Enlarge /. When the battery is fully charged, your new Ryzen 4000 series laptop will only deliver its true performance after about 10 seconds at full throttle.
In an embargoed presentation on Friday morning, Intel's Chief Performance Strategist Ryan Shrout led a group of technology journalists through a presentation aimed at getting AMD's Zen 2 laptop CPUs (Ryzen 4000 series) out of hand.
Intel's latest laptop CPU design, Tiger Lake, is a truly compelling release – but it follows on from some devastating issues in the area that Intel is looking for an angle to keep market share from bleeding over its rival. Early Tiger Lake systems performed incredibly well – but they were configured for a 28W cTDP instead of the much more common 15W TDP found in production laptop systems – and reviewers were allowed to use battery life do not test.
This led reviewers like you to really compare Intel's i7-1185G7 at 28W cTDP to AMD Ryzen 7 systems at half power – and while Tiger Lake generally came out on top, the performance mismatch prevented it from being a conclusive or devastating blow for AMDs, there was an increase in market share from OEM providers who primarily buy laptop CPUs.
Enter the battery
Intel has to improve the perception of Tiger Lake as superior to Ryzen 4000 – and the performance on battery power is a plausible aspect.
Many real workloads – as Intel calls RUGs – have deltas between the performance of AMD connectors and the battery, similar to PCMark 10 Modern Apps.
An angry Intel points out that the most commonly used benchmarks like Cinebench R20 run long enough to cover up Team Red's initial performance lag on battery power.
Intel explains the reason for AMD's performance degradation on battery power – and when these are displayed and not – by graphically displaying the CPU performance and the voltage states of the Ryzen 4000 over time.
Intel's original launch presentations in Tiger Lake were intended to draw attention to battery versus battery differences in AMD's performance, but these attempts have largely gone unheard. Shrout's presentation on Friday was an attempt to retell that story again, this time with enough additional information to get people going.
We can easily see this discrepancy between battery and battery performance in the PCMark 10 application benchmark and also in many of Intel's RUGs – script workloads based on production applications that the company calls "Realistic Usage Guides". However, the same discrepancy between the performance when the battery is turned on and off is not noticeable in more commonly used industry benchmarks such as Cinebench, PassMark, or Geekbench.
Intel's engineering team shows the reason we don't see the discrepancy in Cinebench in the last image of the gallery above. In Intel's tests, the Ryzen 4000 CPUs did not run up to their maximum state until eight to eight, 11 seconds after the heavy workload began.
The x-axis is the elapsed time and the y-axis is the compression throughput in MiB / s. The throughput of Ryzen 4700u jumps up 70 percent between the 11s and 12s. Intel is consistently consistent.
The phenomenon we saw with performance delays in single-threaded workloads is repeated in this quad-threaded workload – a 65 percent increase in performance that increases between the 11 and 12 second values.
The jump in performance isn't quite as pronounced when training with all threads – but it's still sharp, with a jump of 48 percent between the 11s and 12s.
We were able to confirm Intel's results over the weekend using an Acer Swift 3 SF314-42 laptop (with a Ryzen 7 4700u CPU) and an MSI Prestige 14 Evo laptop (with a Core i7-1185G7). In the above diagrams, we repeatedly compress small portions of the Linux 5.3 kernel source and graphics throughput over time on each CPU.
The 4-core / 8-thread i7-1185G7 clearly outperforms the 8-core / 8-thread Ryzen 7 4700u in both single and quad-thread workloads, even after the Ryzen 7 4700u is 12 seconds late Achievement has achieved mark. With the unlimited workload at which the Ryzen 7 is allowed to play its full octa-core muscle, things are much closer – and the 4700u even scores a narrow victory in the last four seconds.
However, there are a few things that we need to point out here. First, and most obviously, Intel claims 100 percent that AMD's Zen 2 laptop CPUs delay power and voltage ramping up to their maximum state. This leads to a sharp, corresponding and diminished performance in the first few seconds.
