Most Valuable Submission of Week 39, 2015: 201.3 points by Splave.ROM (United States)

Last week we received 4760 benchmark results from 829 registered overclockers around the world. The majority of the submissions is coming from Rookie overclockers representing 60% of the active community. They were responsible for 29% of the submissions. We had a peek at the most valuable submissions in a breakdown per league.

As you can see in the table below, the most valuable submission of the week was the Global First Place of Splave.ROM in XTU 4xCPU using the new Core i7 6700K Skylake. With 201.6 points Splave.ROM scores about 110 points more than ObscureParadox from United Kingdom who scored gold in Geekbench3 Multi Core with the Xeon X5698 processor. In the Enthusiast class we find KaRtA from Australia; in the Novice and Rookie class we find Fallen4ngel from Greece and Octogen from Germany!

The overclocking results submitted during Week 39 generated in total 155 World Record Points, 7998.7 Global Points, and 6685.7 Hardware Points. The distribution per League is as follows: 20% for Elite, 36% for Extreme, 22% for Enthusiast, 6% for Novice, and 16% for Rookie. The representation of the active community is as follows: 3% Elite, 15% Extreme, 18% Enthusiast, 12% Novice, and 52% Rookie.



Most Valuable Submissions - Week 39, 2015

League Points Overclocker Benchmark
Elite 201.6 points Splave.ROM XTU 4xCPU Core i7 6700K
Extreme 93.7 points ObscureParadox Geekbench3 Multi Core 2xCPU Xeon X5698
Enthusiast 45.4 points KaRtA XTU 4xCPU Core i7 6700K
Novice 23.9 points Fallen4ngel XTU Core i5 4690K
Rookie 47.7 points octogen XTU 4xCPU Core i7 6700K


Top Hardware Points - Week 39, 2015

League Points Overclocker Benchmark
Elite 35.2 points Splave.ROM XTU Core i7 6700K (1st)
Extreme 24.5 points Zeropluszero XTU Core i7 4770K (36th)
Enthusiast 41.2 points Noxinite XTU Core i5 4670K (34th)
Novice 23.9 points Fallen4ngel XTU Core i5 4690K (60th)
Rookie 31.4points Systemfehler13 XTU Core i7 4810MQ (1st)


24

United States steponz says:

Again no mention of any 3d
. Every 3 d single benchmark has been broken.


No mention of it here...

Belgium Massman says:

3D is expensive -> not much people pushing for it -> lower competition -> lower points Gotta find a way to get more people to get into 3D :( :(

United States steponz says:

well if you keep taking points away like you did with the legacies... we can pretty much see who's fault it is... Im really ready to just post on fb and skip the site completely. Nothing gets fixed here.. about to take my stuff elsewhere.

United States Gunslinger says:

I don't know what the formula's are, but it seems odd to me that the GTPP for a given result are higher than the UGP

If they were at least equal, that would help the overclocker improve his personal ranking, no?

United States steponz says:

Yesh there afraid to touch anything really. I think it's hilarious that xtu gas jumped up 10 pts more in least than 2 weeks..

Czech Republic trodas says:

Massman -

3D is expensive -> not much people pushing for it -> lower competition -> lower points
Gotta find a way to get more people to get into 3D :( :(



The solution is IMHO painfully obvious - more points for 3D benches ;)

United States rtsurfer says:

And make the top rankings focused on the size of the Wallet or sponsorship..?

United States Random says:

rtsurfer said: And make the top rankings focused on the size of the Wallet or sponsorship..?

^This.


And then 1gpu, 2gpu, 3gpu, 4gpu. At 800$+ each? Way to take all the fun out if oc

United States steponz says:

Yeah let's just take 3d out all together. Why not just dumb it down all the way do thay way you guys can reach top level. How about we take all tweaks away.. All here is excuses.... you guys don't have 980tis now? You do realize that all but 2 of the records I have are on retail right.

Czech Republic trodas says:

rtsurfer -

And make the top rankings focused on the size of the Wallet or sponsorship..?


Well, it already is like this way. If you have money, you can bind thru many CPU's to get the golden ones. If you are sponsored, you are in whole different game.

