Lets talk about 32m Low Clock Challenges

[Splave]

This has been bothering me and a few others I know for a while now, to the point Im not even benching 32m at the moment which if you know me is strange.

 

5ghz Low clock challenge has always been fun for some, and its also been a nice playground for freezing memory and learning some new things. With the advent of B-die and skylake everything has changed. Best ICs win, no wiggle room. There are 6-8 timings that really do anything to effect time, and most boards have them set in a profile anyways. You dont need ln2, you just need a fat wallet to play.

 

Now to the real point. If you are in the top 10 of 32m currently you know what is going on. 5002.9 is just for screens. Hold down print screen while you are at 5009 and voila 5002.9 will pop up eventually. (takes a long time if you want the cache to match) 100.187,100.125,100.067. Why dont we make 100.000 the max? There must be a way to level the field because its getting out of hand.

 

5GHZ 32m LCC is turning into XTU where we are hoping for the longest amount of time the CPU is over the limit with BCLK fluctuation. There arent lucky runs, there are runs that the real freq is 5007 for longer than it is 5002. Thats the real "lucky" run.

 

Videos dont work and people wont do that, OC panels and trickery can easily be done still. BCLK raising under load can also be implemented by board manufacturers so the end of a video will show correctly in CPUZ frequency.

 

So what is my answer? Kill it for now, there are no way to verify any result unless ney flys to everyone's house and breathes down the neck while it runs.

 

In the future maybe some sort of watch dog timer can be developed.

 

These are just my feelings and I figure I would share them instead of being pissy about them behind the scenes.

[Noxinite]

I have thought about this recently too. However, it is difficult with large memory dividers as part of the fun is testing the different combinations to push your ram.

Edited September 9, 2016 by Noxinite

[speed.fastest]

To make them work? They need wrapper, but i dont think i will like it because it will take some resource. Now system is fine because Low Clock Challenge is you get like very low point, unless at OC-ESport.

[RULE]

Is not about point, but respect!

I always run 100.067 on MOCF because I have strougle to boot up at 100.000, still in a perfect world should be like 5003.35, but there's also some freq fluttuation.

The asus guy should be 100.00 because the first step after is 100.125 so 5006.25

 

In 4080 scenario is the same 102.125x49=5004.125

 

Total agree with big fella we need a solution

 

Inviato dal mio SM-A300FU utilizzando Tapatalk

[Noxinite]

I think when I was running 5GHz on Z97 OCF a few weeks back I managed to get 5003 (bios setting) to stay at that freq or lower in OS (highest it went was 5003.36 I think). Turning off force OC and speedstep, but leaving on turbo boost or something. I didn't check how it fluctuated under load though.

 

I would recheck now, but my OCF is half dead as the back PS2 and USB ports don't work.

[l0ud_sil3nc3]

I would be down to limit to 100.00 seems to be the only way to keep things perfectly fair. Both Asrock and Asus are over the limit on the first step past 100 bclk can't say for other vendors as I have not tested them. The only thing I have thought about is splitting the bclk steps in half i.e. making them more granular, so instead of 100.0625 the first step would be 100.03125 which should keep things legitimately under 5003 mhz but I am not sure how easy this would be to implement.

 

Also I think not having to freeze ram to get a top score has kinda caused this recent influx of killer lcc submissions. In pre B die days you actually had to freeze memory (and usually cpu) to get a proper sub and now AIO cooler and a fan is all you need to get a competitive result just the way it is.

[GtiJason]

This has been bothering me and a few others I know for a while now, to the point Im not even benching 32m at the moment which if you know me is strange.

 

Now to the real point. If you are in the top 10 of 32m currently you know what is going on. 5002.9 is just for screens. Hold down print screen while you are at 5009 and voila 5002.9 will pop up eventually. (takes a long time if you want the cache to match) 100.187,100.125,100.067. Why dont we make 100.000 the max? There must be a way to level the field because its getting out of hand.

 

5GHZ 32m LCC is turning into XTU where we are hoping for the longest amount of time the CPU is over the limit with BCLK fluctuation. There arent lucky runs, there are runs that the real freq is 5007 for longer than it is 5002. Thats the real "lucky" run.

