[Sammelthread] AMD K7 - Sockel A (462)

[Sparksnl]

I just tested the DFI NFII Ultra (A) as working. At least it is posting. So that makes it a little more rare.
Not recapped (no swollen caps), not modded. Everything seems to be Ok so far.
Maybe that Palomino was a good thing to come with it. That makes it pretty sure OC could not work .

Edit: I see I made a backup of the Win XP install I made in 2018. That will be a fast install if it works...

 

[Sparksnl]

I just dug up my defective Ultra A: These are the mods. Good for archiving.

Mod 1

Mod 2

Resistors

 

[Sparksnl]

@digitalbath

Is it possible to mod the DFI NFII Ultra A BIOS to the latest 619 version which you made for NF7?
FSB so far:
239MHz @1.60v (BIOS)
<245MHz @1.70v (BIOS)

 

[digitalbath]

Hi, possible to mod yes, but LPA BIOS is an unsolved project (second one is AN7 BIOS). While the new added settings in the BIOS work, the option.rom doesn't set the desired values to the system . Currently, I don't know how I can solve this problem. Especially if I don't own this board. Anyway, I will upload the BIOS files and you can test them. Maybe you will get an another result. Please reply if you need more informations to the BIOS files (f.e. the PCI version).

I will create a new version with the new register values from the 619 NF7 BIOS.

PS. I did test the new NF7 optimizations with my other NF7 and got an interesting result. But more to this tomorrow...

 

[Sparksnl]

Actually I do run a PCI GFX card… Was that what you mean?
Thanks so far.

Oh and I boxed my AN7 and archived them. I will look at them in 20 years or so.

 

[digitalbath]

Actually I do run a PCI GFX card… Was that what you mean?
Thanks so far.

No, that wasn't meant 😅. Since the usual used ISA-option-rom seems not to work, I also made an another version with a PCI-option-rom. The diffirence is just the type of the BIOS module.

Oh and I boxed my AN7 and archived them. I will look at them in 20 years or so.

I hope not. Btw, Infrared tested one or some BIOS version(s) and the seems to work. With bugs, but they work. So I have a little hope, that we are able to get a workable version someday.

 

[Sparksnl]

Well they are boxed. Hopeless project so far.
I will be testing those bios files.

I was wondering what happens if I flash an Ultra B bios…

 

[digitalbath]

So, here is a small mod. I added here the 619XT romsips. This BIOS also contains a ISA option rom that change tRC, tRFC, DS, SR, alpha timings and the new register optimization from NF7. This is a test BIOS .The BIOS itself should work, but I don't kwow, if the changes will work. So if you see the memory timing changes in Windows (in tweaker), there is a chance, that this BIOS works like it should. If not, then only the romsips is the change here...

 

[Sparksnl]

Thanks. I will be testing this first. I hope tomorrow.

 

[Sparksnl]

So, here is a small mod. I added here the 619XT romsips. This BIOS also contains a ISA option rom that change tRC, tRFC, DS, SR, alpha timings and the new register optimization from NF7. This is a test BIOS .The BIOS itself should work, but I don't kwow, if the changes will work. So if you see the memory timing changes in Windows (in tweaker), there is a chance, that this BIOS works like it should. If not, then only the romsips is the change here...

I flashed it and this BIOS gives a post and boot. When I look in the BIOS itself I changed menu's:

and

No extra timings here. Just what was there on a different place.
Settings timings still works.

Auto timings for memory:


Change timings from BIOS:


Change timings from Windows (TRC and TRFC):


Edit: I tried booting 245MHz FSB - Yesterday when running stock BIOS this DID NOT boot, now it does:



Not maxxed and running 1.60v where yesterday I needed 1.70v to do more then 239MHz.

Edit 2: For now my new Nr.1 BIOS for this board.

Edit 3: It is flying --> 260MHz FSB 1:1 boot @1.60v

Edit 4: It posts 270MHz 1:1 (no boot so far).

Need anything more?

 

[digitalbath]

So, your board is a good one! booting 260MHz is nice!

I flashed it and this BIOS gives a post and boot. When I look in the BIOS itself I changed menu's:

I reordered some settings in the BIOS to get most used setting into one menu.

No extra timings here. Just what was there on a different place.

extra timings are hidden and not working in this BIOS version.
There's a official BIOS version from 2003 with settings for DS and SR. The newer BIOS Versions doesn't have this setting anymore.

Settings timings still works.

This should be the default memory BIOS settings.
As far as I can see in your screenshots, the romsips (Interface off and on) are working as expected. The ISA-option-rom in the other hand is not working. I expected this. This means no extra memory setting for now.
So our goal should be to get perfect romsips for this board. The 619XT romsips do a fine job here, but I would try to make here a small optimization (change some values) for a better stability.

