knopflerbruce
-
Posts
4290 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Blogs
Everything posted by knopflerbruce
-
That was probably me. I was referring to the chance of an unexplained tweak being a 'cheat' or not. If the gain is low, there's a higher probability that it's legit than if the gain is huge. From my point of view 10%+ tweaks could be legal, but they have to be proven legit - that is, they have to be made public so we can investigate before we accept them as allowed tweaks.
-
Will it affect the possibility of running the actual app in realtime or on a certain core/thread?
-
We're still discussing the fate of the benchmark. We must decide that first... Would be alot easier for us to make a call on each tweak if they were made PUBLIC. Just saying;) As I've said before, any tweak with more than reasonable gains (10%+) must be proven to be legit before allowed, in my opinion. Some people agreed, but no decision has been made. if this goes through, then text edit, transparent windows, audio compression, web page tweaks that boost the score like crazy are all disallowed until someone proves for everyone that these scores are indeed possible without cheating. You can nudge the other crew members if you feel this is a good idea:p
-
Nope, series computing time is way too low. This would definately be reported as "bugged run". You need the same relative gain in each subtest, approximately.
-
I have no idea... but I guess you need ~10% higher speed with a single than a dual. I can 06 at 3550-3600, so I'm thinking I can do a tad more for 01, although it's a weird chip. How much FX57 power do I need to match that+ -
Possibly. I remember fighting my ass off for that one:p 01 will be even more GPU bound I bet. The question is... what CPU to use, and can I get Ln2 in time??
-
I think a decision will come soon, a day or two tops.
-
I don't mean to sound like an ass, but didn't we already have a 939 01 stage? or was that 03?
-
He made this thread, the first post contains a few "tricks" which could (according to him) explain this. If he posts solid evidence that his unstable setup indeed can produce these errors, then I guess it's fine? I've never seen anything like this before, but it would be a bad idea to use the banhammer if he could indeed prove that he's innocent.
-
Why on earth do we have to post official news regarding this case on a daily basis? Should be more than good enough to make a statement when we're done - which is not the case right now. Thumbs DOWN.
-
We're discussion this now, but personally I'd love to see a video where you get missing digits and decimals placed in the wrong places - and perhaps the ~0.01 thingy at good clocks.
-
Well, an you use the text edit tweak to gain a similar raise in scores in a different benchmark? no? Must be bug then
-
All of these are OK under the assumption that the calculations are indeed what they're supposed to be. That's what we have to figure out. LOD and mouse movement should be 100% OK, and at least the powertoy thread count. Not sure if the filters etc change the actual calculation.
-
You can always put the same argument to all the other supertweaks, "supertweak with cheap HW beats expensive setups with highend components". I do agree that it was probably close to be disallowed when we made the decision, but I guess it was still a part of the old times, when we allowed nearly everything. Maybe we need to take a look at it again. Is it virus scan or HDD general usage, btw? if the first of the two it should be possible to verify the performance in some way, as the virus scan engine is in the white paper somewhere.
-
Don't hold your breath. I'm still against tweaks that change the subtest score but a considerable amount, unless they're proven to be allright - wrapper or not. If a wrapper isn't designed to catch a sertain type of tweaks we still need hard evidence to allow it.
-
That's what they do. When they had issues with PCMark Vantage and singlecores they just removed support, after having official support for a few years already.
-
Well, the tweaks that we're talking about are easily detected because of insane boosts. We just have to find them and block them, simple as that. Probably a fair bit of work, but in theory quite simple.
-
No, try 3dmark99
-
The calculation itself shouldn't be changed. If you change the size of the windows, it's obviously in some way messing with the calculation. I have no idea what that flashdesktop thingy does, my post is just general. I am, however, opening up for the surprising(?) fact that we've allowed too much in the past, and that we therefore need to start blocking what seems to be against the rules. If you don't like this, prove that they're legal. We can't give benefit of doubt when the score goes up by a huge margin. Agree?
-
A wrapper COULD work, but I wonder how you'd detect for example non-standard size of the windows in TW, text edit modifications made during the test and so on.
-
"piss on" was not the wrong words, I just used your expression to highlight what I meant:) I've read alot worse in here:D Another problem is that even some of those "approved" tweaks could be wrong. We're not experts, even we do make mistakes. powertoy was disallowed in the beginning, and then allowed later.
-
This is exactly why it's impossible to moderate such results. You can "piss on" other people's scores despite being waaaaaaaaaay behind in terms of clock speed. How you can more than double a score and still not alter the way the benchmark runs is beyond me. Even in superpi 32m and 3Dmark01 you can't do anything close to that. Efficiency should be about making a given calculation run as efficiently as possible, and you simply can't do what have been done recently without making the calculation differ alot from what it's supposed to be. And if you piss on others with yor lower clocks, then you do the same thing. legal or not. Changing thread count as we do in powertoy is fair enough, though. The other settings... not sure. Still, very little we can do against this. Better to just retire it from serious competition here (there's always the ORB if you're not happy fighting for gold cups without points). The moderation crew don't get any $$$ for this, unlike the people at FM. I'm not spending hours and hours on moderating all these funny looking subs for free, that's a fact. Pay me, and I'll consider it. I got paypal:D
-
HDTune bugs. That's for sure. The disk is fine, still HDTune won't run. That's proof of nothing else than a bug in that app
-
We can't ban if something was done in good faith. This is not North Korea
-
Because the dishwasher was too mainstream
knopflerbruce replied to Christian Ney's topic in Offtopic
How about taking a mobo to the car wash?