Upcoming Rule Changes on November 1, 2017

Via this news article the staff is announcing a couple of rule updates which go in effect on November 1, 2017. The rule updates relate to benchmark version restrictions as well as general improvements to include more details in the overclocking results.

  • Rule Update #1 - All Global Top-20 benchmark results must now include a valid verification screenshot and a system picture showing the benchmarked system in action.
  • Rule Update #2 - All verification screenshots must show the settings at which the benchmark was completed. This means it is not allowed to decrease the operating frequency for screenshot purposes. In effect, the staff reserves the right to reject a submission that is deemed out of line for the stated system parameters. For XTU specifically, the data file must contain monitoring information (frequency and temperature).
  • Rule Update #3 - All users are obliged to declare the original source of the hardware used for a submission in case they do not own the hardware. In effect, this is mandatory for all Engineering Sample hardware as they are legally not for sale.
  • Rule Update #4 - Joined accounts, that is an account used by multiple people, are still allowed. However, it is mandatory to disclose who achieved a specific result in the description field. Failure to disclose will lead to removal of the submission.
  • Rule Update #5 - XTU v6.4.1.12 is mandatory from Nov 1, 2017.
  • Rule Update #6 - HWBOT X265 Benchmark v2.1.0 is mandatory from Nov 1, 2017.
  • Rule Update #7 - It is not allowed to publicly, falsely accuse a user of foul play. In case of false accusations the accuser may see their account temporarily or permanently suspended. Foul play is interpreted in the broadest sense of the word.

If you have any questions, feel free to reach out to any of the staff members.


54

France Niuulh says:

Nice change !

France bolc says:

"XTU v6.4.1.12 is mandatory from Nov 1, 2017"

Does not this mean that benching anything older than Haswell (Haswell included) becomes impossible ...?

websmile says:

You might have to re install OS. I asked Luumi to check new version and he did the 853 at sandy bridge 3470 today with new version he said - so there is hope :)

superpatodonaldo says:

Good; just two questions:

Rule Update #1 - All Global Top-20 benchmark results must now include a valid verification screenshot and a system picture showing the benchmarked system in action.

system picture = a photo included mobo-vgas-cpu cooler etc? no more accepted photo with just a vga laying on a table, or so?
do this will be applied for future competitions (when a system picture is required)?
if possible please add some example, a valid photo and a non-valid photo, thank you :)

Rule Update #2 - All verification screenshots must show the settings at which the benchmark was completed. This means it is not allowed to decrease the operating frequency for screenshot purposes. In effect, the staff reserves the right to reject a submission that is deemed out of line for the stated system parameters. For XTU specifically, the data file must contain monitoring information (frequency and temperature).

about operating frequency I noticed that in win 8.1 cpuz shows correct ratio and bus but cpu freq seems to go down also with c-state and speedstep disabled; it is accepted or not?
for example one of my last result (see cpuz cpu): superpatodonaldo`s Aquamark score: 603640 marks with a GeForce GTX 580

Czech Republic havli says:

#1:
So for rankings like this Y-Cruncher - Pi-1b overclocking records @ HWBOT from now on I have to take photo of the system? Well, this will be a lot of pictures... as with these CPUs almost any submission hits top 20 global with less popular benchmarks :D

#3:
I'm not sure I understand what this means exactly. Let's say I buy bunch of old ES on ebay and bench them for fun / cups / HW points... outside of competitions of course. So what changes for me starting with 1st Nov. 2017?

United States yosarianilives says:

superpatodonaldo said: Good; just two questions:

Rule Update #1 - All Global Top-20 benchmark results must now include a valid verification screenshot and a system picture showing the benchmarked system in action.

system picture = a photo included mobo-vgas-cpu cooler etc? no more accepted photo with just a vga laying on a table, or so?
do this will be applied for future competitions (when a system picture is required)?
if possible please add some example, a valid photo and a non-valid photo, thank you :)
[/url]


Further on this question will we require enthusiast and below to show their rad in picture if they're not using air cooling?

