Intel 12th-Gen Core Alder Lake Architectural Benchmark

  • 2021-12-01 06:00
  • |
  • 323
  • |
  • Recompile By Brian - CPU Review
  • |
  • Comments

A few months ago we investigated misconceptions about CPU cores, explaining how the overall performance of a processor is affected not only by the number of cores the CPU has but also other factors including cache level and capacity. Here's an interesting and unique look at Intel's 10th Gen series in an article we titled "How CPU Cores & Cache Affect Gaming Performance". Basically what we did was compare the Core i9-10900K, Core i7-10700K and Core i5-10600K at the same 4.2 GHz frequency, with the same memory, memory timing, loop bus frequency, etc

Then compare the three CPUs with only 6 cores/12 threads enabled to see how much of a difference L3 caching has made when it comes to gaming performance. We then compare that data with 10700K and 10900K with 8 cores enabled and finally 10900K with all 10 cores enabled.

In summary, it turns out that in most games it is not the number of cores but the amount of L3 cache that is responsible for the improved performance on the higher end Intel parts. Of course, tracking the extra cores will see the higher end parts pulling further forward, but at least on today's games it's all about L3 caching.

That investigation then turned to a quad-core version where we included Core i3 models and a similar version for AMD CPUs, where we looked at the progress of AMD CPUs over 10 years and back. back to Intel.

To that end, we thought we should add the new 12th Gen Intel Alder Lake CPUs to the data set, so here we are, and it's a lot more involved than we think. I originally imagined. While all other CPU architectures have one, two, or maybe three different configurations, the 12th Gen Core has three configurations per CPU.

For example, 10th Gen CPUs with 20MB L3 cache for Core i9 models, 16MB for i7 models and 12MB for i5 models, similarly segmented Alder Lake cache capacity, 20MB L3 for Core i5, 25MB for i7 and 30MB for i9. But above all, we have to figure out what kind of core configuration we should test. Four P cores, four E cores or a mixture of both? The correct answer is of course all three configurations, and that has provided us with loads of fascinating data to look at.

To be clear, with four P-cores enabled, we're using Hyper-Threading, so this is a 4-core/8-thread configuration. Basically SMT was enabled when supported for all test configurations. This means that since E cores do not support SMT, the 4-core E configuration is 4-core with 4 threads. Then the hybrid configuration with two P cores with two E cores is a 4-core/6-thread configuration.

For testing, we used MSI Z690 Tomahawk Wi-Fi DDR4 because we wanted to use the same DDR4-3200 CL14 low latency memory used to test all other CPU architectures that support DDR4 . In our testing, the DDR5-6000 didn't show any faster speeds for gaming, but most importantly we wanted to keep the data as good as possible for this feature. Finally, all configurations were tested using a Radeon RX 6900 XT. Let's dive into the data.

Starting with Rainbow Six Siege, there's a lot more to go through, so bear with me. First, take a look at the Core i9-12900K, we see with the four P cores enabled and locked at 4.2 GHz, this configuration performs well at 510 fps, just 3% faster than AMD’s Zen 3 architecture.

Then with two P cores and two E cores enabled, the performance drops by 15% which is quite a significant drop and then with just four E cores the performance drops another 12%, not much and not close. the decline I was expecting. Quite surprisingly, in this title four E cores can match the performance of a Core i9-11900K, even though the 11th gen architecture doesn't bode well for the title, I still wouldn't expect will see any results like this.

When we compare different 12th generation processors, we see that from 12600K to 12700K the additional L3 cache increases performance by 4% with only P cores or 7% with E cores only. Then, from 12700K to 12900K, we're looking at another 5% performance boost for the P cores and a pretty substantial 10% gain for the E cores.

If we compare all the data we have, we see that their E-core enabled 12th gen CPUs are comparable to Skylake as Intel has claimed, at least when looking at the data. 12900K. It's also interesting to note that with two P cores and two E cores, the 12900K is a bit slower than the Zen 3's four cores. So this suggests that a part like the 5950X will be much faster than the 12900K to play with. games, once the games use as much 16 cores... in 10 years from now.

