Frequency detection bug with old Socket 5 CPUs

Strunkenbold · September 4, 2018

We unfortunately encountered a CPU-Z bug beginning from version 1.69 which was released on March 19th, 2014 until version 1.84 which was released March 20th, 2018.
Basically the detection mechanism of CPU-Z for old Socket 5/7 Pentium's, Cyrix, Winchip, K6 CPU's was broken and produced wrong results.

Antinomy digged a little bit deeper, so here is what he found out:

On 4/18/2018 at 12:35 PM, Antinomy said:

O.K., started to check the CPU-Z bug. Looks like CPU-Z assumes stock FSB (lol) and calculates the multiplier. Whereas the logic should be the opposite or better none at all.

Booted MMX at 100*2.5, shows 100*2.5, that's easy. Then changed bus to 103, 95, CPU-Z stays still. Then I set 112*2,5x and CPU-Z suddenly changes to 3*100, a multi changed. And it shows 300MHz with 280 real clock. Definitely bugged. Will specify affected versions and CPU cores and report back.

On 4/18/2018 at 2:54 PM, Antinomy said:

Pentium MMX - versions 1.69.0 to 1.83.0 are bugged. Below and above are fine. Will check other CPUs.

Edit: same goes for P54C and Tillamook.

AMD K6 (both Little foot and Model 6) - versions 1.69.0 to 1.77.0 bugged. Starting from 1.78.1 are not affected.

I experienced this bug already 3 years ago and reported it to the CPU-Z bug thread here in the forums. When I checked results for the Old School is Best School competition I stumbled about the problem again, so unfortunate the bug wasn't been fixed in the last years. I contacted Frank Delattre from CPU-Z and he finally fixed the bug. One minor disadvantage is that we have now no longer FSB detection on these old chips. If you want to submit Motherboard frequency results, you have to use at least a K6-2 from now on.

I went through our db and saw some results which are affected by this bug. Too keep the integrity of our db, I will remove the following submissions:

I know that some of these score are perfectly valid, CPU frequency wise. But keep in mind that we are a small team and I we dont have time to discuss every single result. If a result cant be trusted, it will be removed. It took 4 years to fix this bug and another half year to clean the db.

GRIFF · September 4, 2018

Ok for me. Do you need to delete another subs, or subs of the OldScool competition?

PS: If you need help to check some funzionality or bugs and incompatibilities, I'm at your disposal.

Edited September 4, 2018 by GRIFF

max1024 · September 4, 2018

I'm NOT agree.

1. http://hwbot.org/submission/3331215_max1024_cpu_frequency_pentium_mmx_233mhz_350_mhz

It's normal result 100*3.5 = 350 Mhz under -35C, or may be you think that my Super Pi1M on 350 also fake? http://hwbot.org/submission/3331220_max1024_superpi___1m_pentium_mmx_233mhz_6min_32sec_134ms I'm on the HW bot more then 10 years and nkow about old HW all.

2. http://hwbot.org/submission/3457688_max1024_cpu_frequency_embedded_pentium_mmx_266mhz_500_mhz Read this via google translate https://overclockers.ru/lab/print/82458/retrokloking-world-fastest-pentium-mmx or may be AIDA 64 and other soft also have a BUG?

[/img]

And other my results.

I look Strunkenbold at this issue formally, not understanding the situation, but it would be worth it. It is very sad to see such a lack of competence.

Gumanoid · September 4, 2018

And I felt sore. What's happening on hwbot lately? who are we? Overclockers? benchers? or "screenshoters"? the results are deleted and blocked without a real reason! All perfectly understand that the result is correctly received, but the window is not in the wrong place = ban. What if I can not physically place all windows at 1280*1024 resolution? There are frames of the reasonable and human factor. Therefore, moderators are people, not bots. What prevents ask before blocking? But most of all enrages that everything is formal approach.The man presses the "block" button realizing that the result is good in itself, but there is a "snitch" who instead of overclocking looks at the results stupidly and finds some minor flaws. What's wrong with you guys? Ten years ago, there was no such crap. all overclockers bench for fun! By your actions you discourage motivation for new achievements. Is this the purpose of the hwbot today?

p.s. Turrican did not want the hwbot to be the way it was now. It's a shame...

Edited September 4, 2018 by Gumanoid

zesen · September 4, 2018

After such, hands drop and do not want to anything benching.
What happens to HWBOT?

Mr.Scott · September 4, 2018

8 minutes ago, zesen said:

After such, hands drop and do not want to anything benching.
What happens to HWBOT?

It is what it is.

There will always be people that will play.

