This has been a known phenomenon since the mid 90's when hardware accelerated audio became a thing in windows. The performance impact could be up to 30% depending on the game and hardware used.
I will say, your IDE and floppy cable routing was genius. If you intend to keep using that PC case in the future, which wouldn't be a terrible idea, because it seems to be a pretty lovely, if undercooled, case, then you may wish to either invest in a cheap fully modular PSU, or if you want to make a video about modifying PSUs, you could create your own custom length cables to provide yourself adequate cable management slack. If you want the most pliable cables possible, I suggest finding silicone insulated wire, as it is more compliant than the typical plastic insulated wire, though it is considerably more expensive. I would personally love to see a series of videos about customizing that case for increased longevity. Modding it to include additional cooling, repainting it to eliminate the peculiar text on the front panel, maybe even cutting the side/top panel to add in an acrylic window so you can see all the lovely retro tech hiding inside. I'm not sure about the availability of IDE ribbon cables and their connectors, but if you can find some, it could also be fun to have a short video about building custom-length IDE cables. From my understanding of how they're assembled, it uses punch-through connectors to puncture through the insulation and grip the wires, so the actual connector fixing should be pretty easy. I spent so much time back in the day figuring out how to delicately fold and tuck IDE and PATA ribbon cables and colorful fixed power supply cables, so content like this is really nostalgic for me.
Very nice channel. Also liked the part at the end, used to have a Sempron back in the day, never knew the integrated sound card was probably consuming a lot of CPU resources.
I was in college from 99-03 and this video brought me back to my trusty AMD machines of that time. I really enjoyed the 386 and 486 videos, but this one really got me feeling sentimental. Thanks!
I liked this video in particular because I am trying to build a mid-late 90s semi-retro gaming PC for myself. I think I'd be willing to go to early PCIe instead of AGP, but some of what I learned from this video will be useful! Your 2/3/486-era stuff is also cool!
My main aim is actually on 2/3/486-era. I have much more fun tinkering with that oldies. I built this PC just to have one AGP system to test some of my AGP graphics cards. Glad you liked it.
BTW thanks for that colaboration with Phil, very interesting that you can take 12V from the ISA Bus and transform it to -5V and inject it on the bus like that. As sweet as it is I keep wondering if integrating this on an XT-IDE ISA CF adapter would be as desirable as I imagine. 😁
Thank you, well it depends on how many people find it desirable as well. At least, XT-IDE doesn't need that and I find it very confusing to find circuits on some parts, which are not required. Furthermore, XT-IDE adapter is something, what you want to use in different systems, which probably already have -5V.
@@necro_ware ohh so it seems like it's not such a great idea hahaha... Do you know about this project? x86.fr/category/electronics/atx2at-smart-converter/
I liked the clean assembly with the Flat IDE cable. I loved the repair of the video card cooler's interior, I had never seen this type of repair and the replacement parts, I think it never arrived here in Brazil. Usually when a cooler has a problem we replace the entire part.
Yeah that weird Via southbridge is a pain. The A revision is a bit more compatible (supports SATA 2 without messing around with jumpers on the drive) but still doesn't work with SATA 3 SSDs. I wound up settling on a NOS Sandisk SATA 1 drive (32 GB I think) for the boot drive and a regular 500 GB 7200 RPM SATA 3 HDD jumpered to SATA 1 speeds. By the way, most of the AGP X700 and X800 series GPUs were native AGP and didn't need a bridge chip. If you wind up with weird issues with the X1650 it might be a good idea to hut one down. You can occasionally snag a deal on one. A couple of X800 models (I think the X850GT was one) did have a bridge chip, however, so make sure you get a good look at the back of the PCB before purchasing.
Nice high end early to mid 2000s setup, well, high-ish. I was running a P4b Northwood 2.4 @3.83ghz paired with a Geforce 4 MX440 in 2003 and was drooling over a Radeon 9600x that I never did manage to save up enough for. I was running 1gb of DDR 400 high end RAM though, probably the most expensive part of my setup. In 2004/5 I upgraded to a Pentium D dual core and a Radeon 1600x. Its funny how I can track where I was in my life by the PC I was running at the time :)
From what I remember from those times, motherboards with those VIA chipsets where VERY, VERY picky about the order on which you installed the drivers. If you messed up the "right" order, you ended up with things like the network or sound card not working. IIRC, on a clean XP install... you HAVE to install the VIA chipset drivers before anything else (where those called Hyperion "4 in 1" drivers?). Then network, then sound card, then video drivers. Basically, in order of "complexity" or "size". Don't forget the usual reboot between each driver, just in case, of course :-D
I vaguely remember having a VIA network card that had the exact same behaviour. Even though it connected at 100Mbit, I had to actually explicitly select 100Mbit (instead of auto-negotiate) in the hardware settings to make it work. Worth a try, but I guess you've already tried that.
For the network card, what netstat -i reports in windows ? The packets end up in errors or being dropped ? Have you tried to set the speed to 10 instead of 100 ? Also not sure but have you checked the autocross thing ?
I'm curious how you ran that Mobile Sempron CPU on your board. I have the K8V-MX version of this board and a K8U-X, both of which have absolutely no way to control CPU voltage (thus OCing them to stock desktop voltages). I ran my Mobile Sempron in a Gigabyte board (K8NE) which has CPU voltage settings.
