The official Old School is Best School Season 4 Round 1 thread

[websmile]

So there are results which state dmi set to 0 but show 5x multiplicator plus fsb and valid? Are there magicians here? Any thoughts on this?

[macsbeach98]

Well thats simple DMI=1 I had to rerun mine because I forgot to set it.

You know what to do Michael they cant plead ignorance its in the limitations on the competition page.

[Strunkenbold]

37 minutes ago, websmile said:

So there are results which state dmi set to 0 but show 5x multiplicator plus fsb and valid? Are there magicians here? Any thoughts on this?

I think he meant that he had to unset it because he couldnt get a verification else way.

If DMI would be off, CPU-Z wouldnt read the chipset Registers (which it reads currently wrong) thus showing FSB frequency. Its very likely bugged anyway, Multi is very high.

I also see many Subs not having a picture of CPU and Motherboard...

[GRIFF]

From my tests DMI = 0 allows me to save validated files and validate it online. With DMI = 1 it gives me a freeze launched cpu-z.
As for the photos of cpu and motherboards I would say this: if the cup from the beginning was well organized would be eliminated but with the poor initial organization I would say to take with the pliers this rule only for this round, I hope.

[websmile]

So practically we can guess or throw a dice if a result is valid or not? I removed a 1m because the screen was missing and nothing else, happy discussion about the results and you better pick someone for next round who moderates this, bc I have big doubts Albrecht will do it and I am out.

[TAGG]

Why did my MMX 166 345MHz sbmission get removed from the comp? I ran DMI=0 provided pictures of board and CPU and even a video proving that this CPU is capable of 3x multiplier...

http://hwbot.org/submission/3804783_

Not that it matters for the ranking I just wan't to know why it was removed

[Mr.Scott]

1 hour ago, websmile said:

So practically we can guess or throw a dice if a result is valid or not? I removed a 1m because the screen was missing and nothing else, happy discussion about the results and you better pick someone for next round who moderates this, bc I have big doubts Albrecht will do it and I am out.

It's a shame that even when you spend a lot of time and energy to try to cheat proof a comp people continue to find ways.

Why not apply that energy to just pushing your hardware legit.

It blows my mind. 

[GRIFF]

For me in this cup there are no people who cheat. there are no prizes or middle money. and thank goodness we know each other now.

[Strunkenbold]

I also don't think that someone is intentionally cheating. Yet the problem is that CPU-Z bug. Actually you don't even think about there could be something wrong. 

I added the photo verification rule very early. Afaik there was only one participating at that time and that was GRIFF. His Submissions are correct, so actually everyone's else submissions should be correct too. 

Submissions without DMI off will be blocked. Those simply can't be trusted. And moderation team will not have the time to decide on every single submission if it is legit or not. Hardware categories will get cleaned from bugged results once new CPU-Z is out. I'm in contact with Franck Delattre and he already has a fix. Unfortunately it comes too late for us. 

Photo verification is a reasonable thing for those old hardware. I actually wanted a shot showing the markings of the CPU together with the motherboard. Unfortunately there were again some subs who didn't follow this rule. I handle this gracefully. But subs not showing any detailed CPU picture will also get removed from the comp. 

[TAGG]

@Strunkenbold Sorry about not combining the CPU and the board in one picture, I thought the CPU picture was purely intended so that other participants could see the batch...

I'm still not sure why my 345MHz submission was removed from % OC but not from max freq though...

Max freq submission:

http://hwbot.org/submission/3804788

% OC submission:

http://hwbot.org/submission/3804783_

I had it submitted in both stages as you can clearly see from the 11 comments on the %OC one

Might have something to do with the database beeing messed up, I submitted the max freq later...  If none of the mods removed it please add it to the competition again

[macsbeach98]

It was a fun comp well done everyone

The next time Socket 7 is used though no CPU frequencies its just too much drama with the validation benchmarks like SPI, Wprime and say Cine 2003 would be better.

[TASOS]

For this round , the percentage % stage , went wrong , cause of the cpu-z bug , especially on non-Ali chipset boards.

The good thing , is that Strunkenbold's move , contacting Franck Delattre about it , will provide us a better cpu-z for future use.

I thank them both (Gregor and Frank).

 

As a sidenote , just to laugh a little ... imagine all of us to have picked an other proposal for OSIBS (use of older 486 systems)

 

Personally i feel sad about making a bad choice for this stage ... my goal was to have fun and not cause trouble to participants and moderating team also.

Although i dont think anyone had the intention to cheat in this round.

 

I would have expected more of a cpu variety (regarding AMD K6's) in this round , but it seems that most of us played safe with Intel MMX , instead of experimenting with the AMD's.

Perhaps this round needed some more time ... cause of the general situation of Hwbot , the past weeks.

Anyway ...

 

I  am also sad , because my GA-5AX went bad during testing ... and now i have to try and fix it (if it's possible) or replace it (even harder).

My other board , a FIC , Via chipset , was constantly reporting 1X multi only (whatever Intel MMX cpu i used) and i chose not to present any scores , cause they looked weird.

 

Congrats to all participants ... and see you all in next round !!!

Have fun !!!

 

[Strunkenbold]

14 hours ago, TAGG said:

@Strunkenbold Sorry about not combining the CPU and the board in one picture, I thought the CPU picture was purely intended so that other participants could see the batch...

I'm still not sure why my 345MHz submission was removed from % OC but not from max freq though...

 

It doesnt really matter if you use a 200 or 233 MHz CPU for max OC. But it matters if you use a 166 or 233Mhz CPU for % OC. Therefore the requirement for the CPU picture verfication is higher.

