IanCutress

Ah crap. Stupid reset BIOS took off my 5 core setting. Fecking feck feck!

Single core celerons suck for Globals

Ticket ID: 1610 Priority: Low This chip is a single core w/HT. Thus is should be a 1x CPU, not a 2x. Is a new admin putting in chips and putting them in the database in terms of threads not cores?\r\n\r\ne.g. http://hwbot.org/submission/2327485 thinks it\'s a 2 core chip but it\'s a single core chip

Ticket ID: 1609 Priority: Low There are two i3 3225 entries in the system. One seems to point to a quad core chip, the other to a dual core. This chip is actually a dual core with HT, so please merge the quad to the dual and make sure it\'s all dual Thanks

Wrong benchmark (Spi 32m), didn't put the score in correctly

ugh so CPU limited SR-2 next for you. These 8-core SB-E Xeons at stock do 64k+ CPU score, but really drop the GPU score something rotten.

Didn't we rule against RST? Isn't it technically a bugged run then? Something jumping in and disrupting the time calculation

Using a 3960X, the early BIOSes have OC problems over 4.6GHz or so. They may have changed it with the latest ones. Also, for SSD RAID-0 setups on those SAS ports, the lanes are limited by Intel spec to PCIe 1.0 for coherency. They now have a BIOS update which allows you to change this to PCIe 3.0 - I managed to get 2.4 GB/s read and write speeds using eight ADATA SX910s. My review can be found if you want to search for it.

As you know I deal with ECS on a regular basis. Usually the issue is BIOS compatibility - they need to work with memory vendors to make sure they know how to deal with the newest kits. They need to analyse their markets - when MultiCore Enhancement came out they (and some other manufacturers) were blindsided. There's also an issue with support. A lot of people knock on ECS due to the large amounts of bad press they have had regarding RMAs. Ideally ECS should spin out a company. Retain ECS on OEM manufacturing (which makes 70% of their sales, mostly in pan-Asia), and a new name in the consumer motherboard arena. They should hire someone who knows their competition to start designing and rearranging their R&D team to focus on what really matters, rather than a lot of marketing bumf. The Non-stop testing was a good idea for their OEM branding, but not much appeal beyond that. What would help a lot is if they had a killer product at some point. Something better than the rest, even if they made a loss. But that requires a good team behind the product through design.

Steve, check original post in thread. It has been updated. Also, regarding 'normal' scores, check Gluc's recent Core2Duo sub: http://hwbot.org/submission/2323429_gluvocio_pcmark_2005_core_2_e8400_(3.0ghz)_44665_marks 33k Transparent Windows (with Mouse and Bob80's on 3960X @ 7970 I only get 25k) 50 Web pages/s ~6k Audio/Video (about what I get with a 3960X OC + reg tweak, but he gets on C2D) 345 pages/s Text Edit That's about right, though I'm not sure what he's doing Transparent Windows wise. Some other tweak I do not know probably

Using new drivers?

Redo this with Bob80's tweak

Sweet - in the list of OK/NO, you have put video encoding as no. But Pro said:

Awesome thanks. When the majority is sorted, it's worth getting MM/Rich to update rules and make a front page post if necessary.

At the time, the scores were valid and we've been waiting for formal clarification of what tweaks are valid or not. They will be going in due course, don't you worry. Or maybe people being charged of crimes shouldn't be allowed to vote by your logic? Everyone still gets a voice, this is still an act in motion. I've been asking for a list of yes/no rather than the 'just use your head' arguement, because there will always be different interpretations (or misunderstandings about what certain programs do), but nothing came out in 24-48 hr so I decided to write my own. Again if there is anything to add to that list, please do so. PS I'm all for the results.txt file being shown in all submissions from now on.

Thanks for your input! I agree with most of what you say, but there are conflicting statements that I would like to clear: Much like PowerToy, all F2D does is change the registry. Albeit on the fly like other additional tools you mention. But *anything that changes the size of the windows is illegal*. EVEN IF it can be done by the registry. Taken in isolation and out of context, several of those comments could justify one or the other. Then again, if someone programs into the registry to reduce the window size by 5% for a 5% boost (assuming it's linear), then that will be hard to spot. So Tweaks: Bob80's - YES Mouse Movement - YES Resizing windows (F2D or Registry) - NO IE Settings - YES AMD RaidXpert - YES (for now, new decision pending, may be retroactively removed) Changing Codecs in the directory - NO Changing Codecs outside the directory - ? Script to change files for Web Pages - NO Script to change files for Text Edit - NO D3DOverrider - YES Video Encoding via registry - YES LOD - YES Other registry tweaks - TO BE DISCUSSED BY ADMINS Scripts to change drive letters - YES Scripts to change CPU clock/cores - YES Forcing PC05 to look in other directories - NO Please feel free to add to this list if I am missing anything.