Czech Republic Elkim says:

havli said:
#3:
I'm not sure I understand what this means exactly. Let's say I buy bunch of old ES on ebay and bench them for fun / cups / HW points... outside of competitions of course. So what changes for me starting with 1st Nov. 2017?


Keep invoice and pray for seller's soul.

websmile says:

I checked new xtu on gene 5 and 3770k/ OS windows 10x64 = works
On the rig picture, after we discussed this we want a real picture of the system in use or at least fully built with cooler etc. This is a general decision also for competitions in the future, we need to have a unified guideline again, no matter if we really need it or not. I am not sure we will ask for pictures for all competitions in the future though, sometimes we simply do not need it.
Your result is a borderline case, I can see multiplicator so I know it is 6800 but we prefer full speed settings. I will talk with massman if we accept speedstep
Top 20 means top 20 xd
Old ES is not what we aim for, we have too much complains about new gen, wrong accusations of hardware sharing and we want to end this by more transparency. We will clarify this on old es, but in theory ebay is enough

United States yosarianilives says:

websmile said:
Old ES is not what we aim for, we have too much complains about new gen, wrong accusations of hardware sharing and we want to end this by more transparency. We will clarify this on old es, but in theory ebay is enough


Glad to hear, am currently looking for ES g34 opterons and was worried it may become a pain to sub.

superpatodonaldo says:

well, about my result I can't understand why cpuz shows lowered freq, could it be due to BIOS?
disabled c-state and disabled speedstep should force cpu to run full freq....
and it occurs only in win8.1

United States yosarianilives says:

superpatodonaldo said: well, about my result I can't understand why cpuz shows lowered freq, could it be due to BIOS?
disabled c-state and disabled speedstep should force cpu to run full freq....
and it occurs only in win8.1


Stupid question, but you have windows set to max performance power plan right?

superpatodonaldo says:

yosarianilives said: Stupid question, but you have windows set to max performance power plan right?


Question never is stupid, I have to check

Poland phobosq says:

Does #1 and #2 apply to CPU-Z based benchmarks (frequency/ref. clock/mem clock)? I'm asking as it might impact scores, getting a screenshot at DDR1 400+ is usually 10-15 MHz lower than max valid, I'm pretty sure the same applies to reference clock on 775.

Sweden Rauf says:

Good rules in general. I really like that you have to show correct frequencies. But do you want to block 90% of all 3D scores? Not many (of the less experienced users) even know how to show actual frequencies in GPU-Z. Also speedsstep will cause lots of trouble if it is enforced. What about systeminfo for newer FM benches? Does it have to show correct frequencies? Do we need sys info monitoring enabled during bench? But system picture and source of hw adds even more complexity to make a submission. And totally unnecessary. What good does it do? How can you verify it? Rules that you can't verify are just useless...

websmile says:

CPUZ might be an expection, I do not think we need the screenshot for this.
On frequencies we mainly talk about cpu and downclocking for screen, this is not allowed anymore. We do not talk about the rookie league but global top 20, so showing full cpu clocks should work. The 3D clocks we still have to discuss again, but it makes not much sense nowadays in my eyes.
On the ES issue, this is caused by the community´s permanent accusations of people sharing hardware. And yes, we can ask vendors about es if we doubt source of a sample

United States rtsurfer says:

Rig pics, ughhh..... :(
HATE IT!!!

Maybe you guys should fix your image size too large error to actually show a properly coded error before you go onto apply that rule. I thought the issue was fixed, but then I ran into it again on Team Cup.

United States Splave says:

If the server is flooded by a couple comps ending what will adding 3x the data to every single sub achieve site performance wise.

United States Gunslinger says:

rtsurfer said: Rig pics, ughhh..... :( HATE IT!!! Maybe you guys should fix your image size too large error to actually show a properly coded error before you go onto apply that rule. I thought the issue was fixed, but then I ran into it again on Team Cup.
this times 1 billion It's 2017, no freaking reason 1080p is too damn big

United States yosarianilives says:

Gunslinger said: this times 1 billion

It's 2017, no freaking reason 1080p is too damn big


True, my phone takes 4k by default. I have to actually trow my system pics into an image shrinker so that they're "web" sized. It's just an extra step.