@@necro_ware this is pretty much what it did for me as well on the Asus boards, 1.4 volts. However, I was able to achieve 1.2 volts (the spec for those SKUs) on the Gigabyte board; this allowed me to cool it passively. www.cpu-world.com/CPUs/K8/AMD-Mobile%20Sempron%203300%2B%20-%20SMS3300BQX2LF.html
@@computerguy096 I also think, that it should be possible. But just as you told, this mainboard has no way to setup the voltage manually. It detects 1.4V for the mobile version and you have to live with that. Well, better than nothing, at least it detects the CPU properly.
@@necro_ware they are probably desktop Semprons repurposed for laptops, and have no issues running at 1.4. Anyways, I've had even worse luck with the VT8237R southbridge, most of the boards that I have with this chip do not detect my SATA HDDs at all, especially the older Socket 462 ones. Not even the very early Western Digital drives with molex and sata power on the same PCB. Quite peculiar.
I think the problem with 3dmark 99 is indeed related to Vsync. When you look at the "racing" demo, you can see the FPS is remaining at 60fps constantly. Check in nvidia's control panel to check if vsync isn't enabled. The S3 unichrome is behaving pretty much like a 90's graphic card ... which is kinda what it is, since I think it's the last chip S3 ever made before being bought by VIA integrated into a chipset. I have a similar one in my Windows 98 core 2 duo, and when running Unreal tournament, the software mode was faster than the hardware mode with that thing x) By the way, do you have any tips to fix these old fans ? I see you fixing them all the time, but I'm not super confident at doing that. How did you find the right bearings ? For the network driver issue, I guess you've tried the drivers from ASUS' website ?
This S3 Unichrome in my video is already the integrated version into the VIA chipset. I have to say, I found it quite good for a built-in chip. At least as a DX7 graphics card it makes a good figure IMHO. Not very fast, but sufficient and stable. In regards of the fans, I just measured the old bearing, went to Aliexpress and found a set of the same size. That wasn't actually too complicated, so no magic or secret tips there :) And I also tried Asus network drivers. There are tons of reports from 10 years ago, where people said, that their VIA Rhine cards stopped working one day. They all replaced it through another network adapter and blamed the hardware. As my showcase with Linux confirms, it is obviously just a software problem. I guess something in Windows was updated, what rendered the VIA network drivers nonfunctional one day. Probably, this was never fixed.
@@necro_ware yes I know it's integrated ^^ It's just that mine uses a different driver. I just thought of something that sounds stupid (but considering the nature of this problem, who knows ?). Have you tried to change the date of the os to something close to the motherboard's release date ? Also I guess you've tried each version of Windows XP, or have you tried only a couple of service packs ?
Well, I didn't try another date. I think, I could try, why not?! :D And, yes, I tried different versions of Windows XP - Home, Pro, German, US, SP2, SP3. I didn't try one without SP though....
@@necro_ware since this chipset is from late 2003, and that windows XP service pack 2 is from mid 2004 it means this was tested with windows XP service pack 1 at most. So maybe windows XP sp1 would work ? When the oldest posts about this issue have been written ?
@@DxDeksor Yes, may be. All the reports were made with SP2, as far as I remember. I guess, the oldest one was from 2007. May be I'll try SP1 later, just to bisect the issue. However, I wouldn't like to use SP1 anyway, there would be probably another problems with Graphics Card, DirectX, 4in1 or whatever :)
My third PC had a VIA Rhine II (or III) and it failed on me after a while. Tho my machine used a VIA P4M266A chipset (for an Intel socket 478). Even tho I would've loved to have the board you got, I can feel I would've become annoyed with these issues.
Many people reported 10 years ago, that this network card stopped working one day on Windows. Everybody assumed, that the hardware was defect. As you can see, the hardware is absolutely ok, since it works in Linux just fine. It must be some driver issue on windows.
Hi necroware, i have a Nvidia geforce 8400gs, with passive cooling, i upgraded it to a fan cooling and well it still runs pretty hot, at about 60 celsius in idle and 65 in benchmark, i already changed themal paste with Arctic MX-4
Thank you! Just don't want to call it old, so let's call it semi retro then ;) I usually tinker with much older computers, than this one. So this video is quite unusual for my channel, but I just found this build quite interesting, because I had to handle so many, but obviously common problems. I thought, this could be interesting for the one or another as well.
@@necro_ware regarding your network issues, did you try to set the speed to 10 or 100 on both the card and on the switch instead of having auto negotiation on?
@@KJohansson Yes, I tried that as well. Unfortunately with no success. As I saw on the internet, there are tons of reported issues by users, where this card stopped working from one day to another about 10 years ago. I guess, something happened in Windows, what broke the VIA drivers, which were never repaired, as far as I can see. Almost all reporters switched to another network adapters and blamed VIA Rhine for being broken. As you could see in Linux, the hardware is absolutely ok.
When I used to work in a Computer shop around the mid to late 2000's Via chip sets caused all kinds of nightmares they used to be good but the Duron/Athon onwards Via south bridges were problematic with windows. Like yourself i have been using Linux since the 90's but of late I have been using Windows 10 with WSL2 Windows Subsystem for Linux something I am unsure you may of toyed with it is interesting what Microsoft have done it is a bit limitred but you may already know. I am a Slackware fan and I am about to do a frees build of Linux from Scratch again as I enjoy building my own OS from scratch and often mod the source code along the way but none the less another great video but I suggest stick with Linux with the via chip set at least the Linux kernel the drivers just work out of the box on retro machines Linux rocks.
