It’s going through a usb hub. This is the first time having this issue on this keyboard with any motherboard, idk why this motherboard doesn’t like it.
So this issue is still not fixed.
Cpu-z is still hanging:
Not sure what’s up with that. As you san see it hangs on “monitoring” the whole system fully locks up.
Not only would it not detect the cpu correctly, it simply refused to post at all at anything above 75fsb, 4.5x multi
This is definitely progress!
Running 3dmark 2000 now, it’s getting fps similar to what I got on my p5a which is impressive.
I need to see what the memory bandwidth is looking like now that I updated the bios.
P5a could get as high as 205 mb/sec (timing tweak/overclock) stable 230 ish unstable in memtest which is the highest I’ve ever seen on a socket 7 build. Earlier when I was doing. Memtest I saw 122mb/sec which is typical for mvp3 that was at 66fsb I believe though. So actually may have been better than my other mvp3 board considering I think my other was doing 120-130 at 100mhz
I usually use my trusty 2cas 256, 256, 128 matched chip sticks for a total of 640 but earlier when troubleshooting I switched to single sided cas 3 sticks for a total of 384. The reason for this is it was actually the most mb that I could get to post in that configuration at 4.5x 75mhz
Now that it’s running at 100x6 I’ll revisit the ram and see what can be done there.
Edit:
Yeah, this board is basically on par with my p5a
But agp seems to be far more stable! P5a was having all sorts of crashes with this card on the same config/test
These bios options leave a lot to be desired.. arbitrary turbo, fast, medium, etc settings.
I noticed a interesting option, not sure if the old bios had it, but it’s cpu/pci clock and has options for setting the fsb/pci at various mhz increments between 100 and 133mhz for the fsb
It gives different options depending on where you have the jumpers set. Here are the options it gives if you set 66mhz fsb
I tried it at 112 mhz? I think it was (100fsb via jumpers) and it wouldn’t post anymore. So I set 66fsb via jumper and it posted, so I went in and set 95fsb in bios and it reset the cmos 🤣 oh well…
Woah, this is the highest I’ve ever seen memtest report on a socket 7!!!😳
Test is only at 5% so will have to see if it is stable.
It’s going through a usb hub. This is the first time having this issue on this keyboard with any motherboard, idk why this motherboard doesn’t like it.
So this issue is still not fixed.
Cpu-z is still hanging:image.jpg
Not sure what’s up with that. As you san see it hangs on “monitoring” the whole system fully locks up.
On my Epox MVP3G-M I found that CPU-Z locks up if ACPI is enabled in the BIOS, but works fine if it is disabled.
It’s going through a usb hub. This is the first time having this issue on this keyboard with any motherboard, idk why this motherboard doesn’t like it.
So this issue is still not fixed.
Cpu-z is still hanging:image.jpg
Not sure what’s up with that. As you san see it hangs on “monitoring” the whole system fully locks up.
On my Epox MVP3G-M I found that CPU-Z locks up if ACPI is enabled in the BIOS, but works fine if it is disabled.
Tried your suggestion , didn’t fix it 🙁
I started having stability issues which memtest didn’t seem to want to clue me in on.
I think I solved it by going back to single sided modules.
This may explain why I had so many problems with my fic 503+
Anyway, I recall others saying this about theirs as well but didn’t think much of it because memtest was all cool
Now I gotta start from scratch again with bios and mem timing tweaks after debugging all this.
Latest aquamark: (proof of stability! W00t)
I suppose this raises the question: would single sided 512mb sticks work?
I’ve settled on a configuration that seems to be quite stable here, and since sometimes that’s hard to find, I’m going to follow up with all the parts and settings
For now: halo ( 1024x1280 at max detail settings 🤣)
I just played through the first level. There isn’t much fps difference between minimum settings and max settings, I think because of the 9800 pro doing the heavy lifting on the max settings difference.
Without trying to go over 100fsb or 600 core the best scores I can get are the ones already posted for 3dm200, aquamark.
It seems that the memory timings can get far more agressive if you use single sided ram let alone that the double sided isn’t stable at all, i couldn’t get double sided to pass memtest on anything faster than medium setting.
