Windows 8 has been banned by the world’s top benchmarking and overclocking site

[Dankk]

http://www.extremetech.com/computin...worlds-top-benchmarking-and-overclocking-site

Thought this was interesting. HWBot stops using Windows 8 and removes all previous Windows 8 data from the site because of an unreliable real-time clock (RTC).

 

[ChronoReverse]

I wonder if they simply didn't configure their bios settings properly (i.e., not using 64-bit HPET).

According to them, this occurs with both overclocks and underclocks in Windows 8 but I'm running Windows 8 overclocked and I certainly don't have this issue.


Either way, it seems premature considering they don't seem to have properly tested this enough. Of course, people will eat this up because it's Windows 8.

 

[Kenmitch]

I say read the article before you pass judgment on HWBOT's actions.

The results of the fault in the RTC are insane in some instances. Equates to 100's of MHz of free speed.

http://hwbot.org/news/9824_breaking_windows_8_benchmark_results_no_longer_accepted_at_hwbot

If what they say is true then the only thing they could do was to ban Win8 and remove all the results.

On a side note: Overclockers scramble to purchase/pirate Win7 to regain their ranking!

Guess for now on we'll need to ask what OS your using to get your results when comparing overclocks and benchmarks.

 

[SunnyD]

It would be sort of funny if all of those performance improvements that Windows 8 brought to the world really actually evaporated.

 

[SPBHM]

I had this exact problem with windows xp and some lga 775 hardware

 

[PottedMeat]

so how is a rtc implemented? they don't just have a rtc module ( 32768hz xtal + battery + a ic / southbridge integration )? i thought software just reads a few registers which wouldn't be affected by an oc.

 

[ChronoReverse]

I say read the article before you pass judgment on HWBOT's actions.

The results of the fault in the RTC are insane in some instances. Equates to 100's of MHz of free speed.

I've read the article and the problem is that they didn't do sufficient testing.

I already mentioned how it doesn't occur with my system even though I matched what _the article_ claims causes the problem.



In fact, further testing from other people has shown this only occurs if you overclock after Windows has booted and not if it were overclocked via BIOS. There hasn't been any indication of whether this occurs with older versions of Windows either but if you've used Windows XP in VMware on a system with power-saving downclocking, then you'd know that even Windows XP will experience desynced clocks if the CPU speed changes after boot-time.

Unless more information and testing is done by hwbot, this was a very poor claim to announce like this.

 

[Kenmitch]

I've read the article and the problem is that they didn't do sufficient testing.

I already mentioned how it doesn't occur with my system even though I matched what _the article_ claims causes the problem.



In fact, further testing from other people has shown this only occurs if you overclock after Windows has booted and not if it were overclocked via BIOS. There hasn't been any indication of whether this occurs with older versions of Windows either but if you've used Windows XP in VMware on a system with power-saving downclocking, then you'd know that even Windows XP will experience desynced clocks if the CPU speed changes after boot-time.

Unless more information and testing is done by hwbot, this was a very poor claim to announce like this.

I was just making a general statement. If it was directed at you only I do know how to quote

Looks to be the issue entirely to me also.

HWBOT is all about e-peen.

The problem is how would you separate the results? Not sure if it would even be possible to do it. Seems like the only way to make sure all the invalid results are removed is to remove all Win8's.

 

[Arkaign]

Windows 8 is already banned at my house

j/k I actually have it on my old SSD in my main PC, I just don't boot to it at any time other than testing something.

 

[DigitalWolf]

Well they can ban it.. I don't care to be honest.


I didn't really like it at first but then I got used to it and it certainly isn't any slower than if I toss in my other SSD (same model) with Win 7 on it. However, I will admit it took a long time for me to decide it was ok and there are a few things I still like better in Win7.

/shrugs

I don't benchmark anyway.. I just play games personally.

 

[Idontcare]

HWBot isn't exactly the bastion of solid testing methodologies in the first place.

Remember how badly borked that hyper-Pi program was, and how the author kept iterating the software without issuing trackable revision numbers or change logs?

And yet HWBot kept the benching program on the list of accepted/validated benchmarking programs.

I do hope Microsoft figures out a way to make time less relative in these self-referencing computer systems.

 

[ViRGE]

HWBot isn't exactly the bastion of solid testing methodologies in the first place.

In this case HWBot's problems are real, but they also are only of consequence for them. Outside of the competitive overclocking community no one is regularly doing realtime BCLK adjustment. At the most they find a stable BCLK and then program it into the BIOS for future use. So the impact on the common overclocker is nil; even AT and other hardware reviewers shouldn't be affected. This is solely an edge case that can be used to cheat in end-user competitive benchmarking.

