Jump to content
HWBOT Community Forums

Frequency detection bug with old Socket 5 CPUs


Strunkenbold
 Share

Recommended Posts

11 hours ago, Strunkenbold said:

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. 

I carefully read all your arguments and want to say the following. I've tried almost all the sockets, chipsets and processors since socket 3 and I can say that the final FSB frequency depends on the mass of factors such as the actual capacitor power, the degree of their wear, the current that is applied to the main elements of the board including the cloker, the temperature that acts on the elements of the motherboard. As a result of these and other physical processes, FSB often "swims", even in real time. The deviation from the set frequency by jumpers can exceed 1 MHz. I think many are aware of how from the "tired" motherboard  in addition to squeeze out extra performance. For those who do not know, I tell: need to give a good load on the processor, running a benchmark. As a result of increasing the voltage at all nodes of the system, the FSB will also be tightened, depending on the physical state of the components of the motherboard and even the used power supply, which can deliver different currents along different lines, both to the large and to the smaller side. Somewhere such a method reduces performance and the FSB fall sdown, even downt the limit. CPU-Z in some cases just creates a load on the processor. As a result, from 66 MHz, you can get 67 MHz or even 65. All my results are in this corridor. This is the normal behavior of old systems.

If you look at the result of other utilities such as AIDA 64, NOBODY knows the algorithms for changing the CPU frequency of this and other utilities, and there are no guarantees that they are not subject to the same errors as certain versions of CPU-Z. Therefore, one should not assert anything without knowing the exact Facts.

I understand that Antinomy also owns the information, but let me refer to the opinion of one person is not reasonable, there is still logic, physics, mathematics and what only the creators of programs and processors know. And if the frequency as you write differs by 0.5 MHz, is this a Bug? This stupid frequency, as I wrote above, can be different at any moment of the time, and a competent OVERCLOCKER who knows 100% of the behavior of his old components will make the right decision to increase this frequency by different methods. Bug is when exhibited 150 and shows 450, for ALL of the my results, the measurement error does not reach even 5 MHz. If you say there is not 500 and 495 MHz, different tests have different load, who wants to do better - will take the next FSB stripe.

5 hours ago, yosarianilives said:

It is the only way. If a rule is applied you can't selectively choose the ones you "feel" are cheating. 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.

As for the application of rules, laws and so on. I consulted today with lawyers and that's what they told me. In civilized, democratic countries, when adopting a new law or rule, this law has no inverse legal force. In jurisprudence, there are cases when, when adopting new rules and laws, its validity extends to legal relationships that arose before the adoption of the law. In such cases, the civilized and democratic norm is the indication of all aspects and terms from when the new legal relations will affect the old order. Typically, such situations arise when one law is canceled, and the second has not yet entered into legal force and for this period (for this period) when the old law does not work, and the new one has not yet been adopted, the retroactive rule on legal relations arising in the past . In this case, in advance in the mass media, this information should be published in public access.
What I see here, no news on this fact was published on the main page of the site, no information came to the email. The present edition of the rules still applies, no transitional periods are observed. From the point of view of common sense and the law, such actions are simply not legal, not democratic. If you follow such logic, then the first idea is to delete ALL results from Cinebench 2003-R15, where the screenshots close the visible part of the screen. But this is not done. Further it is necessary to delete all results without screenshots (which were lost when moving the database). We want to establish the same order? Using the same approaches?
If the administration intends to introduce NEW RULES, first publish these rules for ALL. Write a date from which these rules will apply and which versions of programs or tests to use for each family of processors. It will be fair to all. Old results, as the results without screenshots or closing the windows of programs leave as is. There will still be many attempts to establish more than one record and the existing results will be improved, but there is no need to touch, then a lot of energy, energy and time were spent.
As for the help, I am ready to help in identifying and correcting the wrong results. I often look through the old results and see as much garbage. In a series of cases, which can be corrected by replacing the Mhz by Ghz, or by moving the comma several characters back. Write to me in PM I do not refuse help.

