I.M.O.G.
-
Posts
869 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Blogs
Posts posted by I.M.O.G.
-
-
Could be some sort of recalc - GTPP didn't change for our team, but HTPP is changing quite a bit. My league rank changed quite a bit too.
Probably just RB breaking stuff though:
I made some aggressive optimizations to the code to decrease the load on the server when people are making submissions in popular categories. Hope I didn't break anything. Submitting should go 10x faster now. -
I'll do it. You can pay me in hardware.
-
Write-back cache is standard functionality of many software, comes with both on-board and discrete RAID-controllers. Intel RST, AMD RAIDExpert, Promise WebPAM, etc. It is already used for ages for PCMark scores, way before someone discovered third-party caching software like Fancy Cache or Super Cache.
There is a big difference between that two types of software with caching ability.
Third-party software can use so much RAM for cache that you want, and it gives you performance very close to RAM Disk driver (if have enough RAM installed).
"Bundled" software that comes with hardware not allow you to specify amount of ram for cache and don't have "mirroring partition" feature. You can only enable or disable that option. And if enabled it gives the same boost no matter how much system RAM you use - 4 Gb or 64 Gb.
Thank you SAV, very informative.
-
I'm not sure why you ask, nothing is the matter though. Bob80 asked me what I thought of Steponz's score so I explained - I think its interesting, but there are multiple scores which are similar in nature, so I mentioned those as well. I was just replying to his inquiry - I don't know why he's using the C word as I clearly didn't.
I called steponz's score a new tweak or exploited glitch - I don't know how to do it, so those are basically the options. I didn't call it a cheat, bob80 called it that, and maybe he took the wrong implication from my statements.
When I called the other types of scores not well understood, that is all I meant - I didn't mean they weren't legit. They are just exceptional, with only a few players understanding how to do them currently.
-
Dear Mr Stevero, please go away! lol
Nice work.
-
Nice work David!
-
Same here, fine again now.
-
The main site has been crawling for me and another US based user for the past 24 hours or so. Something going on currently?
The forums are responding much more quickly.
-
I think its interesting - either a new tweak or an exploited glitch. Scoring higher on a subtest than anyone in the top 20 is just reason to look closer, nothing against steponz or you for figuring it out or pushing further - thats a good thing. Someone should help hwbot staff (or massman at least) understand how its done as much as possible, so we know if its a tweak worth investing time on, or if once the tweak is understood its not going to be permitted in the rules. Often times submissions slip through though that aren't valid, so personally, I try not to invest much time on tweaks that I don't know for sure if they are accepted.
I think this transparent window score and the 50+/60+ web page rendering scores are the only submissions right now which aren't fairly well understood. Only gluvocio knows how to do 50+ on web page rendering, and mtech 60+ on web page rendering (other than youngpro who has done 100+ on XS but not submitted to hwbot, because he was using a different browser - most don't know gluvocio's or mtech's trick for 50/60 however EDIT: Sav submitted a 56 web page score yesterday too now)
EDIT2: I also think the high SSD-based virus scan scores are interesting. Sav just posted a 1200+ virus scan yesterday using 2x SSDs from onboard RAID0. That is higher than the actual disks are capable of, and is done through ramcaching of onboard raid software. Ramcaching software is disallowed in the rules, but these submissions are being allowed - not sure where the line is drawn on ramcaching software. These scores are getting the same virus scan performance of hardware raid cards with onboard RAM, but its done through software.
-
Can't blame a guy for guessing.
You did have shell32 on your desktop. 
-
Bob80: I think i got it now. That tweak is using resource hacker, or similar tool, to alter shell32 (the windows driver library for customization, like hacked theme files and stuff). We haven't gotten a ruling as hwbot staff was just investigating, but that seems like what is going on? Not sure what I think about that, seems kinda like changing out codecs or something.
-
bob80: If its a 0x7F error message its related to detecting your storage drivers after installing Intel RST. Most commonly, people get 0x7F when IDE/AHCI/RAID is set incorrectly in bios.
Dennis: The highest win7 transparent window score in the top 20 for 4x cores is currently 15K - I hadn't looked further, thanks for mentioning steponz. I was with Janus67 when he put up that score (he was using my storage), and we were indeed running vista32 - we knew vista32 could do higher transparent window than anyone else in the top 20, so we wanted to see what sort of score we could put up with it. That was using all commonly known tweaks, most/all of which are mentioned in this thread already I believe. It turned out to be a bad tradeoff, with win7 other scores would have been higher, and the transparent window score would have been lower, but his total score would have been better.
