In this article, we'll be taking a close look at CPU gaming performance, pitting the venerable Intel Core i5-760, stock and overclocked, against the Intel Core i5-4670K. These two chips have something important in common - they are the ~$230 value leaders of their times, mid-2010 and mid-2013, respectively. Three years have past - four years, in fact, if you start counting from the i5-760's predecessor, the i5-750 - since the first modern mid-priced quad-cores hit the market. Amazing how time flies...but what kind of performance do we have to show for it?
This is one in a series of articles that will give you the information you need to make informed buying decisions - that's what this website is all about, after all! The basic question we'll be focusing on is this: if you have around $200-250 to spend to upgrade your circa-2010 system, which component should you upgrade, and how much performance can you expect out of that upgrade? Will an i5-760 "bottleneck" (hold back) a modern video card in the $200-$300 price range? We aim to find out!
OK, first we're going to set a baseline target for our comparisons. If you spent around $200 on a video card in mid-2010, you almost certainly bought the Nvidia GeForce GTX 460, which was a fantastic card at that time. Today, $200 buys you the Radeon HD 7870. We can tell you based on having tested both of these cards that the HD 7870 is twice as fast.
Now let's get to the more interesting part - if we want to replace our circa-2010 ~$200 CPU with today's reigning mid-range champ, the 4670K, what kind of performance boost can we expect? Is the performance jump equal to the one we see in video cards? To find out, we're going to be running an HD 7870 through our benchmarking suite on both an i5-760 and an i5-4670K. We believe the HD 7870 closely approximates the type of card many mid-market gamers are buying today. No, it's not a Titan, but frankly, if you're buying a Titan (or any high-end Crossfire/SLI setup in the $800-$1,000 neighborhood), you probably shouldn't be running them on a $230 CPU.
So, with that out of the way, here are the three test benches we used:
- Core firstname.lastname@example.orgGHz (stock), Asus P7P55D Evo, 8GB DDR3-1600, Sapphire HD 7870@1050/1250 (Catalyst 13.4)
- Core email@example.comGHz (overclocked), Asus P7P55D Evo, 8GB DDR3-1600, Sapphire HD 7870@1050/1250 (Catalyst 13.4)
- Core i5-4670K@3.6GHz (stock), Asus Z87 Gryphon, 8GB DDR3-1600, Sapphire HD 7870@1050/1250 (Catalyst 13.4)
These systems are identical other than the CPU and motherboard - and both of these can impact performance. In regard to the CPU, the most obvious factor is clock speed. The i5-4670K is rated at 3.4GHz, but under load, it actually operates all four cores at 3.6GHz due to Intel's Turbo Boost technology. The i5-760 is rated at 2.8GHz, but it runs at 2.93GHz under load using its own version of Turbo Boost. Right off the bat, the 4670K has a 23 percent clock speed advantage. Additionally, due to process improvements, the 4670K has an instruction per cycle (IPC) advantage of an additional 30 percent or so, bringing its total theoretical advantage to 60 percent. Overclocking the 760 to 3.5GHz, the best we could do with this chip, makes up for most of the 23 percent clock speed advantage, but none of the IPC advantage. Now, we just said that both the CPU and motherboard matter - the motherboard's contribution is its use of PCIe 3.0, providing double the graphics bandwidth of the PCIe 2.1 video card slot used on the i5-760's P55 motherboard. In practice, this typically only leads to a 2-3 percent advantage, but we should see this even where the CPU is not a bottleneck to performance. In other words, the new platform should always be slightly faster. We'll soon find out if that proves true!
As always, we try to account for variability as much as possible, so all of our tests were run three times, and we provide the mean result for both averages and minimums. Every benchmark in this article, other than 3DMark, was drawn from real-world runs in a game world, all in 60-second increments, all repeating the same loops. We use this more time-consuming method, rather than built-in benchmarks, because they much more accurately represent a gaming CPU load, whereas built-in benchmarks often over-emphasize graphical elements while requiring little CPU processing power.
Two additional notes:
- we really hoped to bring you true "clock-for-clock" data on the i5-760, but we simply could not get the CPU stable above 3.5GHz - overclocking truly is a "silicon lottery" and you can't take anything for granted!
- we did not test with more powerful video cards, even though the i5-4670K would likely stretch any potential lead over the i5-760 given more GPU power. Our rationale is simple - we wanted to give our cost-conscious readers something to chew on as they plot their midrange upgrades.
