It's time to go back to CPU benchmarking, and we're not looking at an old CPU, but the first 10th generation processor from Intel. As you probably know, there are no 10th generation desktop parts in sight yet, but Intel has done some interesting things on the laptop side.
Admittedly, the 10th generation from Intel is confusing because architecture and manufacturing are different. A group of processors uses the long-awaited 10 nm process from Intel, codenamed Ice Lake, and brings two major improvements: a new Sunny Cove CPU core with IPC boost and Gen11 graphics with more execution units and new functions .
The other group of processors is Comet Lake, this is another refinement from Skylake to 14nm. We are not quite sure how many pluses we have to add in the end, but there are certainly a few now. Comet Lake doesn't bring any major updates to the micro-architecture or integrated graphics, but we do get faster storage and Wi-Fi 6 support.
Now both product lines exist in the same generation for the same product class. Comet Lake consists of 15 W parts of the U series, while Ice Lake consists of 15 W parts, together with some 9 W and 28 W models. So if you buy an ultra-portable laptop, you can choose between Comet and Ice Lake, depending on the model.
Each CPU row has its own advantages. Ice Lake brings a big leap in integrated graphics performance, but is limited to four CPU cores at modest clock speeds and is 3.9 GHz for the 15 W models. Comet Lake has the same crappy built-in graphics we've had for generations, but with much higher CPU clock speeds (up to 4.9 GHz) and a new six-core configuration for the top-end core i7. Intel somehow managed to put six cores in a 15W package.
On paper, it looks like Comet Lake is the option for the best CPU performance and Ice Lake for the GPU performance. However, this is confused by some vendors, including discrete GPUs, so it's not that easy.
The naming scheme is terrible. Ice Lake processors are referred to as the Core i7-1065G7 for the top 15 W Core i7, while Comet Lake receives the equally terrible Core i7-10710U. Six letters or numbers for the model name of each processor are too many and for the average buyer, how are they supposed to tell the difference between a 1065G7 and a 10710U? These names have to be massively simplified or split into two lists to make the difference clearer.
Aside from all of this information, we'll be looking at Comet Lake first, and hopefully we'll get to that in a few weeks' time when Ice Lake machines arrive in our labs.
Now you might think another 14nm product range isn't that exciting, but that's not really the case. As with Kaby Lake Refresh, where ultra-portable laptops jumped from 2 to 4 cores and increased performance, Comet Lake jumped from 4 to 6 cores.
At the moment, this is only with one product in the series, the top-end core i7-10710U. Other Comet Lake CPUs, like the Core i7-10510U and the Core i5, remain four cores, and then Core i3s and below are dual cores. Today's review is a full benchmark breakdown of the Core i7-10710U, so we can see the benefits of installing 6 cores in such a narrow performance range. A follow-up examines how the Core i5 models improve things a notch too.
Compared to other Core i7s of the last generation, the 6 cores and 12 threads are the obvious main feature. However, in order to have as many cores within the same 15 W TDP as previous generations, the clock speeds had to drop. The base clock is now 1.1 GHz compared to 1.9 GHz for the Core i7-8665U, and also for boost clocks there was a slight reduction of 4.8 GHz in the previous generation and 4.9 GHz in the new Comet Lake Quad-core noted possibility.
What about all-core turbos? Intel doesn't usually advertise it, but we know that the Core i7-8565U operates on four cores at 4.1 GHz. The Core i7-10710U reaches a maximum of 3.9 GHz across six cores, but still reaches 4.1 GHz on four cores, so we don't lose anything on workloads with the same number of cores. It's just that the 10710U can be pushed out onto six cores if required, even at lower clock speeds.
The other advantage that the six-core model brings is an increase in the cache, which is now 12 MB and matches Intel's six-core Coffee Lake processors. However, there are no upgrades for the GPU, so we get another UHD 620 with 24 execution units and clock speeds up to 1150 MHz. This is a slow GPU. So don't be surprised if many OEMs pair this CPU with a discrete, low-power option like the MX250 from Nvidia.
The test platform for today's benchmark is the MSI Prestige 14 A10SC. This well-groomed laptop includes the Core i7-10710U and an Nvidia GeForce GTX 1650 Max-Q GPU and 16 GB of dual-channel DDR4 memory, resulting in a powerful 14-inch slim and light machine.
We have had access to a technical example as opposed to a final retail model of the Prestige, which means that there are some early issues with this laptop that we expect to be resolved with the retail version. However, this is not a laptop test, but we are focusing on the performance of the 10710U.
It should be noted that there are several different performance profiles. The "balanced" mode in the MSI Creator app sets this laptop to the standard configuration of the CPU of 15 W. This is 15 W at the long-term PL1 power limit and 45 W at the PL2 limit. When using the "high power" mode, the maximum configuration of 25 W is reached: 25 W PL1 and 51 W PL2. These performance figures are similar to those of previous generation laptops, so we can make some comparisons.
The Turbo Boost duration differs between the laptops because each OEM can configure this individually. Therefore, there are some differences between the models, especially for short-term tests. However, these numbers that we're going to go through should still be representative of the typical performance of the i7 10710U, especially for longer workloads where the CPU is at its PL1 limit.
