ragnargd

https://www.tomshardware.com/news/windows-raspberry-pi-xp-linux-raspbian-professional

[had to correct this, the "listdevices.txt" was misleading, only exactly ONE driver supports GTX 970 and GTX 980. Other GTX 9x0 / Titans were mentioned in later drivers, but were not in the inf files.] Interesting find while explicitely searching for drivers older than those on the NVidia-Site, but after the launch, on www.forum-3dcenter.org http://uk.download.nvidia.com/Windows/344.11/344.11-desktop-winxp-32bit-international.exe More infos: https://www.forum-3dcenter.org/vbulletin/showthread.php?t=556959 I tried it, and my GTX 970 was detected and runs fine (no SLI, though, even though it is active, and detected on W10 64bit). Excerpt from ListDevices.txt: NVidia mostly lists drivers AFTER this one on their site (there may be one or two more inbetween containing the 970+, but i did not detect any more. Jaclaz?). Oh, and one further insight: There exist inf-files for OEM cards (i.e. nvmii.inf for MSI) that indicate later drivers (347.25, 347.88) support these cards as well, as long as they are from said OEM. (look i.e. here: https://forums.laptopvideo2go.com/topic/30969-modding-nvidia-oem-inf-files/ ) Using these may be possible, but as these may have other settings (assuming OC, requiring better cooler, etc.), it is always a bit risky. If they can be used for your card without inf-modding, you just pinpointed a driver tailor-made for your card, congrats... Starting with version 350.xx, 970+ are neither mentioned in the listdevices.txt, nor in inf-files. I suspect, this has something to do with XP 32bit dropping out of support with Microsoft on April the 8th 2014, so NVidia possibly joined the bandwaggon here... or was "encouraged" to do so by Microsoft, and only OEMs that were complied to deliver drivers for a certain time got some support. But thats just guesswork... For modding inf-files to have 970+ GPUs use the GTX 960 drivers, which seems to be working well, search on MSFN (maybe later i'll link to those threads - give me some time...). According to an analysis of the later drivers, of which the 368.91 (iCafe) seems to be the latest, while 368.81 is the latest generic, remember, that there exists more than one GTX 960, so choose well: NVIDIA_DEV.1401 = "NVIDIA GeForce GTX 960" NVIDIA_DEV.1406 = "NVIDIA GeForce GTX 960"

I had a look at this thread on guru3d again (Jaclaz find): https://forums.guru3d.com/threads/nvidia-geforce-icafe-expresso-368-91-whql-desktop-driver.408957/ A user named maur0 sayeth this: >> what driver is more good this or 368.95 hotfix? >>17/07 368.91 >>20/07 368.95 >>3 days older So there may be even a hotfix... But the only thing i located was from microsoft update (in a cab file) for W10 32bit... i guess something like that doesn't necessarily even exist for XP, so don't hold your breath... Ah, btw, I pulled the XP 32bit iCafe-driver 368.91, as per the guru3d thread, both from nvidia china, and softpedia, and guess what, neither the 10x0 are in the inf, NOR are the GTX 970/980/etc. ! (the drivers are identical) If you want to use the GTX 970 +, you have to inf-mod, eos. The only decision you have to make is wether you take the .81 or the .91 as your base. (i did NOT succeed 100% with the .91 driver yet, but that may be due to my limited skill-set, i get exceptions, and SLI is not working) I just found this one: uk.download.nvidia.com/Windows/344.11/344.11-desktop-winxp-32bit-international.exe Supports GTX 970 and GTX 980. Works. No SLI detected by the driver, though (which is active on W10). Somehow, NVidia is inconsistent with their support-politics... Now, look at this:

Nice find! Truly bizarre games... And that's exactly what i'm after....

I might try. But for my curiosity: Why and what is that supposed to help with? Only certain NICs are slow, not all of them, and i don't ever overclock...