So, with all of that introduction out of the way, on to the benchmarks!
3DMark Fire Strike Standard Preset
3DMark Fire Strike helps set the bounds for what we can expect throughout our benchmarks. We highlight the Graphics Score and the Physics Score, as they are likely to be the most purely GPU and CPU bound, respectively. The Graphics Score shows just a slight advantage for the newer system - 5751 versus 5530, or a 4 percent bump. That is likely mostly due to the use of PCIe 3.0 on the newer platform's motherboard. Now, the Physics Score shows something entirely different, being a nearly ideal test of theoretical CPU processing power. The 4670 scores 7240, a healthy 40 percent higher than the 760's score of 5170. Remember how we said that the 4670K should have a combined clock speed and IPC advantage of 60 percent? Well, we're not giving away too much by saying that this delta will be the exception rather than the rule in our benchmarking suite.
Looking at the overclocked numbers, only the Physics Score goes up, and it scales perfectly with the overclock - a 19 percent improvement for a 19 percent clock increase. It also makes up half of the deficit versus the 4670K, slotting in 17 percent below that chip. On a clock-for-clock basis, we could extrapolate that the 4670K is actually only 15 percent faster.
Deus Ex: Human Revolution - Hengsha Landing Pad
Deus Ex: Human Revolution, released in August 2011, is the oldest game in our benchmark suite, and also has the least sophisticated game engine in this test, meaning it presents the lowest load to the video card, and potentially shifts more of a burden to the CPU to keep up. In other words, we may see the graphics card idling as it waits for the CPU to finish what it has to do to render the game to the screen. And the results echo this - we have the clearest sign of CPU-bottlenecking among all of our gaming tests. The i5-760 hits 99 frames per second (fps) on average, and drops to 57 as a minimum. In comparison, the i5-4670K hits 113fps on average, with a minimum of 83fps. There's just no doubt about it - with an average that 14 percent higher and a minimum that's 45 percent higher, the 4670K absolutely changes the game, no pun intended.
Another sign that the CPU was holding the HD 7870 back - we found that the GPU usage was hovering around 92 percent during our multiple runs on the i5-760, versus the ideal 99 percent GPU usage we'd get in a fully-optimized game experience, such as we had on the i5-4670K.
Looking at the overclocked i5-760, we see a huge uptick in the minimums, all the way to 73fps, an above-theoretical 28 percent boost from the stock clock speeds. It also nearly catches the 4670K by hitting 108fps on average.
Battlefield 3 - Swordbreaker (Single-Player)
We now turn to one of the most popular games of the past two years, Battlefield 3, or BF3 for short. It's still a graphics powerhouse, one of the showcase DX11-based game engines of 2011, and continues to look great in comparison to current games, although it is soon to be replaced in the echelon of great games by Battlefield 4.
These results are actually incredibly close, with the average of the i5-4670K coming in at 62.3fps, versus 59.7fps for the i5-760 - a 4 percent advantage. The minimums are actually dead even at 42.5fps. We'd guess that the difference here is again a result of the use of PCIe 3.0, as there just isn't a lot for the CPU to do in BF3's singleplayer mode.
In regard to the overclocked i5-760, the minimums strangely drop to 40.7fps, but the average jumps to 61.3fps, again nearly catching the 4670K. This is among the most GPU-bound tests in our suite, so there isn't much room for the overclocked 760 to move.
Battlefield 3 - Caspian Border (Multi-Player)
Now this is where things start to get interesting: BF3 multi-player, a true test of any CPU's mettle. While it doesn't ask much more of video cards than its single-player counterpart, it absolutely hammers CPUs, so much so that dual-cores can barely run the game. Luckily, the i5-760 still has what it takes, averaging 53.2fps with a minimum of 43.3fps. That's a bit slower than the i5-4670K, which hit an average of 56.2fps and a minimum of 46.3fps, a 5.5 percent and 7 percent difference, respectively. The i5-4670K is definitely pulling ahead here due to its greater processing power, and presented with a lot more GPU power, we'd likely see any even bigger difference. But remember, this test is about the ~$200 upgrades, and we can comfortably recommend a $200 video card for users of the i5-760 who might otherwise be concerned about losing too much potential performance.
The overclocked i5-760 more than makes up the delta, actually surpassing the i5-4670K here by hitting 58fps. Keep in mind that this is in a live multi-player round, so there's a pretty significant margin of error, but even so, it's pretty clear that overclocking the i5-760 brings it up to the same basic level as the 4670K.