yosarianilives

I haven't been able to go over about 320 without dropping a channel on any of the sempys I binned so far on any board I own, but I've also only had the cpu on ss with chipset on air. Perhaps I need to put both on dice. I did find that slamming all the secondary voltages helped a bit, including dram, so perhaps I need some d9s to push above the 2.2v I feel safe with on my hypers.

No dropped channels!

They've been missing for a while, somehow certain subs that are subbed as "no pts or rank" get included in 1c rankings even when not 1c. So it's so far below "1st" that it doesn't get pts. Also I've been hemorrhaging hwpts for some inexplicable reason since the start of this weekend. I'm gonna wait until the dust settles with all this recalculations to complain too loudly though.

I wouldn't say that there is as far as performance, although I can understand someone not wanting to sub it as custom watercooling. Perhaps we should just combine custom air, aio, and custom water into ambient (custom) or something?

Sadly this was getting a 2600 for price of 2600k. So I did logical thing and benched at safe voltages then sent back.

Well done man! I will have to rerun next time an ebay seller thinks they can send me 2600 instead of 2600k ;)

All I know is if you don't it's locked to 60. If you force triple buffering in driver you can get 63. Nothing you do in driver will prevent that. Perhaps the rules should be updated to account for this, as you can't just force vsync off in the driver. It doesn't change the actual rendering of the scene so it should be allowed.

You have to set single buffering in the app, then use d3doverrider to force triple buffering.

#RIPLTT

I did find the issue, I think. It doesn't show if you look at the rankings but for some reason there are some non 1c scores as part of the 1c ranking. So the #1 1c score shows up as 5th globally, and it's far enough behind the "1st" score so as not to receive pts. http://hwbot.org/submission/3770515_yosarianilives_hwbot_x265_benchmark___4k_celeron_g470_0.61_fps

Ooooh that hurt absolutely savage. But more seriously is the algorithm having issue with that?

Looks like x265 is getting better, although for some reason no pts for 1c...

They're looking like less bugged pts for x265, but did you pull 1c globals? http://hwbot.org/benchmark/hwbot_x265_benchmark_-_4k/rankings?cores=1#start=0#interval=20

Should be 3x hd 3870 afaik

You can delete it yourself

It's probably just broken rankings like we see for many other things rn. Just look at x265 4k rankings...

It's not just march subs, my 1700x x265 4k sub from last year is getting the 64.8 pts for example and my 1c gold in the same category from february is showing half the pts it was a month ago on the sub page and doesn't even show up as counting towards my score. Once the db is sorted you need to take the site down for several days and do a complete recalc of ALL scores.

Naw, those timings look similar to how my psc clocked. If OP wants to be sure he can look at the serial number. The first 3-4 digits will be a year and production week. After that the next 4 digits will be IC. 1040 is psc, 0640/0660/0680 is elpida and could be bbse based on clocks.

Probably the IMC. I'd say you can probably get $20/stick on the high end.

Luckily LOD is allowed on the bot http://hwbot.org/news/889_application_17_rules/

Crazy mem straps my friend, and apparently a good imc. I was told that the bclk is high cause it's the only thing that worked.

That's really great, I fear I may need to bin far more 2600k's than I have as of yet if I want a 5.8+ chip. Or perhaps the golden 6 even.

Would you believe that was one of the chips I haven't binned? Didn't even buy to bench specifically, originally went into a machine for one of my friends, then I got it back when I upgraded him to my crap 1500x. Was happy about the imc though, when I tested it on my ocf a few months ago it seemed to get worse the higher the bclk and wouldn't break 2000 really. Now it does 2400 for me and gets some very nice results. Much better experience after all the ivy pentiums that my mve murdered in cold blood trying to train unstable mem settings in ln2 mode.

Perhaps I should pull my gma 950 board out for them sweet, sweet gpu glowballs