Cold air is heavier than hot air so, hot air goes up and cold air down, thats basic, next, the cpu fan needs enought space to take fresh air. The cpu coolers for Socket 462 are going to be to cloose to the PSU. A blower fan o a PCI bracket wont help much because it is on the botton of the case and so it will be blowing mostly cold air. Fron the photos, I think it is not posible bacause there isnt space, but I think that the best option if you are going to mod the case, is to put a fan on the top of the case blowing air outside, that would help to take the hot air outside the case.
And the front fan on the botton of the case could help a bit to push fresh air inside, even if its fairly obstructed.
With a modern psu I think that your best result if the cpu fan is very close to the psu, try to reverse the CPU fan, if you dont do anything more, hot air always out.
Thats my opinion.

A fan in a PCI slot does help a little bit since it transports air out that is warmed up by the Voodoo so that it doesn't reach the CPU. In fact I think the V2 should be cared about more than the CPU 😀
I wouldn't revert the CPU fan. The bottom part of the cooler that touches the CPU gets a lot warmer than the surrounding air and that heat must be transported away as fast as possible. A fan that blows onto this metal does this a lot better than one that sucks it from a distance.

But you could install those 5.25-drive-bay fans and make them blow out the air since the sit at the top of the case.

It's practically impossible to do anything substantial without modding, i.e. drilling additional fan placements. Easier to get semi-modern case with better internal structure, unless height is limitation.

Inverting CPU fan with modern style PSU is a decent idea though. Slot 1/Slot A won't need that though, placement is already optimal.

HanSolo wrote:

A fan in a PCI slot does help a little bit since it transports

You need to maintain positive pressure in case or it will be some nasty dust magnet. Typical minitower design from late 90s usually has only one intake 80mm/92mm fan vs PSU exhaust fan of similar size, to balance things out.

But you could install those 5.25-drive-bay fans

Those are practically useless, unless cranked to absolutely ludicrous noise levels. Cable obstructions near 5.25 inch bay will also reduce any benefit.

P.S.
All thing considered, optimal choice would be to drill two new fan holes - one near hot expansion slots (Voodoo PCI, AGP card) for intake and one for exhaust above CPU cooler. Practically, Antec Nine Hundred case approach. Also to install SFX style PSU (or maybe TFX) with adapter, to reduce airflow obstruction.

Thanks to everyone for their input here.

I have the hardware needed to go one way or the other on this but now there does not seem to be a way to monitor CPU temps accurately outside of the BIOS.

I've tried all the usuals:
Speedfan
HWMonitor
AIDA32
SIV

They ALL show a constant 122 degrees celcius at idle while the Bios stays between 48 and 50.

The system stays on with no issues for hours and the 122 never waivers or changes one degree.

Start a stress test like Prime 95? Still stays at 122.

There is no way that this is correct. The CPU cooler is cool to the touch and stays on for hours.

If I can't find a way to monitor temps there are going to be other issues here.

Ideas?

The motherboard and CPU model ?

bogdanpaulb wrote on 2023-04-26, 23:22:

The motherboard and CPU model ?

Gigabyte GA-7VT600
Athlon 1 GHz - A1000AMT3B

HWInfo.

tomcattech wrote on 2023-04-26, 22:58:

They ALL show a constant 122 degrees celcius at idle while the Bios stays between 48 and 50.
The system stays on with no issues for hours and the 122 never waivers or changes one degree.

Typically, the MIO/SIO does not have descriptors for assigning temperature and voltage monitoring inputs.
By tradition, the temperature of the CPU is set to TEMP3, the motherboard - TEMP2, not used - TEMP1.
An unused input is either not connected and the readings jump, either to the ground - then it will be +125 degrees, or to the reference voltage - then -55 degrees.
On some motherboards, the CPU is route to TEMP1 or TEMP2 because it was easier to route the PСB that way.
The BIOS takes this into account and the data is taken from the correct input.
In programs - only if the program knows the layout for this particular board.

Do not mind it.

I think that with that combination of motherboard and CPU, you will never get a reliable CPU temperature reading. Thunderbird and earlier cores did not have their own temperature sensor inside, this was introduced with the Palomino cores.
From the images I see on the internet of the different versions of the 7VT600, it does not have a NTC inside the processor socket like older motherboards. But older motherboards weren't reliable, anyways.
So, I think that the BIOS misleads with the CPU temperature reading simply because there is no temperature sensor and the motherboard is oriented to more modern processors, the Athlon XP.
For this reason, as already mentioned, the software and the BIOS will always show the same, because the CPU does not have a temperature sensor. Maybe the board has a small SMD NTC, but that won't be reliable because it won't be even touching the CPU it will read the temperature inside the socket and that's the BIOS reading, wile the software is looking for the on-die diode.
I may be wrong, but the lack of a temperature sensor on the CPU and the negligence of the motherboard manufacturers were what made possible the famous video of Toms hardware of an Athlon Processor burning up when removing the heatsink while it was running. (https://www.youtube.com/watch?v=YYQSHXNFvUk) and (https://forums.tomshardware.com/threads/does- … -second.315598/)

Nobody would think of doing such a thing, but it is true that the heatsink can unintentionally come loose for different reasons.
It was something very commented at the time.