Since we only focus on CPU and iGPU performance in this test, we disabled the discrete GPU for these tests. The GTX 1650 Max-Q gives this laptop a huge boost in computing or graphics workloads, but that's not our concern at the moment.
We start with something that Intel reviewers asked not to use: Cinebench. While you probably won't be using a 15W laptop CPU to render 3D models, Cinebench is still an excellent benchmarking platform for examining the performance of single and multithreads and how they compare to each other other products on the market.
The Core i7-10710U scores impressively on a multithreaded workload for just a 15W processor. It is by far the fastest 15 W chip on this list and 33 percent faster than the Core i7-8565U. Yes, we get 50% more cores, but the clock speeds have been reduced to 15 W, so it is very decent to get this extra performance boost.
You will also see that the Core i7-10710U is only slightly behind the 35-45 W quad-core Core i7-7700HQ from a few generations ago and essentially corresponds to the 25 W configuration of the Core i7-8565U. Although Intel was stuck at 14 nm in the three years since Kaby Lake was launched, they have managed to bring a similar level of multithreaded performance with more cores and optimizations to a much lower level of performance.
Now you may be wondering, how is that possible? Does Intel cheat the TDP and run long-term workloads outside of 15W as we saw with desktop processors? The answer is no. As far as we can tell, the CPU gradually drops to 15 W for longer-term workloads. Sure, there has been a boost for short-term workloads that now consume up to 45W of power, but a significant portion of Cinebench is limited within the PL1 limit.
The reason for this lies in the non-linear nature of the voltage pulse-speed curves. If you reduce the clock speeds slightly, depending on the process node, you can significantly improve power consumption.
If you look at this timing diagram for Cinebench, you can see that the i7-8565U in its 15 W configuration runs at around 2.4 GHz in the long term or at 25 W over 3.0 GHz over four cores. The Core i7-10710U works with only 1.9 GHz at 15 W over six cores. That is 500 MHz less than the quad-cores of the previous generation or more than one GHz less than 15 W to 25 W.
If Intel had lowered its last-generation quad-core chip to 1.9 GHz, the power consumption would have dropped significantly. They then used the electricity budget that was freed up to introduce two additional cores and bring the chip back to 15 W. The equation works so well that the extra cores more than compensate for the sunken clocks in this type of workload in terms of performance. Therefore, we get more power with the same power consumption, simply through increased parallelization and more efficient operation of these core clocks.
In this Cinebench diagram, you will notice some other things. The single-thread performance is about the same compared to the 10th generation with the 8th generation because the single-core turbo frequencies are approximately the same. At 25 W, we also see a big performance gain over the Core i7-8565U, but not at the level of the 45 W six-core i7-8750H or i7-9750H. These pimped chips with higher power limits are 18% faster, so you can see that the 25W 10710U is more efficient but can't keep up with the performance of these high-end chips.
The x264 encoding is another good multithreaded benchmark. Video encoding is a popular task, even if only files are converted for playback. Once again, we see that the 10710U is about 25% faster than the 8565U, whether it's 15W or 25W configurations. However, for those who do a lot of video coding, it still makes sense to purchase a 9750H system with up to 37% more power compared to the 25W version of the new 10th generation Intel chip.
In Handbrake, despite switching to the more intensive x265 encoding library, things don't differ too much from what we've seen with x264 encoding. Generation after generation, we see 24% more power compared to 15 W models and 20% compared to 25 W models. Again, the increase from a 107WU 25W model to the 9750H is still 33%. Given that you can now achieve 7700HQ-like performance in a slim and light system with 25W cooling power, we are very impressed with what Intel does.
Let's go through a few more coding benchmarks. Blender is an interesting computational benchmark because we do this on the CPU and GPU and get the lowest score, which gives us an indication of the overall computing power of a laptop. After all, we don't want to disadvantage systems that can perform the task faster on the GPU than on the CPU.
Most of these ultra-portable results work on the CPU, so we see margins similar to what we talked about. About 30% gene to gene improvement from four to six cores. However, the interesting results here may be these Ryzen comparisons. As we know, Ryzen Mobile has a much more powerful GPU than Intel's Comet Lake, and we can see that these Ryzen laptops run this benchmark on the GPU. However, the Core i7-10710U can intercept it almost alone on the CPU, which shows the strength of this configuration with six cores.
Our Premiere benchmark is based on CPU and GPU performance at the same time, since we encode a 4K file with GPU-accelerated Lumetri effects. Usually these ultra-portable CPUs throttle under the GPU requirements, which is why we don't see a big performance difference between the different 15 and 25 W models without a discrete GPU. As soon as you add a discrete GPU to the mix, this coding flies absolutely.
In our non-Lumetri benchmark, which slightly reduces GPU requirements, we see a modest gain for the i7-10710U over the i7-8565U. However, you still want a discrete GPU here, and in general, your decision to buy Premiere with this type of laptop at Premiere may be due to the GPU rather than the CPU.