Hi, my Gigabit-NICs are not fun. My "compatible" realtek and broadcom NICs work, but deliver just about 50MBit, which is abysmal performance for NICs calling themselves PCIe Gigabit (on all W10, XP and W98SE). Is suspect incompatibility between the drivers and the Mobo (ASRock 970A Pro3 R2.0), as even the onboard Realtek (that only works on XP and W10) delivers just about 50MBit. I remember quite a bit better performace under W7 on otherMoBos, delivering about 450 MBit (I don't remember on which MoBos, though, i suspect those MoBos are long gone). (Other HW: FX-4320@stock, NV 7900 GTX 512MB PCIe, SB Audigy 2 Platinum PCI, 4 x 4GB DDR3 1333, W98SE on Sata III 60GB SSD, XP and W10 on Sata III 120 GB SSD each) Of course i tested with other NICs in the same PCIe-Slots: - An Intel 1000 CT Desktop PCIe 1x works about right under W10 and XP, delivering 936 MBit - an Intel Aquantia NBase-T PCIe 1x (by StarTech) delivers blazing 2.5GBit (the NIC is too expensive, though) - an Intel Aquantia NBase-T PCIe 4x (by Asus) delivers trail-hot-blazing 4-6GBit in the electrical 4x PCIe slot, saturating the SSD (my Fileserver with its M.2 Raid5 can deliver 8GBit). And my trusted Intel 1000MT PCI ist just held back by the PCI-bus, delivering about 220MBit (which, as you think about, makes the realtek and broadcom PCIe look even worse). I will stick to latter, as, by all means, from all NICs working under all three OS, it has the best performance on that MoBo. ------------------------------------ A little research on the "Intel CT Desktop Adapter": I showed a little confusion why i did not get that adapter to run, as there seems to be a driver for W98SE. Reality is, of course, not giving in to what i think, no matter what i try. Here is the inconvenient truth: There has never been an "Intel PRO/1000 CT Desktop Adapter". If you find a product by that name (it happens a lot), it is a wrong label. There WAS an "Intel PRO/1000 CT Desktop Connection" or "... Network Connection", having a driver for W9x, but that is an entirely different thing! That "thing" is purely an onboard adapter, and is connected to the CPU by a DMA-technique called "CSA", chipset (probably!) is Intel 82572EI. It provided Gigabit, by using a direct connection to the memory controller hub of the then-new Springdale- und Canterwood-chipsets (at about the year of 2003). There were some popular P4-Gamer boards using that OnBoard LAN. But there never was a "Desktop Adapter" of that type, not for PCI, nor for PCIe, and, basically, never could have been, because CSA was (probably) not meant for that, and, for sure, never happened. Since 2008, Intel produces an "Intel CT Desktop Adapter", GigaBit, PCIe, but there is no "PRO" in the name, it has the Intel 82574L chipset, and isn't even remotely compatible to the old driver-set from 2006 for W9x. It somewhat belongs to the Intel PRO set, as Intel says itself, but that just adds to the confusion. That one is a fine PCIe GB Adapter, and you CAN probably get that one run with W9x, by using the DOS-Driver (see the sticky), but that is an entirely different matter. Cheers, Ragnar G.D. ------------------------------------- ignore the rest down here, this is just for documentation purposes... ;-) ------------------------------------- What unnerves me, is, that i don't get the Intel 1000 CT to work under W98SE, even though it nominally has drivers for W98SE, and the drivers "semi-detect" the card. I see that Ruthan had no luck with them either, while back in the days spaceheeder got it to work on W95: https://msfn.org/board/topic/141402-windows-95-21ghz-cpu-limit-broken/page/7/?tab=comments#comment-948852 Peter mentioned them here (while i found them in the list of infs in the latest ProSet sofware for W9x): https://msfn.org/board/topic/85037-new-intel-1-gbit-ethernet-cards-and-w9x/?tab=comments#comment-575835 Does anyone have one of them running nowadays? I see there a different variants of that NIC, maybe the part i bought is not covered by the driver any more. Anyone? ------------------------------------- Cheers, Ragnar G.D.

@ryandoesnstress: The crackling sound is a problem many audio cards have, this is not restricted to Audigy2 (i had this as well, unfortunately), i had this with i.e. onboard audio as well. It is a manufactoring problem, with bad insulation. When it appeared with a soundcard, i replaced it. When it happened with onboard-audio, i was only able to "repair" it by replacing the motherboard with an identical (but intact) model. It's a nuisance. You can dampen the effect, if you crank up the volume of the computer to 100%, while lowering the volume on the external audio equipment (i.e. the amplifier). Still, sometimes i had the effect, that the noise appeared when i moved my mouse... that was where i replaced the motherboard... an expensive "solution"... :-(

Welcome, defuser, to MSFN! My main motivation to run W9x are old games, as a matter of fact, the more obscure, the better. But even though, i know of only a very few games that only run on W9x, and not on XP. Most others run better on XP than on W9x, which is my only reason i have XP running as well. Can you tell me the names of those games you are trying to run on W9x, that don't work on XP (steam or not)? I'm collecting those games, if i can get a hold of them... Cheers, Ragnar G.D.

@ryandoesntstress: For a nicer experience, choose a card with exactly 256MB of Memory. If you don't want an additional power cable, the GF 6600 GT is nice, supported under any circumstances, and there are good passively cooled models. I prefer the NV 7600 GT, though, as it is (quite nearly) as fast as a GF 6800 GTX (which was the fastest supported GPU for W9x), needs no auxiliary power cable, and can can be cooled passively as well. It needs an inf-modded driver, but that is comparably easy to accomplish. There are models that still have a VGA connector, which comes handy in certain situations, so choose one of those, if you can. The EVGA Superclocked 7900GTX is the fastest GPU you can get for W9x, but that one definitely needs a modded driver, because it only is available with 512MB of memory, of which W98SE is not fond of. There is one single 7900 GTX with 256MB of memory, by BFG, i think, but it is neither easy to get, nor cheap, If you want a reasonably fast GPU, look out for a 7900GS 256MB, which are a bit easier to get. If you set your target-lock on any of those, PM me, and i will assist.