It is like I get a retail Pentium 4 and my friend get promo Pentium 4 - from best waffer and with completely unlocked multi. Who wins is, at that stage, almost already determined no matter how hard I try. He can always bump FSB with lower multi, gaining SPEEEEEEEEEEED ;)

So there is nothing like fairness in the world. If you can afford modern hardware and you are not completely dumb, then you get good results. That it is. And getting fastest, that is what this is all about. Or you want to get most points to this score? :))))
http://forum.hwbot.org/showthread.php?t=146557

United States rtsurfer says:

steponz said: Yeah let's just take 3d out all together. Why not just dumb it down all the way do thay way you guys can reach top level. How about we take all tweaks away.. All here is excuses.... you guys don't have 980tis now? You do realize that all but 2 of the records I have are on retail right.
1) I have never owned a 980Ti. 2) Okay your cards are retail, but they are still expensive, which is one of the issues with 3D. 3) Don't get me wrong, I'm not trying question your 3D skills or saying that 3D is easier that 2D. For a good 3D record you need to do all the work you would do for 2D & then more. 4) I think the model of points being based on amount of competition in the category is fair, because even if a category is "easy" at first, the more the people are competing in that bench the tougher it gets. Because everyone is trying to outdo someone else & so they keep on trying to find a new tweak that would give them the edge. So no, 2D is not dumbed down & the tweaking is still there.
trodas said: rtsurfer - Well, it already is like this way. If you have money, you can bind thru many CPU's to get the golden ones. If you are sponsored, you are in whole different game. It is like I get a retail Pentium 4 and my friend get promo Pentium 4 - from best waffer and with completely unlocked multi. Who wins is, at that stage, almost already determined no matter how hard I try. He can always bump FSB with lower multi, gaining SPEEEEEEEEEEED ;) So there is nothing like fairness in the world. If you can afford modern hardware and you are not completely dumb, then you get good results. That it is. And getting fastest, that is what this is all about. Or you want to get most points to this score? :)))) http://forum.hwbot.org/showthread.php?t=146557
Sadly it is. But with CPUs you have some advantages, 1) Its easier to bin & resell binned CPUs. 2) You don't have to completely forsake your warranty when overclocking CPUs. If you kill a CPU during a session, most of the time you can get Intel to give you a new one. With the current GPUs (with the Exception of the Kingpin series) usually there is no software that fully allows you to maximize the GPUs unless you have access to a vendor. So you end up having to do various mods to get around the gimps & there goes your warranty. I think the points being based on amount of competition in the category is the right model. Maybe inject some more points into 3D. Tbh I don't care much about points/rankings, but I understand that it is important to other people.

United States steponz says:

Well its skewed.. If you look at the top 10 xt 4 core... theres not much competition there... The only reason why its going up in so many points is because of all the air submissions... and really all of those count for the top amount of points.. if so ill just create a bunch of spam bots that submit ultra low scores and create so called competition... Just because there are more submissions, doesn't mean there is actual competition to what is going on......

United States rtsurfer says:

You do have a point there, a lot of people just ran XTU because they downloaded it as tuning/stress testing utility from Intel.
Maybe Hwbot should "nerf" the 4 core category to make it more reasonable.

We can justify increasing points in 3D because its harder, but that will not increase participation & we don't really accomplish anything more than making a few people, more happy.

Edit:- P.S there would be more "action" in top 10 X4 XTU if good CPUs/Mems were easier to come by. Will happen eventually.

United States steponz says:

Exactly.. why does that deserve points at all.... My main concern with 3d is you don't want to have little points for it... because then you basically not giving anybody the effort to try.... Meaning 3d will eventually die because there is no reward to do and 2d is basically the easier approach. Im not saying give 3d the ultimate amount of points.. but do give it what it deserves.. It takes alot to do and should be rewarded... When I see GPUPi for cpu and for gpu records giving more points then FSE... There is something really wrong with the points. At this point, it might be better to go somewhere else or startup another site as they seem to scared or unwilling to make changes.

United States rtsurfer says:

I wouldn't go as far to say why it deserves points at all, but.. One of my teammate suggested breaking the 4Core category into 4-Core & 4-Core HT (This is obvious, Idk why its not this way from the getgo). That should decrease the points on an individual 6700K submission. Increasing points in 3D might incentivize more people, but the costs will always be a big wall. A better way would be to have higher points for 3D benchs where you don't need the grunt of an 6/8 core. That will decrease the overall cost of the setup & breed more competition. Do away with FS & enable points on FSU. You'll still have to do the hardwork of 3D, just without an 8 Core.

Sweden lanbonden says:

One more thing to mention in the 2d vs 3d discussion is the publicity aspect, you never see those XTU records geting mentioned outside of HWbot but a great run of 3dmark yielding a worldrecord gets posted on alot of hardware related sites and draws attention to overclocking, losing out of the 3d overclocking would ressult in a huge drop of exposure! Same thing when you end up in a discussion about your hobby and the other guy got somewhat of computer knowledge they never ask about how fast you can run SuperPi or you XTU score, its all about the marks.

United States xxbassplayerxx says:

If anything, we need a cap on scores. No more point inflation... just shut down the ridiculous ones. Cap a score at 50 hardware, 50 global, 50 global team, and 50 hardware team... Max of 200 points for one submission, as opposed to the 166.4+37.2+478.1+51.8 = 733.5 points for this score. To show how uncompetitive XTU really is, graph out all of the scores. I bet you'll see that ~90% of the scores are only ~50-60% of the current WR, if that. I bet it'll look like this, if not steeper.