United Kingdom GeorgeStorm says:

Whilst I like the idea of rig pics (mainly because I like seeing people's setups :) ) I do agree the image uploads need to be better, either better (more detailed) error messages or just increasing the size limit. Although as Allen said, what that'll do to the performance of hwbot worries me (since it's hardly speedy at the best of times, maybe you can give Intel a poke to throw some more servers your way for keeping xtu around?)

United Kingdom mickulty says:

Rauf said: But system picture and source of hw adds even more complexity to make a submission. And totally unnecessary. What good does it do? How can you verify it? Rules that you can't verify are just useless...


To be fair given those rules will only affect people who have probably spent hours learning the hardware and optimising to get the best possible score, I don't think 2 minutes to snap a photo and write a comment is that harsh.

Argentina Alan_Alberino says:

Have a question about events: lot of people is getting globals using same CPU, which on general rules it's not allowed... You shouldn't disable points on that cases? Like it or not, it's lot of people getting points from same hardware

United States Splave says:

If you are a paying sponsor it is allowed but you must promise to not sub all on the day of the event and save 2nd place run for a week after ;)

websmile says:

I think this is an interesting theoretical question, but only the amateur world tour team can answer why this is handled the way it is. Of course you cannot ask people to byod if they are new and learn at high priced cpus, but the question of rewarding points for this is legit

Antarctica Trouffman says:

GOOD to have the rules update.
I really like #7.

says:

rest in peace HWBOT

Australia zeropluszero says:

Rule Update #2 - All verification screenshots must show the settings at which the benchmark was completed. This means it is not allowed to decrease the operating frequency for screenshot purposes. Yeh this is rough.

Australia avalanche says:

Rule Update #6 - HWBOT X265 Benchmark v2.1.0 is mandatory from Nov 1, 2017. Is this version even fast?

United States Gunslinger says:

avalanche said: Rule Update #6 - HWBOT X265 Benchmark v2.1.0 is mandatory from Nov 1, 2017.

Is this version even fast?


It hasn't run as smoothly for me as the original.

Australia avalanche says:

Less score then?

Belgium leeghoofd says:

Gunslinger said: It hasn't run as smoothly for me as the original.


you got any data to compare Gunnie ? which platform ?

websmile says:

I had similar results on Z170 tests at 2,10 to 2.00, margin of error excluded they were same speed, no idea about other platforms

United States Gunslinger says:

Leeghoofd said: you got any data to compare Gunnie ? which platform ?


X299, 7740X, the benchmark would struggle to run, both 4K and 1080 preset. Talking single digit FPS, 5.8-6.4GHz

Belgium leeghoofd says:

K I'll test tonite. Which OS did you use ? Win8.1?

United States Gunslinger says:

W7 64 bit I think, don't have it set up at the moment or I'd check.

Sweden Rauf says:

websmile said:
On frequencies we mainly talk about cpu and downclocking for screen, this is not allowed anymore. We do not talk about the rookie league but global top 20, so showing full cpu clocks should work. The 3D clocks we still have to discuss again, but it makes not much sense nowadays in my eyes.
On the ES issue, this is caused by the community´s permanent accusations of people sharing hardware. And yes, we can ask vendors about es if we doubt source of a sample

Rules say only nr1 only applies to global top 20 scores. So frequencies will apply to all other subs as well. I actually think GPU frequencies are the most interesting/important ones to show. Should be mandatory for global top 20. If you don't show GPU freq, how can we verify plausibility for 3D?

As for ES, I mean that if someone have multiple sources of ES hw. How can you tell which CPU/GPU/mem is used? Also, some might now want to reveal their source publicly. Might be sensitive in some certain cases.
Also, what does it prove anyway? If I say Asus or Asrock or whatever provided ES, what diff does it make? It's not like they have one ES cpu anyway, theoretically cpu can be shared anyway...

mickulty said: To be fair given those rules will only affect people who have probably spent hours learning the hardware and optimising to get the best possible score, I don't think 2 minutes to snap a photo and write a comment is that harsh.

