pro max 18 plus Your heart? Analysis and Speculation on the DELL Pro Max 18 Plus for the Year 2025, Part 5

[song_1118]

My space for uploading files has been used up, so I am unable to display images.
If you need to view the pictures, please go to the link below, but it is only available in Chinese.

https://www.toutiao.com/article/7450361997889176098/

 

 

Your heart!
Yes!

Attack on Titan !

The DELL Pro Max 18 Plus has the huge body of the Max 18, and its heart must be strong, otherwise it will be inferior to the 18 inch body, unable to showcase the charm of the 18 inch 3840 x 2400 resolution serial OLED screen, and cannot be paired with the title of flagship mobile workstation!

 

The heart of Max 18 refers to its independent graphics card.
The flagship model of DELL Precision mobile workstation's discrete graphics card has been using DGFF for 6 years.

 

What is DGFF?
In 2018, starting from DELL Precision 7730/7530, its GPU began to use a new specification module called DGFF, which stands for "Dell Graphics Form Factor". This is a new graphics card modular design launched by Dell after years of self research.

DGFF graphics cards, like MXM graphics cards, can be easily maintained and replaced, but the biggest improvement is the optimization of circuit connections and installation methods. Its purpose is to replace the MXM standard that has been used for many years on high-performance mobile computing devices (including mobile workstations), while continuing to have the same modular advantages as MXM and avoiding the inconvenience of using GBA to package GPUs on motherboards.

In 2017, the flagship model of DELL Precision that last used MXM graphics cards was the DELL Precision 7720, which used the MXM B graphics card. The second flagship model 7520 used the MXM A graphics card.

 

However, the newly launched DGFF specification by DELL, which holds 100% of the patents, requires additional fees for other manufacturers to use, so it has not been seen that other major manufacturers have adopted it for a long time.

At present, only Framework Modular Laptop Company in the United States has adopted DGFF technology specifications for its 16 inch model graphics card and hard disk expansion module, named Framework Laptop 16 Expansion Bay module, which uses DGFF technology. Its independent graphics card module and hard disk expansion module can interact with each other, as shown in the following figure, which is a hard disk expansion module designed according to DGFF specifications. It can install two M.2 2280 SSDs on top and also has line expansion to support the power supply and management of two cooling fans:

 

In addition, there are rumors that Zhitong International will release multiple laptops with 15.3-inch screens in 2025, which will use graphics cards similar to DGFF specifications. Zhitong International is a wholly-owned subsidiary of Softtek and owns the "Mechanical Revolution" laptop brand. Therefore, it is highly likely that at CES 2025 next month, Mechanical Revolution will release models that use graphics cards similar to DGFF specifications:

 

In fact, DELL's own history of using DGFF is also constantly exploring and developing:
Firstly, it is only used within its own Precision family;
Secondly, not all models in the family are universal, and the shape of DGFF graphics cards varies between different models;
Then, every two years, the body mold will be replaced, and DGFF will also change accordingly.
Take the DELL Precision flagship 77xx series as an example, its development history is as follows:

 

 

DGFF1
In 2018-2019, the 7730 and 7740 generations shared the same body mold and used the first generation DGFF1:

As shown in the figure below, DGFF1 is a rectangle with a small corner cut off;

At the long edge of the rectangle, there is an HDMI and a mini DP port, which are the two video output ports at the back of the body;

The power supply for the graphics card is obtained by connecting an L-shaped dedicated cable to the motherboard, as shown in the lower left corner of the figure below;

The circuit connection between the data signal and the motherboard uses a silver gray connector with dense contacts on the inside, as shown in the lower right corner of the figure. One end of the connector labeled MB on the outside is connected to the motherboard, and UFK 1902 Rev 0.8 is labeled on the inside. The number of connectors used varies depending on the graphics card model, with the highest level graphics card using three.

 

DGFF2
In 2020-2021, the 7750 and 7760 generations shared the same body mold and used the second-generation DGFF2:

The shape of DGFF2 is still a rectangle with a small corner cut.

But there have been changes that are completely incompatible with DGFF1:

As shown in the above figure, the placement direction of the rectangle has been rotated 90 degrees, and an HDMI and a mini DP port have been placed on the short side of the rectangle;

As shown in the figure below, at the same time, the installation position of the connection pieces of the data signal circuit is also rotated to the other side, and the maximum number of connection pieces remains unchanged at three:

The annotation on the inside of the connector used is slightly different, it is "UFK1902 Rev 0.8", but from the shape and contact arrangement, it should be the same as the one used in DGFF1.

The connection method and cables of the power supply circuit are also similar to those used in DGFF1.