Been experimenting with a Socket 754 build too recently, but I wanted to build an unnecessarily powerful Windows 98SE gaming PC on a budget, so I wanted to use Socket 754, an Athlon 3400+ and a Geforce 6600GT, but I'm having some stability problems. Tried both a VIA chipset mobo like yours, and an Asrock Sis chip board, but still no luck with Windows 98. It can be rock solid for hours, and next time Windows won't even boot and will show Windows Protection Error.
I had a lot of problems especially with the network card. I almost went crazy trying to get it running in Windows XP. Well, at least luckily the system ran stable....
Via Rhine on those asus K8 something, the Atheros on the Asus P5GC something motherboards... the integrated ethernet always acts funny after sometime, I always had in hand a pci network card in hand when working with these
@@necro_ware very nice glitch ! I love this! look at glitch I came across years ago on a laptop it was RED from the beginning (out of the box for some reason) and then I restarted the settings to the default and it become BLUE as it shoud be at least on all other AMI SETUPs I have seen and never came back to the r, fortunately I made the recording before trying that :D th-cam.com/video/ik99xjfubhM/w-d-xo.html
Regarding LAN, did you try various chipset drivers for this VIA (especially latest VIA Hyperion)? Since you are using external driver manager, that may be the issue. I remeber having trouble with LAN back in a day that in VIA based mobos that was solved by installing those 4in1 VIA packs.
Yeah, as I told in the video, I tried 20, may be 30 different network and 4in1 drivers combinations. Also windows XP SP3 has a built in driver for that network card. That one didn't work as well.
One downside to using an IDE cable like that in a non standard orientation is that cable select will no longer work for drives. Cable select is done by cutting a pin on the cable between the master and slave drives, so relies on the cable orientation. I guess if you put the middle one on the motherboard it'd still work. Though of course just setting the jumpers on the drive isn't much of a hassle
I never rely on the cable select feature anyway. Not back then and not today, since most of the cables I had in my hands didn't support that, that's why I just always prefer to set the jumpers to master/slave ignoring CS altogether.
I remember having this exact same issue with the same VIA lan chip. The solution, I kid you not, was using the driver prodvided on the CD of the motherboard. If I used the latest one from the VIA website or the one from the ASUS site, the thing would not work. If you have a driver CD of a motherboard with such lan chip, I suggest you try that driver.
@@PJBonoVox You probably mean the notebook videos? Well, what shall I say, I never thought about that..... Just watched once again, at least you can see how I put thermal paste onto the CPU :D
oh, I forgot to mention in the video, that I analyzed the traffic with the wireshark. I can see outgoing packets from the machine, but incoming packets seem to be ignored.
Ich hatte ein ähnliches Problem mit meiner Netzwerkkarte auch mal. Bei mir schien diese keine gültige Mac-Adresse mehr zu haben. In den Eigenschaften der Netzwerkkarte (Geräte-Manager) konnte ich unter "Physikalische Adresse" manuell eine Andere vergeben. Da ich die Ursprünglicher noch wusste, nahm ich die. Ich weiß nicht, ob es auch mit einer "generierten" Mac-Adresse klappt.
Vielen Dank für die Idee! Mein Adapter hatte eine gültige MAC Adresse, doch ich habe auch schon versucht eine andere einzutragen, aber leider wollte das Teil auch nicht. Wie gesagt, unter Linux alles kein Problem, einfach hochfahren und es geht direkt ohne Weiteres. Ich befürchte, dass in den VIA Treibern für Windows irgendwo ein Bug drin ist, was den Netzwerkadapter dort unbenutzbar macht. Ich frage mich nur, wie es damals niemandem aufgefallen ist?! Wenn man im Netz danach sucht, findet man hunderte berichte, wo die Leute sich über eine kaputte VIA Netzwerkkarte beschweren und überall ist die Lösung die selbe - eine andere Karte nehmen. Und, dass eine andere Karte geht wurde überall als Beweis wahrgenommen, dass der VIA Adapter kaputt sei. Doch unter Linux zeigt sich eindeutig, dass es nicht so ist.
Because I wanted a better compatibility to the old stuff in Windows XP. There were issues on 64-bit Windows XP back in the days. Linux on the other hand, which I'm using on that machine as well is 64-bit.
I know the rest, yes. I bought this power supply 10 years ago and since that time it was not used very much. I didin't have any issues regarding this PSU. I opened it once though, around 2 years ago, it was fine.
the SATA 2 (or 3) Drives on SATA1 Controllers is NOT a VIA Problem, its a General Problem with all SATA1 devices I have tested. VIA VT8237, nForce 2. SIL 3112 might be the same thing, dont remember I sadly don't have an Intel to test but I doubt it works better... That Issue was fixed with later iterations of the VIA VT8237. I thin it was the S-Version (VT8237S). PRO TIP: STAY AWAAY from the VT8251!!
Well, it's a known bug of VIA. You can even read about it on Wikipedia. I have no issues with other chipsets here. Using the SSD, which I showed in the video, with almost everything. Runs everywhere with sata1, but at least it runs. VIA is just crap :D Unfortunately for this CPU it was the most used chipset and almost the only one you can get today.
@@necro_ware Its not, as said it also is in other Chipsets at the time. The mentioned Nforce 2 for example. The MCP-S has the same issue and some SATA Controller too. So for it to work you need a SATA2 PCI Card. VIA on a decent quality board is better than whats said. And there are also some S754 Boards without VIA CHip, for example the ASUS K8S-MX. And with S939 you find more nForce Chipsets - which are even worse than VIA!
