Hi all

I'm sharing a solo run on the Ivy Brifge platform. have problems with the automatic hwbot submission for this specific linux build.

Benchmark: y-cruncher Pi 5b (5,000,000,000

Total computation time: 1479.532 seconds (24.659 minutes)

Multi core efficiency: 97.18% (kernel overhead: 0,24%)

CPU Intel Core i5 3570k 4.44 GHz (Fixed)

Motherboard: MSI Z77 MPOWER (LGA 1155)

OS/tuning: Linux/11-SNB - Mina =Haruka (build 0.8.7.9547-static)

The run comleted with stable temp of 68° C. The eficiency here is roughly 2x faster than average Windows optimized run for this cpu class

I've attached the full screenshoot showing the Spot Check Goog status and the final time. i also have the .txt log available for verification. if any1 can help me with the .ycv validation for this linux build? Please let me know!

Pi - 20260403-200113.txt

Hello,

We took this decision for Windows only Operating Systems as it's output it moderateable as scores, performance boosts are reproduceable by the moderators. Even with Windows, checking some 3D subs is sometimes an utter nitemare.

The optimisations which each new big Linux Kernel release can destabilize the rankings time after time. Which implies users will continuously need to rebench, spending too much money on eg liquid nitrogen to maintain their Hardware or even Global points or get asswhooped by benchers on air/water clocks and far lesser clock speeds. We know there there's a small group of benchers at HWBOT that would applaud this fact, but we took a stand and prefer to highlight Hardware OC over pure Software overclocking.

I assume you can sub these scores Linux based scores at Y-cruncher site, sadly we won't allow them here.

Have a nice weekend

Hi leeghoofd,

Thanks for the detailed explanation. I understand the challenges regarding moderation and kernel variance.

However, seeing the results I’ve achieved with my 'Ice Tractor' project—reaching a 97.18% multi-core efficiency with only 0.24% kernel overhead on an Ivy Bridge i5—it feels like a wasted opportunity for the OC community to ignore this level of system engineering.

Would HWBOT ever consider creating a dedicated Linux OC Section in the future?

This would keep the Windows rankings 'pure' for the hardware-only crowd, while providing a space for those of us who want to push the absolute limits of hardware-software synergy. It would be a great way to highlight true technical excellence without 'destabilizing' the current global points system.

I believe many advanced tuners would welcome a category where efficiency is as important as raw MHz.

Best regards,

I understand your point, but it feels like a paradox. It’s true: in this community, I might look like someone 'cheating' the benchmarks by using Linux. But I always thought that overclocking meant reaching the highest possible performance by mastering every single aspect of the machine.

My 'Ice Tractor' project isn't about 'software trickery'. It’s about System Engineering. I’ve spent months achieving a 97.18% multi-core efficiency and reducing kernel overhead to 0.24% precisely because I want to take 100% control of the hardware I’ve assembled.

If we limit ourselves to one OS, we aren't testing the true limits of the silicon; we are just testing how well the silicon survives the inefficiency of Windows. I thought the spirit of HWBOT was to push boundaries, not to set artificial ones on the software that manages the hardware.

Isn't the goal to find the absolute performance? If overclocking is about pushing the limits, why do you accept the limitations of an inefficient operating system like Windows?