 

[BallaTheFeared]

My thoughts exactly, besides it's unlikely many hwbot users would even use Windows 8 now that the "cheat" has been exposed.

Last I checked they all run custom modded versions of XP for just about everything, and a custom modded version of Windows 7 for DX11 tests (very few on hwbot anyways). The custom XP isn't even a real functioning OS, so many services are disabled it's just a shell host for shaving milliseconds off scores, they don't even have networking services installed.

 

[Idontcare]

My thoughts exactly, besides it's unlikely many hwbot users would even use Windows 8 now that the "cheat" has been exposed.

Last I checked they all run custom modded versions of XP for just about everything, and a custom modded version of Windows 7 for DX11 tests (very few on hwbot anyways). The custom XP isn't even a real functioning OS, so many services are disabled it's just a shell host for shaving milliseconds off scores, they don't even have networking services installed.

Exactly, equally as relevant to reflecting an end-user's experience as, say, an OS timing loophole

Once people are allowed to "customize" the compute environment, when does it become cheating versus acceptable customization?

If I can customize my WinXP to be practically useless outside of running a benchmark program, how are the results any more meaningful than if I took advantage of a timing loophole that is also useless outside of running benchmarking program?

IMO all HWBot has accomplished here is calling attention to how irrelevant their entire database of "world records" really is.

 

[FalseChristian]

And we need Windows 8/8.1 exactly why?

 

[Arkaign]

And we need Windows 8/8.1 exactly why?

For all of our touchscreens and desire to have metro apps default to taking our entire screens in a non-resizable manner? To turn touchpads into nearly unusable minefields of misinterpreted user input? To hide 'shut down' as a "charm" inside SETTINGS? LOL!!

 

[Dankk]

And we need Windows 8/8.1 exactly why?

It's not needed, but the performance optimizations sure are nice.

 

[Idontcare]

And we need Windows 8/8.1 exactly why?

The word "need" is subjective of course. Purely a first-world problem.

Does anyone in the consumer space really need anything beyond Win3.11? Probably not, it is all a toy of convenience really. After all, the world got along just fine in 1890 before the invention of the PC.

So who is to say that Win8 is needed?

Personally I prefer Win8 over Win7. But this sort of dissension over GUI style is not unfamiliar territory for Linux users. Talk about your passionate debates over GUI style and kernel

 

[hans030390]

For all of our touchscreens and desire to have metro apps default to taking our entire screens in a non-resizable manner? To turn touchpads into nearly unusable minefields of misinterpreted user input? To hide 'shut down' as a "charm" inside SETTINGS? LOL!!

Desktop mode + regular Windows applications. You have access to metro apps if you want them. You have the same options you did before and then some. Also you can split metro and regular/desktop apps on your screen at the same time. That functionality is further expanded on in 8.1. You are never forced into either setting. The choice is yours, and it's better to have that choice than not when you have Windows 8 on your desktop, laptop, and tablet. Problem not found. The unified OS is quite nice.

Touchpad and tablet input work great for me on Windows 8, even on cheap devices. No worse than 7. Problem not found. Perhaps your hands and fingers don't work properly?

Moving something to better accomodate a new UI =/= hiding something. It took me all of, well, 5 seconds to figure out how to access the power options on my desktop. If it took you longer, you might want to reevaluate your computer skillz before complaining and posting crap like that. Have you ever thought about the "Start" button and menu? Why is it called "Start?" Why "Charms?" Why "File," "Edit," "View," and so on? When you get down to it, a lot of UI naming conventions don't make sense (though some based on early UI designs might have made sense in the past).

While I would have preferred to see the power options outside of settings, such as right when you open the charms menu, it's really not a big deal. I actually thought it was fairly logical to place it in settings along with everything else in there.

Again, problem not found. I get by perfectly fine with Windows 8 and now prefer it from a usability standpoint to 7, largely because it works seamlessly across all of my devices, desktop to tablet. OMG I had to learn some new things here and there! What a TERRIBLE thing that is. My only complaint with Windows 8 is the lack of tutorials for people who don't know how to use computers and/or can't learn how to use computers (which, given the large animosity to Windows 8, it would appear a staggering number of people fall into that category). Anyone I have personally taught how to use Windows 8 has picked it up VERY quickly, and none of these folks have said they prefer Windows 7 over it. I think I'm up to a whopping 4 or 5 people now that I've introduced Windows 8 to. Not many, sure, but you'd expect otherwise given how the majority of people seem to respond to it.