Horun wrote on 2023-04-25, 23:08:

That case is definately NOT designed for any high wattage system. I quickly see two things wrongs. 1. PSU is vertical mount so cannot suck as much air out as a horizontal at top of case. 2. No extra back fan mount area. Even if the front fan was made wide open it would not allow the heat build up to escape near top where it collects. Here is a quick pic of other late 90's - early 2000's case backs designed for 1/2 decent air flow. I suggest figuring a way to do as Bitwrangler said and create a blow hole either in top mid of case or in top mid of the side panel.

^ This.

Cases with PSU mounted on top of the CPU area were designed for slot 1 systems where the PSU placement does not impact the amount of air the CPU fan can suck in. Socket 7 and super 7 ATX boards as well as some socket 370 builds will also do fine in such a case with a side mounted PSU since they don't generally make much heat - for example a 1GHz pentium 3 with a slim heatsink and large(ish) fan - say 80mm should do ok.

I recommend upgrading to a case with a top mounted PSU.

Socket3 wrote on 2023-04-27, 06:04:

Cases with PSU mounted on top of the CPU area were designed for slot 1 systems where the PSU placement does not impact the amount of air the CPU fan can suck in. Socket 7 and super 7 ATX boards as well as some socket 370 builds will also do fine in such a case with a side mounted PSU since they don't generally make much heat - for example a 1GHz pentium 3 with a slim heatsink and large(ish) fan - say 80mm should do ok.

I recommend upgrading to a case with a top mounted PSU.

it's definitely not super7 socket system.

Hoping wrote on 2023-04-27, 04:54:

I think that with that combination of motherboard and CPU, you will never get a reliable CPU temperature reading. Thunderbird an […]
Show full quote

I think that with that combination of motherboard and CPU, you will never get a reliable CPU temperature reading. Thunderbird and earlier cores did not have their own temperature sensor inside, this was introduced with the Palomino cores.
From the images I see on the internet of the different versions of the 7VT600, it does not have a NTC inside the processor socket like older motherboards. But older motherboards weren't reliable, anyways.
So, I think that the BIOS misleads with the CPU temperature reading simply because there is no temperature sensor and the motherboard is oriented to more modern processors, the Athlon XP.
For this reason, as already mentioned, the software and the BIOS will always show the same, because the CPU does not have a temperature sensor. Maybe the board has a small SMD NTC, but that won't be reliable because it won't be even touching the CPU it will read the temperature inside the socket and that's the BIOS reading, wile the software is looking for the on-die diode.
I may be wrong, but the lack of a temperature sensor on the CPU and the negligence of the motherboard manufacturers were what made possible the famous video of Toms hardware of an Athlon Processor burning up when removing the heatsink while it was running. (https://www.youtube.com/watch?v=YYQSHXNFvUk) and (https://forums.tomshardware.com/threads/does- … -second.315598/)

Nobody would think of doing such a thing, but it is true that the heatsink can unintentionally come loose for different reasons.
It was something very commented at the time.

This is some excellent information!

Do you think that the following motherboard would have the same issue with this CPU?
AsRock K7S41GX

I would transplant the contents of that rig to a different case with better airflow and not ruin that fine looking case by substantially modifying it. That case is made for slotted CPUs of an era where CPU and graphics card TDPs were way lower. They are essentially ATX-ed AT minitowers with the PSU tilted 90 degrees to allow for the case to have decreased height while retaining airflow from a slotted design.

You may not like it, but I wouldn't try to shoehorn such a system into such a case.

If you really absolutely want to put it in there, perhaps at best you could find some sA CPU HSF that moves air the other way around (oh well) and have a PSU with 12cm fan that can also move a lot of air. However you work it, chances are it's gonna end up being a noisy system.

Tetrium wrote on 2023-04-27, 23:12:

I would transplant the contents of that rig to a different case with better airflow and not ruin that fine looking case by substantially modifying it. That case is made for slotted CPUs of an era where CPU and graphics card TDPs were way lower. They are essentially ATX-ed AT minitowers with the PSU tilted 90 degrees to allow for the case to have decreased height while retaining airflow from a slotted design.

You may not like it, but I wouldn't try to shoehorn such a system into such a case.

If you really absolutely want to put it in there, perhaps at best you could find some sA CPU HSF that moves air the other way around (oh well) and have a PSU with 12cm fan that can also move a lot of air. However you work it, chances are it's gonna end up being a noisy system.

All things being equal you are probably right.

But honestly, If I can find a way to monitor CPU temp I'm up for the challenge.

(Going through the parts bin to see if I still have those "round" IDE cables as well...)

In the worst "case" scenario (LOL), I have another case that would be perfect for the situation except it doesn't scream "Retro" like this one does.