  • Like 8
Link to comment
Share on other sites

I have results with this bug, made an 115x3 instead cpuz marks 100x3.5. It's a 100% bug.
However!!!!!!!!
We must be careful to remove the outdated results because behind each result there is time and money lost. And we as bencher do not have the blame because cpuz did not work properly 2-3 years. Or did you have to wait? In these cases the minimum is to write the annotations in the submisions with explanation of the real settings. More than anything else for the following generations for those who want to refer to different results and make certain researches.

I personally do not lose anything but some of us lose weeks and months of work.

I'll give you a simple example: You build a two-storey house with red bricks after two years out that the two-storey red-brick houses could not be built in that area. They demolish it. You like it? Are you happy?

  • Like 3
Link to comment
Share on other sites

what the hell is going on here again?deleted my 3 result but nothing say to me and deleted 37 (Сarl!) result of many other people...people who use ln2 or dry ice, who was wasting their time, binning a lot of processors,who could or have already burned them cpu that would reach 350mhz or higher and then you take and delete everything,just like that...it epic face palm,why such disregard for people? me and others what to do now? cpu-z have bug?ok why did you allow this to be used? why you not allow some other program?you need validation? (cheaters everywhere) ok,why not write your program for old hardware?(gpupi,wrapper aquamark,x265 etc you can when you want)

hwbot 2018 for any participant or it now fanclub w9 and them henchmen's but the other nobody and nothing? i watch,them like despotism,no democracy...no wonder they so aggressive every time.

Edited by Remarc
  • Like 3
Link to comment
Share on other sites

  • Crew

Okay guys I restored now all of your results because its simply not worth it to discuss this any further.
If you would provide a minimum amount of fairness: Please write down your actual FSB / multiplier combination in the comments section and correct your score accordingly. If you dont know it anymore, please be honest and delete your score. If you dont, its also okay as whole result db is already a mess with or without those scores doesnt matter anymore.

BTW, as these were almost _all_ Air or Water based results (except one SS) despite someone claiming Dice and LN2 in the time you write lengthy essays, contact your lawyers (zu viel Tagesfreizeit?), try to explain why possibly any CPU-Z result can differ up to 20% on those old systems and explain why no one never faced this bug despite everyone continual claiming how much time they invested on those old systems, you could simply put back your air coolers on that systems and make a new validation with current CPU-Z.
Its really funny to see how creative people can get when they need to find excuses. 
Somehow I feel that you guys miss entirely that it just was just about the CPU frequency scores. Its not like we moderated SuperPI scores or something despite these scores also show wrong frequencies. Or in other words these were 'just' 37 submissions with old hardware which you can get for free from scrapyard or for very little money on ebay. Of course I know its still hard work and its not a great feeling when this work disappears over night. But what really makes me thinking is the fact that some of you still dont understand that this is something which needs to be done to provide a fair competition. But everyone just look what he looses. No one sees what the community gains by clean rankings which can be trusted.

About the screenshot issue: If you would mind to take a look at rules section of the Cinebench benchmarks, you see an example image how your result screen has to look like. And there you see that the render scene is always completely visible. Thats the red line around the CB window btw. So this is no new rule. The only thing which changed was the way how we enforced this rule. In the past we just approved those scores despite we knew that they were actually against the rules. After it came to our attention that some people altered the rendered scene we changed our behavior and moderated all submissions with overlapping windows or similar. Because we dont have the men force anymore to reproduce suspicious results, we are stricter now for good reason. Old subs stay in as we dont have the time to went through every category and clean results but if we are in doubt about the correctness of a score we moderate it too without thinking too long. Because you agreed to the rules, it is your duty to obey them. 
Edit:
In fact it seems Christian Ney changed the example screenshots in February last year. I have to admit that there wasnt any news about this which is shame. On the other hand this rule is now over one year old. Actually time enough.

To quote our general rules:

Quote

2. Benchmark Submission Verification

As you probably understand, claiming a certain benchmark score is nothing without having the proper verification to back your claim up. At HWBOT, the verification plays a bigger role than on, for instance, your overclocking forum. Why? First of all, our competition contains more than 35,000 overclockers from all over the world, which means that most likely you will not know most of the people who you are competing against. The biggest problem of the unknown is … the fear of the unknown, or better put: because you don’t know the person behind the nickname, there’s no trust between you and that person. This implicates that chaotic entries with messy validation will probably be considered illegal rather than legal, although the opposite is true in most cases.

