Massman

I guess would could give it a whirl in the next Fanboy competition. Any benchmark that would be perfect for OS comparison?

Yup, I did. Was on a hotstreak for about 6 games

FYI - you need to submit to the contest here to be eligible for the lucky draw!

VRM tuning seems to help nicely - thanks for the hint. Was this a 7.6G chip before?

Awesome! Link about the Overclock Masters Party?

Fyi, there are five types of CPUs: - Retail: bought from shop. - Media samples: chips that are sent to media, usually not preselected for overclocking. - QA samples: chips that are shipped to mainboard vendors to do development and QA testing. No pretesting (usually average). - Marketing samples: chips reserved for marketing events, usually not pre-binned but upon request they can be pre-selected. - R&D samples: chips picked from the production line (or even especially designed) matching certain characteristics, usually very handpicked. The "worst" handpicking takes place in the R&D sampling requests, which is basically scanning the production line for chips that have special characteristics (eg: current leakage). I'm not sure how it is with AMD, but for example Intel has a bunch of special CPUs in their labs that were produced only for internal R&D. For Intel CPUs, in general, on long-term some retail samples will match the internal R&D. For AMD, not sure. But, as you can see with the 3770K, as long as you bin enough CPUs you can find a really good one. So, it's not like you really need the industry contacts to get the best one; in fact, sometimes these special handpicks are bad overclockers too! You can select a bunch of CPUs with the best potential on a scientific basis, but how far they can overclock is not guaranteed.

Long time no see!

E8200-E8600 is a perfectly legal combination. Not the best though ... Yes, I really wish it was that simple. At the time of competition configuration the GDDR5 category did not exist yet, hence there was only one GT 630. The HCE needs to be updated to support competition-specific hardware groups, yes, but that's not specific for this competition. The competition is configured to say "use 5 different cards from these 5 categories". I did add the GDDR3 to indicate the correct GT 220 category and for the other 3 you can find a category 100% matching the name as shown on both contest and contest stage page. Anyway, I updated the contest page. Within a stage, a user can only contribute once to the country score. So, no, you can't do all five with AM3+. So when am I the most right and moral: if both GDDR5 and DDR3 are allowed, or just the original category? Change, in this case, would be trying to get the HCE to quickly support groups of cards so both DDR3 and GDDR3 would be supported. The GT 220 was indicated as GDDR3 on the contest as well as contest stage page so that's hwbot's fault.

Impressive for just watercooling!

I understand the problem with the categories here. The issue lies in our database structure and the code for the HCE. Essentially, each vga model has its own gpu_id and the HCE allows me to add limitations on vga model by using the gpu_id or a string of gpu_ids. In its current form, the engine does not support grouping vga models, so adding multiple GT 630's would allow users to use 3x GT 630 in the competition stage which conflicts with the idea of having to use 5 different generations. I've just discussed this problem with Dennis and we have two possible solutions for this. He's looking at the code as I'm typing this to see if we can solve this within a reasonable timeframe and work effort. Keep you posted! Fixed!

pieterhwbot

I'm not talking about HWBOT, the website, but the idea we launched in January 2011 (http://hwbot.org/forum/showpost.php?p=212183&postcount=21). That's essentially the proof-of-concept of what the ROG Knock-Out software is; that is the idea they copied ("stole" lol) and what Der8auer says we should've patented. I know why they are saying it's "an evolution of online sport" too - I used the exact same phrasing when discussing the idea internally with Frederik too . In my (humble) opinion there are three problems with this approach to overclocking - one is self-inflicted, two are fundamental to the concept - that could prevent it from kicking off like online gaming did. Those two fundamental problems is why we decided not to persue this type of overclocking at hwbot as it would take up a lot of resources (=time) to develop. Instead we added the Challenges, which is a non-realtime implementation of the same concept. To be honest, I'm really interested to see if the ROG team will be able to figure out a solution to those fundamental problems. They have a lot of talent on board, so I assume yes. They definitely have more possibilities to solve it than we had! But I do agree with what you're saying: if this brings more people to overclocking and benchmarking, it can only be good for our community! .

The opposite actually as the amount of submissions that need to be taken into account to calculate the weight is lower.

Not all enthusiast forums are like that. In some forums, you still can get a question answered before 10 posts are made and have the thread pretty much done. I hope that answered your question /thread done.

Damn strong! Well done man!

Did you guys like the Fanboy contest? Any suggestions on what other Fanboy battle you want settled?

It's this card: http://hwbot.org/hardware/videocard/geforce_gt_430/ So I think the answer to your question is "yes" - one person can only contribute to a stage once - 940BE would be "AM2+", correct Still looking into it.

Interesting points - I'll bring them up at our next staff meeting (which we hold every day). I tend to agree that even though a time-based algorithm might be more elegant, it's not really bringing to the table. Just more frustration about dropping points.

Small update. This development sprint, the team is working on the Benchmark Ranking API. As mentioned in the opening post, this will allow benchmark developers to easily get a ranking up and running for testing benchmarks software. Additionally to the API itself, we're also updating the benchmark application structure of the HWBOT database to allow more flexibility in the benchmarks we host.

But ... but, but, but ... it's yellow!

You're welcome!

Did the lucky draw through Random.org again - all 136 overclocking results were included . Intel Lucky Draw Winner: Toyone [hwbot=2329073]submission[/hwbot] AMD Lucky Draw Winner: Bartman1973 [hwbot=2329211]submission[/hwbot] I will forward the contact information to Enermax! Congratulations!

I'll get on it today .

Nice!

Nice! Did you use slow mode to cool down before running 1M ?