-
Looks glitched, top 10 scores don't score that with higher CPU freqs, and thats primarily a CPU test. A good transparent window score on SB should be 15K to 20K on windows 7 (this is the highest transparent window score in the top 20 for 4x cores: http://hwbot.org/submission/2250621_janus67_pcmark_2005_core_i7_2600k_46410_marks). The other CPU based tests seem like they are inline with good scores for his frequency - its only that subtest which seems out of whack.
Most notable on that super high transparent windows score is that its on windows 7, windows 7 typically scores 1 or 2K less than vista, tho 7 scores better in some other tests. If I were trying to maximize a transparent window score, I would not do it on 7.
I'd check if he can do it more than once. If he can, then he may have uncovered a new tweak. If not, its a glitched score and shouldn't be submitted.
Just my opinion, hope it helps.
install OS in sata mode : raidthen install Rapid Storage Technology
( this should be the link )
Put SSDs in raid by this utility and with write back : enabled
you should gain much in general usage , virus scan and something in memory access
If you tweak this a certain way, I've seen a guy from Poland score 600 gen usage and 9000 virus scan. I don't think that kind of score is legit though, its essentially the same result as using a software ramdrive essentially.
EDIT: Correction, that Poland user was using raidxpert, not sure if Intel RST can be tweaked that way.
-
Oh, and another thing:
From usability point of view mandatory fields should be easily distinguishable.
The most popular solution is to mark all mandatory fields with a red asterisk after the label, like this:
Note that you have a legend next to submit button - [*: required field], but none of the fields are marked that way. Only the fieldset legend for processor is marked
The way it is done currently and because the fact that not all fields have labels, it might be easier to add an explanation to the legend of every fieldset, as it's done for [verification (mandatory)].
For CPU you could have something like this in the legend text for the fieldset "processor (processor and cooling required)".
The point is that the user should have a visual information about required info he needs to enter on the initial stage, not only after Submit button clicked.
Aside from that, one thing that had been noted before. Please change the tooltip position, so it appears on top of the field, not on right and covering the next field that is supposed to be filled. IMO it would be better.
$(function(){ $("input[title]").tipTip({maxWidth: "auto", activation: "focus", defaultPosition: "[b]top[/b]", delay: "100"}); });good idea
-
+1 Keeping internal lists for specific tasks is difficult, but worthwhile. Having a list on HWBot would make it more useful.
Going one step further, I would like it as well if I could maintain my own hardware library. For example, if there were a checkbox or something I could clear to indicate if I still have a part or if I have sold it... Then if we encourage our teammates to keep their hardware library updated, we can look at all hardware the team has run, as well as all hardware the team currently still owns.
-
2GB effective, haven't found a 4GB stick of RAM that the Areca will detect as 4GB. It either detects them as half, or it won't detect them at all. One of our teammates contacted Areca, who tried to work on the card but ended up bricking it - his card was RMA'd as a result and he's waiting to get it back.
Don was over my place last weekend for a team benching session.
-
Correct, it kicks out when the XP startup test finishes, if configured correctly.
-
By the way, my xp startup is currently around 800MB/s. I can dial it in to score between 205MBs - 220MBs consistently.
Using IOmeter I use the following settings:
Size (sectors) - 204800
Access Type - 32k 100% read
Time (seconds) -15s
# of workers - 1 or 2
# of I/O - adjust as necessary, 5 to 20 is a good starting point
The rest you can leave at default, or configure as you wish.
Addendum: You may need to manually edit access specification file size to tune things in - iometer effectiveness at pulling down your XP startup score can vary a lot if your unpulled down score is 500MB/s+. In testing, it took five minutes to dial it in on an e8400 and a 2600k, however an e6600 on the same platform as a e8400 took like half an hour... Sometimes the effectiveness of pulling the score down varies, even when using the same iometer settings. So you have to adjust.
-
Great thing never had a coach/teacher with that mindset while growing up.
I didn't mean to offend you, I was making a joke a bit. To be more serious though, I don't think most people have met many coaches from other teams that tell others how to beat their team. I was just sharing some tips anyway, and making a light-hearted comment.