I am just saying it sounds like this. I am surprised that the staff did not choose to put Pro or Genie to head this. It is a very complex issue and a lot of the OP in the thread was regurgitation of rules that haven't been updated for the little exclusions since, rather than something concrete with a simple yes/no system. As IMOG says. RAIDXpert also does something odd that gives better scores than RAM. But it had been approved my Massman before this thread was started, and this thread is all about the new tweaks being allowed/disallowed. It was odd to me that you had never heard of it if you had been put as head for PCMark. Also do not forget the situation with AMD and Tessellation, which is a little more subtle. It's been allowed on AMD because it is freely changeable in the drivers (and sort of easy to spot). This means that NVIDIA cards have to work harder to get the same scores. I'm all for global consensus with benchmarking, or a clear defined list of yes/no. Just as long as it is clear PS It's almost 4am. I'll leave until I wake up

Regarding that point, if you were part of the PCMark scene, you would have read this thread that came out of the Team Cup (you remember, the large team event): http://hwbot.org/forum/showthread.php?t=49082&page=7 Please read page 7-10, especially the top of page 10. My issue is that person overseeing all of PCMark05 should know that these discussions take place. It would help for clarity.

I find this extremely offensive. There is no need for this, especially from administrators. I find this insulting and confusing that as a business HWBot would approve of this mentality being projected. If you wish to publicly call out my integrity or honesty, please do so. Nothing is stopping you in this regard.

I have never manipulated your words. I quote you directly. If you disagree with what you say yourself, then that is up to you. I will say again, just in case it was missed. I have no issue with you blocking the subs I made. I have an issue with your use of words, from an understanding viewpoint. The tweaks have to be declared officially invalid. You have declared that this is a retroactive rule, taking into account all previous submissions. Until that point it was declared, it was still a tweak being discussed. The word cheat implies malicious intent. There was no malicious intent. The rules have been made such that it is now illegal to use the tweak. Have I used the tweak since I found out it was declared illegal? No. But there is *NO* list of what tweaks are allowed, or specifically what tweaks are not allowed. Out of the known tweaks, there should be a simple yes/no list. What will it take for an official, clear cut, yes/no list? I want clarity, consistency, and honesty above all else. With one thing being said, and then something else being said, people can get easily confused. Is it that hard to be succinct and affirmative? There was an easy way to sort all this from the beginning of the thread, with said list. Having reiterations of vague rules with numbers of exclusions is downright complex and confusing. This is compounded by the fact of understanding all the current tweaks in circulation (such as AMD XPert, which was previously discussed elsewhere) or the ongoing recent history with this benchmark. Having been given a chance to ask (and answer some questions), I feel it is only bringing up more issues. However, I like the opportunity to debate, as long as points are taken on board from both sides.

Thank you. If you would like to include any of those comments in the public domain, please feel free to do so. We would love to hear what comments you have to say.

I respectfully request that you do not put words into my mouth. To suggest I start calling people names is offensive and defamatory. The post is not an attack on you. I repeat. The post is not an attack on you. I do not make attacks on people. I request and urge clarity from the system from the top level. Please be consistent in your reasoning, and write a list of what exactly is legal. If you want a 'clean' benchmark, then we need a wrapper that installs an unalterable virtual machine. No SetLOD. No mouse movement. You are declaring you would like a fine line between legal and not legal. At the minute, all we have is a grey field. Some tweaks can be done legally one way, but not legally the other. The main example here is CherV, which I quoted twice in my post. Making it clean is one thing, but honesty is another. I am very honest, but I am also not afraid to speak my mind. Without regularity or clear instructions, this is going to be a minefield of heartache for benchers and moderators alike. My comments are all perfectly valid. I have provided explanation and quotations necessary to ensure we have a full debate about this issue. I would like to have that debate. In order to debate, we need a dialogue to which the goal is that community is served best. I only want what is best for the community.

