But before we get into our tests, we should first take a look at what we are actually dealing with here in detail. For this, as always, a look at the SPD with the software Thaiphoon Burner helps. As always, the full SPD export is available as an extra file linked here, open in new tab, zoom and scroll please. 😉
Besides the manufacturer and product number we also find the confirmation that these are single-rank modules with 8 memory chips per rank and module, as Kingston already exemplarily states in the data sheet. The board layout is the “A2” reference specification with 8 layers. Furthermore, we already find a hint on the installed memory chips, 8 Gbit DJR or D-Die from Hynix. Since the software is often mockingly referred to as Thaiphoon Guesser, we’ll take another physical look at the chips just to be on the safe side.
Finally, the two XMP profiles are read out, the first of which contains 2667 MHz at tCAS 20, tRCD 30, tRP 30, tRAS 52, tRC 78, tFAW 56, tRRDS 14, and tRRDL 15 at 1.6 V. The second, much more conservative profile with 2000 MHz at tCAS 19, tRCD 23, tRP 23, tRAS 42, tRC 94, tFAW 42, tRRDS 9 and tRRDL 10 and 1.35 V is a commendable addition and second performance floor before JEDEC, in case there are compatibility problems with the first XMP profile and other hardware. Unfortunately, a temperature sensor is not installed on the modules.
Heatsink Performance
Now we come to the temperature test, because RAM modules advertised for overclocking must of course also be able to be cooled reliably, because otherwise the just mentioned instabilities can occur. As the stress test we use Testmem5 v0.12 with the profile “Extreme1@Anta777” because of its very high thermal load.
For this purpose 2 external type-K thermocouples are used, which are sampled with an Elmorlabs EVC2 and KTH with 64 Hz. One thermocouple is placed in the of the outer and warmer module in contact with the board and heat sink pad, while the other thermocouple measures the room temperature. The delta is then formed from both values and plotted over the approx. 40-minute test duration.
For reference, I’ve also included the Crucial Ballistix MAX and the DJR-based G.Skill Ripjaws V modules from past tests in the diagram. In all three cases, the modules are tested with passive airflow and without distance between the modules.
Although the G.Skill and Kingston modules use the same board layout, memory chips, and operating voltage, the difference between them is stark. While the former settle at around 27°C above room temperature, the modules from today’s test remain significantly cooler at 16°C delta. The Crucial modules are pretty much in the middle, although they also have much bigger RevB chips with more waste heat.
So the Fury Renegade heatsinks not only look good, but can also prove decent cooling performance. Of course, a single temperature sensor can only measure the average temperature of the module. Whether the memory chips at the edge now run hotter due to their poor contact to the heatsink cannot be determined beyond doubt with our test setup. It should also be said, however, that at least in the XMP no heat-related instability occurs.
- 1 - Introduction and SKUs
- 2 - Packaging and looks
- 3 - SPD and heatsink performance
- 4 - Teardown and PCB analysis
- 5 - Test systems and methodology
- 6 - XMP compatibility and overclocking
- 7 - Synthetic benchmarks – AIDA64 and Geekbench 3
- 8 - Gaming – Cyberpunk 2077 in UHD, QHD, FHD
- 9 - Final thoughts and conclusion
8 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Urgestein
Veteran
Veteran
Veteran
Mitglied
Moderator
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →