VOGONS


Gateway 2000 P5-90 Restoration

Topic actions

Reply 20 of 92, by derSammler

User metadata
Rank l33t
Rank
l33t

Nice system. I've built a very similar one not long ago. Same mainboard, P75, 16 MB, ATI Mach64.

However, a word of warning: the on-board IDE controller is full of hardware bugs and will corrupt data written to the drive, especially when any DMA transfer happens (e.g. copying from/to floppy). The latest BIOS will disable some features of the controller to prevent that from happening, but some bugs will remain. To be on the safe side, update to the latest BIOS, do not use any OS older than 1996 (they know the controller and use work-arounds), and only use a single partition not larger than 528 MB.

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

Reply 23 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
derSammler wrote:

Nice system. I've built a very similar one not long ago. Same mainboard, P75, 16 MB, ATI Mach64.

However, a word of warning: the on-board IDE controller is full of hardware bugs and will corrupt data written to the drive, especially when any DMA transfer happens (e.g. copying from/to floppy). The latest BIOS will disable some features of the controller to prevent that from happening, but some bugs will remain. To be on the safe side, update to the latest BIOS, do not use any OS older than 1996 (they know the controller and use work-arounds), and only use a single partition not larger than 528 MB.

Thanks for the warning, but the FUD here is overblown and short on real details. The RZ1000 PCI IDE chipset was known to have data corruption bugs as noted but they did not affect DOS, Windows, or Windows 95 on Intel motherboards. Your prescribed cure is worse than the disease, and even if there was chance of data corruption, this isn't exactly a mission-critical machine.

PCI motherboard support: RZ1000 IDE Controller

Reply 24 of 92, by derSammler

User metadata
Rank l33t
Rank
l33t

Are you aware that you linked an old website from Intel..? Of course they say: "hey, no real issue here". They did the same with the FDIV bug. You may want to read this: http://mindprod.com/jgloss/eideflaw.html

DOS and Win3.x are only safe if you use the latest BIOS which disables some of the controllers features. But even with that, it constantly destroyed my second partition under DOS, which was beyond the 528 MB boundary. Also keep in mind that the latest Intel BIOS is safe, but your board may not use the Intel BIOS. Compaq for example used that board as well but with their own BIOS. Gateway probably, too.

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

Reply 25 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
derSammler wrote:

Are you aware that you linked an old website from Intel..? Of course they say: "hey, no real issue here". They did the same with the FDIV bug. You may want to read this: http://mindprod.com/jgloss/eideflaw.html

DOS and Win3.x are only safe if you use the latest BIOS which disables some of the controllers features. But even with that, it constantly destroyed my second partition under DOS, which was beyond the 528 MB boundary. Also keep in mind that the latest Intel BIOS is safe, but your board may not use the Intel BIOS. Compaq for example used that board as well but with their own BIOS. Gateway probably, too.

I can tell you're very passionate about this, but you're still overstating the extent of the issue. This is from your own link: "In theory the flaws could do damage under DOS (Disk Operating System), DESQview, Windows and Windows For WorkGroups in 16-bit mode, but so far there have been no damage reports. Windows-95 contains code to bypass the flaws." And later "PCI machines with Intel BIOSes that run only DOS, DESQview, Windows 3.1 or Windows-95 are safe." This is because DOS was not a multi-tasking OS, which meant opportunities for the bug to surface were rare. You can find the same sentiment in the Usenet FAQ about the issue. Of course any users of this board should endeavor to use the latest BIOS available for their OEM and properly patch vulnerable OSes, but the rest of the nonsense you've suggested is a waste of time. It's likely you experienced some other problem, especially since the data corruption issue related to the RZ1000 was a lot more subtle than destroying whole partitions.

The Gateway systems used the Intel AMIBIOS with the "T" OEM signifier, as in 1.00.12.AX1T.

Last edited by JudgeMonroe on 2019-08-07, 21:04. Edited 1 time in total.

Reply 26 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie

Let's talk performance. Expectations for a P5-90 system with 16MB RAM are pretty broad given the contemporary software requirements. Games generally didn't "require" more than a 486DX/66 until the later 90s when MMX became popular, the Pentium II came out, and titles started running in Windows instead of DOS. This P5-90 system should comfortably run most titles through 1997 or so. It should run most of the second generation shooters (e.g., Duke 3D and Quake), and 2D games of the era (Diablo, Civilization II) without much issue. Titles that really push the envelope will be problematic, and upgrading to 32MB should provide plenty of memory headroom for Windows apps (even with Windows 95, 16MB is a practical minimum). Older games that run too fast should be okay in "deturbo" mode that turns off the L2 cache and introduces some NOP cycles. Failing that, conventional "slowdown" utilities can get it even lower.

Basic TopBench benchmark scores around a 250.

IMG_1487.jpg
Filename
IMG_1487.jpg
File size
1.93 MiB
Views
256 views
File license
Fair use/fair dealing exception

In deturbo mode (CTRL-ALT-"-"), this drops to around a 68, which TopBench compares to a high-end 386, which is a sweet spot for a lot of games through 1990/1991. Even demanding Origin games like Wing Commander 1/2 and Ultima VII will run well at this speed.