Hi, i'm trying out Chocolatey, a windows packaging software. I like "choco upgrade all -y", like with "apt upgrade", as many of the everyday software i use is updated, and i can install all those packages on a new machine (including choco itself) with one small script. But. I don't know anyone using it as well, did not find many critical voices (always good to listen to), and whatnot. Does anyone here use it? Know it? Hate it for good reason? Cheers, Ragnar G.D.

Comparing it by this links... (always GTX vs. Quadro) https://www.techpowerup.com/gpu-specs/geforce-7900-gtx.c154 https://www.techpowerup.com/gpu-specs/quadro-fx-5500.c1335 Memory clock: 800 MHz/1600 MHz effective vs. 505 MHz/1010 MHz effective, which will result in slightly better gaming benchmarks for the GTX. VRAM: 512MB vs. 1GB, which favours the Quadro in some very few games on i.e. XP, but which will not offset the better memory clock of the GTX in benchmarks. Power usage: 84W vs. 96W, which is slightly in favour of the GTX (the Quadro has a slightly better cooling for the VRAM on the backside of the card, though) At the moment, the Quadro is much easier to get, and cheaper, but as we are talking about used GPUs here, that may change occasionally. My budget is gone at the moment, and i already own two GTX, so i will not experiment here, but i'm pretty sure both GTX and Quadro are equally worth it, both being the fastest cards ever for W9x, and for sure both need rloews patch. SLI is only an option on XP (and definitely NOT on W10), and there, only on certain MoBos that are not usable by W9x, with only certain old drivers, so all chaps here can safely ignore that. I tried that, but be my guest if you succeed none the less.

True, WinME is a different story. It's always good to point out that with all similarities, they (95, 98, ME) still do differ.

Hi, I got a used ASRock 990FX Extreme4. I installed an AMD FX 4320, 1GB Ram, and an NVidia 7900GT 256MB, all connected to my VGA/PS/2 KVM switch, at a 1920x1200 Dell. Naked W98SE installion: Check! System shuts down clean via "Start" or ATX-button. Well-behaved (means: Detected, does not work on W98SE, but does not trash the system, so can be used to the full by XP/W10 in Multiboot): - USB2, USB3, Power-saving CPU-states, onboard LAN, onboard HD Audio ... - I disabled LAN and Audio, as i will use a compatible PCIe 1x Gigabit LAN and SB Audigy(2) PCI later on anyway, for XP/W10 as well. Caveats (same as with any FX chipset): - SATA channel 5 and 6 can (and have to) be set to IDE mode, so an SSD and a DVD drive can be used ... - but the AHCI-channels 1-4, and drives at them, are not even detected by W98SE, which is what i like most about the FX chipsets ... :-) - Using himemx is mandatory, even with 512MB - Deactivate the Marvell SATA controller! (Did not test the IDE/Firewire controller yet) - Will not work, and send system into the dreaded compatibility-mode, disabling any CD/DVD-Drive. Maybe later with rloews driver... WIll have to check AHCI-mode later ... Installed: IE, DX, Realtek LAN Driver, patched inofficial NV Driver, SP3.0c. Looks good, but because the board seems to have a problem, strange behaviour... Never get to the point to install NUSB.

I doubt 1GB will be more of a problem than 512MB - W9x will crash in almost all cases (a very few exceptions on some specific NV 6200 exist, but they are what they are: exceptions). rloew wrote a working patch to limit VRAM to 256MB, which always worked for me for W9x, while on every other OS, the amount of memory was unimportant (except, W10 needs 256MB, so 256MB is the lowest common denominator, so to say). @Goodmaneuver: Using that patch of rloew is a good idea, just in case you didn't know of it...

Thank you for your input! I never bothered for the X2 cars of any kind, because they only profit from SLI-like scenarios, and have the same problems (you already mentioned heat, which already in engineering resulted in lower clockrates). On W9x, i assume the 7900GTX will win, as the modded W9x driver does not support the second on-board GPU (SLI really is from the Windows XP age), and there never was an official driver for W9x. Or am i wrong? To see if XP or W10 profits from the second GPU in Dual-Boot scenarios where you cannot install a faster GPU in a second slot may actually be worthwhile. Did you (or did anyone) have benchmarks for this? Links? P.S.: Oh, well, i remember testing SLI with my two 7900 GTX on the MSI 970 Krait, and reviewed my mails - it didn't work, on no operating system, not XP, not W10, and neither MSI nor NVidia provided assistance... well...