Need anything more?

yes, one more thing. could you make a dump file of your PCI register? This could probably help me to understand some values.

 

[Sparksnl]

Thanks, I will be making that dump file.

Yes it seems to be a good one. Those 619XT changes really work.
Would it be better then to not use the latest bios and use the older bios? I can test a few to see which has the options.

Thanks work the work and re-ordering the BIOS. It did annoy me that it was not logically ordered.

 

[digitalbath]

Those 619XT changes really work.

Just be sure that you only tested the "pure" romsips. I put my "NF7" optimizations into the non working ISA-option-rom. Since the option-rom is not working, the optimisations should also not work.
I will modify some few values in the 619XT romsips to get a slightly better stability.

Thanks, I will be making that dump file.

thanks.

-------------------

Da wir hier schon bei den Optimierungen sind, Ich habe diese NF7 Optimierungen auch bei meinem anderen NF7-S v2.0 getestet. Es ist ein älteres Exemplar noch mit schwarzen Slots und ohne dem "Ultra 400" Schriftzug.

NF7-S

v2.0

2003/11

A1

SPP

?

A3

MCP-T

dunkelbraun

v0.51

246

keine

digitalbath

Link

schwarze Slots

?

Die Optimierung hat auch hier geholfen. Ich bein ehrlich gesagt immer noch sprachlos. FSB 264MHz bei 1,7V. Board ist ungemoddet.
Das Ergebnis finde ich super, aber mit den Optimierungen bin Ich noch nicht zu 100% zufrieden. Es bedarf noch weitere Tests und Anpassungen, da die OC Ergebnisse nicht immer reproduzierbar sind.
Bei diesem Ergebnis habe ich schon einen Wert gegenüber der vorherigen Version abgeändert.

@stunned_guy kannst du den OC Wert in der Liste anpassen? Danke.

Btw. Die FSB 265MHz Grenze habe ich auch hier.

 

[Sparksnl]

Just be sure that you only tested the "pure" romsips. I put my "NF7" optimizations into the non working ISA-option-rom. Since the option-rom is not working, the optimisations should also not work.
I will modify some few values in the 619XT romsips to get a slightly better stability.

I am not sure what you mean that I should test just the pure romsips. How do I do that!

 

[digitalbath]

Well, I did two kind of tweaks for my NF7 tests.

-The first are the tweaks in the romsips. These tweaks can be included into the romsips (BIOS) and you're able to use them.

-The second kind of the tweaks are some PCI register values for the Memory Controller (see post #6.469 ). I can change them in Windows or i can include them into the BIOS via ISA-option-rom. Since the option-rom is not working in the LPA BIOS, we're not able to usem via BIOS. The origin of these register tweaks are values from a DFI LPB BIOS. So maybe you don't need this kind of the tweaks. Send me the dump and I will check these values for you.

edit:
You have only to test the BIOS files as you want. I try to include all tweaks in the BIOS I can.

 

[Sparksnl]

So it would be wise to run from BIOS and not just boot, but run PI 32M as well?
Or is maximum 1:1 boot enough?

 

[digitalbath]

So it would be wise to run from BIOS and not just boot, but run PI 32M as well?
Or is maximum 1:1 boot enough?

For the dump file, a bootup at 1:1 should be fine.
Or do you mean the FSB tests? I guess this depends how you test your boards. The romsip tweaks will work with the bootup, the other register tweaks only after bootup. You test, as far as I know, most with bootup tests. So the bootup tests should be enough to see a difference. (I am not sure if I did understand you).

 

[Sparksnl]

What I meant was if I should test differences in performance. Like less time in PI 32M.

I tried booting 264MHz and 265MHz @1.60v. This works but causes issues.
I noticed that although my PS2 keyboard seems to be working, my USB ports are giving issues.
Could this be one of the issues for the 264MHz border?

 

[digitalbath]

What I meant was if I should test differences in performance. Like less time in PI 32M.

Ahh okay. I am not yet in performance tests. My intention is at first try to reach max stable FSB and then doing performance tests with Pi and 3DMark2001. I you want to do some performance tests, you can do this and I will try to do this with simmilar settings and other board (to have comparable results). Until now I only did some PI runs for stability test.

I tried booting 264MHz and 265MHz @1.60v. This works but causes issues.
I noticed that although my PS2 keyboard seems to be working, my USB ports are giving issues.
Could this be one of the issues for the 264MHz border?

Yes, I get same problems with all of my NF2 boards at 264/265MHz. Some boards are able to run 1M before the USB or VGA freezes. Using PS2 mouse and Keyboard helps somtimes. Disable APIC can also help.

 

[Sparksnl]

I must note that I use a HD4350 PCI GFX card. So using PCI GFX does not seem to be making a difference if others were using AGP GFX.
I added the RW everything file.