 

Videos dont work and people wont do that, OC panels and trickery can easily be done still. BCLK raising under load can also be implemented by board manufacturers so the end of a video will show correctly in CPUZ frequency.

 

So what is my answer? Kill it for now, there are no way to verify any result unless ney flys to everyone's house and breathes down the neck while it runs.

 

In the future maybe some sort of watch dog timer can be developed.

 

These are just my feelings and I figure I would share them instead of being pissy about them behind the scenes.

 

Your opinions seem to be on point with what myself and other friends who truly understand Efficiency, Memory OC and prefer to bench 32m or GB3 think. My last LCC 32m sub using B Die was the very first run on MOCF because of this reason. I feel that my kits are top 5% of all retail but would rather keep my run times to myself because having just one person thinking I cheat is too many. 32m llc has always been about honesty and integrity and the veteran OC'rs can tell when a run just isn't right or "looks funny". I'm sorry but if I know someone cheated even just one time, my mind will never see his lcc scores as legit unless there are some built in protections like GpuPi. We have also been discussing the point made about records being done on air, and I feel like B Die has been both good and bad for our community. It's nice to see more guys becoming interested since Samsung has made this easier but at the same time feel things were far better in 2012-2013 where air was somewhat competitive but if you want to be on top ln2 was a necessity.

[Rauf]

Lol, what would limit to 100.00000000 bclk help. It would still be a waiting game for some to see when the lowest bclk variance hits 100.00...

 

And we would probably see an upswing with vendor skinned cpu-z as they only show one decimal.

[RULE]

There's 2 solution, killing our efficiency with a software like hwinfo running during the benchmark or video proof with all the system, not just screen.

 

And since are we talking about all LLC is time to kill XTU, for have a good score I need temp throttling? Realy? Time to introduce geekbench instead.

 

Inviato dal mio SM-A300FU utilizzando Tapatalk

[speed.fastest]

SuperPi HWBot Edition with Wrapper like System Info or something like GPUPI

[Splave]

I wouldn't mind a wrapper since it would hurt everyone's efficiency equally. Good points being made

[RULE]

Wrapper? No man please we can't kill best benchmark stand. U know wrapper means net framwork and probably we can't run so tight with mem like now.

 

Inviato dal mio SM-A300FU utilizzando Tapatalk

[Mr.Scott]

You aren't killing anything. The exploiters are. You're just collateral damage.

[RULE]

This collateral damage may shift the focus of memory performance on xtu a well know bugged benchmark.

 

What about video prof with cpuz open, will affect just the efficiency of the benchmark not also the performance of the memory and will be the very same handicap for everyone.

 

Inviato dal mio SM-A300FU utilizzando Tapatalk

[coolhand411]

hahah...why you want to kill it now bro ?..why didn't you bring this to attention few months back when we all knew that the bclk fluctuates that much ? mater of a fact is I think you were the first one to submit with cache at 104.9,this was well known fact ,you didn't had the problem while sitting on ES modules and leading 5.0 challenge ,yes I agree that changes could be made if we all agree but lets start with the next chip-set...or remove all subs with cache above 103 ,no one is cheating man ,you just haven't found the golden setting yet that's all

 

"xtu a well know bugged benchmark"..I don't thinks so ,just because you can't match something it doesn't mean it's bugged,as of now I hit 5x742 on two different boards and 2x743 on Mocf with the same timings ,it doesn't look "bugged" in my opinion and what about all the boards with 102.7 bclk limit ..maybe they should remove all the subs above that limit ?

Edited September 12, 2016 by coolhand411

[RULE]

Edited September 12, 2016 by RULE

[newlife]

I fail to see how 5mhz matters at all and I think some of guys really need to try AMD because you ain't know what real fluctuations of the bclk is until you try AMD

[speed.fastest]

Yep, FX with default HTT (200mhz) with 25x Multiplier is fluctuate like 15mhz or more

I think is that doesnt matter, what matter is the cheater, we cant detect it as far as i know. But who the hell want to cheat at low clock challenge if not in competition?