What about something that doesn't code? Once again, six cores are a big improvement on 7-Zip and offer around 30% more performance compared to Whiskey Lake. Since this is a short-term workload, which is mainly operated in the boost clock area, the steady increase in boost clocks and turbo performance limits by Intel has also helped here, especially compared to CPUs from a few generations ago like the core i7-7700HQ.
Adobe Photoshop Iris Blur is an effect that is mainly limited to CPU, cache and memory. We only see modest increases of ~ 10% compared to the 8565U for the 10710U, which is slightly lower than in other, more intensive tests. It is also less advantageous to switch to Intel's higher TDP six-core CPUs, although the advantage is still there.
Photoshop's Smart Sharpen filter goes in the other direction and is limited to the GPU. Given that we're not getting anything new with the UHD 620 GPU, it's no surprise that the Core i7-10710U matches other 15W processors.
MATLAB is a workload that only depends on memory and cache. Therefore, we don't see any big moves or performance improvements here. Comet Lake supports LPDDR4x, but our laptop test system still uses DDR4.
Before we finish this, take a look at the GPU performance again. Without an improvement in the integrated GPU, we see no improvement in CompuBench computing power in the optical flow. Adding an MX150 is required for respectable GPU performance.
We'll start with a generation-by-generation look with the Core i7-10710U compared to the Core i7-8565U in its 15W configurations. The performance varies somewhat depending on whether the benchmark is multithreaded or single-threaded. In general, however, multithreaded performance is expected to improve 25-30% from four to six cores. With such a tight performance limit, that's pretty good.
Core i7-10710U vs Core i7-8565U
Single-core gains are almost non-existent, mainly because the CPU core architecture is unchanged even from previous generations and at least with this six-core design, no improvement in the boost clocks is achieved. Not that the CPU even reaches the boost clock for longer single-core workloads, since the performance budget is simply not sufficient for this. At best, we get 2-3% more performance, and we would say all of this is due to process optimization and binning.
The integrated GPU also remains unchanged, so that there are no increases in performance here either. This is important for workloads like Premiere that are based on GPU acceleration. We expect many designs to come out with discrete GPUs – the MSI Prestige 14, for example, has a GTX 1650 Max-Q – that will significantly increase the performance of the GPU.
It looks similar when comparing 25W models without the single-thread workloads increasing so much. Here, the six-core may even be slower in some situations, but we're still getting ~ 30% improvement on multithreaded workloads.
If you're wondering about 15W versus 25W, and you're lucky enough to get a laptop that can run at 25W, you'll see a decent 30% gain on some workloads or a more modest single-digit gain on others. Given that long-term power consumption with the cTDP Up configuration is 40% higher, we're moving away from the efficiency sweet spot, but for some, the extra power is all that matters.
Core i7-10710U vs. Core i7-7700HQ vs. Core i7-9750H
These next comparisons are, in my opinion, some of the most interesting. The 15W i7-10710U is not that far from the i7-7700HQ on most workloads. The performance is not quite equivalent, but Intel was able to replicate the 45 W power from 3 years ago with a 15 W chip without major advances in process technology. People love to laugh at 14nm ++, and that applies to the desktop, but with these performance-optimized CPUs, it seems to make a little difference if Intel can add more cores to the mix.
Intel has not yet reached the point where its U-series 25-W chip can compete with 45-W-H series models. The i7-9750H is still 30% faster on important workloads like video coding. So don't drop your 15-inch H-series laptop yet, as a 6-core U-series CPU won't stack up as expected.
Core i7-10710U against Ryzen 7 3750H
The launch of Comet Lake is not good news for AMD. The Core i7-10710U is, in our opinion, a much better option than even the Ryzen 7 3750H, and we're talking about a 15 W CPU versus 35 W. These CPUs exchange beats, even though the 3750H uses a lot more power, and the 10710U impressively surpasses it even with workloads like handbrake and especially with single-threaded devices. The only advantage of the 3750H is a much faster integrated GPU, but it can easily disappear as discrete GPU coupling becomes more common.
It's even worse if you give the Core i7-10710U a little more power to play with. Ryzen is hit hard here and we don't even compare the U series to the U series, which is even more of a bloodbath. AMD can't get Zen 2-based designs fast enough here, as we expect the significant changes in architecture and the introduction of 7nm will bring massive gains in the low-power area.
Since Intel is reaching the limits of its 14nm process, we really didn't expect the Core i7-10710U to be that good. With 15 W, we are just right for an efficient six-core design that enables up to 30% more performance with multithreaded workloads. We don't get many other improvements, but this is very decent for buyers of ultra-portable systems and with the increasing use of multithreading.
Anyone wanting to upgrade their lean and light machine should be able to achieve performance gains over a quad-core of recent generations, though you'll need to look specifically for a Core i7-10710U as this is the only six-core Model is. If you come from a dual-core system, you will be taken to a new dimension with all the extra power you get. Seriously, these Intel U series dual-core processors are slow these days.
It will be interesting to see how Ice Lake builds up. We know they're clocked lower, but since the U-series chips typically hover in the mid-2 GHz range with four cores, this may not be a problem. And the improved integrated graphics can be useful, but they can also be easily surpassed by a Comet Lake CPU with a discrete GPU, such as that found in this MSI Prestige notebook. We will see.