It's the only game in town.

yosarianilives · September 4, 2018

And I felt sore. What's happening on hwbot lately? who are we? Overclockers? benchers? or "screenshoters"? the results are deleted and blocked without a real reason! All perfectly understand that the result is correctly received, but the window is not in the wrong place = ban. What if I can not physically place all windows at 1280*1024 resolution? There are frames of the reasonable and human factor. Therefore, moderators are people, not bots. What prevents ask before blocking? But most of all enrages that everything is formal approach.The man presses the "block" button realizing that the result is good in itself, but there is a "snitch" who instead of overclocking looks at the results stupidly and finds some minor flaws. What's wrong with you guys? Ten years ago, there was no such crap. all overclockers bench for fun! By your actions you discourage motivation for new achievements. Is this the purpose of the hwbot today?

p.s. Turrican did not want the hwbot to be the way it was now. It's a shame...

I think we have only ourselves to blame for this. If people didn't cheat then we could trust every result. This end up meaning we have to enforce the letter of the law instead of only intent.

Edited September 4, 2018 by yosarianilives

havli · September 4, 2018

If the scores are bugged, I don't see another way.

Leeghoofd · September 4, 2018

Lol the Bot ended up this way due to the acts of many of your fellow overclockers, who couldn't play nice and keep it fair. Because they all thought they would end up on the payrole of some big vendor and greed took over.

The verification screenshots are like they are and will be imposed as is. Sorry but the 1280x 1024 resolution is a mute call, I have done most of my 2D subs on that resolution. Secondly we are 2018 already aren't we ?

For another of you arguments: If you think we have the time to ask to 20-30 people daily to redo their screenshot or to ask for other verification , how on earth can we keep track ? The rules have been laid down by our predecessors, again in a reaction to the aforementioned reasons.

Also if we don't maintain a strict handling of subs and allow the score for person A , but not for person B... how big of a mess will that be after a few months...

I keep on repeating this but OC is a selfdestructive something... if everybody played fair we would not be here... if you have a score that is out of proportion, you know something is wrong, yet many take the chance and hope it will slip through... one day somebody will spot it and it will get removed...

max1024 · September 4, 2018

10 minutes ago, yosarianilives said:

I think we have only ourselves to blame for this. If people didn't cheat then we could trust every result. This end up meaning we have to enforce the letter of the law instead of only intent.

Who is the LAW? I remember the movie "Judge Dredd" and the famous phrase of Sylvester Stalone: "I'm the Law!". It seems to me that the games of the Gods are not relevant here.
... blah blah ... I talked with a friend and we took it and decided that it turns out that some results are not valid. Wow! And not all, but some selective ones, and only a few people know this universal secret.

Apparently those who decided so (and decided immediately for everyone) have exceptional knowledge in ALL old HW, and everyone else was sitting and drawing screenshots and photos of their results. They made stands, freezed the processors and everything for the sake of making a FALE?,... it is fundamental!

All benching at a certain time, not tied to the versions of CPU-Z. What version of cpu-z acted at the time of filing, this one was used, so it was relevant at the time of filing and only it could be accepted for validation.

I can say that ALL of my results are real, and are obtained by putting jumpers and settings in BIOS and corresponds to what programs shows!

I look and I can not understand everything logically we take a multiplier of 3,5 multiply by the bus 100 we compute 350. Or is there another mathematics?

Well, let the other, apparently some believe that 2 + 2 is five, but then explain how on the motherboard without the possibility of overclocking, having a pair of Pentium 2 MMX 233 get any other numbers except those close to 233? In fact, I have 231 MHz. I understand if that CPU-Z showed 587 MHz, then you can still dream about how such figures were obtained, but 231 on a motherboard that has a bus at 66 MHz and is no longer capable. With two 233 MHz processors, OMG we get 231 MHz - no words, to those who made the decision to delete.

http://hwbot.org/submission/3372942_max1024_cpu_frequency_2x_pentium_mmx_233mhz_231_mhz

33 minutes ago, havli said:

If the scores are bugged, I don't see another way.

I'm also for this! But first you need to also think about it, No?

36 minutes ago, Mr.Scott said:

It is what it is.

There will always be people that will play.

It's the only game in town.

This is not the word of a real overclocker, who thinks about his colleagues in the shop. Football is a team game, we all are the Team, if in the future you want to see one player on the field, please go on. In the past, this is not the only case and many overclockers are not satisfied with this state of situation. It may be instructed that many will go away deleting their accounts, I do not think that the Team of like-minded people will be better off.

yosarianilives · September 4, 2018

3 minutes ago, max1024 said:

Who is the LAW?

How about the mods?