@@Stefan_Payne As far as I remember, Nforce2 was used with Socket 462 and it had no SATA at all. I might be wrong in this regards. In regards of VIA with S754, as I told it is ALMOST the one chipset you can get, at least with AGP. I have another board with PCIe and Nforce3, there SATA works just fine, but I wanted a system with AGP to make some tests. Something other, than VIA is really not easy to find. And in regards of SAT in VT8237, it is really a well known and documented bug. The guys grounded one "reserved" connection of the SATA controller in the southbridge. This connection had to be left floating and was used in SATA-II for extended initialization. Since it was grounded, the SATA-II couldn't initialize properly anymore. What shall I say? RTFM, especially, when you design a chipset :) Please pay attention, that I never told, that other chipsets didn't have other or similar problems, but that doesn't make VIA any better, especially with this one and with the famous data corruption bug. External SATA-I Adapters from that time for the PCI slot do work just fine with all my SATA-II and SATA-III drives. I'm using them actually quite often. But the whole I/O in the VIA chipsets from that time was an absolute disaster. I just briefly googled for detailed bug evaluation, but they are all in German. In English there is just mentioning, but no detailed description. Funny enough, the bug is mentioned as it is only on German Wikipedia, but not on the English.
@@necro_ware There were later Revisions of the nForce 2 that came with SATA; as there was a new Southbridge introduced: the MCP-S with SATA 1. Also has Issues with SATA2 as a lot of other things at the time. nForce 3 was AGP too, though had issues with nV40, the PCIe ones are the nForce 4 ones. There was also a 2nd Revision of the VT8237 which had the SATA2 Issues fixed: VT8237*S* Also its not a Bug of the VT8237 if you find the issue elsewhere and you do! THe nforce is one, I think the Intel ICH5 could have the issues too as a lot of other SATA1 Controllers. Just test it yourself, if you don't believe me. You find it on others! Its just the way it is with early specified components where there is only one speed and not multiple like later. And if you google my name, you'll find that I was very active in German Forums when the VIA 8237 was in use. As sad, there is a later Revision of the SOuthbridge that has the SATA2 compatibility fixed - or even SATA2.
I tried that. There is obviously a bug in the drivers, but nobody will fix an old driver for windows xp. As I mentioned in Linux everything works out of the box, so it's definitely a software issue.
oh and since its an AMD K8 based system, you might want to test Registred ECC RAM. Especially with an ASUS Board (Gigabyte works too, ASROCK no idea and MSI does not). That sounds weird but I have 2x2GB Infineon Registred ECC in my ASUS A8R-MVP!
@@necro_ware I didn't mean ECC, I meant buffered Server Memory. HYS72D256220GBR is the model number. And some AMD K8 CPUs with some Boards work with it, others do not. I think I used one of the two Registred ECC Sticks I had in a Sockel A Board at the time (with VIA CHip)...
As far as I know, there was no (or not many?) nvidia based mainboard with AGP and Socket 754. I needed this combination for another reasons. Unfortunately most mainboards from that time were already PCIe and I have two such mainboards here. But I needed it with AGP.
@@necro_ware This is not true, practically every manufacturer made at least 1-2 boards ATX/MATX if not more. I think it's more of an issue of you running into the fact that they're harder to find now and more expensive then these crappy VIA boards.
@@gravitone That's why I wrote "not many?". I also assume, that there are probably some rare ones. But all I could find were VIA and some SiS based. I had this VIA in my parts, so....
Where are you located that they're so hard to source? I typed "nforce3" into ebay.com, and the screen was instantly flooded with 754 AGP boards. Just because you already have a VIA board, doesn't make it a good idea to actually use it, since they're always terrible.
@@gravitone yeah, I'm desperately sorry, that I did that terrible mistake. There is probably nothing I could do to fix this again.... I think, I'll drain my depression in a glass of alcohol :D
I particularly like the under the motherboard with the IDE cables trick!
It was the only way to have tidy cables back in the day, cases were not built with cable management in mind.
I just found this channel tonight, and it's incredible. What a wonderful information hub for retro hardware!
Thank you very much. Yeah, unfortunately it gets more and more complicated to reach the community through YT.
the performance difference at the end with the soundcard surprised me no doubt, interesting stuff
This has been a known phenomenon since the mid 90's when hardware accelerated audio became a thing in windows. The performance impact could be up to 30% depending on the game and hardware used.
I will say, your IDE and floppy cable routing was genius. If you intend to keep using that PC case in the future, which wouldn't be a terrible idea, because it seems to be a pretty lovely, if undercooled, case, then you may wish to either invest in a cheap fully modular PSU, or if you want to make a video about modifying PSUs, you could create your own custom length cables to provide yourself adequate cable management slack. If you want the most pliable cables possible, I suggest finding silicone insulated wire, as it is more compliant than the typical plastic insulated wire, though it is considerably more expensive. I would personally love to see a series of videos about customizing that case for increased longevity. Modding it to include additional cooling, repainting it to eliminate the peculiar text on the front panel, maybe even cutting the side/top panel to add in an acrylic window so you can see all the lovely retro tech hiding inside. I'm not sure about the availability of IDE ribbon cables and their connectors, but if you can find some, it could also be fun to have a short video about building custom-length IDE cables. From my understanding of how they're assembled, it uses punch-through connectors to puncture through the insulation and grip the wires, so the actual connector fixing should be pretty easy.
I spent so much time back in the day figuring out how to delicately fold and tuck IDE and PATA ribbon cables and colorful fixed power supply cables, so content like this is really nostalgic for me.
I've been building, fixing and working with computers for almost 25 years and I just learned so much in that 20 minutes.
Thank you.
Very nice channel. Also liked the part at the end, used to have a Sempron back in the day, never knew the integrated sound card was probably consuming a lot of CPU resources.
