I believe that many old players have had this experience, when buying mainstream positioning memory, they accidentally found that the product used higher quality particles, which can be ultra-high frequency to improve performance, and this kind of good thing was encountered by the housekeeper again, recently to the old platform upgrade when starting a pair of Black Panther series DDR4 3600 from Apacer, to find that it uses SK Hynix AFR particles, here to give you a simple overclocking test sharing~
Unboxing section
Apacer Panther DDR4 3600 8Gx2 memory set strip (hereinafter referred to as “Panther 3600 set strip”) uses common blister packaging, the packaging is mainly black and gold color matching, packaging style takes a simple route, from the packaging information can be seen, the product positioning of the Panther series is GAMING gamer group.
Packaging front
PANTHER Panther series
TARGET GAMING GAMERS
The back of the package
Featuring high frequency and low latency
That’s probably it
Like most mainstream positioning memory products, the Panther 3600 set is not equipped with accessories, after all, the memory product is ready to use and does not need any accessories.
Product details
Black Panther 3600 set of strips to black and gold color matching, metal heat dissipation vest using stamping process, vest shape to golden leopard as the theme, vest itself has no lighting effect, and the top of the heat dissipation vest retains some hollow space, when using a single tower air cooling heat dissipation can be its auxiliary heat dissipation, a bit similar to the role of the early comb vest.
Panther 3600 set of strip body
The golden panther
The identification parameter is DDR4 3600 CL18-22-22-38
Top cutout position
Thermal vest fixed snap
Here’s another side one
The timing of the Black Panther 3600 set that the housekeeper started this time is CL18-22-22-38, which is consistent with the setting of mainstream products on the market, and the memory itself uses an 8-layer PCB design, which is no problem in the face of general overclocking, but if you plan to exceed the limit, it is recommended to add money to the entire flagship.
The memory PCB uses an 8-layer board design
Close-up of PCB thickness
Ready to go on board
It should be noted that the Panther 3600 heat dissipation vest is slightly higher, when using a two-tower three-fan large radiator, players need to pay attention to whether it conflicts with the memory height, the housekeeper personally recommends the use of high-performance single-tower heat dissipation and integrated water cooling heat dissipation designed with memory avoidance, so that even if 4 Panthers are installed, there will be no compatibility problems.
The memory vest is slightly higher
One more one
It looks pretty handsome
Overclocking performance & stability testing
In this test, the butler chose the 5800X with X570 CARBON motherboard as the test platform, and the BIOS was updated to the beta version of AGESA 1.2.0.2 (support memory 4000/FCLK2000 upper limit, fix PCIE 4.0 and USB2.0 conflicts), while turning off PBO and manually locking the CPU frequency 1.25v 4.5Ghz, using the heatsink for the regular version of Limin TR TA120EX. The test environment was 30 degrees in early May in Guangdong (no air conditioning).
CPUZ screenshot
From the data read by the typhoon software, we can see that the Panther 3600 set in the butler’s hand uses AFR particles from SK Hynix, the default frequency is 2133MHz when XMP is not turned on, the memory has an XMP profile, and the memory production date is the 4th week of 2021, and the date is relatively fresh.
SK hynix AFR pellets
Produced in January 2021
8-layer board design
Onboard XMP information
Default memory settings
CPUZ information (4.5G + default memory setting)
As in other memory tests before, in the memory and cache performance of AIDA64, 2133MHz silent frequency with 4.5GHz 5800X memory read and write speed is very general, read at 31224MB/s, due to the 5800X single-core die design, writing only 17638MB/s, and memory latency is also as high as 87.6ns, although it is not the first time to say this problem, but the AMD platform is really necessary to exceed the memory.
AIDA64 (4.5G + default memory setting)
Turn on the XMP function in the BIOS, the memory frequency successfully exceeds 3600+FCLK 1800, and after successfully entering the system, you can see through the CPUZ that the timing part has not been adjusted by AMD gear down mode, and the timing is maintained in the CL18-22-22-38 1T of the XMP overclocking file.
Turn on XMP 3600
The NB/SOC voltage is locked to 1.15
CPUZ screenshot (4.5G + XMP 3600 memory setting)
After XMP overclocking, the memory performance of the 5800X has been greatly improved, memory read and copy have been increased to 50153MB/s and 45469MB/s, and the memory latency has also been reduced from 87.6ns to 62.9ns, which is a significant improvement. The memory write is 28784MB/s, because the single-core die structure of the 5800X is related, and if you want a higher memory write score, you must match the 5900X and above processor.
AIDA64 (4.5G+XMP 3600 memory setting)
Due to the relationship between mainstream overclocking memory, the housekeeper does not dare to exceed 4000 in one step, so through the BIOS to load MSI preset memory overclocking file memory try 3866 in it, the file timing is CL18-22-22-42 1T, the memory working voltage is 1.40V, and the FCLK corresponding to the CPU is 1933MHz.
Load memory try it super 3866MHz/FCLK 1933MHz
The restart was successful
CPUZ screenshot (4.5G + memory overclocking 3866 setting)
When the memory exceeds 3866MHz/FCLK 1933, the memory performance of the 5800X is further improved, the memory read/copy is increased to 53822MB/s and 48679MB/s, the memory latency is reduced to 60.5ns, and the memory write is increased to 30885MB/s, and the latency part of the performance is close to the performance under the Intel platform.
AIDA64 (4.5G + memory overclocking 3866 setting)
Considering that the AFR particles have achieved outstanding results in overclocking in the past, the butler directly uses the same timing setting file to exceed 4000MHz/FCLK 2000, and sets the memory voltage to 1.46v, the NB/SOC voltage is 1.1875v, and the system restarts smoothly into the system after F10 is saved.
4000MHz/FCLK 2000MHz
The process went quite smoothly, there was no card self-test
CPUZ screenshot (4.5G + memory overclocking 4000MHz setting)
Through the CPUZ, we can see that at this time, the system is working at the frequency of 4000MHz/FCLK 2000MHz, and the memory part has not been divided, and the timing part has not been adjusted by AMD gear down mode, still working under CL18-22-22-42 1T; In the AIDA64 memory running score, memory read increased to 54995MB/s, memory copy 50056MB/s, memory write 32921MB/s, and memory latency reduced to 57.2ns, the overall performance is very impressive.
AIDA64 (4.5G + memory overclocking 4000MHz setting)
Under the memory working at 4000MHz/FCLK 2000MHz setting, after 3 hours and 10 minutes through the memtest pro4.2 oven, 430% of the test was successfully tested without error, and the surface temperature of the memory vest was slightly warm.
MT oven 3 hours 10 minutes 430% error-free
Summary: A small surprise from AFR particles
In this test, the butler’s set of Aeropacer Panther DDR4 3600 8Gx2 memory strip performed quite well, as an XMP overclocking memory positioned by mainstream gamers, using metal vest + SK hynix AFR particles, which is thicker than the DJR granules RGB strips commercially available in the same period; By default, XMP 3600 can successfully exceed 4000MHz, and has passed the memtest pro 430% reliability test in 3 hours and 10 minutes, with high overclocking potential and high playability.
The housekeeper believes that the Apacer Panther DDR4 3600 8Gx2 memory sleeve is a memory product with considerable overclocking potential and good cost performance, suitable for hardware and gamers interested in memory overclocking.
PANTHER Panther series
That’s probably it