Should Sandy Bridge Users Upgrade? Evaluating the 2600K Against Modern CPUs

The funny thing about legendary chips is that it’s not always clear at launch which CPUs will go on to power enthusiast systems for years to come. Intel’s Sandy Bridge, like the Celeron 300A of old, has been an incredibly long-lived enthusiast part. SNB was the first major overhaul of the Nehalem architecture Intel debuted in 2008. It included a number of enhancements and improvements to the first-generation Core i7, including a micro-op trace cache, AVX instruction support, Intel Quick Sync video support, and a new GPU integrated directly on-die.

Eight years later, Sandy Bridge has proved quite popular — popular enough, that Intel has struggled to haul enthusiasts off of the product. But the question of just how much of an improvement enthusiasts might see if they did so hasn’t gotten tons of investigation. Anandtech decided to find out how invincible the 2600K is, eight years after launch. To capture the fact that the CPU was quite popular for overclocking, the site also included a graph of the chip running at 4.7GHz.

The results of the cumulative testing are shown below, though the full article is absolutely worth a read. Keep in mind as you look that Kaby Lake is actually running well below the Sandy Bridge clock on the overclocked variant (~4.0-4.2GHz, compared with 4.7GHz).

If you don’t overclock, the improvement from Sandy Bridge to Kaby Lake is considerable. Given that scores are normalized to the OC’d variant of the chip, we see that Kaby Lake is far faster than Sandy in certain cases — 1.65x faster than a stock-clocked SNB in rendering, and 1.39x faster in compiling. Gains, if you are overclocking, are very small, and generally in the 1.14x range.

This isn’t entirely fair to Kaby Lake, which is operating at a substantial frequency advantage (4.1 – 4.2GHz max, versus constant 4.7GHz), but since it was generally harder to get a 7700K up to 4.7GHz than Sandy Bridge, it’s still a fair look at the situation. If you’re running a 7700K at, say, 4.7GHz, however, you can make a reasonable assumption of a significant uplift in KBL’s favor.

I’m only going to discuss the gaming results in frames per second; if you want to see frame time data, hit the original story. This chart illustrates why there’s tension in the community between testing CPUs at low resolutions, where meaningful appearances in performance appear and high resolutions, where those differences are compressed.

At 720p, the 9700K is 1.85x faster than the 2600K. At 1080p, the gap has fallen to 1.54x. (the gap between the OC’d 2600K and the 9700K is 1.26x). At 1440p, there’s just a 1.21x gap between SNB stock and 9700K. At 4K, the gap is five percent — generally speaking, considered below the threshold we’d expect you to consistently notice.

Either way, it’s an interesting look at how much additional performance can be gained by upgrading one of the most popular CPUs of the past decade. It also shows off just how much the rate of improvements has slowed. A 1992 machine — even an absolute top-end one — was a wheezing, arthritic wreck by 2000, not matter how aggressively you deployed upgrades in the interim. Today, a 2600K is still running games just fine, in at least some cases.

Continue reading

Musk: Tesla Was a Month From Bankruptcy During Model 3 Ramp-Up

The Model 3 almost spelled doom for Tesla, but the same vehicle also probably saved it.

Tesla Model 3 Crash Hurls Battery Cells Into Nearby Home

Plenty of people wreck their cars. Most don't take out two trees, break windows in multiple apartments, and smash the plumbing in a third.

How L1 and L2 CPU Caches Work, and Why They’re an Essential Part of Modern Chips

Ever been curious how L1 and L2 cache work? We're glad you asked. Here, we deep dive into the structure and nature of one of computing's most fundamental designs and innovations.

Apple Is Building Its Own 5G Modem

What was rumored is now official — Apple is building its own 5G modem.