IMG_1488.jpg
Filename
IMG_1488.jpg
File size
1.88 MiB
Views
256 views
File license
Fair use/fair dealing exception

It was uncommon for game installers to do hardware benchmarks, but once such title is The 7th Guest, aka "Spooky Myst". It tests both CD-ROM and Video Memory speed. For the latter, it doesn't really tell you anything interesting except your performance relative to its own mystery-meat benchmark:

IMG_1486.jpg
Filename
IMG_1486.jpg
File size
1.78 MiB
Views
256 views
File license
Fair use/fair dealing exception

The CD-ROM test is more interesting in that it at least gives you real numbers:

IMG_1485.jpg
Filename
IMG_1485.jpg
File size
1.91 MiB
Views
256 views
File license
Fair use/fair dealing exception

This test is not very impressed with the "24x" CD-ROM drive in this system. Neither am I, really. It's slow, it's loud, it vibrates, and, well, is generally as horrible as all CD-ROMs were at the time. That "300K/sec" benchmark, by the way, is the expected performance of a 2X CD-ROM so either the test is goofy or the CD is.

Finally there is what might be the pickiest DOS-based multimedia game I've ever met, the 1994 Tex Murphy game Under A Killing Moon, which makes use of SVGA VESA modes and recommends at least 12MB RAM.

IMG_1493.jpg
Filename
IMG_1493.jpg
File size
1.89 MiB
Views
256 views
File license
Fair use/fair dealing exception

It has a much better opinion of the CD-ROM drive. It sets the threshold of "fast" for the Video Speed at 5000k/sec and the Mach32's result blows that away. The Mach32 is a bit more problematic than its raw performance, which I'll talk about soon.

Reply 28 of 92, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I don't think I've seen a better looking desktop/monitor combination. Are you using the RZ1000 for the harddrive and the SMC SuperI/O for ISA IDE w/CD-ROM? Your system looks very similar to my Gateway Batman's Revenge socket 4 board, which I outfitted with a Pentium 133 overdrive chip. I've not been able to an IDE drive working on the RZ1000 controller. What about you?

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 29 of 92, by AvalonH

User metadata
Rank Member
Rank
Member

PC Mag September 1994 reviews this system as part of 40 Pentium 90s covered in the issue (one overclocked to 108mhz that later failed despite a humongous heatsink). The list price for Gateway 2000 P5-90 in that issue is $3999.

Reply 30 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

I don't think I've seen a better looking desktop/monitor combination. Are you using the RZ1000 for the harddrive and the SMC SuperI/O for ISA IDE w/CD-ROM? Your system looks very similar to my Gateway Batman's Revenge socket 4 board, which I outfitted with a Pentium 133 overdrive chip. I've not been able to an IDE drive working on the RZ1000 controller. What about you?

Yes, that's right. The Plato motherboard uses the RZ1000 for the PCI IDE and the SMC 37C665 Super I/O for the secondary ISA IDE.

I am using the PCI-IDE for the CompactFlash adapter without issues. I don't have a mechanical hard drive small enough to use in this system -- the BIOS has the 2.1 GB limit (4096 cylinders) but with cheat codes I have been able to get a 4GB CF card working. It's just not practical for me to stick with it, so I'm primarily using a pair of 2GB cards I can swap at will. Not sure what you mean by "not been able to [get] an IDE drive working" on that controller, but depending on what drives you've been using it could be a BIOS limitation. Some of these BIOSes, like the Intel BIOS on the Plato, respond to unsupported drives by hanging during boot.

The CD-ROM is on the secondary controller and it works about as well as you can expect a CD-ROM drive to work.

Reply 31 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
AvalonH wrote:

PC Mag September 1994 reviews this system as part of 40 Pentium 90s covered in the issue (one overclocked to 108mhz that later failed despite a humongous heatsink). The list price for Gateway 2000 P5-90 in that issue is $3999.

That was the price point for a high-end PC through most of the 90s. I don't know exactly when the commoditization started in earnest and the bottom fell out, but everyday PCs then were priced like Workstations are now. The world is spoiled for computer prices. Post 1995 I always built my own systems so I wasn't really keeping a pulse on retail prices until I switched to Mac in the mid 00s.

To be fair, that $3999 price would have been for the big tower case with a 17-inch monitor. This desktop/15-inch combo would have been $2999-$3499 depending on add-ons. At the end of 1995 the prices were the same but the top CPU was a P5-133 in the Gateway 2000 10th Anniversary system. I kind of wish I still had that thing.

Reply 32 of 92, by feipoa

User metadata
Rank l33t++
Rank
l33t++
JudgeMonroe wrote:
Yes, that's right. The Plato motherboard uses the RZ1000 for the PCI IDE and the SMC 37C665 Super I/O for the secondary ISA IDE. […]
Show full quote
feipoa wrote:

I don't think I've seen a better looking desktop/monitor combination. Are you using the RZ1000 for the harddrive and the SMC SuperI/O for ISA IDE w/CD-ROM? Your system looks very similar to my Gateway Batman's Revenge socket 4 board, which I outfitted with a Pentium 133 overdrive chip. I've not been able to an IDE drive working on the RZ1000 controller. What about you?

