This is one of those side switching PCs but sadly, it's missing the plastic shroud that sits around the switch. If anyone has a solution to this, let me know as it attaches to the power supply via two threaded holes in the PSU shell.
This is one of those side switching PCs but sadly, it's missing the plastic shroud that sits around the switch. If anyone has a solution to this, let me know as it attaches to the power supply via two threaded holes in the PSU shell.
On my ast Premium they simply used an IBM clone power supply with the same switch and shroud.
The 3D print turned out awesome. ... Maybe the holes for the screws could be a little bit tighter, i had to use slightly thicker screws.
Anyway, XUB 625 386L rom is doing pretty well in the tests, second only to the QDI dos drivers.
Thanks again! Very impressive indeed... Now the question is what uses more conventional memory, the XT-IDE or the DOS driver... XT-IDE doesn't use conventional memory per se, but the memory it's using for the ROM could theoretically be freed up as conventional memory.
The 386L Rom takes 12KB, but I think that means it's going to take a full 16KB UMB, yes?
Both the 6500 (v2.4) & 6580 (v3.6) drivers take about 3KB, which is pretty lean. There's reason to do both a ROM and the DOS drivers because the DOS drivers don't try to update the drive geometry, they work with what they got. So when XUB or the DTC LBA bios was installed, speedsys saw and tested the full 32GB. Some VLB drivers cap you at 528MB or at 8.4GB.
I ran the drivers with two flags:
F - For fast hard disk, its data transfer rate is higher than 6.1MB/s.
M - For hard disk that supports multiple sectors READ/WRITE.
With the QDI controllers, the Consumer CF and the Industrial CF performance profiles where not like what I saw with the Promise VLB IDE drivers. The industrial CF usually is the one that gets promoted to the higher transfer rates and achieves low latency while the Consumer CF is stuck in low speed. Here, I see the latency drop when the DOS drivers are installed, but it doesn't seem to get the PIO4 throughput boost. Curiously, the consumer CF is able to get some respectable benchmarks with the 6580 driver. (Note: While the Promise controllers get better bench marks, they don't like drives > 8.4GB and some Promise controllers just refuse to talk to the JMicron Sata bridge.)
The other thing is that the Sinitechi device usually gets very low latency. I expected values < 0.30ms. I tried two different SD's and got the same results, but they were both 512MB SD cards. I wonder if I should have tried a higher capacity SD.
No, it wasn't a cut and paste error. MR and DTC bios got the same scores on the consumer CF. I double checked.
douglarwrote on 2023-10-15, 21:40:The 386L Rom takes 12KB, but I think that means it's going to take a full 16KB UMB, yes? […] Show full quote
The 386L Rom takes 12KB, but I think that means it's going to take a full 16KB UMB, yes?
Both the 6500 (v2.4) & 6580 (v3.6) drivers take about 3KB, which is pretty lean. There's reason to do both a ROM and the DOS drivers because the DOS drivers don't try to update the drive geometry, they work with what they got. So when XUB or the DTC LBA bios was installed, speedsys saw and tested the full 32GB. Some VLB drivers cap you at 528MB or at 8.4GB.
I ran the drivers with two flags:
F - For fast hard disk, its data transfer rate is higher than 6.1MB/s.
M - For hard disk that supports multiple sectors READ/WRITE.
With the QDI controllers, the Consumer CF and the Industrial CF performance profiles where not like what I saw with the Promise VLB IDE drivers. The industrial CF usually is the one that gets promoted to the higher transfer rates and achieves low latency while the Consumer CF is stuck in low speed. Here, I see the latency drop when the DOS drivers are installed, but it doesn't seem to get the PIO4 throughput boost. Curiously, the consumer CF is able to get some respectable benchmarks with the 6580 driver. (Note: While the Promise controllers get better bench marks, they don't like drives > 8.4GB and some Promise controllers just refuse to talk to the JMicron Sata bridge.)
The other thing is that the Sinitechi device usually gets very low latency. I expected values < 0.30ms. I tried two different SD's and got the same results, but they were both 512MB SD cards. I wonder if I should have tried a higher capacity SD.
Untitled.png
No, it wasn't a cut and paste error. MR and DTC bios got the same scores on the consumer CF. I double checked.
I am using a 6580 controller with XT-IDE Bios.
I need the Bios anyway, because I use 16GB flash cards, the Asus SV2GX4 maxes out at 8GB (or 7.8GB to be exact)
So do you think, it makes sense to use additionally the 6580 Dos driver? Will it work?
I am using a 6580 controller with XT-IDE Bios.
I need the Bios anyway, because I use 16GB flash cards, the Asus SV2GX4 maxes out at 8GB (or 7.8GB to be exact)
So do you think, it makes sense to use additionally the 6580 Dos driver? Will it work?