Therefore, we have set a certain amount of rules of verification. Please understand that these rules have been evaluated and discussed more than once. Furthermore, they are meant to serve YOU, not the HWBOT moderators; in other words: the rules are there to make the game more enjoyable for you. Not sticking to the rules may result in a score being blocked or, in worst case scenario, you account being suspended.

Quote

HWBOT moderators may apply a not suspicious exception when moderating scores which do not earn global points. When a submission has minor mistakes or incomplete verification, it does not mean by definition it has to be blocked by a moderator. If these submissions are not suspicious in any way, and do not receive global points, the moderator has the right to approve them. This rule is to prevent users reporting each and every submission which have minor mistakes, in order to get a better rank for their own submission. Reporting submissions which are not suspicious but have minor mistakes is against the fair play spirit, and abuse of the reporting functionality will be acted upon.

So clear speaking we have the right to approve your submission. But its not that you have the right to say we must.
This is not democracy, we have no democracy on hwbot. Keep in mind that this is a service where you agreed to accept its rules. It is a free service, we don't get any money for this. You _can't_ compare this to a government who needs to care for his citizens.
And that is the reason why we need to make decisions which no one likes but are essential to keep this service running. As I said already before we are a very small team and need to take for thousands of submissions and its hard to keep up just with the reported stuff and competitions. How do you think this all ends if we need to permanently think about which submission with a verification issue we approve or not approve? Especially all those discussions afterwards "Why do you accept submission from member A and moderated submission from member B". This is just nothing what a free service with a small team is able to provide.

I have also big headache just when I think about those numerous issues the bot has. Lots of unresolved smaller and bigger bugs that exists for years, a single developer left that you need to beg just to fix at least the high priority ones if at all, the Windows 8/10 issue where no one found a proper solution yet, cheat apps for benchmarks, bugged benchmarks, users finding bugs and marketing it as a tweak and and and...
If there will be ever again a strong team behind hwbot trying to fix those issues, it wont be possible without cleaning the result database, it wont be possible without removing buggy benchmarks, it wont be possible without people loosing points or results, I wont be possible without making some people angry but if it helps to consolidate the bot, it needs to be done.

But if its already a problem for such a crystal clear case and such a low amount of subs like this one I see no future for the bot.
And that is the reason I retire today from hwbot duty.
There is simply no point in spending any more time on this.
Id like to thank all the guys who have the understanding how a community works and how its necessary that the rankings arent a chaos and stay meaningful.

I would advice the wise guys to write an application to Alby for volunteering as HWBot admin.

After all those years I'm out,
sincerely yours,

Gregor 

Edited by Strunkenbold
  • Sad 19
Link to comment
Share on other sites

46 minutes ago, yosarianilives said:

Too much more of this and we'll run out of mods. How do the members not see that this self destructive trend cannot continue for long?

Just my 50cc - you see now former USSR is not compatible with democracy?)

 

Strunkenbold thanks a lot for all your efforts and passion!

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

This topic received a fairly wide response, a lot of dissatisfied with the overclokers, some even left their teams. It is sad. All the below written will reflect the collective point of view of Overloekers from different teams, I just summarized the information.

Question 1. Why do people start to write on the forum only when the administration blocks their results and why they do not read the forum regularly.
Answer. The main problem is the language barrier. Not everyone is fluent in English and has to spend considerable time translating text with the help of translators. Text information in a foreign language is not instantly perceived. That creates a lot and highlight the most important information and translate it is not for everyone. (Try to read personally foreign forums) Second, the difference between time zones and the specifics of each work. Not everyone can hang at the forum all the time, someone comes there once a day, someone once a week. Among the overclockers there are a lot of people who, due to their profession, can not be constantly connected to the Internet. Many overclockers wrote to me that they do not go to the forum on the above mentioned problems, they just do overclocking and spread the results. I hope I understandably wrote the reason with the help of an online translator, why things are in a number of cases just like this. I apologize if something was not completely correct.