I was in college from 99-03 and this video brought me back to my trusty AMD machines of that time. I really enjoyed the 386 and 486 videos, but this one really got me feeling sentimental. Thanks!
Great content! You should be getting 100x more views and subs
I liked this video in particular because I am trying to build a mid-late 90s semi-retro gaming PC for myself. I think I'd be willing to go to early PCIe instead of AGP, but some of what I learned from this video will be useful! Your 2/3/486-era stuff is also cool!
My main aim is actually on 2/3/486-era. I have much more fun tinkering with that oldies. I built this PC just to have one AGP system to test some of my AGP graphics cards. Glad you liked it.
BTW thanks for that colaboration with Phil, very interesting that you can take 12V from the ISA Bus and transform it to -5V and inject it on the bus like that. As sweet as it is I keep wondering if integrating this on an XT-IDE ISA CF adapter would be as desirable as I imagine. 😁
Thank you, well it depends on how many people find it desirable as well. At least, XT-IDE doesn't need that and I find it very confusing to find circuits on some parts, which are not required. Furthermore, XT-IDE adapter is something, what you want to use in different systems, which probably already have -5V.
@@necro_ware ohh so it seems like it's not such a great idea hahaha... Do you know about this project?
x86.fr/category/electronics/atx2at-smart-converter/
I liked the clean assembly with the Flat IDE cable.
I loved the repair of the video card cooler's interior, I had never seen this type of repair and the replacement parts, I think it never arrived here in Brazil. Usually when a cooler has a problem we replace the entire part.
Yeah that weird Via southbridge is a pain. The A revision is a bit more compatible (supports SATA 2 without messing around with jumpers on the drive) but still doesn't work with SATA 3 SSDs. I wound up settling on a NOS Sandisk SATA 1 drive (32 GB I think) for the boot drive and a regular 500 GB 7200 RPM SATA 3 HDD jumpered to SATA 1 speeds. By the way, most of the AGP X700 and X800 series GPUs were native AGP and didn't need a bridge chip. If you wind up with weird issues with the X1650 it might be a good idea to hut one down. You can occasionally snag a deal on one. A couple of X800 models (I think the X850GT was one) did have a bridge chip, however, so make sure you get a good look at the back of the PCB before purchasing.
Nice high end early to mid 2000s setup, well, high-ish. I was running a P4b Northwood 2.4 @3.83ghz paired with a Geforce 4 MX440 in 2003 and was drooling over a Radeon 9600x that I never did manage to save up enough for. I was running 1gb of DDR 400 high end RAM though, probably the most expensive part of my setup. In 2004/5 I upgraded to a Pentium D dual core and a Radeon 1600x. Its funny how I can track where I was in my life by the PC I was running at the time :)
That's funny indeed, but we are all broken in the same way :D
From what I remember from those times, motherboards with those VIA chipsets where VERY, VERY picky about the order on which you installed the drivers. If you messed up the "right" order, you ended up with things like the network or sound card not working.
IIRC, on a clean XP install... you HAVE to install the VIA chipset drivers before anything else (where those called Hyperion "4 in 1" drivers?). Then network, then sound card, then video drivers. Basically, in order of "complexity" or "size". Don't forget the usual reboot between each driver, just in case, of course :-D
I vaguely remember having a VIA network card that had the exact same behaviour. Even though it connected at 100Mbit, I had to actually explicitly select 100Mbit (instead of auto-negotiate) in the hardware settings to make it work. Worth a try, but I guess you've already tried that.
Yes, I tried that, but had no luck with that too....
@@necro_ware Yeah, I figured you'd have tried everything. I guess sometimes you have to say "oh well" and move on.
Nice build! I would have installed an intel gigabit network adapter to improve the speeds. Or would it not matter?
For the network card, what netstat -i reports in windows ? The packets end up in errors or being dropped ?
Have you tried to set the speed to 10 instead of 100 ?
Also not sure but have you checked the autocross thing ?
I'm curious how you ran that Mobile Sempron CPU on your board. I have the K8V-MX version of this board and a K8U-X, both of which have absolutely no way to control CPU voltage (thus OCing them to stock desktop voltages). I ran my Mobile Sempron in a Gigabyte board (K8NE) which has CPU voltage settings.
Hi, it automatically detected and dropped the voltage down to 1.4V The mainboard itself has no way to set it up manually.
@@necro_ware this is pretty much what it did for me as well on the Asus boards, 1.4 volts. However, I was able to achieve 1.2 volts (the spec for those SKUs) on the Gigabyte board; this allowed me to cool it passively. www.cpu-world.com/CPUs/K8/AMD-Mobile%20Sempron%203300%2B%20-%20SMS3300BQX2LF.html
@@computerguy096 I also think, that it should be possible. But just as you told, this mainboard has no way to setup the voltage manually. It detects 1.4V for the mobile version and you have to live with that. Well, better than nothing, at least it detects the CPU properly.
@@necro_ware they are probably desktop Semprons repurposed for laptops, and have no issues running at 1.4.
Anyways, I've had even worse luck with the VT8237R southbridge, most of the boards that I have with this chip do not detect my SATA HDDs at all, especially the older Socket 462 ones. Not even the very early Western Digital drives with molex and sata power on the same PCB. Quite peculiar.
@@computerguy096 Yeah, very poor job by VIA....
I think the problem with 3dmark 99 is indeed related to Vsync. When you look at the "racing" demo, you can see the FPS is remaining at 60fps constantly. Check in nvidia's control panel to check if vsync isn't enabled.
