AMD laptops have a hidden delay in performance of 10 seconds. Here’s why

In a banned release Friday morning, Intel’s Chief Performance Strategist Ryan Schrott walked out to a group of technical reporters with the intention of taking AMD’s Zen 2 (Raizen 4000 Series) laptop CPU down a peg.

Intel’s latest laptop CPU The design, Tiger Lake, is a really exciting release – but it has been on the heels of some crushing upsets in that space, and Intel is looking for a corner to curb its hemorrhaging market share. Early Tiger Lake systems performed incredibly exciting – but they were configured for 28W CTDP, instead of the more common 15W TDP seen in production laptop systems – and reviewers were banned from testing battery life.

These left-wing reviewers like you have correctly compared Intel’s i7-1185G7 with half the power consumption in the 28W CTDP with AMD Raizen7 systems – and even though Tiger Lake has generally come out on top, Increase the market with OEM vendors who have really prevented it from being hit by AMD’s decisive or crushing blow Shopping Laptop CPU in the first place.

Insert the battery

Intel’s original Tiger Lake launch presentations tried to draw attention to the -f-battery versus -f-battery discrepancies in AMD’s performance, but those efforts are largely unheard of. Schrout’s release on Friday was an attempt to retell the story, with additional information enough to put people fired at this time.

We can see this discrepancy between battery non-battery and -f-battery performance in the PCMark 10 application benchmark and also in many RUG-scripted workloads from Intel, which surround product applications, which the company calls “Realistic Usage Guides”. However, similar discrepancies between -n-industry f-battery performance do not appear in more used industry benchmarks such as Cinebench, Passmark or Geekbench.

Intel’s engineering team points out the reason Why We don’t see the discrepancy in the Cinebench in the last picture in the gallery above – in Intel’s test, the Raizen 4000 CPUs don’t advance power and voltage to their maximum position until somewhere between eight and 11 seconds. Then Heavy duty workload started.

Independent confirmation

We were able to confirm Intel’s findings over the weekend, working with the Acer Swift 3 SF314-42 laptop (with the Raizen 7 4700u CPU) and the MSI Prestige 14 Evo laptop (with Core i7-1185G7). In the chart above, we frequently compress small portions of the Linux 5.3 kernel source and graph throughput on each CPU.

The 4-core / 8-thread i7-1185G7 8-core / 8-thread Raizen 7 4700u easily outperforms both single and quad-threaded workloads, even after the Raizen 7 4700u is scattered in its full performance around 12-seconds. Icon. In an unlimited workload, where the Raisin 7 is allowed to flex its full octa-core muscle, things are very close – and even the 4700U will get a narrow win in the last four seconds.

However, here are some things we need to point out. First and foremost – Intel is 100 percent correct in its claims that AMD’s Zen 2 laptop CPU delays ramping power and voltage to their maximum states. This causes a sharp, consistent and reduced effect during those first few seconds.

We reached out to AMD representatives to comment on this design decision. Although AMD representatives asked further questions about our observations, we received no response to the record at press time.

The devil is in the details – so is the heat

But Intel still plays games with its own power consumption. In the screenshot shot above, we can see MSI Prestige Evo 14 with Core i7-1185 G7 during Cinebench R23 run. We haven’t had enough of this laptop to do a full review of it in a long time – and especially to review its battery life, which we’re very curious about since we’ve been banned from testing this state. Two Previously i7-1185G7 systems.

But we can see that – than the CTDP dial of the i7-1185G7 Below Something to estimate near the typical Raizen 7 4000 CTDP 15W, which is widely expected – MSI has chosen to dial in this laptop even further than what we saw in previous prototypes. This product is a variable PL1 in the i7-1185G7 system that makes the Cinebench R23 run as high as W 36W51 in addition to its PLP51, which is the same as the prototypes.

During this Cinebench R23 run, the laptop spent its first 10 to 15 seconds running at a full PL2 power limit of 51W, with temperatures up to a blistering 98 ° C. After that initial, extremely high performance, power, and heat generated explosion, the CPU went down to maintain the average power consumption of 34W. In contrast, the 8-core / 16-thread Raizen 7 Pro 4750U 29 consumes an average of 27.9W with a high of 29.9W at CTDP up to 25W.

As we move from the CPU to ourselves and the laptop-design field, it is worth noting that the system-fan activity also differed significantly between the MSI Reputation 14 Evo-which almost immediately reached the gaming-laptop levels of fan noise- and HP elitabuka, which took more than a minute of time to reach Max and getting more quiet than the MSI fans throughout the run.

The war is on

While Intel did not specifically tell us what conclusions we should draw against the quick performance of the Tiger Lake Zen 2 laptop CPU, the company clearly hoped that “AMD is gambling on the benchmark” and “it turns out, Intel was the winner with all . “

We don’t think there is any such cut and dry conclusion to draw here. Intel’s findings about the slow performance ramp of the AMD Zen2 laptop CPU are, apparently, correct in the facts – we have no trouble confirming it, and it does not explain why many of Intel’s preferred benchmarking techniques show a big performance Delta in favor of Team Blue. And more widely used industry benchmarks than ever before.

But this neglects the greater efficiency of APD systems, in and out of delayed shifts in the state of maximum performance (and battery consumption) in the CPU. When we run the Cinebench R23 for a full five minutes, the Raizen 7 Pro 4750u system renders more scenes than the Intel i7-1185 G7, And It does so with low total consumption electricity. There is no clever trick to explain it.

We also believe that both sides will be tuning into the argument. It has some real-world benefits with moving more quickly to Intel’s most influential state, but we’re not sure the charts are as compelling as they seem. In practical terms, we’ve now spent some time with both the Zen 2 and Tiger Lake laptops – and Tiger Lake systems aren’t really Seems Fast in terms of the subjective experience of seat-of-pants. This strongly argues that the CPU is often focused on advancing power profiles that run faster – if the human piloting system does not notice a delayed upgrade, it is better to save the battery instead.

… and continues

Some exceptions exist – the most notable being the boot time. Tiger Lake systems start quickly and quickly from suspended, and we feel that there is a lot to be done with their readiness to maximize performance immediately. A developer with whom we speculated that the integration of JavaScript Only-in-Time (JIT) could be another short workload that was nevertheless easily human-experienced.

We suspect that the best news for consumers is that it is difficult to answer definitively in the first place for the “which system is better” argument. This level of competition means that neither team can rest on its laurels, and customers are less likely to end up buying systems. No one Want full information about the differences.