VOGONS


Reply 20 of 40, by feipoa

User metadata
Rank l33t++
Rank
l33t++
squiggly wrote:

Will this work with other CPUs? Like if I dropped a Pentium 100 in there what would happen?

I'm also curious about this.

Overall, this sounds like good news. I'm waiting for you all to determine if there are any caveats.

Did you try CPU-Z-v1.57 for Win98? This is what I use on my older systems. If you can't find it, let me know and I will read the readmes to ensure it is freely distributable.

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

Reply 21 of 40, by FesterBlatz

User metadata
Rank Member
Rank
Member

Unfortunately I haven't had much time mess with this lately, but I do seem to be struggling with a stability issue since the mod.

I can let quake 2 run for hours without a hitch, but as soon as I start navigating around Windows 98 the machine will just randomly reboot. Never a blue screen or lockup, it simply acts like someone pressed the reset button. I don't know if it's related to the mod, or if there's some other issue at play. I sort of doubt it's the mod because the problem seems to occur more often once the system has been on a while...

There's also been some "oddness" going on with Vcore jumper settings. Even though I'm using a K6-2+ with a nominal Vcore requirement of 2.0v, I've discovered that I need to set the jumper position to 2.1v to get the board to post and boot. And the hardware monitoring menu in the BIOS shows 2.0 volts. If I set the voltage to 2.2 volts, the monitoring shows 2.1 and the system becomes barely stable enough to even browse around Windows. I suspect either the power supply I'm testing with to be at fault, or perhaps some caps in CPU VRM section of the motherboard are allowing too much ripple. They look OK and are a mix of Rubicon and Sanyo but you never know. When I received the board the area surrounding the CPU socket was caked with dust from the CPU fan, so I suspect it's got tons of hours on it...

Once I have some free time I'll dig into this a little deeper, because as long as the VRM behaves strangely it's impossible to know for sure if the mod is responsible for any stability issues or not.

Reply 22 of 40, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Do you have a non-plus AMD K6 2 or 3 chip which runs at about the same current load to compare stability with?

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

Reply 23 of 40, by FesterBlatz

User metadata
Rank Member
Rank
Member

Yep I have a regular K6-2/500 I could try. In fact, being of the larger 0.25 µm process it should draw even more current than the K6-2+.

When I get some time I'll drag a scope over and do some poking around to verify the VRM circuit is working properly with either CPU as a load.

Reply 24 of 40, by Stedman5040

User metadata
Rank Newbie
Rank
Newbie

I did the same mod to my 1.06 board with a 1k resistor.

I have found the same issue with regards to the voltage. At 2.0v the system will not boot with a K6-2+550. At 2.1v the system does boot up. Have also found that the memory performance is still not giving the same figures as the 1.04 board using the same bios settings. Got to say though the memory performance as per Everest Home 2.20 is much improved over not having the mod. Tried a vanilla K6-III 400AHX in the board and everything fires up OK at 2.2v. I have not given up on it yet, but there are still issues with this vesion ogf the board regards K6+ cpus

Stedman

K6 III+ 500
Epox MVPG2
512Mb Hynix CL2 SDRAM
40MB WD HDD
Creative GeForce2 Ti
CMI8738 Sound card

Reply 25 of 40, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Are the memory throughput results of the non-plus, K6-III 400AHX, the same with and without the resistor mod on your v1.06 board? Apparently the non-plus chips work fine on v1.06 and later boards, so just wanted to ensure that there was no loss of performance or stability.

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

Reply 28 of 40, by squiggly

User metadata
Rank Member
Rank
Member

Why are some people reporting issues and some not? Is the mod being done the same way? I have 2 of these mobos which I was planning on throwing of a cliff once I found out they couldn't handle K6+s.

Reply 29 of 40, by feipoa

User metadata
Rank l33t++
Rank
l33t++

airs: Do benchmarks confirm that your + chip works the same as your non-plus chip at the same freq, e.g. compare a K6-III-450 to a K6-III+ 450?

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

Reply 30 of 40, by airs

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

airs: Do benchmarks confirm that your + chip works the same as your non-plus chip at the same freq, e.g. compare a K6-III-450 to a K6-III+ 450?

Unfortunately I only have a K6-2 500 and K6-2 300 to test with. Would that be sufficient?

Reply 31 of 40, by feipoa

User metadata
Rank l33t++
Rank
l33t++

You can compare benchmarks of the modified board before the mod with K6-III-non-plus and after the mod with K6-III+. Or, you can get some idea for proper performance by using a second, but similar, motherboard which is known to work properly with the K6-III or K6-III+

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

Reply 32 of 40, by airs

User metadata
Rank Newbie
Rank
Newbie
feipoa wrote:

You can compare benchmarks of the modified board before the mod with K6-III-non-plus and after the mod with K6-III+. Or, you can get some idea for proper performance by using a second, but similar, motherboard which is known to work properly with the K6-III or K6-III+

Sounds good - I also have a Epox MVP3G-M with the 2MB BIOS (but that has a VIA chipset).

