meljor wrote:

Nice score on the cpu mark. I am REALLY curious how it scores in the cpu tests of everest, can you run those for me please? Especially interested to see what the fsb does for the fpu tests, as it is usually a lot slower as p2 in these.

If you want to run these: On the lavalys page there are older versions, please use 2253 beta. (the program now has the name aida, older versions are everest)

Shame about the ti500, still would like to see how it goes compared to the v3 pci.

I will try the version of Everest you linked later, I only had a version of Aida32 that would not support the K6-3+ when I did the benching.

But I did run PCMark 2002 😀

The scores are not great but at least almost even with a PII 450.

No UDMA for the Pentium II neither for the K6-3+, the HDD performance is for some reason much better with the Gigabyte Ga-5AX. The exact same type of 40GB drive was used.

PII 450

K6-3+ 4.5x130

K6-3+ 5x120. Onboard cache activated.

K6-3+ 5x125

And when it comes to the Geforce 3 ti500 it performs just fine. I do not have any AGP drivers installed. Forceware 43.45.

K6-3+ 5x120 MHz no onboard cache.

GLQuake 1280*960 32bit colour: 126 FPS. Click for screenshot in full resolution!

Quake II Open GL 1280*960: 109.4 FPS. Click for screenshot in full resolution!

Quake III 1280*1024 32 bit colour: 53.9 FPS. Click for screenshot in full resolution!

EDIT

I got very bad FPS (~30) in Unreal Gold running D3D so I thought that I would install and tweak the AGP driver. With the AGP driver I get 1% better D3D performance, 0.5% more FPS in Quake II and III and 10% worse performance in GLQuake. Totally worth the hassle 😜. The bad performance in Unreal Gold with anything other than Voodoo cards or really fast CPUs isnt a big surprise.

I improved the Gigabyte GA-5AXs non existent cooling a bit.

Sadly it did not bring back stability with the onboard cache activated using the Geforce 3 at 120 MHz FSB but now Windows at least starts... its an improvement I guess. This board seems more stable with a PCI video card.

That MOSFET with the Macgyvered heatsink screwed to it did not get cooler when the whole PCB isnt cooling it anymore, I got a nice burn when I was using my finger as temp sensor.

I'd get a decent fan (as in house fan), keep the case open (it its even on a case) and just cool everything with it just for testing. If that doesn't get it stable with the cache activated, then no amount of improvements to the system's cooling will.

alexanrs wrote:

I'd get a decent fan (as in house fan), keep the case open (it its even on a case) and just cool everything with it just for testing. If that doesn't get it stable with the cache activated, then no amount of improvements to the system's cooling will.

With a PCI video card the cache is 100% stable at 120 MHz for some reason, even with no cooling at all and 30C ambient.

The issue right now is that the last couple of days has been the hottest so far during this summer but as it hasnt been that hot up until now I have not bothered with setting up my AC this year so the air I blow over the board which isnt even in a case at the moment is 30 degrees C hot.

Im sure it will work later tonight with an open window but that isnt what I call stable 😀

Skyscraper wrote:

With a PCI video card the cache is 100% stable at 120 MHz for some reason, even with no cooling at all and 30C ambient. […]
Show full quote

alexanrs wrote:

I'd get a decent fan (as in house fan), keep the case open (it its even on a case) and just cool everything with it just for testing. If that doesn't get it stable with the cache activated, then no amount of improvements to the system's cooling will.

With a PCI video card the cache is 100% stable at 120 MHz for some reason, even with no cooling at all and 30C ambient.

The issue right now is that the last couple of days has been the hottest so far during this summer but as it hasnt been that hot up until now I have not bothered with setting up my AC this year so the air I blow over the board which isnt even in a case at the moment is 30 degrees C hot.

Im sure it will work later tonight with an open window but that isnt what I call stable 😀

30 degrees is hot so it can be the problem. But did you try another agp card? Maybe the ti500 (and the overclocked cpu) draws too much power from the board? Don't know if that can influence the cache but i would take another agp card and see how that goes. Even if it is a very slow card it would atleast rule out the problem about agp.

