VOGONS


Slow AMD K6-2+ 550 MHz performance?

Topic actions

Reply 20 of 31, by Blzut3

User metadata
Rank Newbie
Rank
Newbie
cyclone3d wrote on 2020-03-23, 00:36:

What chipset does the USB 2.0 card you have use?

Since it's unclear who "you" is in this post, my experience was with an NEC chip (don't have exact part number). I've heard that VIA controllers behave the same in this regard though.

bloodem wrote on 2020-03-23, 13:02:

Yes, I've disabled the USB 1.1 controller since the beginning, along with the Serial and Parallel ports. I would have used the 1.1 controller, but unfortunately I don't have a USB bracket for it.

Interesting, so if you're not using an on board USB controller, then that means when you disabled the card in device manager you disabled all USB support? Strange that it didn't improve your performance back to normal since it did when I was testing these things. Can't easily test my K6-III+ to re-verify at this time though.

bloodem wrote on 2020-03-23, 13:02:

Further updates: Meanwhile, I started playing with another SS7 board: DFI P5BV3+. This one does not work at all with the PCI USB 2.0 adapter. The Windows installer throws a BSOD during hardware detection (even though it doesn't even have a driver for it). Without the USB 2.0 adapter, the installation works perfectly. Also tested adding it after the installation: Windows freezes during boot. I checked the IRQ table, all devices have a separate IRQ, so there's no IRQ conflict going on. It just seems that the SS7 boards don't like this adapter at all (and some are worse than others).

Although I think it's very unlikely that you're seeing a similar issue on a different board with a different chipset, I do have to wonder if your blue screen could be even remotely related to this issue I solved with my K6? If nothing else something to keep in the back of your mind in case you find that the system is perfectly stable except one thing.

Reply 21 of 31, by foil_fresh

User metadata
Rank Member
Rank
Member

man i got a similar setup (in my sig) and now i gotta compare stats. i also have an NEC usb card that's pretty much always got a mouse and a USB hard disk connected... interesting stuff

Pentium MMX 166MHz / Shuttle HOT591-p / S3 Trio3d2x / SB 16
Pentium II 266MHz / Lucky-Star 6ABX2V / Riva128 / AWE64 / AW744L
Pentium III 950 MHz / SY-7VBA / Voodoo 3000 / Aureal Vortex 2
Pentium IV 3.4GHz/ GA-8S655FX / Geforce 6800 GT / Audigy

Reply 22 of 31, by derSammler

User metadata
Rank l33t
Rank
l33t
Blzut3 wrote on 2020-03-22, 20:49:

The USB card doesn't prefer anything.

I know, but the PnP routines of the BIOS prefer IRQ sharing based on device classes. And apparently, "USB Controller" and "Display Adapter" get a shared IRQ assigned most of the time. This is true for three machines I've built with an USB 2.0 card added.

cyclone3d wrote on 2020-03-22, 14:25:

I think I know exactly why it is causing the lower scores with the USB card installed.

USB polling. It can use a decent amount of CPU and especially for a lower speed CPU, the CPU usage needed is going to be significant.

Can't be the case here, as he has that issue even when the card is disabled in the device manager.

http://retro-net.de/blog.html

Reply 23 of 31, by auron

User metadata
Rank Member
Rank
Member

this might sound ignorant, but is there really a need for these cpu/pci hogging cards in the first place when even AT boards from that time usually have 2 onboard 1.1 ports? the pinout is afaik always standard unlike PS2, brackets are readily and cheaply available, and i've yet to notice any performance degradation from using a mouse on that. for highspeed transfers, ethernet would be another alternative.

Reply 24 of 31, by matze79

User metadata
Rank Oldbie
Rank
Oldbie

Why you don`t use MVP3 Onboard USB ?

On 486 you can feel the impact on performance if you use a USB Mouse.
The Overhead is significant.

