Temperature behavior
Let’s now move on to the temperature behavior and first a few words about the new test system. It was often criticized that the last Ryzen 9 3900X used was too easy to cool and had too low a power limit to provide meaningful results. I had already explained that not all testers have an i9 14900K(S) at their disposal that can be turned up to meltdown and often simply have to use what is available for the tester at the time. After all, I have now been able to upgrade to a current AM5 platform and with PBO enabled, the otherwise so efficient Ryzen 9 7900 (88W) quickly becomes a real drinking buddy, which tries to rapidly reach the temperature target of 95°C set by AMD “by design” with plenty of power consumption (>160W). Since CPU/GPU boost algorithms and dynamic limits make it difficult to make concrete statements, I will probably increasingly note the power consumption and clock rates in addition to the temperatures in the future in order to be able to better evaluate my observations in the respective cases.
Test system:
The fresh B650 platform with Ryzen 9 7900 was used as the test system. PBO is always activated for the tests in order to make power consumption and temperatures a greater challenge for the cooler and case. The DDR5 RAM kit from Corsair clocks at 6000MHz, and the optically matching RX 7700XT Pure from Sapphire was once again used as the graphics card. The CPU was cooled with an NZXT Kraken Elite 360 RGB AIO.
CPU | AMD Ryzen 9 7900 PBO on |
CPU cooling | NZXT Kraken Elite 360 RGB |
Mainboard | NZXT N7 B650E White |
Memory | CORSAIR Vengeance RGB 32GB DDR5 6000 CL36 |
Graphics card | SAPPHIRE AMD Radeon RX7700 XT Pure |
SSD | HP FX700 M.2 2TB |
Case fan | NZXT F360 RGB Core x 2 | F120 RGB Core x 1 |
Power supply | NZXT C1500 Platinum 1500W |
Other accessories | NONAME Cable Extension Kit Black/White |
General conditions:
AIO fans: 1150rpm
Case fans: 900rpm
Pump: 1350rpm
Room temperature: 23.1°C
CPU Torture
For the CPU stress test, I again used Cinebench R23 in continuous loop. For almost the entire test, >160W was applied to maintain an allcore boost of ~4825-4850MHz. The average temperature of the Ryzen 9 was 93.8°C, so it could be reliably kept below the 95°C target. The AIO fans were set so that they were just loud enough in a load scenario. The system drew 250W from the socket in Cinebench.
Gaming workload
Borderlands 3 was again used for the gaming test, as it practically fully utilizes the graphics card throughout and therefore generates a lot of waste heat. Especially with AIOs mounted in the lid, this has a significant effect on the CPU temperature, as the waste heat from the graphics card is partially transported outside through the radiator. A lot of compiling obviously has to be done at the start of the game, which is why there was a brief rise in temperature to almost 80°C, after which the CPU stabilized in the low 60°C range. The system drew 350W from the socket during the gaming session.
The graphics card was also able to keep a cool head in the H7 Flow. With a GPU temperature of just 53°C, the fans only had to run at just over 1500rpm and a fairly stable clock rate of ~2500MHz could be maintained.
The HotSpot was also absolutely harmless.
Interim conclusion
Basically, the temperatures in the new H7 Flow are rock solid. The idea of repositioning the power supply unit to accommodate three 120mm fans specifically for the graphics card obviously works. Thanks to the size and the decent Aiflow, good results can be achieved overall. Unfortunately, the pre-assembled F360 case fan is a stripped-down 3-pin version, which can only be regulated in three stages via the NZXT controller used: off, ~1000rpm and “full throttle” (well over 2000rpm), whereby they are already clearly audible at ~1000rpm. By controlling them directly via the board, they can be adjusted a little more finely, but at tolerable volumes they no longer move huge amounts of air. Saving a few cents for the missing PWM control costs the case a small devaluation in the end, because in view of the, in my opinion, visually and haptically very successful single-frame fans, savings were really made in the wrong place here.
41 Antworten
Kommentar
Lade neue Kommentare
Mitglied
Neuling
Urgestein
Urgestein
Urgestein
Urgestein
Veteran
Urgestein
Urgestein
Moderator
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Moderator
Alle Kommentare lesen unter igor´sLAB Community →