DGFF3
In 2022 and 2023, the Precision 7770 and 7780 generations will use the third-generation DGFF3

DGFF3 has a larger rectangular angle in shape, and the graphics card body no longer has a video output port:

The position, quantity, and connector specifications of the data signal connection circuit interface between DGFF3 and the motherboard have changed again - this is only natural, as DGFF3 no longer integrates HDMI and mini DP ports - as shown in the following figure:

The orange arrow indicates the data line interface, which has only two, JDGFF1 and JDGFF2. (J stands for Jack (plug/connector means);

The PJPDC1 marked with a blue arrow is the interface of the power supply circuit, which is no different from the one used by DGFF1 and DGFF2. It still has a five channel design, but the position is different.

As a result, the number of connection pieces used between the DGFF3 data line interface and the motherboard has been reduced to a maximum of two;

As shown in the following figure, the new connector has a light yellow PC board appearance, with Unimicron/Neoconix and 80P PCBridge marked on it:

It can be inferred from this that:
1. This connector is produced by Neoconix, a subsidiary of Unimicron Electronics, which was acquired by Unimicron in 2016;
2. The 80P PCBridge indicates that this connector has 80 contacts/pins, and the PCBridge - "PCB Bridge" - should be based on Neoconix's PCBeam technology.

 

Continue to use?
The above is the continuous exploration and development history of DELL's DGFF over six years and three generations.

As can be seen, with Precision replacing molds every two years, DELL is constantly adjusting and improving DGFF;

Even after NVIDIA forcefully increased the power supply of the graphics card, DELL abandoned the built-in video output port and freed up space to add a power supply unit, allowing DGFF3 to continue moving forward;

It can also be seen that DGFF3 seems to have further reduced component costs and reduced assembly processes in production by adopting the 80P PCBridge produced by Neoconix.

The author believes that:
The DELL Pro Max 18 Plus will continue to use DGFF graphics cards.

So, is the DELL Pro Max 18 Plus using DGFF3?
Or is it the new DGFF4?

 

DGFF4？
To answer this question, you need to first refer to the following materials.

A. Desktop RTX 5090
Release: Expected to be released on the eve of CES 2025, at 18:30 Pacific Time on January 6, 2025, and may be available in mid February 2025;
Parameters: GB202-300 core is used, with 21760 CUDA cores and 32GB of GDDR7 memory. The bit width of the memory is still being confirmed, with a frequency of 28GHz and a bandwidth of 1792GB/s;
Power consumption: The overall power consumption of the graphics card will be reduced, but the specific amount of reduction has not been determined yet;
Technology: Introducing neural rendering, advanced DLSS and other technologies to enhance GPU performance, the improved RT core can present more realistic light and shadow effects and reflections in games, and will also improve the performance of AI assisted tasks, better integrating AI into the game and content creation process. In addition, AI will optimize its power consumption and thermal management to achieve more energy-efficient operation.

 

B. Desktop RTX 5080
Adopting GB203-400 core, it has 10752 CUDA cores and 16GB of GDDR7 memory. The memory bit width is 256 bit, the memory frequency is 30Gbps (higher than RTX 5090), and the memory bandwidth will reach 960GB/s.

 

C. Professional graphics cards of the same level:
There is currently no news.

 

D. Mobile RTX 5090
Currently, some news about the mobile version of RTX 5090 lp:
Core: Adopting GB203 core, which is the same as the desktop RTX 5080 core;
CUDA： Perhaps similar to the desktop RTX 5080, with 84 SM units and 10752 stream processors;
Video memory: using 16GB or 24GB GDDR7 video memory, with a bit width of 256 bits;
Power consumption: The top of the line mobile version RTX 5090 may still have a power consumption of 175W, similar to the previous generation mobile version RTX 4090;
Performance: Some believe that the performance of the mobile version RTX 5090 will be significantly improved compared to the mobile version RTX 4090, and the performance gap with the desktop version RTX 5090 will further narrow;
Release: Expected to be released by the end of the first quarter or early second quarter of 2025.

 

Based on the above four points of ABCD, the author analyzes and speculates that:
At present, there will be no DGFF4,DELL Pro Max 18 Plus graphics card, DGFF3 will continue to be used.

 

The reasons are as follows:
Firstly, due to the delayed release of NVIDIA's new generation of mobile professional graphics cards, the Max 18 may continue to use the RTX 5000 Ada from the previous DGFF3 in the early stages of its release;
Secondly, by mid-2025, after NVIDIA releases the new generation of mobile professional graphics cards, the Max 18 will be immediately adopted, and it is impossible to make major modifications to the body and internal layout of the Max 18 for this purpose;
Thirdly, the flagship model of NVIDIA's new generation of mobile professional graphics cards should have the same hardware parameters as the mobile RTX 5090; This means that the flagship model of NVIDIA's new generation of mobile professional graphics cards will maintain the same power consumption as the previous generation, so DGFF3 has no power supply issues;
Fourthly, the only issue is the graphics memory - NVIDIA's flagship mobile professional graphics card model uses GDDR7 graphics memory, is it 16GB or 24GB?

