I think most of the instability problems boil down to old components, people don't care about the state of the motherboard capacitors, and depending on where you live, the board can suffer from very high temperatures, or too low, too much humidity or too little humidity, and that causes the capacitors to decay at different times and hours, and to different proportions.

I live in Brazil, and here it is very hot, and most people live in places with high humidity due to the proximity to the sea, it is quite common for me to find swollen capacitors, or leaked in equipment from 10 years old, sources ATX suffers A LOT (partly because of our terrible power grid full of noise), there have been cases of cards that weren't terribly unstable, but were a little "slow" for my liking, so I took out the capacitors, and measured some of them , I even found 2200UF capacitors that in the measurement had just over 300UF!

Also sometimes I find inductors that lost their varnish (due to the heat) and ended up accumulating oxidation (very common in coastal regions)

I believe that in North America, it is easier to conserve equipment, because houses are generally thermally insulated, and have a temperature and humidity control system, in addition to a cooler climate (even in regions like California, or Florida, due to most people use air conditioning). Here in Brazil air conditioning is a luxury item, and it consumes a lot of electricity, so only high-end people have one at home.

Sphere478 wrote on 2021-12-27, 12:40:

[...]

Can you share your jumper and bios settings?

Sorry for late reply, was away for the holidays. Now in quarantine for the next week, so should have time to look into this. I recently acquired a Winchip2 which needs putting through its paces, so more than enough reason to dig out the P5A anyway.

Speaking to OP topic, I’ve encountered some interesting results on my new freeway fw-ti5vgf 1.00

It’s actually outperforming the p5a memory seems to be performing faster on it. It may have to do with the 2mb of 4ns cache😳

dionb wrote on 2022-01-04, 23:52:

Sphere478 wrote on 2021-12-27, 12:40:

[...]

Can you share your jumper and bios settings?

Sorry for late reply, was away for the holidays. Now in quarantine for the next week, so should have time to look into this. I recently acquired a Winchip2 which needs putting through its paces, so more than enough reason to dig out the P5A anyway.

...aaand my P5A appears to have died 😦

Won't power on, forcing Power Good with a paperclip just gives 00 on the POST card.

Certainly not giving up on this board, but focus will be on basic functionality, not fancy settings with K6Plus. Afraid I can't give you BIOS or jumper settings anytime soon.

soviet conscript wrote on 2021-12-27, 21:24:

Doornkaat wrote on 2021-12-27, 21:12:

Thanks for the feedback!👍 Stability is up to the individual chip I guess. Do you have the cache chip and the rev H northbridge? […]
Show full quote

soviet conscript wrote on 2021-12-27, 21:01:

I have a GA-5AX with 512KB of L2 cache and the revision H northbridge. I've been running a K6-III+ @ 550MHz in it for years, rock solid though it's unstable at 600MHz, maybe I'm doing something wrong.

Thanks for the feedback!👍
Stability is up to the individual chip I guess.
Do you have the cache chip and the rev H northbridge? Is there no internal cache on the NB like on rev G?

It is the rev H NB and the board does have 512KB of L2 cache on-board. Board revision is REV. 5.2

Sorry, I missed your reply! Would really be interesting to see if there was a 1MB rev H northbridge GA-5AX.
I was confused about internal tag vs. cache though. 😅

What kind of scores are people getting with their GA-5AX's and P5A's in Sandra 99 Memory Benchmark?

I have a GA-5AX R4.1
1 x 256mb PC133 CL2-2-2
Running 5.5 x 115 (633Mhz) on a modded K6 II+ into a K6 III+

Currently getting:

CPU/Memory Bandwidth 184 MB/s
FPU/Memory Bandwidth 200 MB/s

For the CPU Benchmark:

CPU Dhrystone: 1964 MIPS ... can't break 2k 🙁
FPU Whetstone 781 MFLOPS

I'll do some testing on my P5A as well. I have a rev 1.03

magicmanred wrote on 2022-06-14, 15:00:

What kind of scores are people getting with their GA-5AX's and P5A's in Sandra 99 Memory Benchmark? […]
Show full quote

What kind of scores are people getting with their GA-5AX's and P5A's in Sandra 99 Memory Benchmark?

I have a GA-5AX R4.1
1 x 256mb PC133 CL2-2-2
Running 5.5 x 115 (633Mhz) on a modded K6 II+ into a K6 III+

Currently getting:

CPU/Memory Bandwidth 184 MB/s
FPU/Memory Bandwidth 200 MB/s

For the CPU Benchmark:

CPU Dhrystone: 1964 MIPS ... can't break 2k 🙁
FPU Whetstone 781 MFLOPS

I'll do some testing on my P5A as well. I have a rev 1.03

In the k6-3+ aiming for the stars thread there is a user who made a modified 5ax bios that has the p5a timings I believe btw

This is not an asus p5a but it does have an ali Aladdin v chipset and i am curious to see how it matches up with better systems. It is an IBM Aptiva 2153, it came with a 3xxmhz k6-2 of some kind which i swapped for a pentium mmx overclocked to 250mhz with a 100mhz bus. I have always felt that it feels weirdly fast compared to a lot of other systems. The latency seems very low, when i click on something in windows 98 there is almost no hesitation at all, everything just pops up almost instantly. It does have an sd card boot drive but i dont think that has anything to do with the speed because i have a 600mhz pentium 3 system on an abit board with an identical sd card setup and it still feels like it takes longer for things to happen despite the system being overall way faster and more powerful. My sandra 99 results are attached below, weirdly sandra thinks the processor is at 238mhz even though its actually 250 as reported in the bios.

