In addition to the Epyc Genoa-X processors with huge 3D V-Cache, AMD also announced the brand new Epyc Bergamo.
These are also server CPUs, they also rely on the Zen 4 architecture, but they are based on the Zen 4c cores. They can be called small, by analogy with those in Intel CPUs, but Intel’s small cores are based on a different architecture from large ones. The architecture is the same here, and the Zen 4c cores differ from Zen 4 only in a reduced amount of cache memory.
Due to this, physically on the crystal, the core takes up less space. In addition, Bergamo, unlike Genoa, are produced according to 4 nm standards, which made it possible to further reduce the size of the crystals. As a result, Bergamo became the first AMD processors to have not eight cores on one chiplet, but 16 at once.
|CPU||Number of cores/threads||Frequencies, GHz||L3 cache, MB||TDP, W|
There are only three models, apparently, so as not to overlap with the Genoa line at all, although these two lines have slightly different positioning.
As you can see, despite the presence of 128 cores in the top CPU, the TDP of this model does not exceed the performance of Genoa flagships. True, the frequencies here are lower.
As for positioning, Bergamo is focused on cloud servers and data centers, but not just in general, but for those systems where maximum density is important, that is, it is necessary to fit as much computing power as possible in as little volume as possible. Nobody else offers anything like this among x86-bit CPUs.
By the way, the top 128-core Bergamo processors consist of 82 billion transistors.
Compared to Intel Sapphire Rapids, AMD’s new products, according to AMD itself, provide performance up to 2.6 times higher. Intel, judging by its presentation yesterday, does not agree with this. However, a full-fledged competitor for Bergamo will be Intel’s Sierra Forest processors, which will offer up to 144 small cores, but they will not be released until next year.
The new CPUs support PCIe 5.0 and have a 12-channel RAM controller. They use the same SP5 socket as Genoa. It is also worth adding that the new processors are available from today.
12 processor cores and iGPU of the Radeon RX 6500 XT level with a TDP of 25-35 W. This will be the mobile Ryzen 8000
Specifications of AMD Strix Point processors, which will replace the current Phoenix, have appeared on the Internet. It is expected that such APUs will be called Ryzen 804x.
According to insider All The Watts, such processors will be produced at TSMC facilities using the N4P process technology. The crystal area will be 225 mm2.
The configuration will include four Zen 5 cores and eight Zen 5c cores, meaning Ryzen mobile processors not related to the high-performance HX line will have more than eight cores for the first time.
The APU will also include an iGPU based on RDNA 3+ architecture with 8 WGP units. In more common form, this means 1024 stream processors. Let’s remember that the Radeon 780M has 768 stream processors with RDNA 3 architecture. That is, the new iGPU, which will probably be called Radeon 880M, will be more productive thanks to the updated architecture and thanks to a significantly larger number of computing units. By the way, the Radeon RX 6500 XT also has 1024 stream processors. The source also notes support for DDR5-5600 and LPDDR5X-8533 memory and TDP 25-35+ W.
Insider All The Watts previously accurately described the Ryzen 7000 CPU before their presentation.
Intel’s new Meteor Lake processors won’t be all that new. The company admitted that there are few architectural differences from Raptor Lake
At the end of this year, Intel will introduce Meteor Lake processors, which will offer a new process technology, new architectures for large and small cores, and a new iGPU. However, as it turned out, regarding architectures everything is not so clear.
Intel itself has reported that the large and small cores of Meteor Lake processors are very similar in architecture to the cores at the heart of the current generation of Raptor Lake.
Yes, the new cores will be called differently in both cases (Redwood Cove and Crestmont), but there are apparently not too many technical differences. At the same time, we recall that the Raptor Cove cores in the current Intel CPUs are an improved version of the Golden Cove cores in Alder Lake.
Intel notes that it is not the architecture update that plays an important role, but the new Intel 4 process technology, which will allow the new CPUs to be faster and more energy efficient.
Compared to the previous generation of Raptor Lake, the microarchitecture of the P and E cores in Meteor Lake has some improvements. However, we have transitioned Meteor Lake to Intel Process 4. In general, when we introduce new processes, we strive to reduce risk from an architectural perspective. Once we achieve a stable flow of processes, we tend to take a bigger leap by improving the architecture. So, since this is a new big node upgrade for us and we can also get the power efficiency benefits of Intel 4, we made very few changes to the core architecture
That is, the tick-tock approach that was previously familiar to Intel takes place.
At the same time, it cannot be said that future Arrow Lake in this case will carry truly new cores, since this generation will also change the technical process (to Intel 20A).
GeForce RTX 3060 is officially the most popular video card among gamers on Steam
Valve has published another statistics for the Steam service. This time based on the results of the end of September.
And the changes this time are quite noticeable. GeForce RTX 3060 has finally become the clear leader in the rating. Now this card is available to 6.27% of gamers on Steam, and in just a month its share has grown by 1.4 percentage points, which is a lot.
Let us remind you that formally the leader has been the GTX 1650 lately, but, as we have said more than once, in fact it was the RTX 3060 about six months ago, because before the RTX 30 line, Steam statistics combined mobile and desktop adapters of the same name into one position. That is, the share of the GTX 1650 is the total share of the mobile and desktop GTX 1650. While in the case of the RTX 3060 there is a desktop card, which occupies the same 6.27%, as well as the mobile RTX 3060 Laptop, which has another 3.83% at this time moment.
If we return to the rating, the GTX 1650 is in second place with a share of 4.78%, and the RTX 2060 closes the top three with a share of 4.69%, which has surpassed the GTX 1060.
Formally, there are no questions to Nvidia even after the raid. The European Commission denies the fact of conducting an antimonopoly investigation
EU competition authorities are not conducting a formal investigation into chips used for artificial intelligence, the European Commission announced on...
Samsung will release the latest Quasar chips to compete with Nvidia
Samsung’s contract chip manufacturing division has a new customer in the form of a Canadian startup called Tenstorrent, which develops...
Chinese telescope FAST discovered 76 new faint pulsars
These pulsars are special in the sense that they periodically emit a pulse as they spin, which is why they...
Toyota Hilux 2024 presented
The 2024 model year Toyota Hilux pickup truck was released on the home market in Japan. The car retained its...
News6 days ago
The Firefox browser now has a built-in page translator that works even without the Internet
Electric Cars5 days ago
“We have a huge problem,” Volkswagen and Audi plants around the world shut down
News7 days ago
110 inches, 8K, 240 Hz and more than 40 thousand backlight zones. Hisense TV UX unveiled – one of the most advanced TVs in the world
Laptops7 days ago
Microsoft’s CASO feature eliminates the need for iGPU and dGPU laptops to have a multiplexer to switch between graphics cores