Connect with us

GeForce RTX 2060 with 12 GB of memory will go on sale on December 7, and GeForce RTX 3070 Ti with 16 GB of memory and RTX 3080 with 12 GB of memory in January

Published

on

Rumors about new Nvidia video cards of the Ampere and Turing lines began to take on sharper features – the timing of their release was finally announced. And although there is no confirmation of this data, and Nvidia after the leak may “replay” something, we still have more specifics regarding the timing of the release of new products.

Reportedly, it will be possible to buy a new version of the good old GeForce RTX 2060 first of all – with twice as much memory as the regular version of this model. The announcement of the GeForce RTX 2060 with 12 GB of memory will take place on December 7, and sales will start on the same day.

Exactly 10 days later, Nvidia will finally officially unveil the GeForce RTX 3070 Ti with 16GB of storage and the RTX 3080 with 12GB – both of which have been talked about a lot. But this will be a paper announcement, and real sales start only on January 11th.

What will happen to the current models after the release of new products is not clear. But if it makes sense to keep the GeForce RTX 3070 Ti with 8 and 16 GB of memory in the line, in general, then it is very strange to offer the GeForce RTX 3080 with 10 and 12 GB, so the current model will most likely simply be replaced. new. However, if the GeForce RTX 3070 Ti with 16 GB will not differ from the current model in anything except the amount of memory, then the GeForce RTX 3080 12 GB is credited with the new GPU GA102-220 with probably more CUDA cores and a wider memory bus – bit width 384 bits instead of 320 bits.

Click to comment

Leave a Reply

Your email address will not be published.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

Components

$ 5,000 Nvidia graphics card with no DirectX support or video outputs. A detailed review of the CMP 170HX has appeared

Published

on

The Nvidia CMP 170HX is a unique graphics card, and today we have the opportunity to take a look at its full review.

Recall that the CMP 170HX is a specialized video card for mining. Only, unlike other adapters in the line, this one is based on a GPU that is not intended for the consumer market.

$ 5,000 Nvidia graphics card with no DirectX support or video outputs. A detailed review of the CMP 170HX has appeared

The heart is a huge GPU GA100, but in the case of the CMP 170HX, it seems that partially defective chips are used. Firstly, there are only 4480 active CUDA cores (out of 8192), and secondly, there is only 8 GB of memory against 40 or 80 GB of Nvidia A100 accelerators. But this is HBM2e memory with a bandwidth of 1.5 TB / s! As a result, Ethereum mining performance exceeds 160 MH / s.

We can learn some very interesting things from the review. For example, the GPU along with the memory is covered with a giant heat-distribution cover. Perhaps the largest among video cards. And the CMP 170HX is not only devoid of any video outputs, but in general it is not technically capable of running games, or even speeding up any calculations, since there is no support for the DirectX or Vulkan APIs and calculations with CUDA support do not work. In Windows, the card is defined as an RTX 3090, and it is allegedly even loaded by the system, but it is not clear what exactly, since applications cannot use the accelerator.

It remains to add that such a card costs about $ 5,000.

Continue Reading

Components

Linux kernel code already knows about AMD 96-core processors

Published

on

The latest Linux kernel patch has received the first confirmation of the existence of AMD processors with 96 cores.

More precisely, it talks about the new AMD Family 19h 10h-1Fh and A0h-AFh CPUs, which can contain up to 12 chiplets. Recall that each chiplet currently contains eight cores.

The code does not indicate which processors this information applies to, but, of course, we are not talking about consumer Ryzen. In this case, we are dealing with either server-side Epyc or Threadripper. And with regard to the former, everything agrees with the data of AMD itself.

Recall that the company recently talked about the Epyc Genoa processors on the Zen 4 architecture, which will just have up to 96 cores. As a result, we now know that the new Epyc chiplets will still contain eight cores, but the number of chiplets will increase. And in Bergamo processors, their number should reach 16 at all, if AMD does not change the chiplets for these CPUs.

Continue Reading

Components

Dual processors, dual 4K Sony Micro OLED screens, eight cameras and Mac-like performance. Apple headset is expected in the fourth quarter of 2022

Published

on

Recently, rumors have become more frequent regarding Apple’s upcoming mixed reality headset. Prominent analyst Ming-Chi Kuo of TF International Securities describes it as a standalone device that does not require an iPhone or Mac connection.

The mixed reality headset will feature two 4K Micro OLED displays from Sony, as well as two processors. The main processor will have the performance level of the Apple M1 single-chip system, and the auxiliary processor will process information from the sensors.

It is also reported that the headset will receive up to eight cameras that will be used in augmented reality applications. In addition, the device can be used as a full-fledged gadget for virtual reality applications.

Previous rumors have indicated that the headset will weigh just 200-300 grams. At the same time, how much the device will cost is still unknown: a very wide range is called from 1000 to 3000 dollars.

It is confirmed that the device could be released as early as the fourth quarter of 2022.

Continue Reading

Most Popular