TaPaKaH

Drop down and give me 3min37, private !

The switch happened only on kits whose specs allowed for it. As much as G.Skill would like to use cheaper ICs on, say, 2400C8 PI, there simply aren't any alternatives out there. It's not impossible to find 1600C7 with X-series that are new, but chances are those won't be good. Normally most 2200 9-9-9 kits are BBSE, but from the seventh digit of the bottom marking, I see it's a "C" or a "D", indicating Elpida BCxx (which officially never existed) or BDBG/BDSE.

1337 !

You managed to get global points with SDRAM in 2021, haha :D

Another one bites the dust

Yes. All of my 3D are with stock VRM

You made me look like an inefficient retard. Good job! :)

That's been some looong sleep, padre

Awesome Hyper clocks!

@ground: Still waiting for 300 BCLK X58

Northwood style I've only seen Wolfdale-6M chips on E5200 and E7200 and heard about Celeron E3200/E3300. Not sure if other models ever even existed.

For E5x00/E7x00 you need PLL under the default 1.5V, which only Gigabyte P45 and PLL-modded boards can set. Without PLL mod, the current #1 E5200 stops at 390 air, 480 LN2. Also, you can immediately discard Wolfdale-3M derived ones (96-98%) and focus on Wolfdale-6M ones. You can distinguish these physically. I've tried around 100+ of the latter, which would yield same results as 2000-4000 random ones.

This is still very lucky I tried 1500-2000 pcs E2140, E2160, E2180 and my best chip did 565 LN2.

Your luck is absurd. I tried about 3000 (of E6300, E6400, E6750) and found only three that would barely do 600 on air. Nowhere near 615.

How many E6300 did you try?

Are these any good in 32M? Are these single- or dual-ranked sticks?

Awesome!

Try another board. GA890FX aren't the fastest 2D boards out there...

Looks like I got out of 775 (and benching altogether) at the right time :D

So we're back to 1800-1900 mem clocks for max 32M efficiency now?

We made two local supermarkets run out of weissbier. Hence why P3 only ...

awesome FSB for ambient!

700x9.5 on E8500 ... http://noisecutters.com/wp-content/uploads/2020/05/You-did-it-Product-Hunt.png

Awesome clocks! But I guess you can do 7min46-7min48 with better efficiency

Another argument for allowing to disable / downclock cores: continuity and simplicity of rules. Imagine a new person trying to submit a score in 2021 and having to learn about differences in rules for pre-2010 and post-2010 hardware? Perhaps a separate hardware point category for all-cores-same-clock would suffice? It's not like there aren't hundreds of these already per each piece of hardware.