If you use DOS and you have a free 3KB upper memory block, I think you should check it out. It would be interesting to find out if it improves the performance of your CF or not. Might also be worth while to check to see if it interferes with 32 bit disk access under Windows.
ok guys. here's some more photos so you can better see what I'm dealing with. sadly, I already have the plastic shroud mentioned previously. I illustrate that in these photos. It seems this Turbo-Cool 300 AT Power Supply has a different version of it. See photos to see what I mean. Also the holes above and below the switch are fine threaded and accepts screws. I also include photos of the place the black shroud came from and why I cannot use the black shroud.
I am using a 6580 controller with XT-IDE Bios.
I need the Bios anyway, because I use 16GB flash cards, the Asus SV2GX4 maxes out at 8GB (or 7.8GB to be exact)
So do you think, it makes sense to use additionally the 6580 Dos driver? Will it work?
If you use DOS and you have a free 3KB upper memory block, I think you should check it out. It would be interesting to find out if it improves the performance of your CF or not. Might also be worth while to check to see if it interferes with 32 bit disk access under Windows.
Does not work for me, I tried with the DOS from Windows 98SE, so 7.1
When I put device=c:\qd6580.sys into config.sys, the execution stops.
I would not say "crash", because it still reacts to CTRL ALT DEL.
I bought one of the PSUs from that link in my previous post. It was bad even though it seemed to have never been used so I got a return with a full refund and to my surprise I was told to keep the PSU! I then harvested the switch bracket and switch bezel with the included screws. All I can say is WOW! This machine is incredible! The pictures speak for themselves.
As you can see, the bracket was needed as well. Thank you every one for your sugestions! I hope these photos help someone in the future.
Now for the reason this was posted in this forum of all places. This machine now features a dual 16-Bit ISA card IDE setup with XT-IDE. I needed two due to the way the case was designed. I brought this relic into the more modern age and what a PC it turned out to be! Featuring 256KB SRAM L2 Cache, an AMD 5x86 clocked at 133MHz, and a Cirrus Logic GD5434 16-Bit ISA card upgraded to 2MB it really handles Duke Nukem 3D well at 320 x 240 resolution. Now, take its IDE controller and remove ground pin 20, the card accepts keyed 80-wire IDE cables. Toss in a Promise Technologies EIDEMax 16-bit ISA card set to secondary with the ROM replaced with XT-IDE r625 repeated 5 times in a Winbond W27C512 64kb DIP28 EEPROM and a second keyed 80-wire IDE cable, you now have a similar setup as the card originally discussed in this post. The first card needs to be left as master for this to work. The nice thing is the first card can be any card with IDE pin 20 removed. The second card must be configurable as secondary for this to work. As for XT-IDE, the IDE channel must be set according to the jumper settings on each card and everything just works. In Windows 95, I had to scan for new hardware but that was fine. Now I have dual IDE channels just like on my VESA Local Bus card. Hope this helps someone in the future as well.
Planning on doing the same on my QD6500, but want to flash XTIDE. I have M27C512's here I could use.
Sure the other sockets don't need to be populated? looks like the bios address jumpers lead to those.
If they need to be populated, what chips would I need?
edit: looks like those are a needed logicchip and a buffer. 74LS138 & 74LS244
douglarwrote on 2024-01-09, 17:34:The Microchip 27C512A EPROM has a working copy of XUB 625 burned on it that works on my SMC network card. […] Show full quote
I removed and resocketed all of the socketed chips and tried setting JP to "Jumper C" but still no luck.
Do I have an incompatible rom? Did I get the wrong glue chips?
6500.png
its working for me now.
yea its weird, but the jumper defenitely has to be moved somewhere else than A otherwise the ROM gets no address at all.
your ICs also look different, my 74LS138 also has no P on the end, so no idea whats different here.
Eimerwrote on 2024-01-14, 10:31:its working for me now.
yea its weird, but the jumper defenitely has to be moved somewhere else than A otherwise the ROM gets no […] Show full quote
its working for me now.
yea its weird, but the jumper defenitely has to be moved somewhere else than A otherwise the ROM gets no address at all.
your ICs also look different, my 74LS138 also has no P on the end, so no idea whats different here.
Also try doubling the 8KB ROM file 4 times to get it up to 64KB if you use a 512KBIT EEPROM
Glad you got yours working. It gives me the motivation to keep working on mine.
As far as I know, the suffix on an IC part number indicates something about the packaging, so an N or a P would indicate that it is a plastic DIP package instead of a ceramic package, so I expected that a 74LS138p would be compatible with a 74LS138. But I'm not an expert at these things.
I'm using the 12KB "386L" image of XUB. Can you get the native support for the QDI Vision QD6500 and QD6580 VLB IDE controllers in any of the 8KB images?
I can try replicating the ROM image. It's odd because the rom works in my system using this card below, but I guess there's different bus driver chips on it. Dont' know if that makes a difference.