I’ve always been a big fan of CPU round-ups and the test results that come out of them are always interesting to look at. Over time you can see trends slowly forming, but having an entire decade’s worth of work in one lab for testing can reveal things that even seasoned hardware reviewers could have missed when looking at the chips inside their original launch week. Techspot’s Steven Walton took out some time to do just that, and compared thirteen CPUs from Intel’s past and present lineup. This kind of roundup is especially useful if you’re still on older hardware and haven’t seen any concrete proof that an upgrade will be a good idea, so follow me after the jump to see a few of the results and grab a link to the source article to see all the other interesting things that came out of Walton’s work.
The jump from Conroe to Haswell has been a very long road indeed. We’re still using the basics of the Core architecture that Intel created back in 2005, where they repurposed the design and architecture of the Pentium III to eventually create the Pentium M, a Frankenstein chip composed of the remains of the Pentium III and all the good things that came out of Pentium IV. Conroe belonged to the Core 2 Duo family, which were the first dual-core processors in Intel’s lineup. Core Duo is technically ahead of Core 2 in that regard, but it was merely two physical processors mated to each other on the same die and package, whereas Core 2 was a native design and much more efficient.
Over the years, Intel has reiterated upon the Core architecture no less than nine times in the past ten years. From Yonah through to Merom, then Penryn, then Clarksfield, Arrandale, Sandy Bridge, Ivy Bridge, Haswell, and then finally Broadwell, Intel’s been on a tick-tock release. A “tick” for Intel is where they shrink a previous architecture down to a new production process, like 32nm to 22nm, while a “tock” is a major architecture change. In the past, a “tock” implied significant CPU performance increases, but these days that isn’t the case.
Intel only really targets 5% peformance improvements these days and pays more attention to improving the performance of specific instruction sets, or adding in dedicated hardware to improve efficiency, or even make changes to the GPU instead, as seen in the shift from Haswell to Broadwell. I personally also view the Iris Pro family as its own set of “tocks”, because the eDRAM embedded into the chip, which serves as L4 cache, is a huge change to how the CPU accesses and works with larger data sets now.
Looking at Techspot’s results, the shift as you move up the ladder is either going to be very apparent, or not at all. I’m only looking at three of their graphs here (there is plenty more data in the article itself), but the basic idea is what you’ve probably been hearing for years – when you’re in a GPU-limited situation, there’s still some benefit to using a faster CPU, but its not really a huge leap.
That becomes more pronounced when you take a look at the Core i5-760 results. The Core i5-760 is still pretty damn functional all these years after launch. Tomb Raider’s results at 1080p are really not unplayable, and its placing on the PCMark 7 graph isn’t bad either. It still gets a run for its money by a lowly Celeron G1820, but that’s as a result of Intel’s improvements to the Core architecture over the years. Single-threaded performance plays quite heavily into the G1820’s performance, so its not surprising to see it do as well as it did in these tests.
Finally, going into the power consumption graph, some context is required for a few of the results. AMD’s A10-7850K almost always finishes with the same or better result to the Core i5-760, but it only uses more power in Prime95 because it also has to feed a bigger and more complex GPU. If there was a Athlon X4 860K in this round up, you’d see power consumption drop to the same level as the Core i5-2500K.
So if you’re on an older rig and you’re looking for something to help you figure out if you should upgrade or not, I’d highly recommend giving this article a read. While it won’t be a comprehensive answer to your particular question of whether you should upgrade your system, it can’t hurt to arm yourself with more information and make a more informed decision as a result.