Massman

I think OVP kicked in around +150mV. What do you need? IGP and Vcore?

Any other boards that are going to be used for the IGP contest? If yes, please let me know what models and I'll try to push MSI for providing more extreme bioses

No, it means that many members not being able to submit old results is not a problem for me. As these kinds of challenges are meant for people to make new scores

Crazy!

W00t! Found a new bios in my mail ... 2.081V Vaxg (IGP) Just flashed it ... works fine just stock. Hooking up the single stage tonight to see the effect . Download link: LINK Be careful!

Yes, I just tested it and the voltage goes up quite rapidly, so you might want to use a higher value for VR. Were you monitoring the Vcore while increasing?

I think it's getting okay again?

People who actually take the time to explain ideas to me know that I'm a very good listener. I don't see why I should create an artificial group of 'community leaders' when, at the moment, very few of those leaders throw in suggestions. http://hwbot.org/forum/showthread.php?t=12005

The powerteam principle can be applied to ANY HW/BM ranking. The idea is to have a powerteam in all rankings: 3DMark03 4x GPU as well as Core i7 920 SuperPI 32M.

All points should be back I have also disabled my access to the database

They will. Restored roughly 97% of the data ... now recalculating the best submissions. (I hate software)

'Powerteam' is a term introduced to differentiate between the overall teams league (which we all know and care about) and the teams league per benchmark ranking. 'Powerteam' can be seen as ANY benchmark ranking, but instead of grouping the results per user_id (= creating user BM rankings) grouping the results per team_id (= creating team BM rankings). The last couple of HWBOT OC Challenges can be understood as 'Powerteam' challenges as well: you bench for the place of your team within a benchmark ranking. Alternative 1 = overall teams league based SOLELY on this powerteam thing. Alternative 3 is a combination of the powerteam concept and the effort of casual overclockers putting a load of results up the bot.

Points may be dropping for a while ... seems like I seriously fucked up

It's a user error on my side ... fixing it ...

?

You are correct: 2 cores and 4 threads go into the 2-core competition. Why? Hyperthreading is a feature Intel implented to improve the multi-threading performance. It's a FEATURE, just like 'amount of L2 cache' is

The idea is: - UFL = top sponsored/corporate overclockers - XOL = Xtreme overclockers - EL = Ambient overclockers The rankings will be split up in 'member rankings' and 'team rankings'. To visualize this, check this page: where the tab now says 'member rankings', it will be split up in two tabs saying 'member rankings and team rankings'. - Member rankings = like they are now - Team rankings = described in earlier posts as 'powerteam' For the different user leagues, we'll use the following criteria: - UFL: global points - XOL: global points + hardware + competition + ... - EL: global + hardware + ... (but filtered on ambient cooling results only) All above rankings are based on the one and only member ranking algoritm; like it is now. The EL will just count all points of your scores with ambient cooling. In case you have a better score with extreme cooling, it will beat your ambient cooling result so the latter will not receive any points and YOU will drop in the EL ranking. For the team ranking, this doesn't change anything as all user rankings are based on the same member ranking algoritm.

P11

How's PCMark scaling exactly? CPU dependant or rather storage dependant?

Again - hyperthreading is a feature of the CPU. 875K = 4-cores (with HT) 530/655K = 2-cores (with ht)

Moved so people can comment . It was a pleasure seeing the German overclockers going at it again ... they really are passionate about overclocking. The environment in the German OC gatherings is always very nice and relaxed (such as AOCM); just a pleasure to be there. Also a big thanks goes out to ASUS Germany for setting up this Extreme OC tech meeting! Concerning the me-failing-and-playing-retard-game ... in my defence:

Let's not get ahead of ourselves. HWBOT staff has asked for more details on a couple of runs, which still is allowed I think, and there has been no feedback yet.

It's got something to do with the nature of the subtest: it's designed to test multi-threaded applications. The Sempron 145 is single core ánd has no L3 cache, which makes it have a buttload of problems in multi-threaded situations: one of the four subtests at once and no cache to store instructions or data. I think your RAID controller's cache might be allowing the CPU to get data quicker or so. I'll try to see how this subtest scales when using an SSD raid config for OS and Acard as PCMark05 bench drive

? It says "use X processor core(s)" ... that's #cores?