The single sided ram will do turbo on all banks despite some of the sticks being cl3
It’s not lag free, but it’s playable in campaign. I can even get regular headshots
You are seeing the second level, which I’ve actually played to in one run without crash.
Software:
Direct x 9d
All Microsoft sp3 drivers
Win xp pro sp3
Halo ce (origional cd)
Sound not installed
Hardware:
Sound not installed
3x 128mb single sided sticks
K6-III+ 550acr
Radeon 9800 pro
Freeway FW-TI5VGF 1.0 (non ultra)
Remarks:
-A better graphics card would likely be a gf3 ti 500
-It could be expected that basically any k63+ model could likely achieve this result you shouldn’t need a 550. A 400 would probably do the same thing.
-The Freeway Design FW-TI5VGF is based on the TMC TI5VGF which may be a good alternate for those wanting to replicate this build but it is missing hardware features that the freeway board has I believe, though don’t know all the details.
I had some time to investigate the possibility of undocumented fsb settings
Ths clock gen is a ICS 9148af-58
I wasn’t able to find a data sheet for it
So I tried all (literally all) the combinations and came up with this using a low multiplier on the cpu, external cache disabled and loose memory timings.
I believe the 225 setting is 112mhz fsb a undocumented setting!
I have now set the multiplier to 5.5x and fsb to 112mhz giving me 616 mhz cpu
But it appears that the motherboard cache doesn’t like running this fast even at fast timing (as apposed to fastest)
Trying 3.4v I/O
No go
Trying 3.5v I/O
No go
Trying 3.6v I/O
No go, went to a black screen instead of usual hang though.
It seems that 112mhz is the only undocumented setting and you have to give up cache in order to use it.
The tests will continue without cache!
Gave 6x multi a try, that’s a no go. It posts to something about boot block.
Not surprising as that’s probably 670mhz. Bummer that it didn’t work, we can all dream though. Maybe I can try it at higher voltage though. This is at 2/2.1v now.
Omg…. 2.4v
I actually got it into memtest at that speed but not for long and rebooting it didn’t want to post again. Back to 5.5x 2.2v and let’s see what we get.
No cache, normal ram timings, 616mhz core clock and aquamark is doing a run!
So far this speed seems stable. The only question is wether or not it’s faster given that I had to give up 2mb of 4ns cache to get here…
616 core 112fsb seems stable because I made it through aquamark with loose memory timings.
Aquamark score dropped of course, 6,500
Now to tighten memory timings and see if we can get a score increase!
Okay, so I tightened timings to turbo for all three, and 2 and did another aquamark.
It’s really missing that cache.
All that for nothing..
Back to 100x6 🤣 and cache!!!
Last edited by Sphere478 on 2022-01-28, 04:24. Edited 1 time in total.
That's a bummer that you have to disable the motherboard's cache to hit 112. How's your five volt rail? Edit: I just scrolled up and saw that it's 4.91, which is good enough, so that's not an issue.
You shouldn’t have to do 2.4v to hit 560mhz what’s up with that?
I’m guessing windows 98se is the reason you are getting a higher score?
It's a 400 MHz 1.6 volt chip, and that's what it takes to be stable at that speed. Yes, I believe the lower hardware footprint of Win98 is the most likely explanation for the higher score. A 9800SE is an inferior card to a 9800 Pro, with only half the pipelines (and this one is only 128 bit), but that doesn't much matter when it's held back by the CPU and memory. Also, for some reason on the P4 board that I normally have this in I can use the Omega drivers and unlock the other four pipelines, but those drivers won't work with this motherboard and CPU.
You shouldn’t have to do 2.4v to hit 560mhz what’s up with that?
I’m guessing windows 98se is the reason you are getting a higher score?
It's a 400 MHz 1.6 volt chip, and that's what it takes to be stable at that speed. Yes, I believe the lower hardware footprint of Win98 is the most likely explanation for the higher score. A 9800SE is an inferior card to a 9800 Pro, with only half the pipelines (and this one is only 128 bit), but that doesn't much matter when it's held back by the CPU and memory. Also, for some reason on the P4 board that I normally have this in I can use the Omega drivers and unlock the other four pipelines, but those drivers won't work with this motherboard and CPU.