Anyways, bitching about something that does not follow the rules being taken down when the rules are established ahead of time is silly. If you have to use a larger monitor or a gpu that supports higher resolution how is this any different than requirement to use windows 7 on certain benches? People bitch about scores being taken down when they knew the rules prior to subbing. If you fail to properly validate your submission then who's fault is it? It's not the mods fault, they don't run the bench for you. When you get pulled over for driving 65 in a 45 do you complain to the judge that the law is stupid or do you recognize your own mistake and pay the ticket? Seems pretty straightforward to me.

max1024 · September 4, 2018

6 minutes ago, yosarianilives said:

How about the mods?

Anyways, bitching about something that does not follow the rules being taken down when the rules are established ahead of time is silly. If you have to use a larger monitor or a gpu that supports higher resolution how is this any different than requirement to use windows 7 on certain benches? People bitch about scores being taken down when they knew the rules prior to subbing. If you fail to properly validate your submission then who's fault is it? It's not the mods fault, they don't run the bench for you. When you get pulled over for driving 65 in a 45 do you complain to the judge that the law is stupid or do you recognize your own mistake and pay the ticket? Seems pretty straightforward to me.

Hey, did you ever click on my results? Which monitor or GPU? At least read carefully and look at the links above. Enough already to write the general phrases, it's time to think carefully and to conclude that this or that system could work at the frequency or not. We are not discussing the resolution with the GPU, although this is also an important issue, and SPECIFIC results. Wake up.

Edited September 4, 2018 by max1024

Mr.Scott · September 4, 2018

How dare you call me not a team player or a real overclocker.

What you fail to realize is that, this site is not a democracy. It's not yours. You are a user and are to abide by the rules laid down by the owner of the site. Period.

If you don't like it, don't play.

That is basically your only choice.

FWIW, 1/4 of the listed subs are from my own team, so it's not like I don't have a dog in this race.

W9 stands behind the mods decisions. Like it or not, they are the law.

Edited September 4, 2018 by Mr.Scott

max1024 · September 4, 2018

7 minutes ago, Mr.Scott said:

What you fail to realize is that, this site is not a democracy. It's not yours. You are a user and are to abide by the rules laid down by the owner of the site. Period.

If you don't like it, don't play.

That is basically your only choice.

I have been following all the rules since 2007, and I can not understand why the rules are different for all and the criteria by which these rules are established and applied at different times in different ways. In this case, they have not yet answered my results specifically what is wrong with them. Just "wrong" does not suit me.

Mr.Scott · September 4, 2018

What makes you say the rules are different for all?

I don't see that at all.

If you see something out of line, help the mods out and report it. They can't be everywhere all the time..........unless you'd rather turn a blind eye and bitch later.

Edited September 4, 2018 by Mr.Scott

Strunkenbold · September 5, 2018

13 hours ago, GRIFF said:

Ok for me. Do you need to delete another subs, or subs of the OldScool competition?

PS: If you need help to check some funzionality or bugs and incompatibilities, I'm at your disposal.

There where some subs from old school competition which needed to be removed but those werent final scores.

Thank you for your offer, we need every help we can get.

Strunkenbold · September 5, 2018

11 hours ago, max1024 said:

2. http://hwbot.org/submission/3457688_max1024_cpu_frequency_embedded_pentium_mmx_266mhz_500_mhz Read this via google translate https://overclockers.ru/lab/print/82458/retrokloking-world-fastest-pentium-mmx or may be AIDA 64 and other soft also have a BUG?

I dont know why you are thinking that you are not affected by this bug?
Your link says that you were running this CPU with 124x4 resulting in 496Mhz. Yet your score on hwbot is 500Mhz with multiplier of 5 (correct me if Im wrong but this multiplier is not possible with this CPU).

Now just think about someone who tries to beat your score with the current (bug free) CPU-Z version. If he just do the same like you did (124x4) = 496Mhz he would be always second place. Yet he scored actually the same. Wouldn't he deserve to share the first place in the ranking with you?

I would really recommend to read twice the quotes of Antinomy. He perfectly describes how changing the FSB can result in bugged scores. The variance between the actual clock speed can be 0.5 MHz up to 40 MHz. In any case its bugged. And as I dont have the possibility to check what your actual speed was, the scores got removed.

Strunkenbold · September 5, 2018

12 hours ago, Gumanoid said:

p.s. Turrican did not want the hwbot to be the way it was now. It's a shame...

Yes its really a shame.
Its a shame that this bug could last for 4 years without anyone, except me, contacting CPU-Z authors.
Its a shame that nobody stands up when he faces those bugs.
You guys say all the time how experienced you are with old hardware. Yet when boot screen says 280 MHz and CPU-Z says 300 MHz you keep calm because its an advantage for you. As you are all so "experienced" you dont think its strange when you set an FSB of 124 but CPU-Z says 100. You dont think its something wrong when old Pentium MMX CPUs have high multipliers? You open AIDA and CPU speed differs, you dont wonder why?
I really wonder how I could spot this problem with my "non-existing competence". What a shame for you.