Yes, that's right. The Plato motherboard uses the RZ1000 for the PCI IDE and the SMC 37C665 Super I/O for the secondary ISA IDE.

I am using the PCI-IDE for the CompactFlash adapter without issues. I don't have a mechanical hard drive small enough to use in this system -- the BIOS has the 2.1 GB limit (4096 cylinders) but with cheat codes I have been able to get a 4GB CF card working. It's just not practical for me to stick with it, so I'm primarily using a pair of 2GB cards I can swap at will. Not sure what you mean by "not been able to [get] an IDE drive working" on that controller, but depending on what drives you've been using it could be a BIOS limitation. Some of these BIOSes, like the Intel BIOS on the Plato, respond to unsupported drives by hanging during boot.

The CD-ROM is on the secondary controller and it works about as well as you can expect a CD-ROM drive to work.

Sorry, I missed a word. I was wondering if you has been able to get an IDE CD-ROM drive working, either in DOS or Windows, on the RZ1000 port?

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 33 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

Sorry, I missed a word. I was wondering if you has been able to get an IDE CD-ROM drive working, either in DOS or Windows, on the RZ1000 port?

It hadn't occurred to me to try it, but I'm not surprised if it doesn't work.

Reply 34 of 92, by precaud

User metadata
Rank Member
Rank
Member

Nice system, Judge. It definitely sounds like it ended up in the right hands.

Earlier today I discovered that I have one of those Plato boards with a P75 in it, and a dead DS12887 battery, as one would expect. Found that I could get into Setup by booting to DOS first, setting the clock, then restarting into Setup. I see you found that as well. Not many settings available in Setup. Its Bios version is 1.00.12.AX1 ,is there a later one?

I plan on taking a Dremel to the Dallas chip and mount an outboard CR2032 holder on it. It looks doable without desoldering the chip. I have a few 2GB-and-less HDD's laying around so will probably put one of them to good use. Not sure what I'll do with the system, I'm not into games, but it will be a fun project.

Reply 35 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
precaud wrote:

Its Bios version is 1.00.12.AX1 ,is there a later one?

There might be. BIOS 12 is the latest I know of for the Gateway version (AX1T) but if yours lacks a trailing letter (AX1) then it's a non-OEM BIOS and there might be a BIOS 16 out there. I couldn't tell you what it fixes/adds or where to get it.

I watched some videos of dudes taking a Dremel to the Dallas chip and decided that was too much hassle. One good thing about the Dallas chips is the the batteries have an electronic version of the "pull tab to activate" battery protector so even new old stock should be good to go! I've seen some Dallas clones for $20 that have the CR2032 holder on a RTC chip, and I suspect I'd go that route before the Dremel. I've had some recent practice with desoldering after repairing a couple of Apple II ram banks, so that helped.

Besides games, I am enjoying trying out some of the software that actually competed with Windows in the pre-95 days; DesqView, GEM, and etc, as well as unexpected recent development projects, a lot of which are in or around FreeDOS. Primarily I have been archiving my collection of 5.25 floppy disks from 1990-1992 and playing around with hardware configurations. I may use my copy of Turbo Assembler to pick up some rudimentary x86 ASM that I never bothered learning. I feel like there's a lot to experiment with in a vintage system. Have fun!

Last edited by JudgeMonroe on 2019-08-16, 16:04. Edited 1 time in total.

Reply 36 of 92, by precaud

User metadata
Rank Member
Rank
Member

Thx for the info. Your past experience with this board is very helpful. I've looked around and found nothing in the way of newer bioses nor the program to flash it. Chances are, the present BIOS will be fine for my needs. I'll most likely put a SCSI card in it, which bypasses the possible RZ1000 issues. I have SCSI HDD's, CD writer, and PCMCIA card readers sitting around waiting for something to do.

Beware of Dallas chip fakes on eBay. A couple years ago I bought a "new" DS1287 from an eBay vendor, to install in a Tektronix digital scope. I wrote the seller in advance and was assured it was authentic and good, overstock of later production runs in the Philippines. It arrived with the paint on the plastic case still a bit wet, and the battery was dead. An obvious fake. Had to dispute the issue with eBay to get a refund.

I have good desoldering rigs (Pace, Ungar) but would prefer to keep the heat off of multilayer pcb's if possible. The side of the Dallas chip that needs to be opened is nicely exposed and accessible on this board, so it should be an easy job.

Archiving 5.25" disks sounds like an excellent idea! I should do that too, I have a pile of them going back to the CP/M days.

Cheers.

EDIT : Did Gateway write their own modified bioses, or contract Intel to do it for them?

Reply 37 of 92, by JudgeMonroe

User metadata
Rank Newbie
Rank
Newbie
precaud wrote:

EDIT : Did Gateway write their own modified bioses, or contract Intel to do it for them?

Intel provided the BIOS for their mainboards and OEMs generally weren't in the business of writing their own. The OEM signifier was used to prevent flashing, say, a Gateway board with a Dell BIOS. The BIOS was the same at a particular revision but there could be cosmetic differences like copyright statements or Logos.