Keep up the good work!

meljor wrote:

Skyscraper wrote:

With a PCI video card the cache is 100% stable at 120 MHz for some reason, even with no cooling at all and 30C ambient. […]
Show full quote

alexanrs wrote:

I'd get a decent fan (as in house fan), keep the case open (it its even on a case) and just cool everything with it just for testing. If that doesn't get it stable with the cache activated, then no amount of improvements to the system's cooling will.

With a PCI video card the cache is 100% stable at 120 MHz for some reason, even with no cooling at all and 30C ambient.

The issue right now is that the last couple of days has been the hottest so far during this summer but as it hasnt been that hot up until now I have not bothered with setting up my AC this year so the air I blow over the board which isnt even in a case at the moment is 30 degrees C hot.

Im sure it will work later tonight with an open window but that isnt what I call stable 😀

30 degrees is hot so it can be the problem. But did you try another agp card? Maybe the ti500 (and the overclocked cpu) draws too much power from the board? Don't know if that can influence the cache but i would take another agp card and see how that goes. Even if it is a very slow card it would atleast rule out the problem about agp.

Keep up the good work!

Im pretty confident its the power draw alright, the extra power lines from the PSU to the AGP port are enabled (JP11 VCC3) but the AGP port still uses its own MOSFET aswell which probably takes its power from the same 5V traces as the CPU VRM circuit. I will test another video card later when Im done messing with this Geforce 3 ti500.

I have tried changing the IO voltage to 3.58V but that made the problem much worse, I take that as another sign that the ti500 is a bit too power hungry. The system is still rock stable at 5x120 with the onboard cache disabled though which is rather good as many other boards from 1998 wont run a Geforce 3 ti500 at all.

I did also run the Everest version you suggested at 5x120 without motherboard cache, the results are what could be expected 😀 With luck it will be faster with higher FSB.

Memory read: 398 MB/s
Memory write: 170 MB/s
Memory copy: 196 MB/s
Memory latency: 102.5 ns

CPU Queen: 1189 <--- this looks OK
CPU PhotoWorxx: 506
CPU ZLib: 2561 kB/s <--- this does also look OK
CPU AES: 622

FPU Julia: 88
FPU Madel: 41 <--- this is really bad
FPU SinJulia: 100 <--- this is not too bad

No changes in Everest scores with the onboard cache activated except for CPU PhotoWorxx which scored 588

Thank you for testing everest.

I compared (again) to my k6-3@600mhz. I didn't write down the memory numbers but i know for sure your read speed is pretty high. The rest i did write down back then and strangely enough the only number that really stands out is photoworx: up 40%! That's a lot 🤣 (417-588)

The rest of the numbers are the same. Fpu julia is up 5 points but that's it (83-88).

The weak spot as you know is the fpu with these k6 cpu's, and i was hoping it would get better with a much higher fsb. But strangely enough it doesn't benefit at all. The numbers stay exactly the same 😕

Maybe it does make a difference on your board as i don't know your number at 6x100 but i doubt it.

Since p3-450 is much faster in some games i think julia and mandel are the most important fpu tests for that, as sinjulia is right up there with the p3-450.

Do you have liquid nitrogen? 🤣 🤣 🤣

meljor wrote:

Thank you for testing everest. […]
Show full quote

Thank you for testing everest.

I compared (again) to my k6-3@600mhz. I didn't write down the memory numbers but i know for sure your read speed is pretty high. The rest i did write down back then and strangely enough the only number that really stands out is photoworx: up 40%! That's a lot 🤣 (417-588)

The rest of the numbers are the same. Fpu julia is up 5 points but that's it (83-88).

The weak spot as you know is the fpu with these k6 cpu's, and i was hoping it would get better with a much higher fsb. But strangely enough it doesn't benefit at all. The numbers stay exactly the same 😕

Maybe it does make a difference on your board as i don't know your number at 6x100 but i doubt it.

Since p3-450 is much faster in some games i think julia and mandel are the most important fpu tests for that, as sinjulia is right up there with the p3-450.

Do you have liquid nitrogen? 🤣 🤣 🤣

No liquid nitrogen but we have winter in a few months 😜

I can report that with a Geforce 2 MX 400 the CPU is scaling with voltage just a good as with the Voodoo III PCI but the gaming performance is much better. It seems the Geforce 3 ti500 used too much power for the system to be stable at speeds over 600 MHz or the motherboard cache to be fully stable at 120 MHz FSB. No such issues with the Geforce 2 MX.