We asked AMD representatives to comment on this design decision. Although AMD representatives asked additional questions about our observations, we haven't received a response to the recording at the time of going to press.
The devil is in the details – just like the heat
This MSI Prestige 14 Evo – and its Core i7-1185G7 CPU – is sure to be a barn burner. The "burn" is a bit more literal than we'd prefer, however: CPU temperatures scream up to 94 ° C and CPU power consumption of 34 W is sustained during the Cinebench R23 tests.
The Ryzen 7 Pro 4750u in this HP EliteBook laptop stayed relatively cool and quiet during the Cinebench R23 run. The maximum power consumption is a whopping 21 W less than that of the i7-1185G7.
However, Intel still plays games on its own power consumption. In the screenshot above we can see the MSI Prestige Evo 14 with a Core i7-1185G7 during a Cinebench R23 run. We didn't have this laptop long enough to fully check it out – and specifically to check the battery life, which we've been very excited about since it was banned from testing this status in two previous i7-1185G7 systems.
But we can see that with this laptop, MSI chose to choose the cTDP of the i7-1185G7 even wider than was common in previous prototypes, rather than choosing it on something that was roughly the typical Ryzen 7 4000 cTDP of 15 W. This production i7-1185G7 system has a variable PL1 that reaches up to 36 W over the course of a Cinebench R23 run – in addition to its PL2 of 51 W, which is unchanged from the prototypes.
During this Cinebench R23 run, the laptop spent its first 10 to 15 seconds at the full PL2 power limit of 51 W and temperatures of up to 98 ° C. After this initial burst of extremely high power, power, and heat generation, the CPU dropped, to maintain an average power consumption of 34 W. In contrast, a Ryzen 7 Pro 4750U with 8 cores and 16 threads – with a cTDP of up to 25 W – consumed an average of 27.9 W with a maximum value of 29.9 W.
As we move away from the CPUs themselves and into the realm of laptop design, it is perhaps worth noting that system fan activity was also vastly different on the MSI Prestige 14 Evo – which almost instantly reached the noise of fan laptops – and that HP EliteBook that took more than a minute to max its fans and stayed much quieter than the MSI throughout the run.
The fight goes on
While Intel didn't specifically tell us what conclusions to draw from the performance lag on Zen 2 laptop CPUs compared to Tiger Lake's immediate performance, the company was clearly hoping for something between "AMD is playing the benchmarks" and "AMD is playing the benchmarks." ". it turned out that Intel was the winner all along. "
We don't think there are any such conclusive conclusions here. Intel's results on the slow performance increase in AMD Zen 2 laptop CPUs are obviously in line with the facts – we had no problem confirming this, and it explains why many of Intel's preferred benchmarking techniques have larger performance deltas in Team's favor Blue exhibit as the most widely used industry benchmarks like Cinebench, PassMark and so on.
However, this ignores the higher efficiency of AMD systems beyond the delayed transition to maximum performance (and battery consumption) in the CPU. When we run Cinebench R23 for five full minutes, a Ryzen 7 Pro 4750u system renders more scenes than the Intel i7-1185G7, and does so with lower overall power consumption. There is no clever trick to explain this.
We also believe that there is a voting point on both sides. Intel's faster transition to the top performing state brings some real-world benefits, but we're not sure they're as compelling as the charts make them seem. In practice, we've spent some time with Zen 2 and Tiger Lake laptops – and the Tiger Lake systems don't really feel any faster in terms of a subjective experience. This strongly suggests that it often does not make sense to start up the CPU performance profiles so quickly. If the system pilot does not notice the improvement in latency, it is probably better to save the battery instead.
… and continues
We suspect that the best news for consumers is that the "which system is better" argument is so difficult to answer. This level of competition means that no team can rest on their laurels and consumers are less likely to buy systems that no one wants if they are fully aware of the differences.