I had the Feeling that some of you didnt notice that Ali boards are affected too. Im sorry, but those get moderated too IF they are Max or procentual OC related.

http://hwbot.org/submission/3764887_kotori_superpi___1m_pentium_166mhz_8min_18sec_537ms

[Strunkenbold]

Im done moderating. Im sorry for everyone which result got removed. 

@Gumanoid
Please pay attention to the CPU picture rule next time.

[Remarc]

On 02.03.2018 at 9:54 AM, TAGG said:

I have the 4.1 I'll share pictures and results of binning when I'm done

can you load pictures with vmod for this mobo (ga-5ax r4.1) pls?

[TAGG]

I didn't manage to voltmod the GPU Voltage because the FB pin is directly connected to Vgpu out, tried it by adding a resistor between vgpu out and FB and by adding a voltage divider between 5V and GND with FB inbetween, both didn't work, also the controller has a internal voltage refference and the maximum voltage you can program via VID pins is 3,5V anyways. The VCCIO mod is easy, just add a 10K ohm VR between the two outer Pins of LX8383A(voltage regulator above the BIOS battery)

[GRIFF]

3 hours ago, TAGG said:

I didn't manage to voltmod the GPU Voltage because the FB pin is directly connected to Vgpu out, tried it by adding a resistor between vgpu out and FB and by adding a voltage divider between 5V and GND with FB inbetween, both didn't work, also the controller has a internal voltage refference and the maximum voltage you can program via VID pins is 3,5V anyways. The VCCIO mod is easy, just add a 10K ohm VR between the two outer Pins of LX8383A(voltage regulator above the BIOS battery)

@Remarc For cpu voltmod need desoldering pin FB and Vsen from pcb, solder these pins together and at the center pin of a 1k trimmer. another trimmer pin solder to the GND and the third last to the Vcore

 

Edited March 17, 2018 by GRIFF

[TAGG]

Which pin do you mean by Vsen ? Do you mean Ifb?

[GRIFF]

7 hours ago, TAGG said:

Which pin do you mean by Vsen ? Do you mean Ifb?

Excuse me: VFB and VREF. Checked at HIP6004, HIP6021, RC5051 and another

Edited March 17, 2018 by GRIFF

[Strunkenbold]

I just checked my Pentium MMX 166Mhz today and the maximum multiplier I could get was 3,5x. Interesting, my 200Mhz MMX just had a maximum of 3,0x.

I also tried to reproduce scores who claimed they had DMI disabled, yet CPU-Z still displayed FSB and Multiplier. At least on my Epox MVP3 and 66Mhz FSB, setting DMI=0 completely prevents the bad behavior. So I dont know what they did.

[macsbeach98]

The 166MMX has to be dated earlier than week 40 to be unlocked mine is week 28 I would guess an early 200MMX would be probably unlocked to 3.5 too. If I tried 4 x multi with the 166 it reverted to 133mhz.

On the ALI chipset GA-5AA  setting DMI=0 takes away bus speed, multi and bios info in mainboard tab I couldnt say with the other chipsets.

I say next time socket 7 is involved just make it benchmarks SPI, Wprime, Cine 2003 or similar and not CPUz validations. Frequency validation is not really a benchmark anyway.

[Strunkenbold]

Thanks testing this also on Ali. 

Just from reading different sources in the net it indicates that 3,5 is Max and only applies to early samples. I can not give a technical explaination but I do think it depends on the internal wiring of some pins of the CPU, thus setting a multi of 4 and up doesn't do any difference because it would need another pin which the MMX completely ignores. In the end it's just another proof that those CPUs cannot get fully unlocked and every result with high multi was CPU-Z bugged.

[macsbeach98]

With my first runs with DMI=1 that I had to redo I did SPI @ 115x3=345 but CPUz showed 100x3.5=350 with DMI=0 it showed 345. I did CPU frequency @ 100x3.5=350

I seen the sub you are talking about 66x5 there is no way that any of them can do 5x multi it could of been set at anything and I wouldnt believe the CPU speed either it was probably thereabouts but wouldnt of been accurate DMI has to be off in CPUz

Edited March 19, 2018 by macsbeach98

[varachio]

I read briefly the posts above...

I must ask you unfortunately from my side , why also my submission removed.

The multi is correct (Cpu 233/66fsd=multi 3,5) , the photo is there, the fsb (83 speed) is the correct, the screenshot is there, everything visible and clear, so what is wrong?

http://hwbot.org/submission/3795358_

I could submit a score of 330 speed but i could not succeed to validate it (cpu-z validation reverted always with rejected), so i prefered to keep the validated lower score and now is blocked as a result?

I really not get it. 

valid = valid

Not valid = Not valid

And I declare my result AS VALID!

[Strunkenbold]

19 hours ago, varachio said:

I read briefly the posts above...

I must ask you unfortunately from my side , why also my submission removed.

The multi is correct (Cpu 233/66fsd=multi 3,5) , the photo is there, the fsb (83 speed) is the correct, the screenshot is there, everything visible and clear, so what is wrong?

http://hwbot.org/submission/3795358_

I could submit a score of 330 speed but i could not succeed to validate it (cpu-z validation reverted always with rejected), so i prefered to keep the validated lower score and now is blocked as a result?

I really not get it. 

valid = valid

Not valid = Not valid

And I declare my result AS VALID!

Because of the above mentioned CPU-Z bug all results displaying FSB and multi are potentially bugged. These runs where made with DMI=1. The rules of comp explicitly stated that DMI must be off. Your submission violated this rule thus was removed from the comp.

We are in the bad position that we can't trust those specific CPU-Z submissions, even if the validator says it's valid. We have enough submissions in the database who proof that the validator verified even bugged results. Sadly this means I can't accept your result even if it may be completely valid and got verified.

I told Franck about the problem of the verification. I hope he has a fix in some days ready.