How many amps can your 5 volt rail supply?

prophase_j wrote:

How many amps can your 5 volt rail supply?

20A

I've updated the bios to 4.1. The CPU is now properly recognized as Duron 1800 MHz. But win98 still reports registry corruption once the desktop is reached.

I am aware that I can't avoid overvolting the CPU. Even with 1.65V, I am 0.15V over specs.

I was doing this to run some benches. I figured a Duron would still outperform an Athlon when being clocked 25% faster.

Get one with at least 30 amps. I can't provide a link right now but if you search the forums you'll are all of is retro heads ran into an issue with under powered supplys.

20A on +5V rail is rather low for this type of mainboard/CPU. Since modern machines generate the CPU voltage from 12V rather than 5V, and new PSUs are rated accordingly (ATX12V specification), it can be difficult to find a PSU with enough oomph on the lower voltage rails. I'd recommend at least 30A for stable operation.

BTW, a Duron 1800 is most likely faster than an Athlon 1400, not only because of the faster clock, but also because it has a newer CPU core with better support for extensions (MMX, SSE).

I'll try my weaker fortron source with 25 A on 5V.

I was wondering if this could be a "win98 meets duron" problem?

I also figured that 1800 MHz would win flat against 256k cache. Especially if I need to overvolt 0.15V anyway, the thing will clock up to 2000+

I remember the old Slot A boards needing quality PSUs. AMD even had an approved PSU list on their site. An Enlight 300W was recommended a lot, I remember.

I'm sure that there were a lot of really poor PSUs in use by people. Those things that come with cases, for example 😳 . Athlon needed a lot more power than other CPUs of the time and likely pushed quite a few cheapie PSUs over the edge. And, well, the early Athlon mobos weren't exactly top quality hardware either and could very well have been quite sensitive to PSU quirks.

I have switched to 25A on a Forton Source PSU. That thing ist heavy / sturdy. Same problem.
And after all, the Athlon 1400 works just fine.
Other tests I have done indicate that the Problem is not really the CPU. I've run prime in dos - no problem.
Once Windows has started, I get registry corruption or bluescreen before I complete the startup.
The memory seems to be the problem. Can't be for real though, cause I tried many sticks, and all worked with the athlon.

Maybe the board is just not compatible with applebred cores?

I don't think 25A is enough. You really want to have 30.

The PSU advice provided by the sage veterans above is recommended. 😀

Also, some suggestions:

1) Uninstall software on your HDD such as George Breese's latency patch, and anything similar. Maybe there's something specific to do with the Athlon CPU in your registry that the Applebred core is clashing with? Edit: Reinstall G.B.'s latency patch afterwards, if this temporary uninstall fixed the problem.

2) If that fails, try a fresh Win98 installation. How about using a cheap and simple compact flash HDD, for example - just to have a "back up / testing" HDD to hand. Edit: Compact flash devices use much less power than a conventional HDD.

3) If that fails, cut out as many power hungry components as possible. For example, use a simple PCI graphics card (eg < 16MB) just to get you to the desktop.

4) If that fails, try a "new" PSU with a minimum of 30A on the 5v rail. (Not first choice, simply because you might not have another PSU to hand right now.)

5) If that fails, try 2 PSUs: one for the mobo, the other for everything else. I do this all the time. Right now in fact, with my KT133A based mobo.

6) Also, carefully examine the CPU for any sign of damage.

If you drop in a more powerful PSU and it still isn't stable, I'd suspect that the mobo isn't fully compatible with the 130nm CPUs. Those CPUs didn't exist for years after that mobo was designed.

If the CPU is unlocked, try underclocking it via multi.

Also, regarding performance, the Durons were usually only about 10% slower per clock than their Athlon relatives. You can thank the giant L1 for that and AMD's exclusive cache. It still has a lot more usable cache than a Celeron, for ex.

For the record:
I have tried the kinetiz 7E-A with a Geode NX-1750. It wouldn't even post. Since the Athlon 1400c does start with a total power drain of 150W and the Geode (same CPU specs) draws only 50W, I guess the (this?) board just doesn't cut it.

Edit: Didn't try hard enough. Got it done.

Also the PSU brand is very important. I'd pick a Fortron 300W over a QTech 400W any day 😉

Why then would the Athlon take off with 100W drain at regular voltage, but an overvolted Geode with a fifth of that power drain fails. In another board, the Geode flies with a 145W PSU.

It depends on the core voltage, frequency and the die size.

...and ofcourse on how you measure power consumption

Edit: The Geode has mobile properties, right? It'll downclock while the desktop models wont

Today, I got both to work!
I discovered undocumented voltage settings for lower voltages and found, that both CPUs need different multi setttings for default. Duron needed auto, Geode needed no settings at all. Used fsb 100 for it.

Yay, well done!

Just a question, can you still change the multiplier of the Geode on the fly or by bios?

And just a hint, a friend of mine (he's got 3 working Athlon XP systems...yes he's crazy too 😜 ) uses CrystalCPU when overclocking his XP's. It's a nice tool when using mobiles on boards that don't really support it right

Update:
Depending on jumper settings for multi, the Geode is recognized as either Mobile Athlon 4 or Athlon XP-M. It can then be manipulated by CrystalCPU! 😁

For a Geode NX-1750, the speed setting of 5.5 would initialize it as an XP-M with 500 MHz (not 550). It seems to work (3DM2000 demo) on 300 MHz as well.

The downside: In order to run the Geode on this retro board, you must also provide the voltage of the old days. At 1.65 V, it will do 1050 MHz, but at 1.55V it won't run 3DM2000 any more. (it's specs are 1400/133 at 1.25V) I guess the bus system is designed for those higher voltages.
I had no stability at fsb 133 at all, not even at 1,75V. I didn't want to ruin the chip, so I stopped there.

Now this is a completely different picture already: 300-1050 MHz, with the ability to change speed instantly, plus ISA, plus level 1 and level 2 cache off option.