This article is the second part of the ThinkPad P16 Gen 1 review, and if there are no surprises, it is also the last. Full text X000 words X 0 pictures, Chinese version will be first published on "Fair Evaluation Chinese Station" (https://www.song1118.com), the English version will be released later in "FairReviews English Station" (https://www.song1118.tech), and will also be issued in the "Fair Evaluation" of Zhihu, Weibo and WeChat public accounts, this article is divided into the following sections:
Total attack: sharpening the gun in battle
Memory: Backwards to advances
Comprehensive: high and low
Pressure: Walking on thin ice
Power consumption: difficult to separate
Temperature: It's hard to put into words
Noise: high and low
Battery: Ten cold and one storm
Evaluation: Narrow red line
Total attack: sharpening the gun in battle
In this article, the P16 Gen 1, the flagship of the ThinkPad mobile workstation and also the performance flagship of ThinkPad---- will launch a final general attack with all the personnel in battle and the barrel fully open, in order to win the core of the "2022 Mobile Workstation Armageddon" exercise ground planned and directed by the author in one fell swoop!
However, before entering the final general attack, the general headquarters of the author's "fair evaluation" exercise, in line with the principle of trying its best to be fair, made a "sharpening gun" arrangement for the ThinkPad P16 Gen 1, and made it further exert its combat effectiveness under the premise of trying its best to maintain the official original factory condition.
Strictly speaking, it is not accurate to sharpen the gun in battle, it should be called light travel---- tactically it can also be called "retreat to advance"!
Memory: Backwards to advances
In the "ThinkPad P16 Gen 1 Review" above, the memory of the ThinkPad P16 Gen 1 has been routinely tested.
Careful readers should have noticed that the memory of this ThinkPad P16 Gen 1 (hereinafter referred to as "P16G1") is fully loaded at the factory---- that is, four 32GB total 128GB memory is installed, in addition to not ECC, it is already the current top configuration, but its regular test results are worrying.
The reason is that the mobile workstation flagships of the three major brands of PC this time, the memory system has been suppressed---- and the good DDR5-4800 is suppressed at 4000 or even 3600MT/s:
ThinkPad P16 Gen 1 can only run at 3600MT/s when installing 4X32GB memory, and maintain 4000MT/s in other states;
HP ZBook Fury 16 G9 full line 4000MT/s---- after the author's actual measurement, this is not the case (for details, please see the author's "HP Zbook Fury 16 G9 Review" Part 2);
Dell was the worst, the Precision 7670 and 7770, all said to be in 3600MT/s condition.
That is, for the P16G1 with full memory, take down two 32GB of memory on the D side, so that the running speed can be "upgraded" to 4000MT/s, and the same memory will still be maintained as dual channels. Although the total memory has decreased from 128GB to 64GB, there is no pressure for daily applications.
After removing two pieces of memory, use the cache and memory benchmark test module of AIDA64 again to test P16G1, and the results are as follows:
The memory read rate has increased from 54000+to 62000+MB/s -- the HP F16G9 is 55000+to 63000+MB/s;
The memory write rate increases from 53000+to 60000+MB/s -- HP F16G9 is from 53000+to 60000+MB/s;
Memory copy rate increased from 57000+to 62000+MB/s ---- HP F16G9 was from 56000+to 59000+MB/s
Memory latency decreased from 94.5 to 88.4 nanoseconds -- HP F16G9 was from 91.2 to 85.4 nanoseconds.
Compared with the performance of many models in the AIDA64 built-in database, the P16G1's memory reading, writing, and copying performance are among the best, significantly better than the P16G1 (54000+, 53000+, and 57000+MB/s) with 128GB of memory in the author's previous test.
The memory delay score is 88.5 nanoseconds, which is at the end of the AIDA64 model with its own database - but it is also better than the 95 nanoseconds of the 128GBP16G1 full memory tested by the author before.
It can also be seen that the measured transmission rate and delay of HP F16G9, whether 4000MT/s or 3600MT/s, is slightly better than that of P16G1.
In this way, the combat effectiveness of P16G1 should be fully exerted by the author on the premise of maintaining the status of the official factory.
The above is also the reason why the release of this article was delayed for one week: the author tested P16G1 in six states (mixed graphics card+independent graphics card) x (128GB memory+two kinds of 64GB memory) one by one, and the workload doubled.
Therefore, in order to avoid the cumbersome text in the article, the following abbreviations will be used for P16G1 with multiple configurations/states:
Hybrid graphics card+2X32GB memory: HG-A
Independent graphics card+2x32GB memory: DG-A
Hybrid graphics card+4X16GB memory: HG-B
Independent graphics card+4x16GB memory: DG-B
Hybrid graphics card+4x32GB memory: HG-C
Independent graphics card+4x32GB memory: DG-C
Comprehensive: high and low
Using PCMark 10 for testing, the six configuration states (HG-A, DG-A, HG-B, DG-B, HG-C, DG-C) of P16G1 are scored as follows:
PCMark 10 Express：5518、6593、5763、6600、6017、6523;
PCMark 10 Extended：8513、10491、8998、10472、9172、10446;
PCMark 10 Applications：13832、13495、14379、13369、14197、13635;
The above test results are compared with the comparison models selected in the author's historical test database, and the results are as follows:
Based on the total score of 38531 for the P16G1-DG-C (4X32GB RAM), there are 5 models in the comparison with a total score, slightly higher than the P16G1-DG-C:
The total score of the P17G2 discrete graphics card status is 103.65%;
ASUS ROG S17 GX703 for 101.90%; ，
The total score of the status of the 7760 discrete graphics card is 101.45%;
The total score of the status of the P1G5 discrete graphics card is 101.06%;
Of the 6 configurations/states of P16G1, only DG-B (2X32G memory) has a slightly higher overall score, which is 100.53% of DG-C.