The side fan is already completed and it looks like I went a little conservative on the size (90mm), but its just so confined in there so it is what it is.

It isn't perfect, but not too bad for a Navy guy...

Long ago, I did a pretty good looking modification to a Aceropen HX45 case by making diamond pattern array of drilled holes, in 3 sizes, increasing in size from top to bottom in front plastic.
That really improved the looks and cut a 120mm opening in back just where the i/o panel is, not easy feat due to a step due in rear. Still have that case in use.

Cheers,

tomcattech wrote on 2023-04-27, 18:38:

Do you think that the following motherboard would have the same issue with this CPU?
AsRock K7S41GX

I don't think you'd have any luck in that regard either, that series of Asrock motherboards was very common around here at the end of Socket 462's life, we used them a lot to replace damaged motherboards when Socket 462 was no longer sold new. The most reliable way I know of to get an approximate temperature reading on a Thunderbird or earlier core CPU is something like the one used on these motherboards https://theretroweb.com/motherboards/s/soltek-sl-75kav or https: //theretroweb.com/motherboards/s/gigabyte-ga-7dxe.
You can see the NTC inside the socket, but the implementation of this system was very erratic as far as I know.
Any motherboard designed for Athlon XP, that is, Palomino cores and later, is very likely not to have it and, on the other hand, I imagine that older motherboards will not use the internal reading of an Athlon XP.
You cannot look at this issue with current concepts on hardware prior to 2002 or so. The system monitor was not yet common or very reliable, and was not always present or correctly functional.
In the case of AMD socket 462, motherboard manufacturers generally didn't care much about this at first.
The best thing you could do is get an Athlon XP, any Athlon XP model will have a frequency higher than 1Ghz and the bonus of the SSE instructions and the FSB at 266,333 or 400 MHz, the slowest would be the Athlon XP 1500+ 1333mhz FSB 266mhz Palomino core, and it would be cooler because of the lower voltage than the 1Ghz Thunderbird core. In theory, this motherboard, GA-7VT600, would be better used with an Athlon XP, since with a KT600 it already belongs to the end of the socket 462 era, being able to use even the latest Athlon XP with Barton core and 400mhz bus.
But in my experience all socket 462 Athlons were very tolerant of the FSB frequency as long as the maximum frequency of the processor was not exceeded, that is, use a 333mhz FSB on a processor marked as 266mhz, or a 266mhz FSB on a processor marked at 200mhz like the 1Ghz Thunderbird by reducing the multiplier for example. Although in the case of Thunderbird 1Ghz I never had problems without reducing the multiplier, that is, 10x133=1333mhz instead of 10x100=1000mhz. They were famous for overclocking well.

tomcattech wrote on 2023-04-27, 23:32:

In the worst "case" scenario (LOL), I have another case that would be perfect for the situation except it doesn't scream "Retro" like this one does.
The side fan is already completed and it looks like I went a little conservative on the size (90mm), but its just so confined in there so it is what it is.

OK looks good but did anyone here say put the case side fan at the bottom ? Hmm bet it does not do much good at dropping mainboard temps. And yes the thermo sensors are not always accurate.

tomcattech wrote on 2023-04-27, 23:32:

(Going through the parts bin to see if I still have those "round" IDE cables as well...)

Most round IDE cables are essentially just normal cables with thick rubber sleeves, which is also not very optimal for airflow, but looks nice for modding. I suggest to modify regular cables with tape.

Horun wrote on 2023-04-28, 03:21:

tomcattech wrote on 2023-04-27, 23:32:

In the worst "case" scenario (LOL), I have another case that would be perfect for the situation except it doesn't scream "Retro" like this one does.
The side fan is already completed and it looks like I went a little conservative on the size (90mm), but its just so confined in there so it is what it is.

OK looks good but did anyone here say put the case side fan at the bottom ? Hmm bet it does not do much good at dropping mainboard temps. And yes the thermo sensors are not always accurate.

Honestly due to the the size\layout of the case there is absolutely no way to get forced air directly to the CPU is it lives literally half an inch from the bottom of the hot PSU.
Due to that I was going for a replacement of the front\bottom fan as it was blocked and an overall movement of air throughout the case from bottom to top.

Update:
With the intake fan and new "side fan" power supply the overall unit "breathes" much better and there is a consistent suction from the intake fan and much improved exhaust (both quantity and temperature) from behind the PSU at the top.

From within the BIOS I can only test idle, but after 2 hours it looks like we have a 2 degree improvement in actual CPU temp (50 to 48C) but what is cool is that the CPU fan is digging the extra space above it and has gained about 300RPM.
(That's over 15% improvement of air volume over the CPU.)

Considering there's pretty much no heat to remove as I can only test at idle, I'm cautiously optimistic....

Like I've already mentioned, look for TFX/SFX form-factor PSU. They are much more compact and can be installed in regular ATX case with an adapter.

Considering there's pretty much no heat to remove as I can only test at idle

Since we're talking about classic Athlon, it has no energy saving features, so it pumps out a lot of heat even while idling.