Hi, has anyone working experience with Quadro FX 5500, or, given that, any other Quadro, esp. with the G7x chipsets? The "patched" NVidia driver lists them, and for the FX 5500, i can imagine it may work, as it uses the G71 chip, which is i.e. the NVidia 7900GTX (which works well, i have two of them). On Wikipedia, it is listed as 1024MB memory DDR2, but i can find PNY models listed with 1GB DDR3, which would be great for Dual-Boot (with XP and/or W10) scenarios with just one GPU, as there is a supported driver for Windows 8 64bit, which works well for Windows 10 64bit. But before i sink some money into that, if you may offer your words of wisdom ... :-) (The driver compiled my Maximus Decimus lists all kind of cards, of which some of them definitely are not working with that driver, i.e. i suspect the NVidia Quadro FX 5600, which is the G80, will not work). Cheers, Ragnar G.D. P.S.: In two days, the ASRock 990FX Extreme4 will arrive. There is hope...

Yeah. I found that. Thank you. I see the "Fair use exception". Does that mean the software is officially legal to use for free? I did not find hints for that on google. But i may be blind...

This is quite a feat! Well done, ruthan!

I would of course use the slipstreaming method, i have tried that successfully already, and it is quite effective. The idea is, to "slipstream on demand". I think i will just start, and then present my meagre attempts. That will take some days, though... My job was, in fact, software packaging (some years ago, for windows servers), so i think i can manage to get one or two things done. I also picked up some ideas from the way it is done for Lookingglass games modding, and Skyrim. That way, we will be able to avoid copyright-problems, which is quite important in the age of lawyers.

Do you have links to that software?

Out of lack of time I will focus on more promising things first, before sinking more time in the GA-MoBo (which i will still do, but later - the memory-problems are something that may be solved). I will get an ASRock 990FX Extreme4 soon, which, being very similar to the 890FX Deluxe5, has a good prospect of working quite well. Let's see. The AHCI-driver is something i will definitely try out on that occasion, now that the drivers are available. Although, honestly, the IDE-Mode was working just fine. The SSD will burn faster, but as i don't work on that SSD every day, just a few times per year, i will most probably never notice. Yeah, now i'll have some sleep. Thank you for your support, this is what got us to where we are.

Hi, some of us install W98SE very often. It is, esp. when SP is involved, a tedious process, for there is no room for errors. Most of us don't install from DVD, but put the installation-files onto disk (or a USB-Stick) in advance. Given that, what do you think about assembling a package of things that will be used anyway (ServicePack), or are usefull while not hurting (rloews memory patch comes to mind), and inject it via script into the installation base of W98SE (or any other system), to speed up things, and make them work reliable and in the right order without fails? It is possible, of course, and i would start working on this, that is not my question. My RFC is: Are there $THINGS that should NOT be in such an injection because of potential problems or incompatibilities? Or which should only be injected via parameters, if the user thinks so? Or $THINGS we should include for good? Cheers, Ragnar G.D.

The BISO-Setup actually is not worse that that of the ASRock, and regarding the PS/S emulation via USB, actually this could help on other boards without true PS/2. Not that it helps with the problem, of course. And yes, i run the SSD used for W98SE on the controller set to IDE Mode (on channels 5 + 6) - a feature only GigaGyte and ASRock offer (MSI does not, and Asus not any more). This will burn the SSD faster, but i don't use W98SE for everyday use, so no big deal here. If i had one wish for free, i'd ask the vendors to put in the bios-option to boot from the PCIe 4x slot (like was possible on the compatible GA-870A-DS3)... I did not try XMGR.SYS yet - i like the fact it is in active development. Worth a try. The low vcache... have to think about this... i don't understand yet, what that would help with memory violation. After these tests, which i will do to learn, and of course to honor your efforts ... ;-) ... , i will sink rather more time in looking for other working boards. With Ryzen taking over the AMD-fanboybase, used AM3+ boards flood the market for cheap... :-D

Ok, tested, fail. It seems, the access violation occurs even before new hardware is detected, during configuration. Thank you none the less, deomsh!

The problem here is that of choice: Do i have to force this board into subservience, or do i work with something that does the job happily, as does the ASRock 970 Pro3 R2.0? The ASRock board is newer, cheaper, and easier to get in shops (ASRock alsways offers their mainstream boards longer than any other company). AND it is much easier to handle. We have a clear winner here, and it's not clear that even with some effort the GA MoBo will pass the finishing line at all. Reason: Even with the parameters given, success on the GA ist not guaranteed, as memory access violation is not the same as components hanging during detection: The offending memory used by the (unspecified) onboard components is already occupied at boot time, and will stay so, while detection and ressource-/driver-assignment happens later. But that's theory, we'll see.