MSI GeForce RTX 5070 Ti Gaming Trio review – A good compromise between Ventus and Vanguard

1 - Einführung und Details zur Blackwell GB203-300-A1 GPU 2 - Testsystem und Equipment 3 - Teardown: Platine und Kühler 4 - Materialanalyse und Wärmeleitmaterialien 5 - Gaming Performance; Rastergrafik 6 - Gaming Performance: Supersampling, RT & FG 7 - Leistungsaufnahme, Lastspitzen und Netzteilempfehlung 8 - Kühler, Temperaturen, Thermografie, Geräuschentwicklung 9 - Zusammenfassung und Fazit

Teardown

When disassembling a graphics card, it is dismantled step by step, whereby the initial state is first recorded. After removing the backplate and cooling components, the board is exposed so that the layout, power supply and soldering quality can be examined in detail. The central components such as the GPU, memory chips and voltage regulator are also analyzed, as are the design of the cooling system and the heat conducting materials used. In this way, not only can the technical implementation be evaluated, but conclusions can also be drawn about the performance and efficiency of the card – a procedure that provides insightful findings for both end users and experts.

Board and components

As usual, the board is quite compact and is based on NVIDIA’s reference design, which relies on three large voltage rails and several smaller ones. The voltage converters for the NVVDD, i.e. the core voltage of the GPU, are already known. What is new, however, is that NVIDIA – similar to Intel and AMD – once again uses separate voltages for the GDDR7 memory and the frame buffer. While dedicated voltage rails for GPU cores and memory are already established, the separation of the frame buffer voltage in this form is a new feature at NVIDIA.

The frame buffer serves as a storage area for the image data required for display on the monitor. Information such as color depth, transparency and resolution is stored here and continuously updated by the GPU. This area is directly connected to the graphics memory and operates under the MSVDD voltage, which supplies the memory chips themselves. While MSVDD regulates the basic operating voltage of the memory chips, FBVDD ensures the stability and accuracy of data transfer between GPU and memory, especially at high clock rates.

The clear separation between MSVDD and FBVDD allows a more precise adjustment of the voltage values to the respective requirements. MSVDD determines the speed and stability of the memory chips by matching the electrical properties of the memory cells and the memory controller logic. FBVDD, on the other hand, ensures that communication between the frame buffer and memory remains efficient. The voltage regulation of the board is clearly laid out: There are a total of 18 control loops, 11 of which are for NVVDD (GPU core voltage, 0.8 to 1.1 V), four for MSVDD (memory voltage, 0.8 to 1.1 V) and three for FBVDD (frame buffer voltage, 0.9 to 1.24 V), supplemented by other smaller voltages for various components.

The power supply of modern graphics cards requires precise coordination of various control and power components. On the back of the board is the MP29816, an efficient PWM controller that takes over the voltage regulation for the GPU core voltage (NVVDD) as well as for the memory (MSVDD) and the frame buffer (FBVDD). This multiphase system (IntelliPhase) ensures even load distribution and precise voltage regulation, which optimizes both the thermal and electrical load.

The actual voltage regulation is implemented using DrMOS modules, with Monolith’s MP87993 being used for NVVDD, MSVDD and FBVDD. These modules convert the control signals supplied by the PWM controllers into the required output voltages. They integrate high-side and low-side MOSFETs as well as the gate drivers in a compact housing, which minimizes switching losses and saves space on the circuit board. All voltage converters have integrated protection mechanisms such as temperature and short-circuit protection to ensure operational safety.

There are no major surprises on the back. Like NVIDIA, MSI relies exclusively on MLCCs under the socket and completely dispenses with polymer capacitors. The reasons for this have already been discussed in detail, as in my investigation into the polymer capacitors of the RTX 3090, I found that their choice has a direct influence on the stability of the GPU, especially at high clock rates. Some models relied exclusively on SP-CAPs, which offer a high capacity but filter high-frequency voltage peaks worse than MLCCs. This led to instability and crashes on certain cards. Models with a mixture of MLCCs and SP-CAPs proved to be more stable, as MLCCs are more effective at smoothing out voltage fluctuations. As a result, manufacturers adapted their designs and increasingly relied on mixed solutions or completely on MLCCs to improve operational reliability. NVIDIA has now also followed AMD and Intel in using a complete MLCC assembly.

In addition to the fuse and shunt resistor for the PEG connection, the large PWM controller for NVVDD and MSVDD and a smaller one for FBVDD can be found here. The whole thing is supplemented by the obligatory supervisor chip, which is responsible for power monitoring.

The board also contains the NCP45492, a high-performance IC for monitoring bus voltages and currents on up to four high-voltage power supplies. This component enables the acquisition and scaling of shunt and bus voltages and allows each channel to be flexibly adapted to specific requirements using external resistors. Particularly noteworthy is the fast settling time, which enables real-time checking of the voltage values. This makes the NCP45492 ideal as a supervisor for the 12V lines, especially for the 12V2X6 and PEG connections of the graphics card.

All relevant components are shown again below in a high-resolution microscopy view:

VHX_000000

VHX_000001

VHX_000002

VHX_000003

VHX_000004

VHX_000005

VHX_000006

VHX_000007

VHX_000008

VHX_000010

VHX_000011

VHX_000012

The cooler

The rear backplate is made of aluminum and also cools the board by means of an attached heat conducting pad, once again in the wrong place, of course. The backplate therefore contributes to mechanical stability and improves cooling. Together with the central cooling block as a load-bearing element, the structural integrity of the card is also increased, guaranteeing stable operation under high loads.

The medium-weight cooler of the MSI GeForce RTX 5070 Ti Gaming Trio is a compromise between cool operation and cost efficiency, which is achieved through a combination of innovative technologies and very high airflow. Cooling is made possible by a continuous, nickel-plated copper heatsink instead of a vapor chamber, which serves as the primary heat dissipation element. This transports the heat directly from the GPU and the VRAM to the five so-called core pipes, three of which run through the middle and then bend back under the area of the first fan. These square-shaped heat pipes behind the chamber optimize thermal contact with the chamber and ensure even heat distribution. The heat is then dissipated through a network of precision-manufactured fins, which are designed for a very high throughput in order to be able to handle the up to 330 watts of waste heat.

The fans of the cooler are each equipped with seven fan blades optimized for throughput, which we already know from the newer Vanguard and Suprim models. In addition, the Zero-Frozr function offers the option of stopping the fans completely at low loads to enable silent operation. Another element is the thermal pads, which I will discuss in more detail, and which ensure additional heat dissipation from critical components such as the voltage converters.

The cooling system represents a very good compromise between size, weight and performance, I still have to measure the rest…