First of all, hi everyone! Sorry to dropping by unannounced, but I thought it would be best to reply here. It turned out to be a fairly long reply, hope you don't mind
// just to clarify: I'm commenting here as Pieter and not speaking for the organization. //
is this the line you're talking about? It's the value of the points not the amount of points awarded that was his intention I believe
So the larger the database grows the HW points are less significant value. I assume in the sense that 500HW points is very little if there are 1 million available.
What I mean is that the more hardware and benchmarks are added to the database, the more rankings there are, and thus the
relative significance of one ranking decreases. This is not a "policy" or something I (or anyone in HWBOT) "want(s) to happen", it just happens naturally.
In 2008, we had 6 3D benchmarks at HWBOT: 3DMark01 to Vantage, and Aquamark3. Including only the ones with global points, today there are 13. So if you were 1st in 3DMark03 in 2008, you were first in
1 out of 6 3D benchmark rankings. Today, you are first in
1 out of 13 3D benchmark rankings. Now consider the hardware rankings. Including the benchmarks with hardware points, there are 24 3D benchmarks. That's an increase by a factor of 4. In 2008, Nvidia launched it GeForce 8000 series. Since then, we have the 9000, 200, 400, 500, 600, 700, and 900 series. Each with a dozen different products. The amount of hardware rankings increases exponentially. That's how we get to a count of 44,000 hardware rankings.
---------------------------------------------------------------
Maybe I wasn't entirely on topic addressing your concern that it appears that "new, expensive hardware + liquid nitrogen = win" is the formula for HWBOT. My personal opinion is that while the perception is certainly undeniable, the intention is not. I, and many others, have spent hundreds if not thousands of hours trying to address this problem. I could write a novel-size book on this topic
.
(For those who are unaware, there is a section on the HWBOT private forums discussing this issue amongst Elite/Extreme. If you ever did subzero, you should have access)
Maybe the perception stems from global and hardware points algorithm, which takes into account two parameters to calculate a submission's points: position and participation.
For the
Global Points, not much changed since the beginning of the HWBOT ranking. This was always dominated by extreme cooling and those with access to the best hardware. Two of the most important changes are 1) splitting up by core count so single gpus aren't competing against the 4way monster rigs, and 2) only the most expensive stuff is competitive. The latter is a problem since Nehalem/Gulftown (980X, first 6-core) and became a larger problem with Sandy Bridge (only 2 overclockable CPUs).
For the
Hardware Points, things are fairly similar too. Major changes are the increase of benchmark and hardware, resulting in a lot more hardware rankings, but also the increase in user base. Consider this: there are more people joining HWBOT today than in 2008, and the majority of those people use new hardware. So the hardware rankings "max out" quicker with new hardware. Resulting effect: it's easier to get a certain amount of points with new hardware. That's why we see that
the top hardware point scoring submissions from last week were all but one from the Core iX 4000 series. Dtaylor.nettech is 11th out of 637(!) with his 4710HQ. I'm
10th out of 170 with my 3000+ on Dry Ice, hand-binned memory, full-out tweaked, and only have 15pts. Do you really think I believe Dtaylor's score is more valuable?
---------------------------------------------------------------
In conclusion,
- there is more hardware, but fewer options to be competitive, and usually expensive and high-end
- there are more benchmarks
- there are more people, who prefer new hardware
The algorithm including parameters position and participation leads to favoring a specific set of highly-popular, highly-overclockable hardware. Note: the basic concept of the algorithm hasn't changed since 2006. The competitions give us a way to value different hardware and cooling. The Old School is Best School competition yields the winner the same amount of points like any of the ROG OC Showdown competitions, even if the participation is 15 vs 150.
Now, is it possible to change the algorithms to favor older hardware more? Yes, it's possible. For example, by simply setting the rule that all hardware golds have 50 points you'd essentially boost the appreciation for the older hardware. But then again, with 44K hardware rankings everyone would have maxed out hardware points in the League. So the question is complex.
To add more complexity, let's throw in the question about defining "the best overclocker" (or: what should the #1 overclocker's profile look like). Some say it's mastering a wide variety of hardware (ie. Knopflerbruce), others say it's the ability to max out contemporary hardware (ie. Rbuass' vga epowering), others say it's performing in live competitions (ie. Monstru). Some say it's the ability to handle liquid nitrogen properly, others find anything beyond ambient cooling silly.
