is an AMD 760G chipset DDR2 board with an Athlon II X3 435 (I think). Not really on the same level as a DDR3 equipped board with a Phenom II X4 980 or X6 1100T...

Now for some blitting from the back buffer.

Reply 27829 of 29077, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Seemed like AMD stopped promoting 3DNow by the time multithreaded advantage was a reality in gaming, so I think you'd want to go for maximum single core speed. So Phenom X2 overclocked past 4Ghz or something. Hard to dig out a comprehensive game list though, here's some...
https://web.archive.org/web/20001109071400/ht … dnow/optimized/

Anyhoo, I'll put it in volume 5 of my todo list for when I've got the x6 1090 up and running, or try some on my x4 955 before that.

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 27830 of 29077, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
BitWrangler wrote on 2024-06-22, 03:02:

Seemed like AMD stopped promoting 3DNow by the time multithreaded advantage was a reality in gaming, so I think you'd want to go for maximum single core speed. So Phenom X2 overclocked past 4Ghz or something. Hard to dig out a comprehensive game list though, here's some...
https://web.archive.org/web/20001109071400/ht … dnow/optimized/

Anyhoo, I'll put it in volume 5 of my todo list for when I've got the x6 1090 up and running, or try some on my x4 955 before that.

Yeah, I was thinking the same thing. Any games (or patches) that would have made specific use of 3DNow! to improve performance would have been produced long before most games benefited from multicore processors.

Thankfully, single core performance was pretty solid for the Phenom II line, even if Sandy Bridge wiped the floor with it. I have no idea how much of an improvement 3DNow! offered overall, but it's interesting to think that there may be some specific pieces of (very old) software out there where a Phenom II is the fastest CPU you could get without having to use some much more modern OS + CPU combination (Windows 10 + Ryzen or Skylake for example).

Now for some blitting from the back buffer.

Reply 27831 of 29077, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Though I would also tend to suspect that the 3Dnow advantage wasn't so great that IPC gain in Intel over 3 or 4 generations from the last equivalent AMD CPU it was optimal on, would tend to wipe it out.

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 27832 of 29077, by Ozzuneoj

User metadata
Rank l33t
Rank
l33t
BitWrangler wrote on 2024-06-22, 04:44:

Though I would also tend to suspect that the 3Dnow advantage wasn't so great that IPC gain in Intel over 3 or 4 generations from the last equivalent AMD CPU it was optimal on, would tend to wipe it out.

Yeah, it is unlikely that there are many (or any) situations where the Phenom II would win over anything faster than an i5 2400, but some of the outlier benchmarks floating around the internet show them within a few percent, which is kind of crazy since others show even an i3 2100 destroying the best Phenom II. It is just something I'm curious about honestly.

Like, how would standard Q2 perform compared to the one with the 3DNow driver on a Phenom II? How would the results look on Intel CPUs?

https://www.anandtech.com/show/160/10

In those benchmarks a K6-2 333Mhz saw almost a 70% improvement.

Obviously a K6-2 relied on hit heavily because it was severely lacking in essential areas without it, but I do wonder how much of a boost a newer CPU would have since a lot of the newer features that modern CPUs benefit from isn't a factor at all (SSE, SSE2, etc.). It's possible the 3DNow focused stuff would actually hurt performance on newer chips since they have been designed to be fast without it.

Now for some blitting from the back buffer.

Reply 27833 of 29077, by newtmonkey

User metadata
Rank Oldbie
Rank
Oldbie

I received a CD-ROM drive today to replace the barely functioning drive in my dedicated DOS machine. It works great... but it's extremely noisy and also suffers from random pauses whenever it spins up/down.

However, ChrisR3tro's CDBeQuiet! tool (http://tools.dosforum.de/#cdbq) works great with this drive, and I was able to slow it down to 4X and resolve both issues. Very pleased!

Reply 27834 of 29077, by ssokolow

User metadata
Rank Member
Rank
Member
newtmonkey wrote on 2024-06-22, 07:45:

However, ChrisR3tro's CDBeQuiet! tool (http://tools.dosforum.de/#cdbq) works great with this drive, and I was able to slow it down to 4X and resolve both issues. Very pleased!

Nice. I was looking for an equivalent to eject -x SPEED on Linux.

I don't suppose you know of one for Mac OS 9?

Internet Archive: My Uploads
My Blog: Retrocomputing Resources
My Rose-Coloured-Glasses Builds

I also try to announce retro-relevant stuff on on Mastodon.

Reply 27835 of 29077, by rasz_pl

User metadata
Rank l33t
Rank
l33t
pjturpeau wrote on 2024-06-21, 21:58:

Today I replaced the original P133 CPU of my Dell OptiPlex GS-5133L by a P200.

I think I'll replace the passive heatsink by one with a fan...

https://turpeau.net/pierrejean/retro/dell-optiplex-gs5133l
That is a nice website, you should totally put a link in your signature. Looking at your pictures PSU is pushing air directly into the heatsink and out the back, so its not so much passive as optimized for the case used. Dell did this a lot all the way to Pentium 3 with success.

