AMD’s new accelerated processors will have between 12 and 15 computing units


AMD’s new accelerated processors will have between 12 and 15 computing units

AMD’s new accelerated processors will have between 12 and 15 computing units

As before, about a month ago, news about the unveiling of the fourth generation Ryzen Series Accelerated Processors (APU)AMD codenamed Renoir was released at CES 2020, new leaks on Twitter also confirmed this It reveals more details about 4 potential variants of Renoir APUs. APU stands for Accelerated Processing Unit and is an advertising name for AMD processors with an integrated graphics card. AMD CEO Lisa Su previously confirmed that the Ryzen 4000 series of mobile processors with 7nm lithography will be manufactured by Taiwan Semiconductor Industries Corporation (TSMC) and will be available in early 2020. In the following, the possible specifications of Renoir series APUs will be analyzed.

Renoir accelerated processors of AMD based on Vega architecture contain up to 15 computing units (CU)

Until now, none of the specifications of the Renoir series APUs have been revealed, and the only information obtained was related to the release time of the mentioned product in early 2020. But now the KOMACHI ENSAKA user account, which is known in the field of leaks related to hardware parts, has revealed information related to AMD’s new APUs. According to the mentioned information, the user of Twitter, Locuza, has raised speculations that if true, it shows that the said processors have a potential performance beyond expectations and a serious competitor for Nvidia and integrated and discrete graphics cards of Intel .

  • AMD Renoir 7nm APUs Specs Leaked

The%20image%20above%20is%20related%20to%20KOMACHI%20ENSAKA’s%20tweet%20and%20all%20the%20speculations%20are%20based%20on%20this%20tweet.%20Using%20these%20specifications,%20Locuza%20has%20calculated%20the%20possible%20configurations%20for%20Renoir%20series%20APUs%20and%20divided%20them%20into%204%20categories%20with%20different%20designs.%20In%20the%20new%20Renoir%20series%20processors,%20according%20to%20the%20mentioned%20user,%20the%20computing%20units%20will%20be%20between%2012%20and%2015%20numbers.%20It%20is%20important%20to%20mention%20that%20the%20accelerated%20processor%20that%20has%2012%20computing%20units%20is%20the%20only%20one%20that%20has%20been%20confirmed%20so%20far%20among%20the%204%20designs%20and%20the%20other%20categories%20are%20currently%20only%20speculations%20based%20on%20the%20available%20information.%20The%20integrated%20graphics%20processor%20in%20the%20aforementioned%20APUs%20is%20based%20on%20the%20Vega graphics architecture and, in most cases, is combined with the main 45-watt processor. They are capable of thermal design.

There is an interesting point about the list published by KOMACHI ENSAKA. The type of APU that has the highest specifications is called Ryzen 9 and supports the B12 design. The B12 design basically means 12 computing units and it can be somewhat assured that it is the only design that will be released in the near future. Among AMD’s current accelerated processors, the Ryzen 5 3400G model has the highest configuration and supports up to 1.9 teraflops of processing power by supporting 11 computing units under the Vega architecture. Due to the number of computing units in AMD’s new APUs and their higher clock rates, these processors will easily be capable of more than 2.2 teraflops. To better understand this value, it is enough to consider the graphics processing power of Nvidia’s MX250 processor, which is 1.4 teraflops at its maximum.

Click on the image to see it in real size

Here is the description that Locuza has published regarding its categories (according to the table above):

  • One of the good details to know about the fifth generation of GCN instructions (Vega Architecture Instructions ) having 3 computing units for each memory level cache is one instruction and each fixed cache.
  • In the configuration of 4 groups of level cache 1 and fixed cache, the maximum number of computing units is 12. In the case where there are 13 computing units, there will be an additional 32 KB of instruction level cache memory and 16 KB of fixed cache.
  • The configuration of type one of 13 computing units is very improbable and the second configuration seems more realistic. However, 12, 14, or 15 computing units are more reasonable.

Whereas the accelerated Renoir processor will probably be produced under 7nm lithography, users will see significant improvements in the clock speed of the said product. Was. So these APUs, which are based on the fifth generation of GCN instructions, will perform much better than the Raven Ridge series when compared based on the performance of the computing units. If we see only 200 MHz improvement in processing speed, with a theoretical calculation, we can say 2.5 teraflops of processing power in the version with 13 computing units and 3.1 teraflops of processing power in the type of 15 computing units in the new APUs. With performance at this level, the Renoir series will definitely beat Nvidia’s graphics cards and Intel’s Iris Pro series.

While Nvidia still uses chips with 16NFF lithography from Taiwan Semiconductor Industries, the difference in lithography used in The blue and red team products are showing up. If Nvidia wants to have a winning card in the competition with AMD, it should use 7nm lithography in its graphics card. The old lithography of the MX250 mid-range processors is very evident, and if this chip really has 1.4 teraflops of processing power, it will not be a justified option for 2020.

Previous AMD has committed to long-term support for the sockets used in the third generation of Threadripper processors
Next What was the fate of Google's hardware purchases?

No Comment

Leave a reply

Your email address will not be published.