pointertovoid

I've used W95a and W95b for years and years, both are rock-stable and identically fast. Run on P1 120MHz 40MB 5400/min, so I can tell you precisely. Blue screens and Dll hell appeared with W98 here. I found a complete 95b on eBay few years ago, so just instruct eBay to warn you when one appears on sale. If you have a licence, it works with any language. But one of (or all?) my 95b install Cd doesn't accept one of my 95a (or all?) licence numbers. DMA works with 95a just as well. This is strictly a matter of proper driver! Being more recent, 95b brings embedded drivers for more hardware and that's all. Use the "Intel Bus Master IDE driver" or the equivalent for your disk host, and you get DMA. I have DMA on a Cmd-0646 disk host and 95a. http://www.intel.com/design/chipsets/drivers/busmastr/ USB on W95b... I fear this is a rather theoretical possibility. You will end up wasting even more time on USB than with W98. It begins to works more or less with WinMe - which consumes much more Ram. I see only Ethernet as a fast port working with W95 and a reasonable effort. So: the real reason for 95b instead of 95a is the Fat32.

About Usbser.sys v7006: I tried to put KB838417 version SP5 properly, using its installer. Usbser lands in Driver Cache, not in system32\drivers, not in Driver.cab, and is not displayed by the Device Manager. I also added Usbser manually in system32\drivers in a second attempt. With and without the file, I get the BSOD when unplugging the USB1 modem. I hence consider that Usbser.sys is independent of this behaviour, which results fully from Usbhub.sys v6935 and v7006, hence both versions of KB838417 are unusable.

I tried again both versions of KB838417 ("latest" and "SP5"), and both are the worst one. This is fully consistent with my previous notes. That is, they bring Usbhub.sys v6935 and v7006, and both produce a BSOD when I disconnect a Usb 1 modem. They also seem to freeze the shutdown of W2k, even from the administrator session - a classical worry by USB. So I'm back to Usbhub.sys v6689, as I don't have KB838921, and since KB838771 seems bugged. This trouble does not happen with my flash card reader, which uses Usb 2.0, with the device manager displaying Usbhub20.sys instead of usbhub.sys. So for people willing to install W2k from a Usb Cd drive, a solution might be to have Usb 2.0 ports and drive, or to copy the Cd to the hard drive and install from there. By the way, replacing the installation Cd by a Dvd is a small effort (don't forget the boot image) and installs W2k a bit faster.

Thank you -X- ! The line didn't exist, but by creating it and filling with the desired folder address, Forefox puts its temporary files there. Well done!

Hello nice people! I've just added a permanent Compact Flash card in my computer to host the paging file. Adata 266x, hence SLC UDMA fixed. Much faster, happy. I have 2GB still available on this CF card, which is volume V: in my Win2k. The agile CF would be perfect for Firefox's many small temporary web files. So: how do I tell Firefox to put its temporary files there? It seems less obvious than with Internet Explorer. This W2k knows 3 Power User sessions in addition to the administrator. Thanks!

Now I have added a permanent Compact Flash card to my usual computer, to carry W2k's paging file. The excellent i815ep +PIII limits the Ram to 512MB. Enough for W2k and the firewall, but Firefox and the Avast antivirus have outgrown the Ram. When Avast updates its virus database, it swells temporarily by 100MB, blocking any other software. But with the fast paging hardware, the computer now stays responsive. With Firefox and Google Earth, I let the machine use 1GB memory, and switching between the applications staid fast - something impossible previously. I can even switch quickly from a video game to Windows Explorer. How comfortable. The Compact Flash is an Adata 266x 4GB, it uses SLC chips, and it fast and responsive. Fixed, Udma as well - only the more reliable Udma transmission makes Transcend a tick better. I formatted it with Diskpar to start the volume at sector 64, this accelerates a lot. The file system should make no difference for a paging file. The CF fits on the secondary Udma100 port at the i815ep, as the main hard disk is on the primary. Sure, an SiI680a Raid controller on Pci and two CF cards would improve performance, but the controller's Bios adds 3s boot time, and I feel the costs excessive for this machine. ----- Hi Benoît, I appreciate and use your work on W95!