Reply 33 of 40, by airs

User metadata
Rank Newbie
Rank
Newbie

3DMark 99 Max results. First is before regedit changes to the Ali AGP driver and second is after.

gZpi919h.jpg

FIUZGX4h.jpg

3DMark 2000 result:

1VONwKCh.jpg

The results look consistent with Phil's SS7 V3 3500 build using a DFI K6XV3+/66 - see here

PC Mark 2002:

c0rrec4h.jpg

System consists of:

Asus P5A 1.06
AMD K6-3+ 450 ACZ @ 550 MHz (5.5x100)
256MB RAM
Voodoo 3 3000 @ 183MHz
Win98SE
Latest 3dfx drivers
Ali Integrated 2.13 drivers from nVidia (2.20 wouldn't install)

Reply 34 of 40, by xjas

User metadata
Rank l33t
Rank
l33t

Since no one's posted one yet, here's what the mod looks like on a P5A-B (AT version.) Note that the traces around the socket are rather different, but of course the pinout is the same. I circled the pins with a black marker so you can see exactly which ones to connect.

20191226_203430.jpg
Filename
20191226_203430.jpg
File size
561.13 KiB
Views
582 views
File license
CC-BY-4.0
20191226_205757.jpg
Filename
20191226_205757.jpg
File size
710.57 KiB
Views
582 views
File license
CC-BY-4.0

Preliminary results: extremely good! This is a 1.05 board & is definitely afflicted with the bug. SuperPi (1M) on a K6-2+/500 without the mod took ~21 minutes, after the mod it dropped to 6:17. My non-Plus K6-2 AFX ran it in 8:16 at the same clock speed. Pretty impressed.

I did a ton of 'before' benchmarks with both Intel & Cyrix CPUs, so I'll be able to determine if this mod has any effect (good or bad) on them. Will post up.

twitch.tv/oldskooljay - playing the obscure, forgotten & weird - most Tuesdays & Thursdays @ 6:30 PM PDT. Bonus streams elsewhen!

Reply 35 of 40, by quicknick

User metadata
Rank Member
Rank
Member
FesterBlatz wrote:

There's also been some "oddness" going on with Vcore jumper settings. Even though I'm using a K6-2+ with a nominal Vcore requirement of 2.0v, I've discovered that I need to set the jumper position to 2.1v to get the board to post and boot.

Stedman5040 wrote:

At 2.0v the system will not boot with a K6-2+550.

Not surprising, since the VRM used cannot provide 2.0 volts (at least, not in the way Asus implemented it). Found out yesterday while restoring the P5A v1.05 that I recently got.
Here's an excerpt from the datasheet, and it's clear that the jumper settings labeled 2.0V on the board and in the manual is giving in fact 0 volts (confirmed with a multimeter), so I guess Asus made a sloppy job with this board.

HIP6004B.jpg
Filename
HIP6004B.jpg
File size
192.17 KiB
Views
360 views
File license
Fair use/fair dealing exception

In the following days I intend to ground the VID4 pin (not accesible through the jumpers) to access the 1.3-2.05V range and of course to implement the resistor mod for using the board with a 400ATZ (@500, don't need it higher). I'll update here when job done.

Reply 36 of 40, by LarkSS

User metadata
Rank Newbie
Rank
Newbie

Hello! Has anyone checked this mod for compatibility with a Pentium 1? I'm considering doing the mod on a P5A-B v1.05, but it's important to me that my Pentium 200MHz still works when I want to switch back.

Reply 37 of 40, by mockingbird

User metadata
Rank Member
Rank
Member
quicknick wrote on 2020-07-02, 23:15:

In the following days I intend to ground the VID4 pin (not accesible through the jumpers) to access the 1.3-2.05V range and of course to implement the resistor mod for using the board with a 400ATZ (@500, don't need it higher). I'll update here when job done.

Did you ever get around to doing any of this?

7ivtic.png

Reply 38 of 40, by quicknick

User metadata
Rank Member
Rank
Member

Yes, actually. But not much testing, due to lack of time. Only tested with Phil's DOS bench-pack, performance is where it should be with the III+ after the resistor mod, and I think I've settled to 1.7 or 1.8V to the CPU after the VRM mod.
Sadly board is back in storage, and it will take a while (weeks, months) until I'll build something around it. And a small correction, my board is version 1.06 not 1.05 as I previously said.

P5A_vrm.jpg
Filename
P5A_vrm.jpg
File size
357.32 KiB
Views
215 views
File license
Fair use/fair dealing exception
P5A_resistor.jpg
Filename
P5A_resistor.jpg
File size
417.83 KiB
Views
215 views
File license
Fair use/fair dealing exception

Reply 39 of 40, by AngieAndretti

User metadata
Rank Newbie
Rank
Newbie
squiggly wrote on 2017-12-12, 01:30:

Will this work with other CPUs? Like if I dropped a Pentium 100 in there what would happen?

If my understanding of what he's doing with the mod is correct, it should still work with a Pentium CPU. Now if you modify the VRM to access the lower voltage-range as discussed later on, that's another story and you'd have to undo that part before running the Pentium chip. I tell you what - this may be three years too late for the O.P. but I can give this a try in about a week when the board I've ordered arrives. This should confirm whether we retain Pentium compatibility with the first mod applied.

If I'm wrong and something terrible is going to happen with an old Pentium CPU, someone please speak up and warn me! 😀