The S3 unichrome is behaving pretty much like a 90's graphic card ... which is kinda what it is, since I think it's the last chip S3 ever made before being bought by VIA integrated into a chipset.
I have a similar one in my Windows 98 core 2 duo, and when running Unreal tournament, the software mode was faster than the hardware mode with that thing x)
By the way, do you have any tips to fix these old fans ? I see you fixing them all the time, but I'm not super confident at doing that. How did you find the right bearings ?
For the network driver issue, I guess you've tried the drivers from ASUS' website ?
This S3 Unichrome in my video is already the integrated version into the VIA chipset. I have to say, I found it quite good for a built-in chip. At least as a DX7 graphics card it makes a good figure IMHO. Not very fast, but sufficient and stable.
In regards of the fans, I just measured the old bearing, went to Aliexpress and found a set of the same size. That wasn't actually too complicated, so no magic or secret tips there :)
And I also tried Asus network drivers. There are tons of reports from 10 years ago, where people said, that their VIA Rhine cards stopped working one day. They all replaced it through another network adapter and blamed the hardware. As my showcase with Linux confirms, it is obviously just a software problem. I guess something in Windows was updated, what rendered the VIA network drivers nonfunctional one day. Probably, this was never fixed.
@@necro_ware yes I know it's integrated ^^ It's just that mine uses a different driver.
I just thought of something that sounds stupid (but considering the nature of this problem, who knows ?). Have you tried to change the date of the os to something close to the motherboard's release date ?
Also I guess you've tried each version of Windows XP, or have you tried only a couple of service packs ?
Well, I didn't try another date. I think, I could try, why not?! :D
And, yes, I tried different versions of Windows XP - Home, Pro, German, US, SP2, SP3. I didn't try one without SP though....
@@necro_ware since this chipset is from late 2003, and that windows XP service pack 2 is from mid 2004 it means this was tested with windows XP service pack 1 at most. So maybe windows XP sp1 would work ?
When the oldest posts about this issue have been written ?
@@DxDeksor Yes, may be. All the reports were made with SP2, as far as I remember. I guess, the oldest one was from 2007. May be I'll try SP1 later, just to bisect the issue. However, I wouldn't like to use SP1 anyway, there would be probably another problems with Graphics Card, DirectX, 4in1 or whatever :)
Yeeey a Necroware video!
On 3dmark99 you need to disable vsync or the score will be limited by your monitor refresh rate.
Ok, good to know, thank you! I thought, that it would be the problem, but I didn't realize, that I have to deactivate it in the program itself.
My third PC had a VIA Rhine II (or III) and it failed on me after a while. Tho my machine used a VIA P4M266A chipset (for an Intel socket 478). Even tho I would've loved to have the board you got, I can feel I would've become annoyed with these issues.
Many people reported 10 years ago, that this network card stopped working one day on Windows. Everybody assumed, that the hardware was defect. As you can see, the hardware is absolutely ok, since it works in Linux just fine. It must be some driver issue on windows.
Hi necroware, i have a Nvidia geforce 8400gs, with passive cooling, i upgraded it to a fan cooling and well it still runs pretty hot, at about 60 celsius in idle and 65 in benchmark, i already changed themal paste with Arctic MX-4
Retro-PC with "easy to get components" are a bit of a contradiction. Great video! :)
Thank you! Just don't want to call it old, so let's call it semi retro then ;) I usually tinker with much older computers, than this one. So this video is quite unusual for my channel, but I just found this build quite interesting, because I had to handle so many, but obviously common problems. I thought, this could be interesting for the one or another as well.
@@necro_ware regarding your network issues, did you try to set the speed to 10 or 100 on both the card and on the switch instead of having auto negotiation on?
@@KJohansson Yes, I tried that as well. Unfortunately with no success. As I saw on the internet, there are tons of reported issues by users, where this card stopped working from one day to another about 10 years ago. I guess, something happened in Windows, what broke the VIA drivers, which were never repaired, as far as I can see. Almost all reporters switched to another network adapters and blamed VIA Rhine for being broken. As you could see in Linux, the hardware is absolutely ok.
When I used to work in a Computer shop around the mid to late 2000's Via chip sets caused all kinds of nightmares they used to be good but the Duron/Athon onwards Via south bridges were problematic with windows. Like yourself i have been using Linux since the 90's but of late I have been using Windows 10 with WSL2 Windows Subsystem for Linux something I am unsure you may of toyed with it is interesting what Microsoft have done it is a bit limitred but you may already know. I am a Slackware fan and I am about to do a frees build of Linux from Scratch again as I enjoy building my own OS from scratch and often mod the source code along the way but none the less another great video but I suggest stick with Linux with the via chip set at least the Linux kernel the drivers just work out of the box on retro machines Linux rocks.
Been experimenting with a Socket 754 build too recently, but I wanted to build an unnecessarily powerful Windows 98SE gaming PC on a budget, so I wanted to use Socket 754, an Athlon 3400+ and a Geforce 6600GT, but I'm having some stability problems. Tried both a VIA chipset mobo like yours, and an Asrock Sis chip board, but still no luck with Windows 98. It can be rock solid for hours, and next time Windows won't even boot and will show Windows Protection Error.
I had a lot of problems especially with the network card. I almost went crazy trying to get it running in Windows XP. Well, at least luckily the system ran stable....
Protection error probably caused by too much RAM for 98.
@@LBXComputers Nope, tried with 128, 256 and 512MB, same problem.