Moving on to the Battlefield V results, we get some interesting insights. First, is that the E cores have attracted the big time in this title, not only is the average frame rate almost halved from what we see when using only the P cores, but the performance is low. % is also broken.

We're looking at a 22% drop in performance with the 12900K when moving from 4 P cores to 2 P cores and 2 E cores. Then we see an additional 31% drop when switching to E cores separately. Worse, that which means that the P cores are 87% faster when looking at the average framerate and 170% faster when looking at the low 1%. So those efficiency cores are horribly slow in this game and anything but efficient.

We also found that when the E cores were enabled, the larger L3 cache capacity of the i7 and i9 models did not result in any additional performance, or at least very little in the way of performance gains. However, with only P cores, the 12700K is 6% faster than the 12600K and then the 12900K is 7% faster than the 12700K.

If we compare that data with the rest of the CPU architectures we've tested, we'll make some notable comparisons. When compared to Zen 3, Alder Lake is up to 12% faster, seen when comparing the 12900K with the 5800X. That said, the 12600K's smaller 20MB cache means it's 2% slower while the 25MB i7 is just 4% faster. So it was that larger 30MB L3 cache that kept the Core i9 stable on the line.

That said, if we force Intel to use E-cores for gaming, we find that the mixed 2-core P/2-core E configuration lags behind the Zen 3. Then if you just use cores E, performance fell off a cliff and now that we're talking anywhere near Skylake's gaming performance, think more about Sandy Bridge.

Moving on to F1 2020, we see that the E cores aren't as bad as what we saw in Battlefield V. We're looking at a 65% performance increase with E cores looking at the 12900K and a 43% increase with the 12600K. The 12600K seems to be suffocating by its smaller 20MB L3 cache as the 12700K is 18% faster when comparing P-core performance, while the 12900K is only 4% faster than the 12700K.

Compared to Zen 3, Alder Lake was slower when limited to 20MB of L3 cache, then up to 10% faster at 25MB and 12% faster at 30MB. For E-core only configuration, Alder Lake is comparable to Ivy Bridge in F1 2020 and far behind Skylake, for example 7700K is 33% faster than E-core configuration of 12900K.

The NPC's heavy Hitman 2 test destroys the E-cores. This is similar to what was seen when testing with Battlefield V. Performance on all three 12th Gen installments is similar and that means we are looking at a 41% performance improvement with 2 cores. P and 2 E cores compared to just using E cores. In fact, if we look at the 1% low performance, it's closer to a 134% increase, which is insane.

We then saw when using only the P cores, the average frame rate improved by 27% when compared to the mixed core configuration.

So again, if we compare the E-core only configurations with the older CPU architectures, we see that the performance is not as good as Skylake. The 1% low performance is just as bad as what we've seen with AMD's Bulldozer, while the average frame rate is much closer to Ivy Bridge than to Skylake.

Even in Horizon Zero Dawn, which isn't particularly CPU-intensive, the E-core-only configurations still struggle, although it will eat up all 4 cores/4 threads, especially if they're slow. If we look at the 1% low performance we see a 104% increase from 4 E cores to 2 E cores and 2 P cores, while going from a mixed core to 4 P core configuration only increases performance by 14 %. . We also noticed very little performance difference between different L3 cache capacities in this game.

If we compare it with older CPU architectures, we find that Alder Lakes' E cores are not much better than AMD's FX series. The 1% low efficiency is almost identical and that means we're miles away from Skylake here.

Cyberpunk 2077 is another game where the E cores can't push 1% lows to 60fps, can't even close. As a result we saw a 100% improvement in performance with the 12900K when comparing the E-core with the mixed-core configuration and then only a 12% increase when using the P-core only. Interestingly, the mixed-core configuration of the 12900K is quite good, while we see a significant drop with the 12700K and 12600K.

Comparing with previous CPU architectures, we see that E cores are much slower than first generation Ryzen and world slower than Skylake. We're looking at the performance level of Sandy Bridge here.

Finally we have Shadow of the Tomb Raider and here we see very little difference between the different Alder Lake CPUs, so cache size has almost no discernible effect here, at least. is for these core configurations. We've seen in the past that with the 10th generation product line, a larger L3 cache provides greater utility with more cores.