I wasn't referring to my scores. Rules have changed, I accept that. We all do. Goal posts can move continuously (ever worked on a scientific research project?!). But please read my post thoroughly. It seems like you have missed the point entirely. Nothing has been set in stone here. Nothing has been formally presented in an official manner to determine what is right and what is wrong. All we have had is vague reiterations of rules which are cherry picked and in direct contradiction to 'legal' tweaks. It doesn't help if the one laying down the hammer is not too concerned with his own understanding of what certain tweaks are or how they work. This is very clear throughout this thread.

Change result from seconds to decimal points per second. That way we won't have the issue like with wP 32 getting down to silly numbers

I noticed one of my scores got blocked, and came to this thread. I have all the best intentions, but I hate to say that this thread is very flawed. Please accept Sweet that this is not an attack on you. I do not want this to be a negative on you. People who know me will know that I am consistantly methodical, hardworking and scientific in my analysis. The following is merely a comprehensive and thought out consistently of what is happening, and it has all the best intentions written throughout. To start, it sounds as if Sweet written the OP of this thread of his own accord without discussing with the admins. I know English isn't his first language, but there's almost nothing to the thread even if it is done as per what has been said in the admins. It doesn't explain the situation fully to those who do not know. It doesn't offer a full and comprehensive solution to the problem. It doesn't provide any guidelines whatsoever on how to deal with it, and what will be done in what sort of timeframe. This is compounded by the fact that Sweet does not understand the tweaks or edits, or has been following the PC05 story. This has been clear multiple times during this thread: ----- i) "Originally Posted by Sweet: Flash Desktops is the first REAL TIME "virtual" desktop manager - Overclocking is real not virtual, then Flash 2 Desktop is "not allowed" answer) This is a gross misunderstanding of what FD does. The way FD is used in PCMark05 is not the intended functionality of the program, so if you go and *copy stuff off the website* (very important), then you will not understand how it is used. F2D is not used as a virtual layer - as CherV has said, and I quote: The way FD is used is essentially as a registry changer but with a timer. Much in the same way we can use a timed program to limit or change the number of cores in a 3D benchmark (or pass the grammar test in PC04). If FD is removed, then registry tweaks are removed. Wholly and forthwith. That also means LOD and D3DOverrider are removed from all other benchmarks as well. FD is *not* used as a virtual environment modification - it is only used as a bit of software that changes the registry on the fly. This makes the following quote null and void: ----- ii) "Originally posted by Sweet: When the screen goes blanck in 3D, its a bug or something else, not a software, instead it would be a cheat in 3D if anyone use a software" answer) This is again not understanding what FD does. FD edits the registry - it detects windows with a certain name and resizes them and changes color. This is just a larger scale Bob80's tweak (more later). In essence this means that FD 'does not cause a black screen' - the windows are there, they are still being processed, but they just can not be seen. I repeat, this can be done in the registry without FD. If the windows were not there, the score would more than likely be 100x higher, not 2-4x. ----- iii) "Originally posted by Sweet: Guys, its easy, enforce rules that are from 2008 and do not ruin a benchmark #no human interaction or software or change codecs or files in Benchmark#, and this will always be so." answer) "Interaction" is the key word here. There are lots of 'interactions' we do - the big one being the mouse moving tweak. Or this could be extrapolated - using the word 'interaction' also applies to adjusting CPU clocks on the fly using onboard tools or software (ROG Connect). If no human interaction is allowed, then no mouse movement tweak, no adjusting cores on the fly, no adjusting CPU speed on the fly. This also means no changing drive letters on the fly. This can also be applied to the rule regarding changing files in the Benchmark. If so, then I guess we should never use the 64-bit patch for Aquamark. Or ever delete the direcp.dll. Or rename Aquamark.exe. This is more nitpicking on the actual words in the rules. No interaction = no mouse movement. Or otherwise we are just cherry picking excluding clauses (which should all be written down). ----- iv) "Originally posted by Knop: If you ask me, the bob tweak seems like a glitch - there's absolutely no reason whatsoever that doing what he did should give a huge boost. However, I'm open for arguments - if someone can actually prove that it does in fact increase the efficiency without glitching the bench, go ahead. Assuming Bob's tweak doesn't cause a bug." Answer) Technically Bob's tweak is neither a glitch or a bug, but rather a flaw in windows. There is an issue in windows when certain windows are open and themes are changed - some non-visible post processing options are not adjusted for programs already open. So when the screen is locked with PC05 open in the different theme, some of those post-processing features of the old theme stay in