Via Rhine on those asus K8 something, the Atheros on the Asus P5GC something motherboards... the integrated ethernet always acts funny after sometime, I always had in hand a pci network card in hand when working with these
WoW how did you get the rainbow colors on the top in the BIOS ? 16:36
This happens only with the X1650 card. Some kind of glitch :)
@@necro_ware very nice glitch ! I love this! look at glitch I came across years ago on a laptop it was RED from the beginning (out of the box for some reason) and then I restarted the settings to the default and it become BLUE as it shoud be at least on all other AMI SETUPs I have seen and never came back to the r, fortunately I made the recording before trying that :D th-cam.com/video/ik99xjfubhM/w-d-xo.html
I can send drivers for VIA. Haven't seen any problems with them in Win98-XP.
Regarding LAN, did you try various chipset drivers for this VIA (especially latest VIA Hyperion)? Since you are using external driver manager, that may be the issue. I remeber having trouble with LAN back in a day that in VIA based mobos that was solved by installing those 4in1 VIA packs.
Yeah, as I told in the video, I tried 20, may be 30 different network and 4in1 drivers combinations. Also windows XP SP3 has a built in driver for that network card. That one didn't work as well.
One downside to using an IDE cable like that in a non standard orientation is that cable select will no longer work for drives. Cable select is done by cutting a pin on the cable between the master and slave drives, so relies on the cable orientation. I guess if you put the middle one on the motherboard it'd still work. Though of course just setting the jumpers on the drive isn't much of a hassle
I never rely on the cable select feature anyway. Not back then and not today, since most of the cables I had in my hands didn't support that, that's why I just always prefer to set the jumpers to master/slave ignoring CS altogether.
I remember having this exact same issue with the same VIA lan chip.
The solution, I kid you not, was using the driver prodvided on the CD of the motherboard.
If I used the latest one from the VIA website or the one from the ASUS site, the thing would not work.
If you have a driver CD of a motherboard with such lan chip, I suggest you try that driver.
I don't have the original CD, but I tried about 20 different drivers with no luck.
@@necro_ware Yeah, I know the pain of not having the CD's. I'd wish I still had mine so I can share with you the driver.
@@MerolaC It's ok, meanwhile I found a board, which is even better. Asrock 939A8X-M works like charm and kicks even a bit more out of AGP ;)
I love how you deliberately obscure the CPU when installing the heatsink so the thermal paste zealots don't start a war 😂
Hehe, that was not even an intention :)
@@necro_ware It's the third video where you've inadvertantly done it. I thought after 3 times you must have had a reason. I guess not :)
@@PJBonoVox You probably mean the notebook videos? Well, what shall I say, I never thought about that.....
Just watched once again, at least you can see how I put thermal paste onto the CPU :D
since the onboard lan worked in linux, try changing the hardware resources (irq, io range) for lan nic in windows
tried, other IRQs, other Address ranges, unfortunately no way.
oh, I forgot to mention in the video, that I analyzed the traffic with the wireshark. I can see outgoing packets from the machine, but incoming packets seem to be ignored.
Some SATA2 drives can be used in SATA1 mode with a jumper. :)
That is true, but unfortunately SSDs almost never. At least the one I'm talking about in the video has no jumpers.
Ich hatte ein ähnliches Problem mit meiner Netzwerkkarte auch mal. Bei mir schien diese keine gültige Mac-Adresse mehr zu haben. In den Eigenschaften der Netzwerkkarte (Geräte-Manager) konnte ich unter "Physikalische Adresse" manuell eine Andere vergeben. Da ich die Ursprünglicher noch wusste, nahm ich die. Ich weiß nicht, ob es auch mit einer "generierten" Mac-Adresse klappt.
Vielen Dank für die Idee! Mein Adapter hatte eine gültige MAC Adresse, doch ich habe auch schon versucht eine andere einzutragen, aber leider wollte das Teil auch nicht. Wie gesagt, unter Linux alles kein Problem, einfach hochfahren und es geht direkt ohne Weiteres. Ich befürchte, dass in den VIA Treibern für Windows irgendwo ein Bug drin ist, was den Netzwerkadapter dort unbenutzbar macht. Ich frage mich nur, wie es damals niemandem aufgefallen ist?! Wenn man im Netz danach sucht, findet man hunderte berichte, wo die Leute sich über eine kaputte VIA Netzwerkkarte beschweren und überall ist die Lösung die selbe - eine andere Karte nehmen. Und, dass eine andere Karte geht wurde überall als Beweis wahrgenommen, dass der VIA Adapter kaputt sei. Doch unter Linux zeigt sich eindeutig, dass es nicht so ist.
Why did you use a 64bit processor and installed a 32bit operating system?
Because I wanted a better compatibility to the old stuff in Windows XP. There were issues on 64-bit Windows XP back in the days. Linux on the other hand, which I'm using on that machine as well is 64-bit.
I just got StarCraft 2 in original packaging , I can't play it on xp??? Anyone know a work around to play it offline
You can't AFAIK. This game is too new for XP, not only do you need to be online, it also needs DirectX 10 I think, which doesn't run on XP.
@@necro_ware oh man that stinks , i bought the physical game box that came out years ago
i suggest to check the caps on the power supply, cheap power supply=> chinese caps, etc; you know the rest...
I know the rest, yes. I bought this power supply 10 years ago and since that time it was not used very much. I didin't have any issues regarding this PSU. I opened it once though, around 2 years ago, it was fine.
the SATA 2 (or 3) Drives on SATA1 Controllers is NOT a VIA Problem, its a General Problem with all SATA1 devices I have tested. VIA VT8237, nForce 2. SIL 3112 might be the same thing, dont remember
I sadly don't have an Intel to test but I doubt it works better...
