AceMagician AD15 Initial Impressions
Posted: Tue May 16, 2023 7:35 pm
Here are some performance testings that I’ve done when asked to test drive the unit.
First things first, in the BIOS settings, the default BIOS from AM capped both PL1 and PL2 to 40W, but it is tunable by the user. I have two sets of numbers for the performance benchmarks, one is the default 40W, and the other is 45W for PL1 and 65W for PL2. I will specify each configuration in the following benchmarks.
All tests are done with Windows 11, OS build 22000.1936
TL;DR: The CPU performance is on-par with Intel i7-6950X or AMD 6800HS while in 40W default TDP. After unlocking the power limit, the CPU will gain some performance to match up against Ryzen 5 5600X, or Intel's own i5-11500 desktop class CPU. As for the GPU side, there's nothing special. I will say it is a very capable iGPU solution for even the most demanding office work, but not for gaming.
###CPU-Z
####Hardware Validation:
[https://valid.x86.fr/i1rx87]
###Cinebench R23
####40W Default:
ST: 1677
MT: 9053
####45/65:
ST: 1681
MT: 9851
###Geekbench 6 CPU
####40W Default:
ST: 2201
MT: 7792
[https://browser.geekbench.com/v6/cpu/1261782]
####45/65:
ST: 2199
MT: 8567
[https://browser.geekbench.com/v6/cpu/1269054]
###Geekbench 6 GPU
####40W Default: 8426
####45/65: 8473
As you can see, increasing the PLs to 45/65 could yield better multithreaded performance, but it does very little to single threaded and GPU performance. Thus, I do not think there’s the need to increase the power cap. However, it is the power to the users to explore whether 10% of performance or lower operating temperature and noise will be suitable for their usage.
###Heaven Benchmark
I know this GPU isn’t gaming-oriented, but I still would like to see how it will perform in a standard 3D benchmark.
####DX9, 720p, basic:
FPS: 40.7
Score: 1025
####DX11, 720p, low:
FPS: 109.3
Score: 2753
####DX11, 720p, medium:
FPS: 52.9
Score: 1332
####DX11, 1080p, low:
FPS: 55.8
Score: 1405
####DX11, 1080p, medium:
FPS: 30.8
Score: 775
###Temperature:
This unit has a pretty good thermal solution. The fan remains quiet even when the load gets heavy. Here is an aggregate temperature chart when I was stressing the system with Cinebench R23.
As we can see, the thermal solution can manage the 40W default TDP quite well, with temperature topped at 80C throughout the Cinebench R23 run. As for unlocking the PL limits, the system first boosted the CPU to use all 65W of power, resulted temps getting as high as 100C while in PL2, and hovered at the low 90s to high 80s after 45W PL1 kicked in. After that, I took the system apart and repasted with Arctic MX-6 TIM. While the highest recorded temperature can still be at the 90s, when PL2 expired and the CPU was conforming to the 45W PL1, the temperature was basically the same, or slightly better, than the stock TIM with 40W power limits.
Nonetheless, during these tests, the fan stayed relatively quiet. It is certainly more quiet than the AM06 Pro I’ve tested. Please be aware, the AM06 Pro uses a 35W chip, while this i5-12450H chip can pull 65W and can maintain 45W. This is impressive to say the least.
First things first, in the BIOS settings, the default BIOS from AM capped both PL1 and PL2 to 40W, but it is tunable by the user. I have two sets of numbers for the performance benchmarks, one is the default 40W, and the other is 45W for PL1 and 65W for PL2. I will specify each configuration in the following benchmarks.
All tests are done with Windows 11, OS build 22000.1936
TL;DR: The CPU performance is on-par with Intel i7-6950X or AMD 6800HS while in 40W default TDP. After unlocking the power limit, the CPU will gain some performance to match up against Ryzen 5 5600X, or Intel's own i5-11500 desktop class CPU. As for the GPU side, there's nothing special. I will say it is a very capable iGPU solution for even the most demanding office work, but not for gaming.
###CPU-Z
####Hardware Validation:
[https://valid.x86.fr/i1rx87]
###Cinebench R23
####40W Default:
ST: 1677
MT: 9053
####45/65:
ST: 1681
MT: 9851
###Geekbench 6 CPU
####40W Default:
ST: 2201
MT: 7792
[https://browser.geekbench.com/v6/cpu/1261782]
####45/65:
ST: 2199
MT: 8567
[https://browser.geekbench.com/v6/cpu/1269054]
###Geekbench 6 GPU
####40W Default: 8426
####45/65: 8473
As you can see, increasing the PLs to 45/65 could yield better multithreaded performance, but it does very little to single threaded and GPU performance. Thus, I do not think there’s the need to increase the power cap. However, it is the power to the users to explore whether 10% of performance or lower operating temperature and noise will be suitable for their usage.
###Heaven Benchmark
I know this GPU isn’t gaming-oriented, but I still would like to see how it will perform in a standard 3D benchmark.
####DX9, 720p, basic:
FPS: 40.7
Score: 1025
####DX11, 720p, low:
FPS: 109.3
Score: 2753
####DX11, 720p, medium:
FPS: 52.9
Score: 1332
####DX11, 1080p, low:
FPS: 55.8
Score: 1405
####DX11, 1080p, medium:
FPS: 30.8
Score: 775
###Temperature:
This unit has a pretty good thermal solution. The fan remains quiet even when the load gets heavy. Here is an aggregate temperature chart when I was stressing the system with Cinebench R23.
As we can see, the thermal solution can manage the 40W default TDP quite well, with temperature topped at 80C throughout the Cinebench R23 run. As for unlocking the PL limits, the system first boosted the CPU to use all 65W of power, resulted temps getting as high as 100C while in PL2, and hovered at the low 90s to high 80s after 45W PL1 kicked in. After that, I took the system apart and repasted with Arctic MX-6 TIM. While the highest recorded temperature can still be at the 90s, when PL2 expired and the CPU was conforming to the 45W PL1, the temperature was basically the same, or slightly better, than the stock TIM with 40W power limits.
Nonetheless, during these tests, the fan stayed relatively quiet. It is certainly more quiet than the AM06 Pro I’ve tested. Please be aware, the AM06 Pro uses a 35W chip, while this i5-12450H chip can pull 65W and can maintain 45W. This is impressive to say the least.