Lynnfield i7 & Skylake i5 Performance Compared

Supernova · 18 Jun 2016 at 01:11

I finally have the answer to the question I've been asking myself and others for months - is it worth upgrading my i7 860 @ 4.0GHz to an i5 6600k? The answer is yes. At least in terms of gaming.

Before deciding to upgrade, I tried to identify if my CPU was causing a bottle neck. Even though it didn't reach 100% in any of my games, I had my suspicions.

I went ahead and bought a 6600k, 16GB of DDR4, and a new motherboard. Before I disassembled my old PC, I ran 9 benchmarks that I hoped covered a broad range of tasks to test the PC. I then ran these again with the new PC and it produced some interesting results. The GPU remained constant. Specs at the bottom for both systems. Bench results in the spoiler tags.

At first I started with Heaven and really didn't see an enormous difference - only small gains. In Cinebench CPU and the 3D Mark physics test, the old i7 pulls ahead, especially in firestrike.

The real interesting scores are the Steam VR test, Cinebench Open GL, Alien Isolation, and DiRT Rally. Alien Isolation shows an increase of 70% in average frames per second, and Cinebench Open GL is a 59% increase. The frame times have come way down in Alien Isolation, and I believe this reduces the appearance of stutter. In Steam VR, out of a maximum score of 10, it jumped a whole point upwards. I'm kicking myself for leaving v-sync on in DiRT rally when testing the Lynnfield system as this would have shown an enormous jump had it been off.

I know it's not strictly testing just the CPUs due to the different OS and RAM, but as the question of whether a CPU upgrade is worth it is rather common I thought it was worth sharing.

Lynnfield PC
OS: Windows 7 SP1
CPU: i7 860 @4006 MHz 8 Threads
GPU: R9 290 @1103 MHz/1403 MHz
RAM: 8GB DDR3 1908 MHz

Skylake PC
OS: Windows 10
CPU: i5 6600k 3500 MHz 4 Threads (stock)
GPU: R9 290 1103MHz/1403 MHz
RAM: 16GB DDR4 3000 MHz

theMilitant · 18 Jun 2016 at 03:28

What resolution were these benches run at?

Supernova · 18 Jun 2016 at 09:15

These were all 1920x1080 except for Steam VR, because that runs within a window.

mickyflinn · 18 Jun 2016 at 11:41

MSTRBKR said:
I know it's not strictly testing just the CPUs due to the different OS and RAM, but as the question of whether a CPU upgrade is worth it is rather common I thought it was worth sharing.

Whats your final conclusion ?

Supernova · 18 Jun 2016 at 12:04

mickyflinn said:
Whats your final conclusion ?

I don't think the RAM capacity is having any effect, but the speed of it could be. For me, it's been a massive improvement to go to Skylake. I ran DiRT on the High preset with 2xMSAA on the old system, and it would frequently dip to 30FPS on demanding courses where there are lots of shadows, god rays, and trees.

Now I run the game on the Ultra preset with 8xMSAA and it's always at 60FPS.

This also shows that benches like 3DMark don't tell the whole story, as you can see from my tests that the new PC score is not that much higher.

Stealthploit · 18 Jun 2016 at 13:46

well having 8 threads makes a lot of diff

Supernova · 19 Jun 2016 at 00:09

Yes, but the point is that real gaming performance has improved massively. Benchmarks prove very little.

koooowweeee · 19 Jun 2016 at 06:58

I'm pretty sure everyone knew the results before opening the thread

Poneros · 25 Jun 2016 at 17:05

The real improvement is the min fps, that's always massive coming from older CPUs.

Pkiely · 26 Jun 2016 at 13:36

A little reassuring. I also run a Lynnfield i7 (870) but at 3.8ghz and have held off for a long time upgrading because it still runs the games I play fairly well.
Upgrading the GPU comes first however, AMD7950 still. The longevity of this build has been remarkable, I almost want to keep it going until it starts to have hiccups whenever that may be.