When compared to older CPU architectures, the E cores struggle with gaming on their own, with average frame rate performance comparable to Ivy Bridge and 1% low performance comparable only to the FX series. of AMD. On the other hand, when using only P-core, Alder Lake is a monster that beats Zen 3 by 11% in this game.

It's eye-opening to say the least. Those E cores don't perform well when gaming, and there's a good reason why, which we'll explain shortly. Now let's take a look at the average scores of the 7 matches we collected.

Across the 7 games we tested, we found that the 12900K is only 3% faster than the 12700K in P-core operation and 8% faster than the 12600K, and those returns are entirely dependent on the capacity difference. amount of L3 cache. The margins with two P cores and two E cores enabled are similar and the same is true for just four E cores.

The interesting story, of course, is the performance difference between different core configurations on the same CPU. So taking the 12900K for example, we saw a 44% increase in average frame rates when going from 4 E cores to a mixture of P and E cores and an 81% increase in 1% low performance. Then from a combination of P and E cores to just P cores, the average frame rate was increased by 20% and the 1% low was 21%.

Obviously you would never run a 12th gen CPU with only E cores, which would reduce performance by around 20%, but let's dive deeper into the analysis in our conclusion.

Intel's 12th Gen hybrid core design is really exciting and it offers some clear benefits for productivity workloads and is sure to prove very beneficial in the mobile space. Now you might be thinking, "sure, I've looked at the benchmarks, I understand that the E cores don't perform well for gaming alone, but why?"

The answer is simple and that's also why the first generation Ryzen is not Intel for gaming when matching at the same core count. Core-to-core latency is very weak - we're talking an average 54% increase.

Typically, P cores take 37ns to communicate with each other while E cores take 57ns and this cripples performance in games and for any other workload that relies heavily on core crosstalk.

The reason for limiting connectivity between Intel's E cores is to make them more efficient, both in terms of power usage and the amount of memory they require. For example, for sequential workloads like what we see in rendering, when there is very little core-to-core communication, the E cores perform well and this is why Intel uses SPECrate2017 to make a statement about their Skylake effectiveness.

If we look at the broader picture, the hybrid design even on the desktop makes sense, at least for Intel. A part like the Core i9-12900K can claim to contain "16 cores total" with 24 threads, because that's technically what it packs, even if not all cores are equal.

On paper, the 12900K looks comparable to the Ryzen 9 5950X, and when put to the test in applications that can take advantage of these heavy desktop parts, the 12900K still looks great, as the E-core weakness that is core-to-core communication isn’t emphasized by those workloads, think Blender as one such example.

Then when it comes to games, the 12900K still shines as no game requires more than 8 Alder Lakes P-cores. Even if a game could spread the load across 16 cores, that wouldn't be a problem. Even in the case of the 12600K, its cores are powerful enough to cope with the load. Otherwise, the game would only be playable with high-end CPUs like the 12900K or 5950X, and that's not going to happen this decade.

Obviously now you will never run a 12th gen CPU with only E cores, but there will come a time when you will have to use E cores for gaming and this can reduce performance by 20% or more, at least based on what we've seen here. But again, I don't expect that time to come in the real life of this series.

Another reason why E-core appeals to gamers is DRM compatibility issues, and I ran into that issue with this benchmark test. In the past, I've tested all CPU architectures with Watch Dogs Legion and Assassin's Creed Valhalla, but both games failed in this testing. Watch Dogs Legion only works with E cores or only P cores, but the combination screwed up the game, strangely stocking 12th Gen CPUs working fine. After that, Assassin's Creed Valhalla failed to load due to DRM detection issues with hybrid 12th generation architecture.

In short, the E cores are a blunder when it comes to gaming and if invoked they drop frame rates. So for gamers, the 12900K and 12700K are 8-core/16-thread CPUs and nothing more. The E cores can support background tasks, but really on a desktop they are better taken care of by two more P cores. There's no reason gamers can come up with the existence of E cores, you're always better off replacing them with two extra P cores.