The PII is known for being less fast on 16 bit than 32 bit, and while 9x OS are theoretically 32 bit, there's a lot of 16 bit under the hood and deliberately 16 bit compatibility stuff. Ergo, AMD, Cyrix and fast P55C feel pretty snappy, while PII/PIII kinda trip over their own feet on that stuff.

BitWrangler wrote on 2022-06-16, 02:04:

The PII is known for being less fast on 16 bit than 32 bit, and while 9x OS are theoretically 32 bit, there's a lot of 16 bit under the hood and deliberately 16 bit compatibility stuff. Ergo, AMD, Cyrix and fast P55C feel pretty snappy, while PII/PIII kinda trip over their own feet on that stuff.

The 16-bit performance is much worse with the Pentium Pro, lacking the improvement that the PII/PIII has. 😉

Yah they bandaided it to make it less awful but it's still at a deficit.

Here are my scores

The GA5AX is one of the slowest boards I have tested. It can not run the same WPCREDIT settings than the P5A.
As for the Rev H of the GA5AX, only the wake up memory after sleep is what changed.

Skalabala wrote on 2023-10-07, 21:58:

The GA5AX is one of the slowest boards I have tested. It can not run the same WPCREDIT settings than the P5A.
As for the Rev H of the GA5AX, only the wake up memory after sleep is what changed.

I was able to hit 650Mhz on my GA-5AX rev 5.2 (Chipset "E", so not even the best) with a whopping 130FSB. 2-2-4-7 ram timings.
2.4v on CPU.

I can do 633Mhz with cache enabled (115fsb) with 2-2-2-5 ram timings (wpcredit used).

I'll attach some pics of my Asus P5A scores (I have a 1.04 and a modded 1.06) One is an "E" and one is a "G". No H's unfortunately.

Here is my P5A 1.06 chipset "G" (modded with 1k ohm resistor to accept the + series chips)
Running at 633mhz (115fsb x 5.5) @ 2.2v
Ram at 2-2-2-5

Memory Bandwidth CPU
ASUS P5A = 11.4% faster

Memory Bandwidth FPU
ASUS P5A = 10.5% faster

Memory Read
ASUS P5A = 10% faster

Memory Write
ASUS P5A = 19% faster

Memory Copy
ASUS P5A = 6.7% faster

Memory Latency
ASUS P5A = 19.8% faster

CPU Score
Gigabyte GA-5AX = 0.2% faster (negligible)

In real world experience, I don't see/feel any difference between these two systems in any game or applications I throw at them regardless of the GPUs I have at my disposal.
Measurable with FPS counters by a small bit, but not noticeable.

I wanted to ask how easy is to replace an Ali Aladdin V revision E to revision G or H?
Besides replacing the bga chip, are there new connections needed on a board like GA-5AX rev 5.2?

CoolDoomer wrote on 2025-02-19, 17:24:

I wanted to ask how easy is to replace an Ali Aladdin V revision E to revision G or H?
Besides replacing the bga chip, are there new connections needed on a board like GA-5AX rev 5.2?

Good question, but the 1.06 PCB is not the same as 1.04. 1.06 even has a different chip with ASUS on it.

Skalabala wrote on 2025-02-19, 20:40:

CoolDoomer wrote on 2025-02-19, 17:24:

I wanted to ask how easy is to replace an Ali Aladdin V revision E to revision G or H?
Besides replacing the bga chip, are there new connections needed on a board like GA-5AX rev 5.2?

Good question, but the 1.06 PCB is not the same as 1.04. 1.06 even has a different chip with ASUS on it.

I meant The gigabyte board though it is my mistake I went to this thread. 5.2 rev has both e and g.

Rikintosh wrote on 2021-12-18, 00:34:

Sounds like a silly question, but it's a total unknown to me. I have some old cards, Pentium II, Pentium III, good cards like SO […]
Show full quote

Sounds like a silly question, but it's a total unknown to me. I have some old cards, Pentium II, Pentium III, good cards like SOYO 6BA+, Asus P2-99, and even a soyo with UltraDMA66, but NONE of them is as agile and fast as this simple K6-2 500mhz on the Asus P5a. I did several tests, and it's not just my delusion, the benchmarks prove it.

My question is, what makes this card so special?

My configuration:

AMD K6-2 500mhz
1x 256MB PC100 Kigston
1x 80GB ATA133 HD
A simple dvd rom drive
Geforce 2 MX200 64mb agp
Sound Blaster CT3670

To test the other cards, I used the exact same hardware, with the exception of the AMD K6. I used it on the other cards: Celeron 300, P2 450, and P3 550E.

I suspect that maybe my P5A has a higher quality IDE controller, which makes better use of my modern HD, but I'm not sure, that's why I wanted your opinion!

My Aopen AX59 blows my P5A's (rev 1.03 and rev 1.06) out of the water with chips that have on-die L2 cache (K6-2+ / K6-III). P5A's tend to overclock better then most of my other super7 boards, but are also extremely finnicky. I have a box of dead P5A's that failed over time (gate A20 error or no post).

Nemo1985 wrote on 2021-12-18, 09:06:

It's no better than the Gigabyte GA-5AX, I will never understand why the p5a is considered better...

It's a lot better. About as fast as a rev 1.03 p5a, but a lot more stable and durable. I still prefer the AX59 pro tough.

What did you use to test the AX59 with? Do you have Voodoo2 with quake 2 bench results? 640x480 demo1 I get 95FPS, K6 3+ 500MHz, P5A 1.06
With a Geforce card I got 160FPS 1024x768.