United States rtsurfer says:

Best option split it in 2, I wonder why the 4 Core non-HT guys never complained, I would.

Why do they have to compete in points with the HT guys when HT gives such a big boost, its just not fair to them.

United States xxbassplayerxx says:

rtsurfer said: Best option split it in 2, I wonder why the 4 Core non-HT guys never complained, I would.

Why do they have to compete in points with the HT guys when HT gives such a big boost, its just not fair to them.


This came up quite a bit when the i7 920 came out. Long and short of it was that a core with HT is still one core. I'd wager that even if 4 core results were split by HT/Non HT they would both be the #1 and #2 benches on the bot and carry 700+ each.

United States rtsurfer says:

xxbassplayerxx said: This came up quite a bit when the i7 920 came out. Long and short of it was that a core with HT is still one core. I'd wager that even if 4 core results were split by HT/Non HT they would both be the #1 and #2 benches on the bot and carry 700+ each.


Cool, I wasn't around back then.
While physically it makes sense, its not an ideal rule for benchmarking considering the boost you get whenever you have HT. Ofcourse this is irrelevant for single core benches like SPi, but it makes sense for other benches.

As for how much points each category would have, maybe [MENTION=47]Massman[/MENTION] could try it on a test server & let us know.

United States xxbassplayerxx says:

rtsurfer said: Cool, I wasn't around back then.
While physically it makes sense, its not an ideal rule for benchmarking considering the boost you get whenever you have HT. Ofcourse this is irrelevant for single core benches like SPi, but it makes sense for other benches.

As for how much points each category would have, maybe [MENTION=47]Massman[/MENTION] could try it on a test server & let us know.


It was a pretty hot debate because the 920/975, etc. were utterly destroying Phenom II and C2Q in multithreaded benchmarks. Test server is a great idea... I'm interested to see how it would play out.

Belgium Massman says:

[QUOTE=xxbassplayerxx;411433]To show how [B]un[/B]competitive XTU really is, graph out all of the scores. I bet you'll see that ~90% of the scores are only ~50-60% of the current WR, if that. I bet it'll look like this, if not steeper. [url]http://i.imgur.com/JnzWsQ8.png[/url][/QUOTE] Mapped out XTU 4xCPU and GPU 1xGPU. Score = user's best score = the score that yields global points. [IMG]http://forum.hwbot.org/attachment.php?attachmentid=3335&stc=1&d=1443766992[/IMG] [IMG]http://forum.hwbot.org/attachment.php?attachmentid=3336&stc=1&d=1443766992[/IMG] For more discussion on "how to measure the difficulty" of a benchmark, I refer you to this thread: [url]http://forum.hwbot.org/showthread.php?t=145745[/url] [QUOTE=rtsurfer;411434]Best option split it in 2, I wonder why the 4 Core non-HT guys never complained, I would. Why do they have to compete in points with the HT guys when HT gives such a big boost, its just not fair to them.[/QUOTE] [QUOTE=rtsurfer;411439]Cool, I wasn't around back then. While physically it makes sense, its not an ideal rule for benchmarking considering the boost you get whenever you have HT. Ofcourse this is irrelevant for single core benches like SPi, but it makes sense for other benches. As for how much points each category would have, maybe [MENTION=47]Massman[/MENTION] could try it on a test server & let us know.[/QUOTE] I read your reply this morning and I chuckled. Threads over cores would make more sense I guess. I think the main issue back then is that the general consensus was that CPU cores should be treated like GPUs. So hyper-threading was seen as a "feature". I'll have to look up the exact formula to calculate the 4c4t-CPU and 4c8t-CPU points. Running this on the test server isn't as easy as it sounds :p. FYI - this would have a major impact on Pentium G3258 users as well, since they are now blocked by the 2c4t Core i3s. I think it's worth investigating how complicated splitting up by threads would be, if we can enable more globals for the really cheap entry overclocking CPUs! For XTU, these are the participants for the (which determines the point scale) Pentium, 4690K, and 4790K: [list] [*]Pentium G3258: 1106 [*]Core i5 4690K: 2604 [*]Core i7 4790K: 4998 [/list]

United States xxbassplayerxx says:

MM, would you mind putting together a pretty graph of 6700K results? Thanks!!!

Belgium Massman says:

Sure. Also improve the other charts. In below charts you see # overclockers compared the the percentage of the top score in XTU, Fire Strike, XTU Core i7 6700K, and Fire Strike GTX 980 Ti.













Important note: comparing the user vs percentage of top result can ONLY be done if the benchmark score scaling is the same. For example, if the scaling is both linear. If the scaling is different, then the comparison is invalid.

Please log in or register to comment.