The MS-5184 can have the FSB up to 130 (or 133?) MHz.
But, I don't want to risk my stuff 😖, I'll just try to overclock by software my K6-III+ 400 to 450 or 500 (Since the boot time didn't changed from the Pentium mmx233, more because the slow IDE interface).
I'm using the same stock cooler kid from the pmmx233, and since it has the same TDP than my K6-III at these use there isn't any problem, the system is really cool, not even hot (the voodoo 3 was, before the extra cooler install), so I think there'll be no problem running it at 500 or even 550.

EDIT: fun fact, I just discovered that when my AT beyboard isn't well connected, the cpu is discovered as -MMX at 16MHz

I'm looking for the latest BIOS for Epox EP-MVP3G5 from 2001-08-07, called " 2A5LEPA2". Can somebody help? Many thanks!

Skyscraper wrote:

... That MOSFET with the Macgyvered heatsink screwed to it did not get cooler when the whole PCB isnt cooling it anymore, I got […]
Show full quote

... That MOSFET with the Macgyvered heatsink screwed to it did not get cooler when the whole PCB isnt cooling it anymore, I got a nice burn when I was using my finger as temp sensor.

Just imagine the temps bare FETs can reach, you could be known forever as "Skyscraper four-fingers" after a careless "digital" temp measurement.

And if you liked FETs don't miss the toroid coils:

My version of GA-5AX ghetto-sinking:

Better get some active cooling/bacon on those chokes!

Hi members, sorry for digging this up 😊
But I would like to know about the chipset revision.
My GA5AX is E and this one in link is H. I am confuzed
http://hwbot.org/submission/992473_1badmojoe_ … _450mhz_750_mhz

Me again 😁

Does any one own one of these?

TMC TI5VGA
FW 5VGF + / Ultra
Asus P5A or P5A-B 1MB cache

Thinking about getting a K6-III+ chip. Which would be better 400ATZ 1.6V or 450ACR 2.0V? I think my mobo doesn't officially go under 2V, so I'll start at 2.0V anyway. Dont' know wheter the low woltage chip overcloks better? I would be happy running 5.5x100 MHz; with 600 MHz I'll feel like a winner 😀

I don't think there is a big difference. 550mhz will work no problem on any k6-3+ imho.

600mhz can be reached by a lot of them, but not all are perfectly stable. Running on 2.1v and using a good cooler mine is working fine at 600mhz. But the 5.5x multi is working better with this chip and is almost as fast, so there is no big loss at 550mhz. 5,5x 105 or 110fsb is by far the fastest option if your board and chip can do it stable.

mv_cz wrote:

Thinking about getting a K6-III+ chip. Which would be better 400ATZ 1.6V or 450ACR 2.0V? I think my mobo doesn't officially go under 2V, so I'll start at 2.0V anyway. Dont' know wheter the low woltage chip overcloks better? I would be happy running 5.5x100 MHz; with 600 MHz I'll feel like a winner 😀

I have had good luck with the 400ATZ 1.6V ones, I own 4 and all can do 600+ MHz.

K6-3+ overclocking depends just as much on the motherboard as the CPU though. I know that some members can't get any K6-3+ stable at 600 MHz so some motherboards overclocks worse than others.

Okay then, orderer an ATZ one from a german seller, he had only two left. So i was in a hurry that i wouldn't be able to buy my dream CPU since I've read this article on THW 17 years ago 😁 Albeit the shipping was a little high.

Skyscraper wrote:

mv_cz wrote:

Thinking about getting a K6-III+ chip. Which would be better 400ATZ 1.6V or 450ACR 2.0V? I think my mobo doesn't officially go under 2V, so I'll start at 2.0V anyway. Dont' know wheter the low woltage chip overcloks better? I would be happy running 5.5x100 MHz; with 600 MHz I'll feel like a winner 😀

I have had good luck with the 400ATZ 1.6V ones, I own 4 and all can do 600+ MHz.

K6-3+ overclocking depends just as much on the motherboard as the CPU though. I know that some members can't get any K6-3+ stable at 600 MHz so some motherboards overclocks worse than others.

I've had two of them and they aren't stable at 600MHz, no matter the voltage or motherboard used 🙁.
Still looking for a decent sample that will do 600MHz at 2.1V or so.