Duke;152658']Ummm do you have a 2700K? hard to beat my score without one. lolNo I don't have a 2700k. Globally I haven't made a good quad core submission yet.
-
^^ Mr. IMOG - well done - but LOL - you should probably be charging for coaching!
I would charge, if I thought I couldn't beat his score anyways. I don't care if he beats other people.
Just kidding, mostly. -
By the way, my xp startup is currently around 800MB/s. I can dial it in to score between 205MBs - 220MBs consistently.
Using IOmeter I use the following settings:
Size (sectors) - 204800
Access Type - 32k 100% read
Time (seconds) -15s
# of workers - 1 or 2
# of I/O - adjust as necessary, 5 to 20 is a good starting point
The rest you can leave at default, or configure as you wish.
-
Duke;152517']Dude did you actually just say that to me? It's the many recent submissions that prove this bench should not exist.
Or at least it shouldn't be classified under a cpu bench. It's not even close to being a cpu bench. Its an OS bench at best.
As far as scaling goes. Really. Come on. Gluvocio benched at 5354mhz. and scored 46898. I benched at 5712 mhz. and scored 40733. Wheres the scaling? Really i can see why you defend this bench so. It's your pride and joy. You do it well. But plz.
3Dmark 2001 is classified as a GPU bench - however top scores have very little to do with GPU performance anymore. CPU performance is more important than the strength of the GPU, but more than that is knowing the tweaks and understanding how to apply them all successfully - in order to claim top global spots in 3DMark 2001, you have to have all your crap together and get one glorious run.
PCMark 2005 is similar. If you know all the tweaks (or most), and you understand how to apply them all successfully - its just a matter of using the right CPU and getting the most clocks you can out of it, having strong storage and getting the most out of it, supported by a reasonably strong GPU, and having everything in line at the same time in order to get that great score.
As for your 5.7GHz 40K PCMark 2005 run, these are the items holding you back:
- It is a good score, anything that I refer to as "low" means compared to top global scores
- 852MHz core is not enough to compete well globally. That is stock on a 5870, so you are at a disadvantage with your 5850.
- Your XP startup should be within 20MB/s of 220, 178 puts you at a disadvantage. 220MB/s is not hard with SATA3, you just have to have good control of your storage performance.
- Your transparent window score is low by 4K or so, so it puts you at a disadvantage. With your CPU and GPU speed, you could do 15K if common tweaks are applied correctly.
- 3D pixel shader is low, GPU core frequency is not high enough
- Web page rendering is CPU speed, you are missing tweaks here
- Graphics memory is ok, could be higher with better GPU mem speeds
- HDD general usage is low, you need stronger storage for PCM05 gen usage.
- Audio/video compression are good, could be higher if tweaks are applied correctly.
- Everything else is good
I hope this doesn't offend you.
If every CPU benchmark was just a matter of frequency determining the best score, then the only benchmark we would need to run is CPUz. I think we run different ones because they require us to optimize our systems in different ways to get the best score in each.
-
Look at the first place rankings, they all have excellent CPU scores, and they are getting better constantly. The work mtech, Stevero, Gluvocio and others doesn't have anything to do with their skill overclocking CPUs?
The GPU clocks are pretty good as well, though not always stellar - they are better than most can do on air cooling, and are usually on SS or LN2. If you criticized the GPU clocks, your comments would hold more water.
Look at the storage performance, they are all using RAID0 and SSDs or hardware ramdrives.
Mr Scott, as someone with about 100 league points, and not a single rank in PCMark 2005 that is competitive globally - you quite possibly could be the least qualified person on HWBot to comment on the validity of this benchmark. Put some hardware together and actually try to do something with it, and I'd give your opinion more thought than the 15 seconds it takes to reply to your ignorant opinion on a benchmark you don't know anything about.
Those who want to complain about the benchmark, please refer to the other thread dedicated to that purpose - this thread is dedicated to sharing tweaks on PCMark 2005 for those who enjoy it. Why go to someone else's party just to crap on it?
Contribute if you know anything, or just quit trolling please.
-
Where have all the points gone ???
in HWBOT Rev.4
Posted
Looks like all hardware points are counting towards OC league ranking now, definitely looks like RB made an oops. Should only be top 20 hardware and top 20 global points contributing to your league ranking. People with a lot of hardware points shot up through the rankings - its like rev3 in the user league.
I'm not sure how that made team points go down across the board, but somehow they did.