I actually had a situation regarding teaching a stranger at Fry's how to turn off a Windows 8 device. They said they'd heard bad things and that it took their friend an hour to figure out how to turn it off (which would help validate my thought on 8 lacking tutorials). I reassured him that the OS was quite easy to use if you were willing to learn and demonstrated a few things, first being how to turn off the machine (explained and demonstrated in both touch and KB/M terms). Afterwards, he just laughed and said that was all much easier and more intuitive than people had been making it out to be. Seems like people are just feeding off the idea that Windows 8 is terrible and needs to be hated, much like people did with Vista (generally silly, unfounded reasons or whining because they didn't like change). Also, I have never and do not work at Fry's...just happened to be a creep and picked up on his conversation with someone else before I jumped in.

UI aside, there are enough tweaks under-the-hood that make 8 worth it for me. Posting benchmark results to one site isn't something I participate in, so that's no issue for me. The logic behind it in regards to saving power on certain devices makes sense from a unified OS perspective, and I honestly can't say I've been negatively impacted by the issues brought up by this benchmarking community site. But, yeah, sounds like it's only an issue if you OC through software post-boot. Surely there is some way to check for this rather than ban Windows 8 benchmarks outright...?

 

[blackened23]

So this affects a few overclockers that wanted to cheat....

Much ado about nothing.

 

[jhu]

Exactly, equally as relevant to reflecting an end-user's experience as, say, an OS timing loophole

Once people are allowed to "customize" the compute environment, when does it become cheating versus acceptable customization?

If I can customize my WinXP to be practically useless outside of running a benchmark program, how are the results any more meaningful than if I took advantage of a timing loophole that is also useless outside of running benchmarking program?

IMO all HWBot has accomplished here is calling attention to how irrelevant their entire database of "world records" really is.

That's odd. Why not just use Linux or some other free OS kernel. It's trivial to compile in only what's needed to boot and for that specific CPU.

 

[rge2]

Anyone who benches knows to get the last 100 mhz you overclock on the fly, not to mention, who benches by restarting the computer each time, that would be incredibly inefficient.

But the kicker is you dont typically decrease bclk when benching, but some people figured out by doing so, you can proportionally increase time the benchmark runs, ie you get faster score with lower mhz. Clearly that is a problem, and the few that figured it out pepper their false scores amongst the real, not to mention those that did so without realizing.

No question HWBOT did the right thing. That being said, doesnt affect anyone but benchers, and only those that decreased bclk on the fly. And will be fixed with windows update.

But no reason to not use windows 8 because of an on the fly benching/timing bug.

I dont use windows 8 because it is a touchscreen interface designed to allow novice users a similar platform across mobile and desktop at the expense of desktop efficiency. But then again, with no tangible benefit over windows 7, really only reason to get the touchscreen version (8) is if you have a touchscreen.

 

[Idontcare]

But the kicker is you dont typically decrease bclk when benching, but some people figured out by doing so, you can proportionally increase time the benchmark runs, ie you get faster score with lower mhz. Clearly that is a problem, and the few that figured it out pepper their false scores amongst the real, not to mention those that did so without realizing.

I doubt that was how the bug was discovered.

What I bet happened is that people who were overclocking bclk noticed their scores were NOT going up as expected, but rather they were regressing and the benches were getting worse. (because the computed time was longer than actual, versus being shorter than actual when underclocking the bclk)

Once HWBot started investigating why the overclocked bclk score were wonky, they discovered you could use the bug to your advantage to cheat as well.

Just my speculation, but I bet that something like that is how this actually came to be discovered.

 

[rge2]

I didnt speculate how they figured it out....you read my statement differently than I intended.

I will improve my grammar so more clear:

"but some people figured out by doing so" substitute "but some people figured out in so doing"

 

[Kenmitch]

I doubt that was how the bug was discovered.

What I bet happened is that people who were overclocking bclk noticed their scores were NOT going up as expected, but rather they were regressing and the benches were getting worse. (because the computed time was longer than actual, versus being shorter than actual when underclocking the bclk)

Once HWBot started investigating why the overclocked bclk score were wonky, they discovered you could use the bug to your advantage to cheat as well.

Just my speculation, but I bet that something like that is how this actually came to be discovered.

I'd agree on your theory.

It's easier to get the last little bit out of a chip on the fly.

On a side note would it be possible to boot at 1.66x strap and downclock to the 1x strap on the fly....Would generate some crazy scores I'd think. Guess you'd need a motherboard with working straps.

Might have to play around and see for the hell of it. I do have a dual boot to Win8.1 setup that I can explode if needed. Gonna wait for 8.1 to officially release before I start the transition over.

I to this day still hold the world record for the 2550k clockspeed and SuperPI-1m on HWBOT....Helps when your cpu is the red headed step child