Don’t Bother Overclocking an i5 13600K…

Let’s be real here, you’ve already watched Linus’, Steve’s or.. Steve’s video on these new chips. You know how they perform and how they compare to last gen and Ryzen 7000. So let’s shake things up and do something a little different. I’ve spent the last week overclocking and testing the i5 13600K to see if it’s even worth doing – and I’ve found some rather interesting results.

If you haven’t seen the specs, Intel basically took their 12th gen chips, doubled the efficiency core count and called it a new chip. Ok there are actually a few other differences, but the extra E cores are the main event here. That means their 13600K now sports 6P and 8E cores, making for a “14 core” chip that can boost up to 5.1GHz. It’s worth noting the i9 will boost to an astonishing 5.8GHz in stock form! I should also mention the new chips now support up to DDR5-5600, up from 4800 last gen, and of course the new chips are accompanied by new motherboards. Specifically Z790 boards – I’ve been using the Asus Z790 STRIX E for my testing with the most recent BIOS version available.

So, let’s get overclocking. This has become ridiculously easy – I did everything through Intel’s Extreme Tuning Utility, or XTU, which was fantastic. I started by just trying to increase the power limit – the i5 now has a power limit of 181W at stock, up from around 150W, although as you’ll find out I’m not sure it got the memo. But still, just increasing the power limit and boost time, much like last generation, did nothing. Actually, it dropped the performance slightly. Amazing. I don’t have any other boards to be able to test if it’s Asus’ BIOS or a bug with the chips in general, but if you do want to push your chip you’ll have to do it manually.

Happily, not only is this thing an absolute monster for overclocking, but it’s stupidly easy with XTU. You just crank the voltage up a bit at a time, bump all the core clocks up a step at a time, then run a test like Cinebench. Rinse and repeat until it stops being stable, then either bump up the voltage some more or back off the clocks. I managed to get 4.2GHz on all 8 E cores, and 5.7GHz on all P cores, and even 5.9GHz on the two best P cores. That meant the chip was running at around 220W of power consumption, up from just 150W at stock in my testing. That meant the Cinebench score jumped from 24,000 points, to 26,622! For a bit of context, the 12900K last gen ran at around 27,000 points in Cinebench, meaning an overclocked i5 is almost as fast as a stock i9. That’s impressive.

Taking a closer look at the results, even the single thread performance jumped, going from 1947 to 2143, a 10% similar 10% bump in performance, albeit for almost 50% more power. In Blender and the BMW scene, the render time dropped from 106 seconds to just 97 seconds, or again around a 10% improvement, and in gooseberry it dropped almost a minute off the render time. After some re-runs thanks to instability, in the Adobe suite we went from 953 points in Premiere Pro to 990, 782 points in After Effects to 840, and 1357 points in Photoshop to 1487. Those are some pretty healthy gains!

If you are wondering about gaming, unless you have a 4090 – in which case why are you buying the i5 – I’m not sure you’ll see a difference. I tested with an RTX 3060 here as it’s the sort of card someone buying the i5 is generally going to run, and the difference in performance is pretty negligible. In CPU bound games, like CSGO, you do get a healthy improvement at 1080p going from 565FPS to 662FPS, or 17% faster, with a near 50FPS improvement in the 1% low figures too. Even at 1440p you net an extra 40 FPS on average with the overclock, and 22 FPS more in the 1% lows.

But moving to any other game that isn’t purely CPU bound, you’ll find a distinct lack of improvement. Cyberpunk on medium settings ran at 130 FPS average on both stock and overclocked configurations, and 105 FPS at 1440p. Shadow of the Tomb Raider ran at 137 FPS average on both configs at 1080p, or 94 FPS average at 1440p. And Microsoft Flight Simulator ran at 112 FPS, although here there was an improvement in the 1% low figures at 1080p.

So unless you have a top tier GPU, or you only do compute heavy work like editing or 3D modelling, you probably don’t need to overclock the 13600K. But what about underclocking? Well I thought I’d test that a little too. I set the power limit to 120W, and unsurprisingly the performance did drop in the compute benchmarks. Cinebench drops to 22731 points, down from 24,000 at stock or 26,662 with the overclock (which draws nearly twice as much power mind you), and the single thread performance dropped a touch too down to 1911 points. Blender also showed a decrease in performance, with the BMW scene taking 7 seconds longer than stock, and Gooseberry taking nearly 30 seconds longer. The Adobe suite performance suffered too, with Premiere running at 934 points, After Effects almost identical to stock at 780 points, and Photoshop taking the biggest hit at 1215 points.

In gaming though, there isn’t much of a difference. CSGO is within margin of error to stock, both at 1080p and 1440p. Cyberpunk is exactly the same as both stock and overclocked, so is Shadow of the Tomb Raider, and again so is Microsoft Flight Simulator. Seriously, unless you do compute heavy work, you are better off saving 30W and just lowering the power limit. That’s how fast these new chips are.

So, in short, you probably shouldn’t bother overclocking the 13600K, even if it can take a whole lot of pushing. Seriously, running at 5.7GHz ALL CORE with an AIO – admittedly a 360mm but still – that sort of frequency was unthinkable without liquid nitrogen. But now anyone with a half decent motherboard, a good cooler and a few hundred in their pocket can smash it. What a world.