tact. This is why the benefit disappears when the benchmark is quit and reloaded. Flaw with Windows rendering and processing, rather than glitch or bug. ----- v) "Originally posted by K404: The user is changing nothing in Windows, they are merely using a feature of Windows as it is designed- changing desktop appearance." answer) Technically Windows is being changed. But much in the same way that the registry can be changed. Windows has software to change the theme, but that is essentially an advanced registry editor, like PowerToy or Flash Desktops. ----- vi) "Originally posted by Knop: On a side note, if IE registry settings makes you unable to use the browser, isn't it also reasonable to say that this indeed bugs out IE9? I mean, it's about opening a webpage, but if IE9 refuses to open a webpage at all, how can anyone say that the test runs as it should?" answer) To run under different settings and to bug out are two different concepts to grasp. Adjusting the IE options to make it unreadable (Javascript off, images off) are just running as different options, and I bet that some people somewhere in the world have one of these options changed like we do. So putting them altogether is a bug? No, a bug is a coding error that could be exploited. All we are doing here is adjusting our IE. ----- vii) "Originally posted by Strunkenbold: Actually we're bugging the test with running it with Vista/7 as there was no Aero when pcmark05 was developed and than we trick it again by running Internet Explorer 7,8,9. In both cases calculation gets done which was not intended by the devs." "Only thing we do here is restrict "super tweaks" but we're still not run the bench like it was intended. And I fear this isnt even possible today." Nice point, well made. ---- vii) "Originally Posted by Sweet: -Well, virus scan is too hight, 2460.21 MB/s You use something like ramcaching ? -Not by now, because to allow this software must be allow ramdisk or ramcache with intel chips -In PcMark'05 ?" This is a big flaw in this whole situation. If Sweet is taking charge (which it seems he is), then he needs to understand *beyond a shadow of a doubt* what is possible before going after scores or posting what is right and what is not. As he seems to be the top admin overlooking the situation, any post he makes could be considered part of the rules. As we know, rules rarely get updated on the main site, so sources in the forums are usually pointed to. In my mind there is only several people who can really know what is going on under the hood. Genieben and Pro are the main admins, with Glucovio, CherV, Moose and Topdog as regular users. The rest of us are left scratching our heads most of the time, and it does not help if the Admin is as well. *Please please please* reorganise this situation to be perfectly clear. Nothing less will do unfortunately. But apparently clear-cut rules are often not a top priority at HWBot ^^ The point I am making. ----- viii) "Quote by Sweet: Powertoy = GUI /// F2D = Software" answer) You misunderstand what a GUI is. Powertoys is software that edits the registry, and FD is software that can be also used to edit the registry, but on a timer if needed. I will reiterate what CherV says to make this point *perfectly crystal clear for everyone* ----- ix) "Orginally posted by Sweet: who use the cheats listed from now, will be punished accordingly" answer) Please please please post a list of the exact tweaks being used, with a yes/no alongside each of them. Genieben has already posted a list of some of them in his wrapper thread. Otherwise the community is still being handed vague overall guidelines. We need something solid, something to be seen. --Overall-- A lot of the reasons given for banning tweaks are perfectly legal for other benchmarks (using programs to interfere with how fast the benchmark is run), or are not understood by the admins (Adjusting IE, or Virus Scan on AMD RAIDXpert, or Flash Desktops 2 is a registry editor with hotkey/timer). I implore the HWBot staff to clear this up once and for all. And I feel that explanation will only be sufficient if Genieben or Pro delivers it. Again, I apologize to Sweet - I know English isn't your first language and this is not meant to be a post towards you. I want what is best for the community, and having detailed knowledge of the underlying benchmark is needed to deal with the issue. And as has been mentioned before, a lot of these scores can be obtained a quick way, or via the registry. If you can edit the registry for 50-100K+ TW, 120+ WP/s and a large text edit, being able to tell the scores apart will be impossible. Especially if the admin doesn't know what is going on exactly. This will make moderation untenable. PCMark05 is a badly written benchmark. The only way to cure most of these scores would be to wait for Genie's wrapper. This would offer two scenarios: 1) Stop all PC05 submissions until the wrapper is complete. The wrapper will either include some tweaks by default, making sure everyone is at the same level, or be able to remove tweaks out of the equation. OR... 2) It's time to remove PC05 from points. Make it a tweakers benchmark First and foremost I'd love it if other admins chipped in that thread with explanations. I'd also like to hear from MM/Rich as to what is going to happen. Much praise for everyone keeping the HWBot community alive and full of vigour. If I need to explain any of my points in a different way for non-native English speakers, please let me know PS. I never even got FD to work properly. I failed hard on that one.