That Issue was fixed with later iterations of the VIA VT8237. I thin it was the S-Version (VT8237S).
PRO TIP: STAY AWAAY from the VT8251!!
Well, it's a known bug of VIA. You can even read about it on Wikipedia. I have no issues with other chipsets here. Using the SSD, which I showed in the video, with almost everything. Runs everywhere with sata1, but at least it runs. VIA is just crap :D Unfortunately for this CPU it was the most used chipset and almost the only one you can get today.
@@necro_ware
Its not, as said it also is in other Chipsets at the time. The mentioned Nforce 2 for example. The MCP-S has the same issue and some SATA Controller too. So for it to work you need a SATA2 PCI Card.
VIA on a decent quality board is better than whats said.
And there are also some S754 Boards without VIA CHip, for example the ASUS K8S-MX. And with S939 you find more nForce Chipsets - which are even worse than VIA!
@@Stefan_Payne As far as I remember, Nforce2 was used with Socket 462 and it had no SATA at all. I might be wrong in this regards. In regards of VIA with S754, as I told it is ALMOST the one chipset you can get, at least with AGP. I have another board with PCIe and Nforce3, there SATA works just fine, but I wanted a system with AGP to make some tests. Something other, than VIA is really not easy to find.
And in regards of SAT in VT8237, it is really a well known and documented bug. The guys grounded one "reserved" connection of the SATA controller in the southbridge. This connection had to be left floating and was used in SATA-II for extended initialization. Since it was grounded, the SATA-II couldn't initialize properly anymore. What shall I say? RTFM, especially, when you design a chipset :)
Please pay attention, that I never told, that other chipsets didn't have other or similar problems, but that doesn't make VIA any better, especially with this one and with the famous data corruption bug. External SATA-I Adapters from that time for the PCI slot do work just fine with all my SATA-II and SATA-III drives. I'm using them actually quite often. But the whole I/O in the VIA chipsets from that time was an absolute disaster.
I just briefly googled for detailed bug evaluation, but they are all in German. In English there is just mentioning, but no detailed description. Funny enough, the bug is mentioned as it is only on German Wikipedia, but not on the English.
@@necro_ware There were later Revisions of the nForce 2 that came with SATA; as there was a new Southbridge introduced: the MCP-S with SATA 1. Also has Issues with SATA2 as a lot of other things at the time.
nForce 3 was AGP too, though had issues with nV40, the PCIe ones are the nForce 4 ones.
There was also a 2nd Revision of the VT8237 which had the SATA2 Issues fixed: VT8237*S*
Also its not a Bug of the VT8237 if you find the issue elsewhere and you do! THe nforce is one, I think the Intel ICH5 could have the issues too as a lot of other SATA1 Controllers. Just test it yourself, if you don't believe me. You find it on others!
Its just the way it is with early specified components where there is only one speed and not multiple like later.
And if you google my name, you'll find that I was very active in German Forums when the VIA 8237 was in use.
As sad, there is a later Revision of the SOuthbridge that has the SATA2 compatibility fixed - or even SATA2.
Now i need to think twice about my P2 SCSI Server :/
Why that? It's a nice project, if you ask me and you got all the problems solved, as far as I remember.
Maybe changing the mac adress on the networking car could helped
Idid this on my msi 865pe neo2-fish2r and it worked but only in 100mbit not gigabit
I tried that. There is obviously a bug in the drivers, but nobody will fix an old driver for windows xp. As I mentioned in Linux everything works out of the box, so it's definitely a software issue.
oh and since its an AMD K8 based system, you might want to test Registred ECC RAM.
Especially with an ASUS Board (Gigabyte works too, ASROCK no idea and MSI does not).
That sounds weird but I have 2x2GB Infineon Registred ECC in my ASUS A8R-MVP!
Unfortunately, I have no DDR with ECC at hand :(
@@necro_ware I didn't mean ECC, I meant buffered Server Memory.
HYS72D256220GBR is the model number. And some AMD K8 CPUs with some Boards work with it, others do not.
I think I used one of the two Registred ECC Sticks I had in a Sockel A Board at the time (with VIA CHip)...
Try to install VIA own chipset drivers, not the stock with windows comes, and then install nic drivers.
Tried everything, different versions of via drivers, different network settings, different MTU, sniffed the network with wireshark. Nothing helped.
There are only losers when dealing with VIA chipsets. Grab a board with an Nforce3/4 chipset.
As far as I know, there was no (or not many?) nvidia based mainboard with AGP and Socket 754. I needed this combination for another reasons. Unfortunately most mainboards from that time were already PCIe and I have two such mainboards here. But I needed it with AGP.
@@necro_ware This is not true, practically every manufacturer made at least 1-2 boards ATX/MATX if not more. I think it's more of an issue of you running into the fact that they're harder to find now and more expensive then these crappy VIA boards.
@@gravitone That's why I wrote "not many?". I also assume, that there are probably some rare ones. But all I could find were VIA and some SiS based. I had this VIA in my parts, so....
Where are you located that they're so hard to source? I typed "nforce3" into ebay.com, and the screen was instantly flooded with 754 AGP boards. Just because you already have a VIA board, doesn't make it a good idea to actually use it, since they're always terrible.
@@gravitone yeah, I'm desperately sorry, that I did that terrible mistake. There is probably nothing I could do to fix this again.... I think, I'll drain my depression in a glass of alcohol :D