I would always go with PS/2 even on Athlon S462 etc. its faster, less overhead..
It even makes a difference on my Athlon 3200+.
Specially if the CPU is very loaded.

https://dosreloaded.de - The German Retro DOS PC Community
https://www.retroianer.de - under constructing since ever

Co2 - for a endless Summer

Reply 25 of 31, by bloodem

User metadata
Rank Member
Rank
Member
matze79 wrote on 2020-03-24, 17:09:
Why you don`t use MVP3 Onboard USB ? […]
Show full quote

Why you don`t use MVP3 Onboard USB ?

On 486 you can feel the impact on performance if you use a USB Mouse.
The Overhead is significant.

I would always go with PS/2 even on Athlon S462 etc. its faster, less overhead..
It even makes a difference on my Athlon 3200+.
Specially if the CPU is very loaded.

auron wrote on 2020-03-24, 16:26:

this might sound ignorant, but is there really a need for these cpu/pci hogging cards in the first place when even AT boards from that time usually have 2 onboard 1.1 ports? the pinout is afaik always standard unlike PS2, brackets are readily and cheaply available, and i've yet to notice any performance degradation from using a mouse on that. for highspeed transfers, ethernet would be another alternative.

As I mentioend earlier, I would use the onboard USB, but I don't have a bracket. I have a bunch of brackets for sound, many for PS/2 mouse, but none for USB. 😀 (I need the USB solely for transferring all my drivers / install kits).
But for sure I will buy a few USB brackets in the near future.

vogons_sig.jpg

Reply 26 of 31, by bloodem

User metadata
Rank Member
Rank
Member

Update 2: Tried the USB 2.0 PCI adapter on a third Super Socket 7 board: LuckyTech P5MVP3 (one of the most stable Super Socket 7 boards out there, also with the VIA MVP3 chipset).
This one doesn't seem to have a problem with the PCI adapter, no errors, no BSODs and no performance penalty. So I guess this will be my go-to board from now on 😀

PS: In the meantime, I also purchased 3 x AT USB brackets, so now I'm able to use the onboard USB.

vogons_sig.jpg

Reply 27 of 31, by kool kitty89

User metadata
Rank Member
Rank
Member

I think most (or all?) MVP based boards use the standard/modern USB headers, so one of those generic add-a-port dongles on ebay should work.

I've also got some 5.25" drive bay caddy/adapters (and enhanced cooling ventilation/airspace) for 3.5" drives, and they have nice front-mount ISB ports with header cables.

Unfortunately, these sorts of dongles/header brackets aren't good for Asus's proprietary MIR ports (which also means no front-panel USB with my P5A-B ... since my USB card also lacks headers on it, though I have another card with internal USB socket ... but I'd need a different type of front panel USB lead for that)

Apolloboy had an FIC VA503A that kept having stability issues, and in hindsight I'd thought it might have been down to poor case airflow along with CPU and cooler mismatch, but come to think of it, it was mostly related to crash-prone USB interfacing. (and he tried several different USB cards, mostly VIA based but of very different vintages: including one I had with firewire also included)

We didn't bother with tracking down USB headers for it at the time given we'd assumed chipset USB would be as problematic or worse (and he got rid of the board before I could fiddle with it more).

So I wonder if the MVP3 chipset in general has issues one way or another. In his case, there's also the 686A Super Southbridge to consider compared to typical MVP3 boards.

I haven't noticed oddly slow performance on my P5A-B system, but I don't think I've done back to back comparisons with the USB card disabled and enabled either.

It's one of the older Aladdin V revisions, too, with the 128MB cacheable limit, but not the weirdly slow K6-2+ issues of the late model. I never updated the BIOS (still a 1998 revision, I believe 1.003) and POST doesn't properly detect the CPU model (or clock rates above 400 MHz ... except with the 5.5 multiplier oddly enough, and that's true for K6-2s as well) but that hasn't seem to have caused any performance issues. (and benchmarks between a K6-III and K6-III+ at the same clock and bus speeds seem to be identical in Sandra 99)

Reply 28 of 31, by rasz_pl

User metadata
Rank Member
Rank
Member
bloodem wrote on 2020-03-23, 13:38:

I would go as far as to say that they are almost as stable as my 440BX boards (blasphemy, I know). 😁

Almost as stable is like almost as pregnant. You can have stable board (440BX), or one crashing (pretty much every non intel socket 7 chipset).