Could it be that this retreat tactic only got a good memory running score? And the overall performance does not have any benefits?
In fact, if you look closely at the 4 sub-scores of each PCMark 10, you will taste the taste:
The highest scores in the four sub-categories are not under the DG-C umbrella, but are scattered among A and B.
In other words, if it were not for the fact that DG-C has the largest memory capacity (128GB) among the six, it has an advantage in some benchmark content that craves memory capacity, otherwise the disadvantage caused by a 10% decrease in memory operating frequency will be completely suppressed by A and B.
Using the industry's comprehensive performance benchmarking software SPECwpc 2.1, the P16G1 was tested in six configuration states and compared with other models, and the situation was:
It is still the total score of P16G1-D-C that leads the way, suppressing other configuration states and almost all other models.
However, there is a heresy:
In addition to the financial services project, the other four projects and the total score of P1G5 are all higher than P16G1 and HP ZBook Fury 16 G9 -- all about 1 point higher, and the final total score is 109.16% of P16G1-D-C!
This is a bit wrong... In fact, in the second part of the "HP ZBook Fury 16 G9 Evaluation" concluded the day before yesterday, the author had doubts about this part of the data of P1G5, but at that time, only HP ZBook Fury 16 G9 was the same platform model, with insufficient reference objects.
Call out the SPECwpc 2.1 benchmark test report of the three, and compare the sub-scores of each lower level under the sub-item one by one, and finally find the achievement of IOMeter of P1G5 hard disk is extremely high.
For example, as shown below:
The scores of Life Sciences of P1G5, F16G9 and P16G1 are 8.82, 6.29 and 7.07 respectively -- P1G5 is about 2 points higher -- readers, do you know how hard it is to improve the scores of SPECwpc 2.1 sub-items by 1 point before 2022?
The author's more than 10 years of test data can fully prove this difficulty.
Looking at the IOMeter score of the lower level of Life Sciences, we finally found the huge gap. The score of P1G5 was 19.76, and the other two were only 3.92 and 5.61.
IOMeter is the benchmark software for testing the performance of hard disk. Regardless of the high performance of P1G5 hard disk, the first thing to be determined is that the performance of P1G5 hard disk tested by the author cannot be so higher than F16G9 and P16G1.
Therefore, the data of F16G9 and P16G1 is normal, while the data of P1G5 is doubtful;
Or is it precisely because the memory of P1G5 runs at 133.33% of that of P16G1-D-C, so there is this miracle?
Since P1G5 was sent to the new owner's house in northern China a few days ago, the author is out of reach and unable to retest ...
Using the industry's comprehensive performance benchmarking software SPECwpc 2.1, 6 configurations of P16G1 were tested, and all tests were successfully completed, completely without the CPU Blender error phenomenon encountered by the author when testing P1G5 and F16G9, so P16G1 has effective Media and Entertainment sub-project results.
Compare with the author's historical test model, as shown in the figure below:
Based on the overall score of P16G1-DG-C, the situation is as follows:
Only P16G1-DG-C (4X16GB) has a slightly higher overall score of 100.52%;
Even without the Media and Entertainment sub-scores, the P16G1's total score for six configurations is higher than all other comparison models.
This comparison result is more in line with the single results shown in the P16G1 evaluation midpart.
SPECwpc 2.1, for the rapid development of hardware, has been unable to cope with the rapid development, can not accurately reflect the full performance of the new hardware---- and PCMark 8 was "abandoned" by UL, SPEC did not use it as the recommended benchmark software.
Coupled with the example of "the P1G5 hard disk IOMeter's performance is abnormally high", the author will no longer use SPECwpc 2.1 as a benchmark software in the future.
Bye! SPECwpc 2.1! expects SPEC to introduce more comprehensive and accurate benchmarking software.
UL Procyon is UL's benchmark software that has been promoted since 2021 and has been updated frequently.
It was used to test P16G1 in 6 configuration states, and the scores were as follows:
Office Productivity scores: 6994, 7328, 6858, 7048, 7037 and 7116;
Photo Editing score: 8317, 8618, 8419, 8661, 8256, and 8291;
Video Editing score: 8264, 8100, 8158, 7921, 8586, and 8429;
Compare the results obtained with the author's test history models as follows:
Because UL Procyon is updated frequently, it can catch up with the pace of new hardware, new systems, and new software, at least compared with the results of the author's tested models in 2021 and 2022, no anomalies have been found.
Based on the overall score of P16G1-DG-C, the results are as follows:
P16G1-DG-A (2x32GB) is the highest at 100.88%;
P16G1-DG-B (4x16GB) followed by 100.18%;
Among the 2021 models, only the P17G2 has a very close overall score of 99.04%;
The P16G1's total score for all six configurations, with the exception of the P17G2, is higher than that of the other comparison models (including the F16G9 and P1G5).
According to the above UL Procyon performance comparison, the comprehensive performance of the P16G1 is slightly higher than that of the previous generation of mobile workstation flagships (P17G2 and 7760).
Post your Opinion