Edit: Is it possible to lower PCI clock in BIOS (modding that)?

 

[digitalbath]

I added the RW everything file.

Thanks!

This is what I see. Green Boxes are the register tweaks. This values look like the default values from Nvidia. Other manufacturers like ASUS, ASRock and MSI used them. LP B uses other values from what I know.
I use here: 48:65; 4C:90; 4D:0B; 50:B2; 74:62

The blue boxes are the offsets I tweaked in the romsips. This values can also be changed in the tweaker. I use same values in the romsip preset "custom#1".

Edit: Is it possible to lower PCI clock in BIOS (modding that)?

As far as I know, AGP clock is tied to PCI clock. Modbin says AGP clock 50MHz is selectable. You could try it. I don't know if there are other ways mod lower PCI clocks.

 

[Sparksnl]

Actually, I did try to set AGP frequency to 50MHz. This did not help.

The added bios file containes these changes?
48:65; 4C:90; 4D:0B; 50:B2; 74:62

 

[digitalbath]

The added bios file containes these changes?
48:65; 4C:90; 4D:0B; 50:B2; 74:62

Yes, but the ISA option rom is not working.
The other way is to set these values manually in Windows (every boot).

Probably worth to ask Infrared if he could integrate them into the tweaker?

Actually, I did try to set AGP frequency to 50MHz. This did not help.

Okay. Did you try Infrared's tweaker?

 

[Sparksnl]

No I did not try the tweaker yet.
So the new bios does not add anything at this point then? I still have to set it by hand?
I can try…

 

[digitalbath]

So the new bios does not add anything at this point then? I still have to set it by hand?

no, nothing except the romsips (blue boxes in post #6.506).
My hope is that we will find someday satisfying a solution.


You could probably try the PCI option rom version to be 100% sure. This BIOS is in this post: #6.489 (If you test this, LAN and LAN Boot have to be enabled.)

To be honest, I don't understand why the ISA oroms doesn't work in DFI BIOSes. LPB should have simmilar problem. All other boards that I modded, didn't have this issue.

 

[Sparksnl]

You could probably try the PCI option rom version to be 100% sure.

I am not sure what you mean here.
The original bios?

Oh and I forgot to tell. I tried 4 memory kits dor dc.
2x BT-D43
2x GEIL pc4000 D43
2x Corsair XL 256MB
2x Corsair XL 512MB

My TCCD 512’s beat my D43 again. All tested on C2.5-4-4-8 1T and 1.70v. This kit was clearly more stable. Something with this kit is different then any other XL kit I have.

 

[digitalbath]

I am not sure what you mean here.

This version. Is has a different type of BIOS module (PCI option rom).

edit.
TCCD dimms more stable then Hynix, nice! With my MSI Delta2 board, I have a simmilar situation.

 

[Sparksnl]

Yes this set is more stable. But from what I know not all are. The good ones are.

Edit This kit runs good on all NForce 2 boards I tested it with. And that is a lot.

 

[Sparksnl]

Thanks!

This is what I see. Green Boxes are the register tweaks. This values look like the default values from Nvidia. Other manufacturers like ASUS, ASRock and MSI used them. LP B uses other values from what I know.
I use here: 48:65; 4C:90; 4D:0B; 50:B2; 74:62

The blue boxes are the offsets I tweaked in the romsips. This values can also be changed in the tweaker. I use same values in the romsip preset "custom#1".




As far as I know, AGP clock is tied to PCI clock. Modbin says AGP clock 50MHz is selectable. You could try it. I don't know if there are other ways mod lower PCI clocks.

I looks like DFI_LPA_2004_619XTR2_309_t4 is a bit more stabel at high FSB speeds (260MHz). It still won't do PI 32M (not even close).

Edit: No post @240MHz+ FSB with BIOS DFI_LPA_2004_EB33_309_B2A_PCI
Extra's are visible. Now testing them.

Edit 2: So far working and tested: tRC, tRFC, tREF, Drive strenght, Romsip Custom 1.

 

[digitalbath]

I looks like DFI_LPA_2004_619XTR2_309_t4 is a bit more stabel at high FSB speeds (260MHz). It still won't do PI 32M (not even close).

Okay, thanks for the information / testing. The register tweaks could probably help (#6.508) .

Edit: No post @240MHz+ FSB with BIOS DFI_LPA_2004_EB33_309_B2A_PCI
Extra's are visible. Now testing them.

Don't forget to activate LAN and LAN boot in BIOS. Maybe you have to boot from the LAN ( I am not sure at this point).

edit.

Edit 2: So far working and tested: tRC, tRFC, tREF, Drive strenght, Romsip Custom 1.

Yay good news! Now we have a way to progess here. Test if you find some bugs to fix. I will create a new mod BIOS version with all tweaks (probably tomorrow).