I should like to catch this opportunity to tell I do agree with CoffieFiend. Especially to prefer fewer cores that overclock a lot, as this corresponds to your needs. As well for the SSD and the uncommonly advantageous Raid-0. On my config with similar needs, I even exaggerated even to an X25-E flash disk. 130€ for 30GB second-hand, but I feel a small improvement over a good MLC disk like the Vertex. The more exotic RevoDrive looks interesting, but I didn't try it. It needs a wide Pci-E and OCZ gives a list of compatible mobos able to boot on it. Even for mass storage, I wouldn't choose a "green" disk. They're far less agile, many even rotate at 5400rpm. When you search for a file lost somewhere in a big folder of subfolders, the difference is huge. I tried the 450GB Velociraptor: much more silent than expected, but not completely as silent as my 7200rpm which I'll keep. I'll give a look to the P67. Some previous Intel chipsets were misleading about their capabilities, especially Pci-E 2.0, which was available at the video ports but not at the x1 ports on the southbridge; their statement was so unclear it could only be intentional. A DVD burner wears out (laser diode) and is a source of worries, hence I suggest to buy it new from a good model. To pick mine, I compared by googling their <modelname>: modelname versus modelname problem and made a ratio between the number of answers.

I've read my notes again and checked which files currently run on my W2k. Excepted KB838417 which must be avoided, I kept the most recent versions of the USB patches and used them for 15 months now: 883528, 838989, 843503, 836111, 890202 notice the sys files can be injected in \System32\Drivers\ without using the patch installer, which saves time at multiple trials, and overcomes language incompatibility. Was this a good choice? Not quite clear! I would trust 838989 because Microsoft issues no warning for that one. The four others didn't improve what I needed (CF card ejection from a USB reader in a user session). I've seen no clear drawback to them neither; once a corrupted Fat32 on a CF, cause unknown. Software has much in common with explosives: the more your know it, the less you trust it. So for me, the situation is: 838417 no, 838989 yes, 883528 843503 836111 890202 unclear.

XP drivers have no reason to work with W2k. Forget that. Now in 2011, indications for drivers working with W2k are getting unreliable. Some do work but aren't tested, others do but any reference to W2k was removed on suppliers' web pages, and occasionally some drivers claiming W2k don't install properly, including for nVidia video. You know the name of your Asus and Msi mobos. The right way is to look at Asus and Msi websites and download the proper drivers! But be sure to distinguish between Intel Sata drivers and Sata drivers for some added chip, among which JMicron for instance would be slow and concealed under an other name like Gigabyte. For the Sata drivers, don't operate the controller in IDE mode, it's really much slower. You do need Ahci mode, hence the proper driver.

Yes, the Hdd is the bottleneck. Even more so with a 5400rpm 2.5" combined with a C2D. Yes, upgrading to 7200rpm will make a big difference. But a good 7200rpm would be even better. Focus on random access time. No 10,000 nor 15,000rpm disk exist for laptops as far as I know. Velociraptors are too thick to fit. Scsi are not compatible. A pure SSD would be much better. I bought a used Vertex 30GB for $100 on eBay (.com or .de). Not all SSD are equal.

This unofficial SP5 for W2k integrates on-request patches that are bugged, more or less heavily, nearly all of them. I tried those for USB, bugs range from not detecting an inserted flash card to not ejecting it to plain BSOD, something I hadn't experienced with W2k. For USB, there are just half a dozen patches. I imagine 50 of them must make W2k unusable. I regret to make this comment to Gurg's admirable work. Integrating all post-R1 patches would be useful, but only the widely distributed patches. To add them to the installation Cd, if you keep the original dX, IE, WMP, you can stick to the standard MS method. To update dX, IE WMP and more as well, use HfSlip (or nLite, they say).

If I got the story properly: - dX9.0c core hasn't changed for years. Each version installs exactly the same. Only security updates change these files. - What changes is the added layer of Managed DirectX, useful only if an application uses this layer. I meant instead dX10 or dX11 on W2k, which are officially not compatible. Anyway, some dX10 cards do have W2k drivers (official ones up to Gtx295 for instance), and then the applications that use OpenGL should get from the driver the functions that dX10 would have brought.

Congratulations! You're grasping piece by piece why these updates haven't been widely distributed. Microsoft writes "not fully tested" you must read "fully bugged".

If I understand properly, you have a W2k installed on a disk on a different machine and hope to transfer the disk and W2k... This operation is well-known to be less than obvious! On all Nt-series Windows, including W2k. The essential reason is that Nt-series load the driver for the disk host (often called: disk controller) as they boot. So if the source and destination machine have the same host, or maybe the same host driver, I've seen it work up to now. If the host differs, the transfer fails. Many processes have been invented, each needs about one page to explain. The most straightforward - but I haven't tried it - should be to transfer the disk, start the W2k install Cd, feed it with the F6 diskette and ask the Cd to "repair Windows". It uses to keep your data but lose some of your settings, and your Windows updates if your Cd isn't slipstreamed. The mode that repairs only the start files could keep more of your installation. Dumb and oversized method, but it takes 20min, less than other methods take to understand.

