Extreme Overclockers Fill Coffee Lake With Liquid Nitrogen

Subject: Processors | October 6, 2017 - 11:44 PM |
Tagged: Extreme Overclocking Competition, overclocking, liquid nitrogen, coffee lake, i7 8700k

A new CPU means new overclocking challenges and with it comes a new batch of refreshed Z370 motherboards. At the high end, the current frequency record for the Core i7 8700K is 7,405.1 MHz obtained by Hovan Yang using a MSI Z370 Godlike Gaming motherboard.

View Full Size

He’s not the only one testing the limits of Intel’s new six core processors though. Asus held an overclocking event a few weeks ago where renowned overclockers Alex@ro, elmor, der8auer, Rsannino, and shamino battled it out. Der8auer got a pre-release crack at the i7 8700K at the event and after de-lidding and replacing the TIM with liberal amounts of Kryonaut thermal paste managed to achieve 6.8 GHz using 1.8 volts and a 68x multiplier (and bumping the cache speed up to 6.3 GHz). With these settings on the monster Maximus X Apex motherboard, he scored 299 in single threaded and 2253 in multithreaded in Cinebench R15. Der8auer compared this benchmark result to Skylake X at 5.5 GHz scoring 237 in the single threaded test. Following the benchmark run, he went for the highest CPU-z validated clockspeed he could hit and managed to push the chip to 7300 MHz (100MHzx73). From there overclocker Alex from Romania was able to overclock his i7 8700K to 6844 MHz and scored 2306 in Cinebench R15.

View Full Size

The overclockers broke 10 new records in the six core CPU category and also managed to break a DDR4 clockspeed record by pushing a single 8GB G.Skill DIMM to 5529.2 MHz at 24-31-31-63-3 timings!

Also of note is that Coffee Lake does not depend of FIVR so overclockers are able to use a full pot of liquid nitrogen (or liquid helium) to cool the processor down to much lower temperatures so that they can crank up the voltage and achieve much higher clockspeeds than Skylake-X which cannot boot if temperatures are too low.

While the ASUS team does not hold the clockspeed record anymore (though they might regain it with some Liquid Helium), der8auer has an interesting video and Asus has a blog post with photos talking about the process, setup, and everything that goes into these extreme overclocking sessions including pre-binning the chips, preparing the IHS and motherboard for the super cold (-185°C to -190°C) temperatures, and keeping the processors and motherboards running. For example, and Josh will be interested in this, part of the process of preparing the motherboard involves slathering it in Vaseline!

If you are interested in this extreme overclocking stuff it gives a bit of insight into all the fun to be had!

Source: Asus

October 9, 2017 | 01:07 AM - Posted by Mr.Gold (not verified)

"Because he can"

Maybe its just me, but I'm more impressed standard overclock.
7.4ghz under nitogen, delided... meh

Show me a stock configuration running prime95 at 5ghz for 24h, I would be more impressed.

October 10, 2017 | 11:59 AM - Posted by GradeTheReviewsAndReviewers (not verified)

I hate CPU benchmarking where many CPU SKUs from both AMD and Intel are tested and then on the graphs the reviwers fail to list the Respective CPU SKU's core/thread counts.

So there will be maybe 10+ CPU SKUs listed on the graphs/charts and no core/thread info included on the graphs/charts to be found. It's one of the most disingenuous things a reviewer can do while trying to compare various CPU metrics such as power uasge on SKUs such as the i7 8700K(6 cores/12 threads)/older K versions and any Ryzen 7 1700/X(8 cores/16 threads)-1800/X(8 cores/16 threads) SKUs.

There has been the usual testing between the Intel 8700K(6 cores/12 threads) and the Ryzen 7 series(8 core/16 thread) SKUs and no per-core power usage metrics to be found and that's no accident that smacks of that Review Manual influnced nonsense that is so subtlety pervasive where Review samples and ad revenues are at stake.

Blender Rendering is such a good metric on one hand while it can load a CPU's cores at 100% in such a reproducable way but sometimes, and sadly much too often, there are not very many per core enegry usage metrics presented when compareing CPU SKUs with differing core/thread counts and that's no accident. I do not trust any review/reviewer who produces graphics/charts and does not provide right next to the CPUs listed those respective CPU SKU's core/thread counts. And core/thread count is a rather important CPU metric especially when comparing CPU SKUs with different processor core/thread counts.

This lack of information in reviews on those charts/graphs/graphics is very telling across the whole interwebs based review industry and it's raises too many suspensions of non objectivity in reviews/reviewers.

Blender Rendering does such a good job of equally loading a CPUs cores/threads at 100% so why can there not be some per core enegry usage metrics provided and it's simply a matter of devided enegry used by the number of cores. And that per-core enegry usage is great when comparind to CPU SKUs of unequal core/thread counts. It's so very easy to spot may naferious attempts at subjective spin when important metrics are not included any many online reviews.

October 10, 2017 | 12:07 PM - Posted by GradeTheReviewsAndReviewers (not verified)

Edit: any many
to: on many

Edit: comparind to
To: comparing two

My brain's wiring is so cixelsyd.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.