2. Culture of communication.
During the discussion in this topic was written a lot of bad. Some users allowed themselves outright rudeness and unprintable replicas. This is very embarrassing. The rules of decency are the same in all countries and continents. Often, there is no due respect, not all oveklokers for 18 years, there are people who are no longer young and have exchanged the 5th dozen - show respect. This topic is read not only overclockers but also news editors of specialized sites, as well as representatives of companies of component manufacturers. This behavior has a bad effect on the reputation of the entire oveklokerskogo community, which clearly does not benefit anyone. Insults are not permissible in principle, for this you need to give a ban. We call for a normal discussion.

3 Order.
If you go to the link http://hwbot.org/article/about_us, you can see that the links do not work, the information is outdated. The news is also not updated, but it's half the trouble. The trouble is in the rules, screenshots and much more than that.
Propose 1. State the rules in a new edition. Make a link to them on the main page. 2. Do not use new rules that will affect old results. 3.All questions that in the future will affect the change in the rules to notify everyone of the news on the main page and make a mailing to all users by email and in private messages on the forum. 4.Review all links to downloadable benchmarks (some do not work). 5. Make a list of all the errors in the tests so that any newcomer can see in the example of the same SR-2 what can be done and what is not.

4.Democracy, Help and a way out of the situation.
Democracy should still be, in conditions of dictatorship and the imposition of opinions of individuals on the interpreted rules, in general, nothing good will come of it. More precisely, it did not work out as many overclockers were simply disappointed in this project, but not everything is lost. Everyone can make mistakes from this, no one is immune. Yes, decisions must be made, but it is better to discuss them in advance, notifying everyone of this. To facilitate the task of establishing order in the database, it is possible to call for volunteers from among the recommended users to grant them the right to adjust the results. Only the correction of minor errors without the possibility of removing it? I think that there are people who want to help.

  • Like 3
Link to comment
Share on other sites

I agree, Strunkenbold is doing a great job to keep the bot running.

Sometimes when problem like this CPU-Z situation shows up... action simply must be taken. And when we don't know if the scores are false or real with those certain CPU-Z versions, then removing all scores possibly affected is the only way to keep the DB in good state. Of course it is some amount of time and effort wasted for those people affected by this. But this isn't really anything new - several benchmarks were removed completely in the past (for example UCBench and others which I don't even remember anymore). Other had points removed (PCM05, Hwbot Prime). I alone lost hundreds points by this... and I'm still here. Cleaning the bugged results from the DB is a good thing and everyone shoud understand it.

  • Like 4
  • Thanks 3
Link to comment
Share on other sites

This is a sad state of affairs and I have to say "Thank You" to Strunkenbold, it's not everyone that's willing to take on the task in this situation - In fact a heart-felt thank you is due to all that's stepped up and made things work up until this point.

I have to echo Splave's statement, the staff are the ones that's supposedly running the show, making the decisions and whatever else - And I'll just say it, it's past time to say what you mean and mean what you say or else related to these problems.

 

 

Edited by Bones
  • Thanks 1
Link to comment
Share on other sites

  • Crew

Thanks Gregor for all the work you did. Much respect bro what you, Carl, PJ, Frederik, Michael and Chris pulled off, to make this free service enjoyable.

I withdrew from the conversation as K5 and all this old hardware is not my cup of tea, however if I look back on 2018 alone many of the OSIBS subs have been heavily discussed over and over again regarding the the validity of many submissions or the hardware being used. You can draw your own conclusions on who's too blame...

Seems the centrum of the universe nitwit wins again by using an abundant, probably in his point of view a democratic usage of words. Give yourselves a big cheer and a tap on the back. Another amazing achievement:  Self Destructive mode is fully unlocked.

 

However this unexpected departure means the bot's evolution might come to a halt, and this from right now: There might not be any more future hardware additions to the Bot's database.

 

Thanks to those that supported Gregor and the BOT for its free service. To the others;  if you don't like it here, why don't you go and play somewhere else where it is much better structurized, better moderated and where you can all have your little saying ... Just let us be happy in our elite biased bubble....

Democracy stopped right here, right now... if one ever pulls this word out of the hat again or starts to copy paste the entire wikipedia to waste my time and lecture me, I'll ban him without even blinking... You all have been officially warned.

 

  • Like 8
  • Thanks 4
Link to comment
Share on other sites

No, it's not a benchmark. Suicide runs are very hard to validate because the result relies 100% on a few MSRs or CPUID registers that are read and interpreted. Remember the FM1 locked multi bug? Very satisfying to see a CPU go as high as you want - but it was only the number in the CPU-Z window of course, nothing real.