Ahci uses to make a huge improvement by enabling parallel requests to the disk - something Nt4 and Nt5 do all the time, more so the Prefetch of XP and followers. Did you observe performance with the right tool? HdTach, HdTune, CrystalDiskMark and others won't tell you, as they issue one request at a time. IOMeter is the tool of choice to see Ahci at work; AttoDisk is less good but easy to use, as opposed to IOMeter.

XP is faster than W2k provided you have enough Ram for Win, the antivirus, the browser or the application you want. And not by little: on my X25-E +E8600 +P45 it boots in 8s, versus 15s for W2k pro (I optimise W2k but know XP little). Once running, XP starts applications much faster than W2k does. This is due to its better Ntldr and Ntdetect (improvable on W2k) and to its prefetch which W2k lacks. Now, why do I stick with W2k? Well, just because XP upsets me within 20s each and every single time.

I installed Win95, WinMe, W2k on CF cards, and: 0) The Transcend Industrial is horribly slow. Use the Transcend 266x or Transcend 300x instead. They are fixed, and have a reliable P-Ata transmission, as opposed to (nearly all) other brands. The adapter doesn't matter. 1) Said Transcend are of SLC technology, whose wearout is negligible, and which have wear levelling. 2 3 4 5) Then you shouldn't care about minimizing writes. Does background defrag exist on W2k3? Defrag is very useful on a CF card and must be done. Aligning the volume on 2n boundaries improves speed, as compared with 63 sectors. But sometimes Win doesn't boot from an aligned volume; this is still unclear to me. I accelerate computers of limited Ram addressing capability with a small CF card which holds the paging file only. But if you can increase the Ram, it's always better than paging, sure. Hope your W2k3 holds in 8GB, because all bigger CF I know are slower. Add more cards if needed. All 600x I know are of MLC technology, whose slow writes of small files make unsuitable for this purpose. The removable flag is overcome by a Raid host, for instance the SiI0680a chip, which alleges the Raid volume is fixed. Such a Raid improves read speed but not write. Maybe an expensive Raid card with much Ram, I didn't try. Many users of CF disks install Win by copy because a normal install can be slow or impossible. BUT: - Slow means your CF is bad on writing small files, and then you shouldn't use it at all; - Impossible means you have P-Ata transmission errors: again, unsuitable CF; so a normal Win install from the CD is an excellent test of your CF card; do it to avoid trouble later. DMA and UDMA should be used as soon as possible on a CF, because PIO is extremely slow, especially during installation. If your controller+OS combination starts in PIO, supply it the right driver right from the beginning with an F6 diskette. For instance with older Intel P-Ata, an F6 version of the Intel Applications Accelerator can be made: from an existing installation on this machine (only), paste on the diskette the corresponding folder containing the txtsetup.oem

I tried some on-request patches for USB on W2k before, and stopped using them, because they brought trouble. see especially my warnings here page__view__findpost__p__836740 page__view__findpost__p__837504 Now I consider that "not fully tested" means "really bugged" and stay away from these patches - except KB838989 in the case of USB..

Here's the kind of performance obtained with Blackwingcat's driver - as good as the XP driver, or even better! An Intel SLC SSD: 30GB X25-E, on an ICH10R host, measured here by AttoDisk with Q=4. And now, a Raid-0 of two X25-E on an ICH10R, again with Blackwingcat's driver. this was with a stripe size of 4kB and an Ntfs volume aligned on megabytes, created by GPartEd. Measurements by IOMeter are equally impressive, especially if its accesses are aligned on multiples of 4kiB for the Raid - not really cheated, since this is the cluster's size. On a single X25-E disk, alignment of volume or accesses brings nothing - as this disk is already a Raid over 10 channels, not 8 nor 16. In both cases, an F6 diskette made with this driver was able to install W2k on the disk or group - but only if the first volume was created by the W2k installation disk, not by GPartEd. Well done, Blackwingcat!