Yes, but what's the point? There are already so many things required of benchers. Remembering how each benchmarks works/validates, what you have to do to not get faulty validation files, using the right software versions, saving validation files, showing the right clocks etc. It's actually very common to find mistakes even in Elite subs. Now they are handled gently because the result can be verified anyway. Seems this will not be the case anymore...
Is it reasonable to remove a submission because it does not have a rig picture? What sort of "validation" does the rig picture do?
If you create a rule, you must intend to enforce it. So are hwbot really going to delete WR's/GFP's because of technicalities?

Trouffman said: GOOD to have the rules update.
I really like #7.

Best make all my accusations on October 31st :)

zeropluszero said: Rule Update #2 - All verification screenshots must show the settings at which the benchmark was completed. This means it is not allowed to decrease the operating frequency for screenshot purposes.

Yeh this is rough.

Only for TeamAU ;)

My main point is this:
Don't complicate things unnecessarily! Hwbot is way to complicated as it is already.

websmile says:

3D clocks for top 20 results are OK because top benchers can handle this, outside of this we will not enforce is strictly because of given reasons. Anyway you know that some cards have more efficient BIOS than others, so clocks will give a hint but differences remain. On 2d full clocks, we have headroom and will not punish rookies but handle this with care as well, common sense is needed. If I see result unlikely to be done I can still ask for rebench and full clocks even at non experienced benchers. We need to be able to check the plausibility of a result, like you mentioned at 3D clocks on which I agree, so this should not be asked too much. I remember a time when downclock for screen wasn´t used, it came with hwbotprime, which is gone, so no problem to show full clocks again^^

On the ES cpus, a remark at the comment field of sub is easy. We want to test this for more transparency and to lower sharing accusations which poison hwbot atmosphere now :( - btw, this also means people lending retail hardware from others should state this, it is no es only rule

United Kingdom GeorgeStorm says:

I would have thought the no downclocking shouldn't affect many subs (especially if there is some leniency), and the stating where you got hw from is to try and keep things open etc I would have thought, to try and avoid the anti-elite viewpoint etc.

Australia avalanche says:

Rauf said: Best make all my accusations on October 31st :)

NOV+. Via private message to mods ... sneaky underhanded always wins the day

Germany Dancop says:

I really like #2 but that means: good bye PiFast at the very limit, since it's been impossible for me to run max freq and open cpuz afterwards :D

Sweden Rauf says:

Dancop said: I really like #2 but that means: good bye PiFast at the very limit, since it's been impossible for me to run max freq and open cpuz afterwards :D

Actually, taking correct screenshots is usually not a big problem. Even for hwbotprime and pifast. Run bench, activate slowmode, load all cpu-z, take safety screenshot, deactivate slowmode, take new screenshot. Always works for me.

Rules seem fine to me, it's just that they come across as super strict. If they are not handled so strict they will be fine.

Belgium leeghoofd says:

Gunslinger said: X299, 7740X, the benchmark would struggle to run, both 4K and 1080 preset. Talking single digit FPS, 5.8-6.4GHz


tested 7900X on 8.1 64bit OS

both run fine on this OS, seems older version has a tiny advantage at 1080P while 4K is margin of error ( tested 5 runs in a row) Newer version is more consistent in output

United States bigblock990 says:

Leeghoofd said: tested 7900X on 8.1 64bit OS

both run fine on this OS, seems older version has a tiny advantage at 1080P while 4K is margin of error ( tested 5 runs in a row) Newer version is more consistent in output


I have done this same test using kabyx and skyx. My results exactly same as what you are seeing.


Rauf said: Actually, taking correct screenshots is usually not a big problem. Even for hwbotprime and pifast. Run bench, activate slowmode, load all cpu-z, take safety screenshot, deactivate slowmode, take new screenshot. Always works for me.

Rules seem fine to me, it's just that they come across as super strict. If they are not handled so strict they will be fine.


Whats the point of implementing rules that are not enforced. Easier just to not have the rules then.

United States yosarianilives says:

Leeghoofd said: tested 7900X on 8.1 64bit OS

both run fine on this OS, seems older version has a tiny advantage at 1080P while 4K is margin of error ( tested 5 runs in a row) Newer version is more consistent in output


Could the different in consistency and performance be related to the switch of time keeping away from system time?