Reply 29 of 31, by bloodem

User metadata
Rank Member
Rank
Member

When I say "stable", I don't just mean "no crashes". Because, to be honest, I didn't have crashes with either SS7 VIA MVP3 boards or 440BX boards (except for the problem that I described in this thread, which, to be fair, is generated by a more "modern" PCI adapter). So, when it comes to this type of "stability", lack of crashes in games & general OS operation, I would say that they've all been rock solid for me.
However, stable in my opinion also means compatibility, knowing that no matter what you try, it will just work out of the box: like, for example, sound cards. I've had issues with many sound cards / DOS compatibility on the VIA MVP3 boards. They mostly seem to work best with ISA sound cards. I even had issues in Windows 98 with some of them. However, on 440BX, I had no issues at all: ESS Solo 1, Yamaha 7x4, Creative Sound Blaster Live 5.1, Audygy 2 ZS etc. They all work properly without any headaches - which is why I said "almost as stable" when talking about my SS7 boards 😀

vogons_sig.jpg

Reply 30 of 31, by kool kitty89

User metadata
Rank Member
Rank
Member

Support for patches and alternate driver revisions (or alternate installation utilities) used to be a lot better for these boards/chipsets. Between official support and user groups (and old forums and chatrooms), I'm pretty sure there were a lot more options to tweak things to general functionality.

Dad had a lot of things hacked together that way on our shared/family/multimedia/HT PC between 1997 and 2000, and some of that stuff became hard to find even a few years later. (he had a working, stable, and compatible cobbled together DVD video home theater software suite that worked quite well in our K6-2 300 + Rage Pro PCI build using a mix of beta drivers and other pieced together components)

That stopped working after a full system upgrade around 2000, or rather he switched to using more standard DVD video software (not windows media player of the time, but something common and brand name, I could probably pick from a list, but forget off-hand) and that turned out to be annoying due to copy protection 'features' preventing us from reliably using the AV output of the Rage Pro. Attempts to resurrect the old software on the new system proved problematic. (It probably still worked 1:1 on the old build, which had been handed down to me at that point, but Dad didn't go that route)

Come to think of it, it might have been an ATi Rage specific software suite we had originally.

I'm not sure what board we had in that machine, but I suspect an MVP 3 one since it got replaced/swapped with a slightly newer VA-503A at some point (or the release date of that board is earlier than I've seen doccumented online, and Dad actually got it new in 1998 as part of the K6-2 upgrade). That board also has onboard sound (VIA 686A southbridge with SB compatibility) and I don't think ever used an external sound card.

I'm not sure what sound board he put in the Soyo ZX (Celeron 366) board that I'm pretty sure went into that family/multimedia PC since it lacks onboard sound. (maybe one of our PC sound cards or fall back to the PAS-16, especially since we only had a 2.1 or 3.1 speaker set-up at the time) Also no idea if he overclocked that thing, though given I actually had it running at 550 MHz for a while without short-term issues a few years back, I suspect so. (2.0V, non-adjustable on that board too, but with the BX style PCI/AGP divider for 100 MHz, though the 75 and 83 MHz settings overclock those; it had the stock cooler and the 550 setting is probably really sensitive to higher temps, so probably 75 or 83 depending on what the video and sound cards tolerated)

OTOH, he may have gotten a PCI based Rage over an AGP one based on the stability and driver compatibility issues some SS7 boards have/had at the time. (that's one thing that seems mostly absent in Windows 98SE and later era drivers, at least for these mid/late 90s cards: AGP Rage 128 Pro, TNT, Radeon, and Geforce cards give more trouble using install utilities, but manually copying the DLLs to the proper location usually works ... it's just a headache to do if you're not used to it)