24GB of video memory?
According to the existing practice of NVIDAI, the mobile version of RTX 5090 is based on the desktop version of RTX 5080, and the video memory of the mobile version of RTX 5090 should also be 16GB of GDDR7;

However, it cannot be ruled out that the flagship model of NVIDIA's new generation mobile professional graphics card may use 24GB of video memory.
So, after adopting 24GB of video memory, will the existing DGFF3 design space not be able to accommodate it?

 

The author believes that:
The flagship model of NVIDIA's new generation mobile professional graphics card, with 24GB of graphics memory, is completely problem free.

 

Because the current layout of DGFF3 arranges the positions of 8 memory granules - the 16GB memory of RTX 5000 Ada lp uses 8 2GB memory granules.
Previously, the single capacity of video memory was usually 2GB or 1GB. For graphics cards like the GeForce GTX 1060 3GB, the 3GB of video memory is composed of multiple smaller memory particles combined, rather than a single 3GB of video memory.
Now, Samsung has announced that it has developed a single GDDR7 memory with a capacity of 3GB (=24Gb), using the 5th generation 10 nanometer process, increasing density by 50%, and packaging size the same as the previous GDDR6 product. It also uses three-level pulse amplitude modulation (PAM3) signal, which can achieve speeds of around 40Gbps and up to 42.5Gbps.

 

So, on the existing DGFF3 layout, 8 GDDR7 VRAM with a single capacity of 3GB can be used to achieve a total capacity of 24GB for VRAM.

As for whether the flagship model of NVIDIA's new generation mobile professional graphics card will use 24GB GDDR7 memory, I believe it should be necessary;
After all, the Asus ProArt Studiobook One W590, which was released in 2019 and only launched in 2020, uses the Quadro RTX 6000 mobile graphics card and already has 24GB of video memory:

 

At the beginning, the author spent a lot of money (9999 US dollars just by placing an order) to purchase Asus ProArt Studiobook One W590:

So, the author will never forget that the Quadro RTX 6000 mobile version has 24GB of video memory:

Now, it's already 5 years later. I don't expect NVIDIA to surpass 24GB, but it should always be possible to catch up with 24GB, right?

This article has reached 3000 words and is about the same length.
So, the content originally planned to analyze and speculate on the cooling system and performance of the DELL Pro Max 18 Plus in this article will be moved to the next installment, "Serial 6".

Coming soon!

[Aaron44126]

Note that you can post images from another site, you just have to drop the image URL in when writing the post and it will convert to an in-line image automatically.

(I personally upload them to OneDrive and use the "embed" button in OneDrive to get an image URL that I can paste here.)

[AL123]

All very interesting based on previous releases I’m doubting we will actually by able to buy these before mid year but who knows with a leak this early. It would fit with the prediction for high end next gen mobile graphics cards from NVIDIA also.

 

Also very curious about the 3 fan layout, I’ve only really seen it on Alienware systems where IIRC the third fan was significantly smaller. I’m hoping for more innovation here from the ex Precision R&D team

[Aaron44126]

3 minutes ago, AL123 said:

Also very curious about the 3 fan layout, I’ve only really seen it on Alienware systems where IIRC the third fan was significantly smaller.

 

This image shows a smaller third fan in the little "infographic", so I am expecting something similar here...

[AL123]

3 hours ago, Aaron44126 said:

 

This image shows a smaller third fan in the little "infographic", so I am expecting something similar here...

Ah good spot would be disappointing if it isn’t better than the Alienware M18 layout. Hope they are going away from the old 7000 series layout where the heat was concentrated at the top left. Even if it was performing fine users would complain about that/ get worried the system was overheating 

[song_1118]

12 hours ago, Aaron44126 said:

Note that you can post images from another site, you just have to drop the image URL in when writing the post and it will convert to an in-line image automatically.

(I personally upload them to OneDrive and use the "embed" button in OneDrive to get an image URL that I can paste here.)

As you wish, I used OneDrive and added pictures.
Within China, OneDrive requires VPN to function properly.

[song_1118]

I will analyze and speculate on the issue of three fans in the cooling system of Max 18 in Part 6.

[Aaron44126]

9 hours ago, song_1118 said:

As you wish, I used OneDrive and added pictures.
Within China, OneDrive requires VPN to function properly.

Doesn’t have to be OneDrive; anything that allows public sharing of images will work. I checked just copying image URLs from your Chinese post but it looks like they add timestamps and authentication tokens to the image URLs, that site probably took steps to prevent this sort of thing…