just going to change the name

  • 34

A week ago NVIDIA surprised us with unexpected news: it had decided to cancel the launch of one of the three graphics cards belonging to the GeForce RTX 40 series that it had presented a few weeks earlier. It is unusual for a company of this size give up launch of a product, but the brand explanation seemed convincing.

On October 14, an entry in the official NVIDIA blog He explained to us that the reason why the launch of the GeForce RTX 4080 12 GB had been canceled was simply to avoid the confusion that the availability of two graphics cards with the same name could generate among users. The launch of the GeForce RTX 4080 16 GB continues, and, in fact, it will arrive in stores on November 16.

However, in a new and unexpected turn of events, Seungkyu Kim, who is the CEO of NVIDIA’s South Korean subsidiary, has introduced a new ingredient in this recipe during a conference addressed to the media in his country. His explanation reinforces the brand’s decision, and yes, it makes sense, but, above all, it gives us a very solid clue that invites us to foresee that finally the canceled GeForce RTX 4080 is comingalbeit under a different name.

This is the perfect tool to squeeze the last drop of juice out of Intel's 'Raptor Lake' processors.

This is the problem: the 16 GB RTX 4080 is much faster than the 12 GB

The two versions of the GeForce RTX 4080 that NVIDIA introduced in mid-September didn’t just differ in the amount of VRAM they included; its graphics processor also had different specifications. The 16 GB-equipped GPU in the review is noticeably more ambitious than the 12 GB-equipped graphics card. In this table we collect these differences, and, in addition, we have included the specifications of the RTX 4090 to put the figures of the two RTX 4080 in context:

nvidia geforce rtx 4090

nvidia geforce rtx 4080 16GB

NVIDIA GEFORCE RTX 4080 12GB

architecture

Ada Lovelace

Ada Lovelace

Ada Lovelace

transistors

76 billion

Not available

Not available

photolithography

4nm TSMC

4nm TSMC

4nm TSMC

cuda cores

16,384

9728

7680

rt cores

128 (3rd generation)

76 (3rd generation)

76 (3rd generation)

tensor cores

512 (4th generation)

304 (4th generation)

304 (4th generation)

calculation units (cu)

128

76

76

level 1 cache

128Kbytes (per SM)

128Kbytes (per SM)

128Kbytes (per SM)

maximum clock rate

2.52GHz

2.51GHz

2.61GHz

dedicated memory

24GB GDDR6X

16GB GDDR6X

12GB GDDR6X

memory bus

384 bit

256 bit

192 bit

memory transfer speed

1008GB/s

735GB/s

Not available

shader tflops (fp32)

90

49

49

rasterization operations

192 ROP/s

96 ROP/s

96 ROP/s

texture map units

512

304

304

texture rate

1290 Gtexels/s

761.5 Gtexels/s

761.5 Gtexels/s

pixel rate

483.8 Gpixels/s

240.5 Gpixels/s

240.5 Gpixels/s

direct x 12 ultimate

Yes

Yes

Yes

pci express interface

PCIe 4.0

PCIe 4.0

PCIe 4.0

hdmi review

2.1

2.1

2.1

display port review

1.4a

1.4a

1.4a

dlss

3

3

3

occupied slots

3

3

2 or 3

maximum gpu temperature

90ºC

90ºC

90ºC

average consumption

450 watts

320 watts

320 watts

recommended wattage for power supply

850 watts

750 watts

750 watts

power connectors

3 x 8-pin or 1 PCIe Gen 5 cable 450 watts or more

3 x 8-pin or 1 PCIe Gen 5 cable 450 watts or more

2 x 8-pin or 1 PCIe Gen 5 cable 300 watts or more

price

€2,099.89

From 1469 euros

From 1099 euros

Inno3D GeForce RTX 4090 X3 OC 24GB GDDR6X

Inno3D GeForce RTX 4090 X3 OC 24GB GDDR6X

As we can see in the table, the GeForce RTX 4080 GPU backed by 16 GB of VRAM adds more CUDA cores and a more ambitious memory bus than those integrated in the RTX 4080 that was going to be accompanied by 12 GB of VRAM. It is even likely that the differences between these two chips do not end here, and that the 16 GB RTX 4080 outperforms the 12 GB in some other feature that we have not been able to contrast.

According to Seungkyu Kim, the performance difference between the two revisions is too large to allow the 12 GB version to continue to be called GeForce RTX 4080.

In any case, these differences justify what Seungkyu Kim explained: The performance difference between both revisions of the RTX 4080 is too large to allow the 12 GB version to continue to be called GeForce RTX 4080. According to this executive, when two graphics solutions share the same commercial name, even if they have a different VRAM map, They should be close in performance. The cut of the memory in the most modest version seeks that more users can do with it.

It is reasonable, but we must not lose sight of the fact that, in all probability, the GeForce RTX 4080 GPUs that were going to be accompanied by 12 GB of VRAM They’re ready. In fact, it is very likely that they have been in the hands of the integrators who were going to launch them on November 16 for many weeks.

And, if so, these graphics cards will end up reaching stores for sure. They will not be called GeForce RTX 4080 12 GB. They may arrive as GeForce RTX 4070 Ti. Or, simply, as RTX 4070 to dry. Or by any other name. But we can be sure that we can get hold of them.

A week ago NVIDIA surprised us with unexpected news: it had decided to cancel the launch of one of the…

A week ago NVIDIA surprised us with unexpected news: it had decided to cancel the launch of one of the…

Leave a Reply

Your email address will not be published.