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 27836 of 29077, by ssokolow

User metadata
Rank Member
Rank
Member
rasz_pl wrote on 2024-06-22, 11:48:

That is a nice website, you should totally put a link in your signature.

*nod* Static site generators backed by Markdown really do seem to encourage that kind of simplicity of design but that one has a particular airy elegance to it.

I'm also working on a static site generator for my blog (which began as something I wrote as a single-page renderer when my brother wanted the family to start maintaining Christmas wishlists and mine had huge "Ask around/check thrift store" sections to be managed) but its current template is a lot more condensed.

(Yeah, I did get a bit nerd-sniped by the project. However did you know?)

Internet Archive: My Uploads
My Blog: Retrocomputing Resources
My Rose-Coloured-Glasses Builds

I also try to announce retro-relevant stuff on on Mastodon.

Reply 27837 of 29077, by pjturpeau

User metadata
Rank Newbie
Rank
Newbie
rasz_pl wrote on 2024-06-22, 11:48:

https://turpeau.net/pierrejean/retro/dell-optiplex-gs5133l
That is a nice website, you should totally put a link in your signature. Looking at your pictures PSU is pushing air directly into the heatsink and out the back, so its not so much passive as optimized for the case used. Dell did this a lot all the way to Pentium 3 with success.

Thanks, I try to keep the website as basic as possible...
About the heatsink, I saw the holes in the side of the PSU blowing to the heatsink. However, the P200 has a higher TDP and I prefer to take no risks...

Reply 27838 of 29077, by Dan386DX

User metadata
Rank Member
Rank
Member
Veeb0rg wrote on 2024-06-21, 23:02:

Picked up this lovely little desktop from the local makerspace.

486dx33, 4meg ram, 165mb hdd. Had to tinker with the video card but it booted right up after that.

Beauty. Would you share a full picture of the chassis? Love those chunky desktop designs from that era.

90s PC: IBM 6x86 MX 233MHz. TNT2 M64. 256MB RAM, 2GB CompactFlash.
Boring modern PC: i7-12700, RX 7800XT. 32GB/1TB.
Fixer upper project: NEC Powermate 486SX/25. 16MB/400MB.

Reply 27839 of 29077, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Because I picked up yet another Radeon 2600pro in a box of stuff last week, and got the 2400pro PCI earlier this year, I was doing some investigation of the performance envelope of these cards... ... ... they don't got one.

Well, they don't got one that seems to belong very well in a "pride of 2007" system, they seem a bit like a SiS6326 or a GF4 440MX, fine as long as you pitch it 2 years earlier.

Where you'd like to guess that maybe a 2600 was about 66% of a 2900, it's the other way, more like 33% at best.

A review on the early drivers, later ones didn't seem to improve things much https://www.overclockersclub.com/reviews/powe … _hd2600pro_512/

I am not even sure crossfiring my two 2600s would be worth it. Certainly I don't want to "waste" a board with two long slots when they can do much greater things. Possibly I might try a hacked up frankenstein with one long slot and a short one... since PCIe graphics doesn't really show a dent at 1x vs 16x until you get to about a HD4670 card.

The 2400 is gonna be worse of course, possibly I'll reserve it for doing something stupid like installing Vista on a Celeron 566.

TechPowerup in the GPU database is somewhat optimistic I feel about their performance level, setting the 2600 equal to HD5450 and HD6450, which seems wrong, those aren't powerhouses, but definitely give you low power elevation over onboard stuff and are capable of some playable framerates in older stuff. Other stuff I am finding here and there has 2600pro more around the level of an 8400GT, which of course was mildly warmed over for 9400 and GT210. TPU ranks them twice as fast as those, I'm not seeing anything to support that elsewhere. So I don't know where the 2400pro on PCI is for real, but gut adjustment based on poor 2600 scores makes me think GF 6600 to Radeon 9600Pro kinda area.

Though I don't know if I'm looking at the right stuff, is it possible they are redeemed by games that can really use 120 shaders, whereas the older stuff would expect only a third or less of that?? Though I am guessing that by the time those games were around, the shader count on everything else had got massive and still made them look horrible in comparison, even if contemporaries were struggling with less.

So I guess, more of a curiosity for representing ATI's first go round at DX10 but not very useful for it, I guess like GF5200 where you can run DX9 BUT.... So maybe a card for odd situations like you want up to 2006 gaming in XP only with a PCIe slot s754 or s775 or something, and you are avoiding the 05-06 cards in their high models that just burn up or have solder probs.... not entirely sure these are free of solder probs but at least they're cheap.... however if looking for something "particularly this fast" you're probably better with a 8400/9400/210 as they use less power and can be fanless/noiseless, or step up to the HD5450/6450 for probably better upward range though not a lot... I guess either are slightly more worth considering in AGP to get more top end out of AGP boards, but of course you can spend a lot more and get a lot more.

Anyway, bit of a shame that a card that has a number that is almost as 1337 as 1337 isn't up to much.

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