Internet Explorer : o yes, I had forgotten that. The site doesn't accept it here neither. Its webmaster, who has a Linux background, explicitly asks to use Firefox. Maybe he got some disappointments with IE, and his opinion about M$ won't have given him patience to solve them? So he probably put something like if navigator==IE then goto hell I suppose this isn't the reason why search engines occult the site, because - I see often their bots in the visitor's list of that site - All sections of the site reject IE, but some are indexed by the search engines, while others aren't listed - My intuition tells that the bots sent by the search engines don't rely on a browser - but I may be wrong ----- Robots.txt (Ciao Jaclaz, grazie!) would be a very credible explanation. Thanks for the link at Wiki! I had already tried to ask the administrator, who avoided giving a clear explanation. Fortunately, his robots.txt is accessible from Firefox, I have it under my eyes. It contains, among others: Sitemap: coughcough/sitemap_index.xml Disallow: / Do I understand properly that sitemap_index.xml then gives the exclusive list of what the bots shall index? The sitemap_index.xml is accessible and contains for instance <sitemapindex etc> <sitemap> <loc>CoughCough/sitemap/0.xml</loc> <lastmod>2011-03-05T12:00:00+00:00</lastmod> </sitemap> And the just mentioned 0.xml contains <url> <loc>CoughCough/viewforum.php?f=66</loc> <lastmod>2010-11-07T17:28:04+00:00</lastmod> <changefreq>daily</changefreq> <priority>1</priority> </url> where viewforum.php?f=66 is the section "Technology" that I'm most interested in, which search engines don't list. What would bots be supposed to do when reading this? Thanks!

I've found two eBay objects looking identical to the one I use with success: paste 150564908127 or 350209471995 in eCreek's search window. Sellers (YvonneWindy and Rexchg) ship worldwide, they were on the first results page of the "CF IDE" search, there are certainly more. Important features : - P-Ata has 44 pins at laptops but 40 pins plus a power connector at desktops, so the adapters to CF differ - Many adapters from P-Ata to dual CF stack the Compact Flash cards, which then don't fit in my laptop's tray. The ones in links are flat. - I can't say if one adapter will fit in your tray, nor if your driver will see both CF. The Bios must see one to start booting. - Dma and Udma should work with any adapter, booting as well. The fastening points don't fit my tray, but I hold the adapter in foam. Even on my P1 120MHz 40MB, W95b boots in 20s from an SLC CF instead of 60s from a 5400rpm. Expect a bigger difference from a faster Cpu. If your Bios knows only CHS as mine does, good luck... Prefer LBA if you can.

The name of the forum there is saposjoint.net Its section called Science gives no hit in Google search, at least from here. But its section called "General discussions" does.

Why? They were useful to readers willing to check Google's behaviour. By the way, it's Sapo's forum, not mine. I'm a member there, no other relation.

Hello nice people! On my W2k, my Adsl connection is configured by the Usb modem's software. No need to use the Internet Connection Wizard: I start the connection software, which ask for an identity and a password if Window's connection has forgotten them, and this software creates or corrects Window's connection as needed. When sometimes Window's connection goes mad, I just suppress it and let the modem's software create a new nice clean connection. Additionally, this software can repair itself from the installer if asked so. This works well, grateful thoughts to the authors. Recently, that procedure did not work. The normal Adsl connection had transformed into a "Virtual Private Network". This happened from time to time in the past, but recently the connection kept changing to VPN every time I established one. It improved after an hour or so. Is this a known weakness of W2k? Or did a malware create this behaviour? Thanks!

Thanks for helping! I get the same results through google.fr, google.com, google.de, google.es used from France, and from other search engines. When searching through Google some text picked from threads in the occulted sections, I don't get results. Examples: - "tinnitus, it is obnoxious" or "technological state just now than Krypton" - But right now, Google finds "remotely operated like Predators and Reapers", which it did NOT on 4th of March (today is 7th) despite the page is from February. Hey, Msfn has real magic effects! - Google doesn't find any "Marc Schaefer, aka Enthalpy" on forum - though I put hundreds there. - The occulted discussions are not the ones where I contributed nor the ones containing my name. But the occulted sections are the ones describing my inventions. More observations: - BAUTforum.com crashed regularly as I attempted to put links there to forum. But other forums did not. - Now I receive emails from myself containing a link to forum, which was still impossible on 13th of February. I had to relay to myself the sent message in order to receive the second try. Same without the antivirus nor the firewall. I checked the message rules for Outlook Express as well. This improved as I began to experiment on it systematically. One other magic effect. And I always received emails containing links to other forums quite normally. I ignore how Php works... The computer hosting forum re-creates the web page each time a visitor asks for a page, as opposed to Html, isn't it? Would it then be technically possible that forum is programmed to detect the bots sent by search engines, and produces for them expurgated pages? Though this alone wouldn't explain the censored e-mails nor the crashes at other web sites.