Indonesia speed.fastest says:

So now we need some capture card for screenshot superpi or pifast at the limit, sometimes some CPU have weird behavior it just bluescreen when screenshot.

Bulgaria I.nfraR.ed says:

Rauf said: Actually, taking correct screenshots is usually not a big problem. Even for hwbotprime and pifast. Run bench, activate slowmode, load all cpu-z, take safety screenshot, deactivate slowmode, take new screenshot. Always works for me.

Rules seem fine to me, it's just that they come across as super strict. If they are not handled so strict they will be fine.


That's what I, as an AMD guy, do all the time.
Finish bench, downclock, normalize temperature in idle if there's a CB, open cpuz windows, take safe screenshot, up the multi to where it was, take another screenshot.
You know, running 32M for about 30 minutes (s.939 for example) and then not surviving the screenshot is not really fun, extra measures need to be taken :D

PS: Still remember ruining a 2h 50m 32M run on K6-2+ CPU, because I didn't downclock after the bench and opening cpuz locked the system. So I had to redo it :D Baby-sitting a pot for nearly 3h on a cold-bugged CPU...

Australia avalanche says:

Luckily I'm not extreme. Machine is 95% stable after bench. Cpu-z's Gpu-z is open from the start. Minimal impact & then print screen software is instant. Don't deserve the score if you are unstable " preserve from BSOD " to downclock & screenshot it. That's just me though :)

Canada WhiteWulfe says:

About time there was a rig pic requirement... Some of us have been always posting one for every submission... Definitely have to agree about the screenshot size error though, as it can be confusing if you've never run into it before.

Edit: I am surprised though that an email wasn't sent out about the upcoming rule changes though, as not everyone checks the forums.

France bolc says:

Hi,
For those benching locked cpus, with C3 enabled, the frequency will increase when on 1 or 2 cores, during gb3 single cores periods, for instance.
Once the bench is over, even though EIST and all other c-state options are OFF, power mamagement is 100%, since C3 is ON, the cpu will go down to minimal multiplier, say 16x on sandy, 8x on ivy/haswell and for 1151 as well I suppose.

According to rule #2, "All verification screenshots must show the settings at which the benchmark was completed. This means it is not allowed to decrease the operating frequency for screenshot purposes. "

Currently, I/we do not actually decrease frequency for the purpose of making a picture, but the motherboard does it once the load goes down. Hence, the cpu-z screenshot represents the settings at which the benchmark was completed, it shows min and max multiplier.

I can't imagine someone not being able to calculate bclk x multiplier, especially for the locked cpus, non extreme/elite benchers may use?

Moreover, which frequency are we supposed to report, the 4c frequency or the 1/2c frequency ? My idea was to report the frequency related to the benchmark, so if using CBR15, I 'd report the 4c value, but if using gb3 single core would be the 1c/2t, and multiscore be the 4c value ?


Nevertheless, in the end, does this mean it is not possible to bench using C3 active ?
Hence, can Hwbot staff make clear in rule #2 about the use of EIST / C-state, in detail?

Thank you very much.

Belgium leeghoofd says:

Rule #2 is more for those that downclock deliberately to mislead competitors or improve efficiency. We have seen scores being submitted at 5.6Ghz while the run was at at least +6.5Ghz

United States Splave says:

XTU required screenshot is missing?

France bolc says:

I understand, although it would be good to know if the moderator staff will be enforcing this rule thoughly or not with enthusiasts looking at low hanging fruits :o

Aleslammer says:

Splave said: XTU required screenshot is missing?


Nice avatar!!!!

websmile says:

If we cannot determine correct frequency we will block scores. This means XTU without working monitoring will be blocked, and low clock results with speedstep as well (for all competitions). If we have c3/speedstep results that we can determine clocks at full out we will not block them, but keep in mind that all global top 20s including XTU need a screenshot, and if you can do a global top twenty you can also show full clocks. We will not accept top scores with lowered clocks or speedstep

Please log in or register to comment.