I'd say it's an impossible job to keep the DB clean in this category. It should be discussed if points are removed there. This is only a marketing category anyway. "Wow, 2999WX on 6GGGG ALL CORES!11" ....yeah.

Strunkenbold, the way I see it, it's impossible to keep all categories clean that rely on user input, only screenshots and simply not enough and unsafe data points.

Edited by _mat_
Typo smiley removed.
  • Like 2
  • Thanks 2
Link to comment
Share on other sites

5 hours ago, Mr.Scott said:

Greg, I can hardly blame you. I told Michael the same when he quit. Huge loss. Thanks for trying.

Again, everybody loses because of a few pukes here.

I hope you fucking clowns are satisfied. If I posted everything I really wanted to say, I would be banned for life.

 

 

Aye captain 😐 Too much passion too much upset

_________

In general writing. This cpu-z thing is minor concern on old hardware. So we have moderator guys blowing up all over the place recently.

Dudes need be bothered on the bigger problems.

________

Strunkenbold. Come back when your are chilled out man. 👍 See from writing above you have support here

Link to comment
Share on other sites

thanks Strunkenbold for your tireless efforts over the years. You've been nothing but fair and it's a shame some in the community can't see past their self importance and recognise this.

Here's hoping you have a change of heart and the whingers learn to take it on the chin for the good of the community, the hwbot database and overclocking in general.

hwbot is a great project let's hope it continues.

  • Thanks 1
Link to comment
Share on other sites

  • Crew

I'd like to remind you what HWBot was initially about - it's a hardware performance database. Think about it, reread it and think again. It's only three words, you can handle it, I'm sure.
Points, rankings, leagues, teams - it's all secondary. It's simply a gamification made for fun. Fun is another thing most forget in their long discussions.
So the main purpose is to keep the DB as accurate as possible, not about your points.

@max1024it's NOT about crystal oscillator deviation you are talking about AT ALL. You simply don't understand the bug then. Xtal deviation affects frequency and will be seen in every realtime frequency utility. The topic is about CPU-Z bug only. If you didn't understand my two little quotes, feel free to ask for details. It's not rocket science.

@Gumanoid, looks like you don't get the situation. The results were blocked not because everyone is a cheater but because we can't be sure they are real.

How dare you mention Carl and talk about rules applied to old results.

I'll give all of you a nice example - a long time ago there was a category called Celeron 350MHz (Covington). You'll never find such a CPU because it doesn't exist. I've found out that CPU-Z couldn't tell a Pentium 2 Deschutes with L2 cache disabled in BIOS from a Celeron Covington (they share the same core) which doesn't have cache at all. How do you think, how many people commented that their scores were bugged, how many reported to CPU-Z? Don't try to "they might not know" - you can't disable L2 cache in BIOS by accident. And Pentium cartridge and PCB looks whole different way from Celeron. So Carl deleted the whole category along with results. And I've reported to CPU-Z and got this issue fixed. And nobody got banned. Some of them were teammates of those who participated in this topic. Would you really continue with this "rules don't apply backwards" and let fake category with false results stay?
Important to not - results weren't painted, neither they were fake in general way. But they were erroneous in terms of hardware performance database.
That's how it was done and how Turrican reacted in such a case.

Back to this issue - it should be sorted out. Maybe we could make exceptions for some cases like non-overclockable boards if a result seems normal and doesn't cross the bug mechanism. Not for me to decide though, it's up to results mods.

  • Like 4
  • Thanks 5
Link to comment
Share on other sites

I want to express my gratitude to Gregor for his work and regret for this decision.
I wish more confidence in my actions and do not take so much to heart claims from users.

Antinomy,  дружище,  imho ты путаешь теплое с мягким. Сообщество сейчас не то. Сравнивать его с зарождением движения - в корне неверно. Равно как и говорить, что здесь всё будет только так как скажут модераторы. 

Популярность нашей дисциплины падает, здесь остались только те кто бенчит всё что движется и те кто на этом зарабатывает. последние своим давлением грозят остаться одни на этом форуме...

Жаль...

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

  • Antinomy featured and unfeatured this topic

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...