People mention all sorts of headaches with VIA based S370 boards, too, but I don't remember having much problem back then (it is a bit harder to get things set up now, or 10 years ago, though, and my retro builds have never been as stable as our home-built systems were back then). I haven't installed anywhere near the number of driver updates and patches for the various devices (and individual games) though, and don't run any of those systems online. (in the event there's any online automatic updates available for those old games)

I remember some of the Factor 5 based Lucas Arts games being much less trouble back then than they seem to be now. (I had Rogue Squadron and Episode 1 Racer working quite well on that K6-2 rig, LAN play worked well on Racer, too)

Intel chipsets (for the most part) from that era seem to have had better 'out of the box' support and also had better built-in drivers for various windows installation CDs or service pack upgrades.

That and I think the actual chipset/memory timing/bandwidth performance tended to be better (socket 7 aside, since that had blows traded several times and intels own CPUs do really well in memory benchmarks on some competing chipsets). OTOH I'm not sure how they compare for overclocking support. (board + CPU stability at various FSB settings, voltages, etc) Sometimes slow chipsets overclock well. And/or they overclock well and any quirks/stability issues don't get any worse than at stock settings. (still not great for I/O bound software)

Oh, and I know the VIA 686B has that HDD (IDE/PATA) corruption issue with SB Live cards, but that's also not S7 relevant since only the 686A was used there. (though I believe there are still specific settings and drivers needed for the SBLive to work properly with the 686A, and not just disabling the legacy audio feature ... I think it's an issue with plug n' play resource allocation) It's nice if you actually use the onboard sound though, with sound blaster compatibility and fairly decent OPL3 clone implementation. (supposedly more accurate than Creative's CQM, not sure how it compares to some of the closer clone FM chips, though or how the signal quality of the output compares ... or how that varies by board: I have a few boards with that chip)

Oh, also as to actual crashing and stability:

I actually run into random system lock-ups far more than error messages of any kind (BSOD or program crash + error message window). The latter is a lot easier to remedy and usually happens with me attempting overclocking, undervolting, or odd system configurations that are generally out of spec. The random lock-ups are more confusing and hard to sort out. (between overheating, disk/OS software bugs/corruption/bad drivers/etc ... the sort of disk/OS issues that scandisk won't pick up either: not actual corrupted data per se, but installation/configuration software that did nasty things it wasn't supposed to do)

Oh that and video cards. I've had so many problems that turn out to just disappear when I swap a card out. Albeit they seem to be worse with power-hungry/hot running cards. (Savage 4 did it, Voodoo 3 does it too: the latter was causing DOS Quake to crash in 640x480 software mode as well as system lock-ups at random points during general desktop operations, especially file transfers, or just after file transfers: both floppy and HDD to HDD, one time after I closed and tried to re-open a disk directory in explorer)

It's a PCI based Voodoo 3 2000 16MB BTW. (I think the other is a Savage 4 Pro, 32 MB)

Rage XL and Rage Pro are pretty rock-solid stability wise ... not great for Direct X 6+ gaming performance, though, and annoyingly have more incompatible synch rates with my LCD monitors. (I liked the Savage for that, particularly as I could fix/limit the synch range used in the driver utility or in control panel) And I have some working CRT VGA monitors, but not the space to set them up currently, especially for a modular test system. (and the horizontal case I'm using gets the monitor stowed on top of it)

That's all on the Asus P5A-B system I've been using for most DOS/9s game and tweaking/testing for the last 10 years.

Reply 31 of 31, by bloodem

User metadata
Rank Member
Rank
Member

The issue you are describing is very common on SS7 boards. Many of them were unable to sustain the power requirements for what were back then "power hungry" video cards. Voodoo 3 is one of them.
I find that GeForce 2 MX 400 is the best video card for SS7: low power requirements (4 - 5 Watts), not too fast, not too slow, great compatibility, the ability to use older drivers like 7.76 which are much faster for SS7 platforms (less driver overhead).

vogons_sig.jpg