And thats the real problem. You guys have the experience with those old platforms. So none of you can tell me that you didnt stumbled about this problem in the past.
Yet none of you started a thread here in the forums or contacted CPU-Z. There are and there will be always software bugs.
You have to understand that it is your responsibility as community to help to fix those bugs.

But you just did nothing. A now you are starting a rant why hwbot is so evil because we removed those subs. How could we dare without asking you?

It makes me sad that some of you fail to see the bigger picture. Its not that I had fun when I removed those submissions. It makes work.

You have to understand that results which cant be trusted because of a known software bug cant stay in the ranking even if it was made with motherboards stock clocks. It also makes no sense to ask for further proof as the rules ask for a CPU-Z validation. They dont ask for an AIDA screenshot. So no trusty CPU-Z validation no score. Period.

I cant tell how Turrican would handle things. I just want to say that in the last years many things have changed.
I just say Windows 10, EVGA SR-2 timer bugs and we currently face a lot of similar issues with buggy benchmarks. And sadly a lot of users who thought they found a "tweak" while it was just bug using.

Your reactions is making me really sad and I don't know what it justifies to spend a lot of time in such a toxic environment. This makes me understand why the whole moderation team is basically gone.

Gumanoid · September 5, 2018

12 hours ago, yosarianilives said:

I think we have only ourselves to blame for this. If people didn't cheat then we could trust every result. This end up meaning we have to enforce the letter of the law instead of only intent.

I agree, but cheating and making screenshots are a bit different things. for example there are 3 results, in the first two cases the screenshots are made completely by the rules, and in the third there are minor violations and the moderator understands this. Enrages that a person does not try to understand what caused these violations. Approach to the issue is your problem, it's wrong. Each situation is individual and you can always ask for additional confirmation and not immediately ban.

12 hours ago, Leeghoofd said:

For another of you arguments: If you think we have the time to ask to 20-30 people daily to redo their screenshot or to ask for other verification , how on earth can we keep track ? The rules have been laid down by our predecessors, again in a reaction to the aforementioned reasons.

Also if we don't maintain a strict handling of subs and allow the score for person A , but not for person B... how big of a mess will that be after a few months...

I get the impression that we are talking about different things. If the overclocker was caught in cheating - I would ban him forever. But if there are no questions to the result, but there are shortcomings in the screenshot, why not specify them at once? Why ban after a week / month / year? People see, wait until the time runs out and then complain - is this fair play you think?

Gumanoid · September 5, 2018

55 minutes ago, Strunkenbold said:

You guys say all the time how experienced you are with old hardware. Yet when boot screen says 280 MHz and CPU-Z says 300 MHz you keep calm because its an advantage for you. As you are all so "experienced" you dont think its strange when you set an FSB of 124 but CPU-Z says 100. You dont think its something wrong when old Pentium MMX CPUs have high multipliers? You open AIDA and CPU speed differs, you dont wonder why?
I really wonder how I could spot this problem with my "non-existing competence". What a shame for you.

Personally, I did not encounter this problem or did not pay attention. I do not have many results with processors below K6-2. Disturbing is not precisely this situation, but the situation as a whole, all this negative atmosphere. Personally, you have no complaints, you do your job and it's always pleasant to compete with you.

TerraRaptor · September 5, 2018

Why claim yourself an overclocker if you get sad with the removal of points?) In most cases it was just a usage of an exploit and one being honest would at least put a comment on his own result that it could be a bug. Staying silent for such a long time led to removal of results. A time to scatter stones and a time to gather them - that simple.

daimons · September 5, 2018

ok, so let's reconsider all the results in regard to the compliance with the rules up-to-date. The availability of screenshots, photos of stands, references??? Is the law going to be applied to everybody and everywhere? This is your great radical sense so far? Is there any sense in these actions right now?

yosarianilives · September 5, 2018

Just now, daimons said:

ok, so let's reconsider all the results in regard to the compliance with the rules up-to-date. The availability of screenshots, photos of stands, references??? Is the law going to be applied to everybody and everywhere? This is your great radical sense so far? Is there any sense in these actions right now?

It is the only way. If a rule is applied you can't selectively choose the ones you "feel" are cheating. That would be a can of worms that nobody wants to open. Much better to just make a blanket statement and say that all results of a certain cpu with a certain version of cpuz could be bugged.

Gumanoid · September 5, 2018

Guys, cheating and a non-ideal screenshot are two different things, but now on the hwbot between these things the sign of equality.

Leeghoofd · September 5, 2018

you got proof for such bold claims...

Frequency detection bug with old Socket 5 CPUs

Recommended Posts

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

Popular Posts

Strunkenbold

Splave

Leeghoofd

Join the conversation