[Splave]

hahah...why you want to kill it now bro ?..why didn't you bring this to attention few months back when we all knew that the bclk fluctuates that much ? mater of a fact is I think you were the first one to submit with cache at 104.9,this was well known fact ,you didn't had the problem while sitting on ES modules and leading 5.0 challenge ,yes I agree that changes could be made if we all agree but lets start with the next chip-set...or remove all subs with cache above 103 ,no one is cheating man ,you just haven't found the golden setting yet that's all

 

"xtu a well know bugged benchmark"..I don't thinks so ,just because you can't match something it doesn't mean it's bugged,as of now I hit 5x742 on two different boards and 2x743 on Mocf with the same timings ,it doesn't look "bugged" in my opinion and what about all the boards with 102.7 bclk limit ..maybe they should remove all the subs above that limit ?

 

LOL Golden, is that why everyone sucks all out and they are kings at 32m LCC?

 

If Im such a cheat why would I want to fix it and not just cheat harder? Maybe its since I have nothing to sell by being #1.

[I.nfraR.ed]

I fail to see how 5mhz matters at all and I think some of guys really need to try AMD because you ain't know what real fluctuations of the bclk is until you try AMD

 

This is the exact same reason I run lower than even 5GHz, because I want to be honest and stay under the limit, no matter in theory it should hurt my score a little:

I.nfraR.ed`s SuperPi - 32M score: 13min 24sec 78ms with a Phenom II X4 965 BE

 

I don't remember the exact fluctuation, but this combination was the closest to 5GHz that did not abuse the 5003 limit. Don't feel like doing the math now though

[TaPaKaH]

If Im such a cheat why would I want to fix it and not just cheat harder?

Because ASRock is no longer the only brand with fluctuating CPU clocks?

[Splave]

It never has been the only brand that doesn't fluctuate. How about any asus result over 100.0 is cheat since they dont have 100.0625 step? Do you wonder why asrock and giga can do 742+ because the bclk is so stable at 102.9** makes you wonder doesnt it?

 

 

was just trying to fix it so it can be fun again, not pointing fingers at vendors, then called a cheat which is lol worthy.

 

Guess I will just be like everyone else and stop trying have fun. /thread

[Guest Bullant]

Is not about point, but respect!

I always run 100.067 on MOCF because I have strougle to boot up at 100.000, still in a perfect world should be like 5003.35, but there's also some freq fluttuation.

The asus guy should be 100.00 because the first step after is 100.125 so 5006.25

 

In 4080 scenario is the same 102.125x49=5004.125

 

Total agree with big fella we need a solution

 

Inviato dal mio SM-A300FU utilizzando Tapatalk

 

Wow if this is true it's seems over ,on asrock I run 100.0625 =5003 bios ,now if asus has only 100.1250 lowest?this is 5006,then at that rate it would be giving a nice boost in eff like 200-300ms boost

 

 

I know they all fluctuate but end of the day it's about what bios is

Edited September 13, 2016 by Bullant

[rtsurfer]

Not to rain on you guy's parade. But on Asus you can adjust BCLK in 0.01 Mhz increments *in BIOS*.

 

I've always ran 100.06 x 50 which should be 5002.9Mhz according to the BIOS.

 

Not to take away from the purpose of this thread, but don't you dare label all Asus 32m Skl results invalid.

 

Sent from my Nexus 5X using Tapatalk

Edited September 13, 2016 by rtsurfer

[Guest Bullant]

Not to rain on you guy's parade. But on Asus you can adjust BCLK in 0.01 Mhz increments *in BIOS*.

 

I've always ran 100.06 x 50 which should be 5002.9Mhz according to the BIOS.

 

Not to take away from the purpose of this thread, but don't you dare label all Asus 32m Skl results invalid.

 

Sent from my Nexus 5X using Tapatalk

 

Think you need to get of yiur asus high horse,I ask the question is it 100.125,only quoting what rule wrote,I've not got the board to tr

 

Simple answer like it can run lower bull is all what was needed

Edited September 13, 2016 by Bullant