Memory benchmarks by IC?

[nnimrod]

First I'll state that I'm sure this has been beaten to death suggested before, so I'm not actually suggesting it, but rather I'm curious what the reasons for it's not being implemented yet are.

Why are memory benchmarks so limited? Memory OC is at least as complicated as CPU OC, and yet we've only got the option to submit memory frequency and read bandwidth? And the categorization of memory is limited DDR/DDR2/DDR3/DDR4.

Memory benchmarks could be split up by IC, and potentially also by chipset or CPU socket. I mean then we'd be able to get a hardware gold for nanya memory in 1156, wouldn't that be awesome  Another possibility is making low clock challenges an official benchmark. I know this suggestion in particular comes around fairly often, and I think it's a really good idea. The most competitive and accessible way to OC memory is the 4/5ghz 32m competitions that have for decades resided in forum threads. 

 

Some of the objections that I can think of:

"It's too hard to verify that a particular memory submission is made with the IC under which the submission was made" - Fair point. CPUz only shows which kit you're using, and many kits can use different ICs. And of course taking a picture of the label on the heatspreader doesn't fix it because you could just take a picture of a different heatspreader if you want to cheat. Same with taking a picture of the IC. And then there's the issue of some labels and ICs being ambiguous regarding which IC's were used. Part of this could be remedied by omitting memory benchmarks from any competitions that have prizes until a better solution is found. 

And lastly I'll state that cheating doesn't bother me that much because for the most part, only the first page of results matters. And I trust that those aren't cheating because 1. Scrutiny for cheating is always applied most intensely to those at the very top, and 2. It's more or less the same names on almost every front page, and those people have reputations that they care about, and are not going to cheat. Also cheating just doesn't really bother me personally, because I'm not competitive with the first page people anyway. So for instance if my 133rd place 5.4ghz 955BE 32m submission is beaten by a cheater a dozen ranks up from me, I don't really care because I'm not comparing myself to the cheater at rank 121 - I'm looking at the first page of subs, which I'm confidant has no cheaters on it.

[Leeghoofd]

It would be far too complicated database wise... Eg Corsair has far too many versions as always... You have to think we always cross link stuff in the database, brand, type, ... Also far too easy to make a mistake or boost a score... Hard to detect and verify (heatspreader removal,...) 

[TerraRaptor]

14 minutes ago, nnimrod said:

I'm looking at the first page of subs, which I'm confidant has no cheaters on it.

Not always that, sad but true.

[nnimrod]

15 minutes ago, Leeghoofd said:

It would be far too complicated database wise... Eg Corsair has far too many versions as always... You have to think we always cross link stuff in the database, brand, type, ... Also far too easy to make a mistake or boost a score... Hard to detect and verify (heatspreader removal,...) 

For whatever reason I always picture you as looking like your avatar. Probably not the case but that's what my mind does 

There's still the low clock challenge idea