Connect with us

EXTRAS

Will AMD RX 7000 GPUs be cheaper than NVIDIA RTX 40?

Controversial topic like few others, and quite speculative in a certain part of the text without a doubt, but we have a problem with the prices of graphics cards that is going to be extended to the entire sector. From processors to motherboards, only RAM and SSD memories are going to be spared, because they are in a difficult moment. But in GPU, will there be a AMD RX 7000 vs NVIDIA RTX 40 price war?

Huang made that clear a week ago, dashing hopes of a price drop. This has some reasons that are what we are going to discuss below, because AMD has followed a different approach that could turn the tables with quite an impact on the market, how do you plan to do it?

Moore’s Law is dead, or is it just desEsports Extrasmismanagement?

NVIDIA-RTX-4090-Jensen-Huang

Let us recall the aforementioned statements by Jensen Huang:

“A 12-inch wafer is much more expensive today,” citing rising chip manufacturing costs. “Moore’s Law is dead… It’s completely over”, furthermore, expectations of doubling performance for a similar cost are “a thing of the past” for the industry.

Why does the NVIDIA CEO say this? Well, we have to look at the 5nm and 4nm wafers to understand that, comparatively speaking and within NVIDIA, the price with reference to Samsung and its 8nm has increased by 2.2 times moving to TSMC. In density a giant leap has been taken with an impressive +177% in just one lithographic jump.

That is to say, from a technical point of view, we now have more transistors for the same or lower price, but the problem does not come so much from the node, which also, but from the list of materials, called in English and with its acronym as BOM , where AMD has quite an advantage.

RTX 40 vs RX 7000 in price: NVIDIA shoots up its costs, AMD maintains them

AMD-Vs-NVIDIA-costs-GPU-RX-7000-and-RTX-40

The bill of materials encompasses many aspects, so they have been condensed into Parametric Yielded, Packaging, Memory and Total, where we understand that each of them does not have to be explained for obvious reasons.

The list shows how NVIDIA and its AD102 spend much more on the GPU, but less on the packaging of it with the PCB, while in memory you pay for the best of Micron and it shows. The Navi 31 has lower GPU costs, but since it is an MCM design, the cost of including the dies and soldering them skyrockets, while in memory it logically cuts costs for NVIDIA.

We went from a BOM of 2.2 on the AD102 in front of the 1.8 Navi 31that is, AMD could theoretically offer high-end GPUs 22% cheaper than NVIDIA cost for cost. Another thing will be the added expenses that both brands have for the hardware, since when we pay for a graphics card, very high associated costs are included.

Returning to the BOM of each chip, history repeats itself or even worsens. The AD103 goes to a total of 1.4, while Navi 32 stays in a 1.1 (-27% for AMD in costs), the AD104 with 1 and the AD106 (not seen yet) with 0.6 are far behind Navi 33 with 0.5which competes head to head with the AD107 with the same value.

Bearing in mind that the Navi 32 will compete with both NVIDIA’s AD103 and AD104 (the green segmentation here looks pretty bad in gamut, since the AD104 has to house three GPUs on paper) it seems pretty clear that If AMD cuts features, it will be competing with NVIDIA on price and possibly performance.

The same will happen with the Navi 33 against the AD106 and AD107. The die of the oldest of the brothers will be larger and more expensive than that of AMD, while those of Lisa Su will be able to get more juice because it is a more profitable intermediate chip and balanced in performance. While the AD107 will be NVIDIA’s low-end chip when AMD only has to use the chips that don’t pass the tests to launch the supposed RX 7500 XT, because apparently due to the cost of the wafers there will be no Navi 34.

Where is the advantage of AMD?

Ray-Tracing-DirectX-Pipeline

Let’s say that it is rather the disadvantage of NVIDIA. Those of Huang are putting everything to the Ray Tracing and DLSS with RT Cores and Tensor Cores, which take up a lot of space in the matrix as they are independent fixed function units. AMD plays another stick and renounces these units using the so-called Intersection Shaders, which is ultimately how the GTX 1000 manage to present ray tracing, for example.

Shaders are used with the Microsoft API where it does not require specific physical units as NVIDIA has with the RT Cores and therefore, although the AMD ones are slower in Ray Tracing, also take up less physical space in the die and they make up for it with more muscle in units FP32a faster internal bus, much more cache and now with MCM architecture.

This is why NVIDIA’s investment in Ray Tracing and DLSS 3 is so strong and they push developers to implement them, while AMD continues to use more common AI technology such as FSR 2.0 and without fixed units to traverse the BVHbut in return, you get more chips and lower failure rateso the price of the set together with the Navi 31 MCD are fabricated at 6nm and therefore they are cheaper and smaller, they can position AMD as a more interesting option in price and performance, as long as we do not want to use Ray Tracing in its maximum splendor.

Finally, it should also be noted that NVIDIA paid $1 billion upfront for wafers with exclusive 4N process at 5 nm, so that is also having an impact on the data logically. As the next step will be the 3nm from TSMC, it is likely that we will not see jumps in RT Core and Tensor Core, since the idea of ​​​​NVIDIA is to use LCM to cut costs, which doesn’t mean cheaper graphics cards, unfortunately. For now and waiting for confirmation from AMD, we will see if the RX 7000 have a lower price than the RTX 40, which, as we saw last week, everything indicates that it will be so.

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Advertisement

More Posts

News12 mins ago

TT Isle of Man – Ride on the Edge 3 announced for March next year

TT Isle of Man – Ride on the Edge 3 has just been announced by RaceWard Studio. As the name...

Games24 mins ago

The mythical Blade of Darkness arrives today by surprise on the eShop

Those of us who wear certain gray hairs are more than aware of the importance that certain games have had...

EXTRAS36 mins ago

This is the best gaming laptop you can buy on Black Friday 2022

The black friday It is one of the most anticipated moments of the year for lovers of technology, and especially...

EXTRAS48 mins ago

France: the deployment of fiber seems worrying in the country

Obviously the deployment of optical fiber in France seems rather nebulous and problematic according to Arcep. Explanation. Laure de la...

News60 mins ago

Kimetsu no Yaiba: how can Nezuko Kamado become human again?

This would be the only way to turn Nezuko Kamado back into a human in Demon Slayer. The plot of...

Games1 hour ago

Nintendo Switch Sports: Golf is out next week

The golf mode for Nintendo Switch Sports was announced before the release, but fans had to wait a long time...

EXTRAS1 hour ago

Warning: Raab vs. Trymacs – “Kick auf Eis” streamer event will not take place! – PIXELS.

The Trymacs event “Kick auf Eis” was supposed to take place on December 29th, 2022. Not only were all the...

News2 hours ago

FIFA 23: 7 strong players under 25,000 coins to improve your team

Are you looking for improvements for your team in FIFA 23, but don’t want to invest a fortune in the...

News2 hours ago

Does Warhammer 40K: Darktide keep its beta progress in the final game? Fatshark does not forget his promise

In just 5 days we will be liquidating infidels in Warhammer 40,000: Darktide. The shooter Fatshark’s crazy game has earned...

EXTRAS2 hours ago

Take a look at these great cosplays of Princess Zelda and Dark Link from The Legend of Zelda

From Reddit we have received curious information about one of the franchises most loved by nintenderos. We are talking about...

Advertisement