AMD Launches Its 2nd Gen EPYC 7Fx2 CPUs With Highest Per-Core Performance – Achieved With Upgrades In Every Category, Featuring Double The L3 Cache

Apr 14, 2020
Submit

AMD and Intel have been trading blows in the server market, which AMD has been gaining an advantage with the EPYC due to the friendlier pricing from AMD. This has forced Intel to cut the price of the Cascade Lake Refresh processors. AMD has taken this battle and flipped it on its head, taking it in another direction. Today, AMD unveiled three new processors, the 2nd Gen EPYC 7Fx2 series which is optimized for frequency and delivers the fastest per-core performance in the x86 server workspace environment.

AMD Unveils 2nd Gen EPYC 7Fx2 CPU That Focus On Highest-Per-Core Perf In The Server Segment - Slight Increase In Price That AMD Warrants With A Performance Increase That Could Cut Operating Costs By 50%

All of this is looking to take the market by storm, challenging Intel's position at the fastest per core server processor. Thanks to an increase in base frequency by 500MHz and some extra L3 cache, the processors have increased performance and can do tasks such as rotational databases, high-performance computing (HPC), and hyper-converged infrastructure (HCI) applications with relative ease compared to the competition. The silicon used for these chips has been premium binned, which gives AMD the best possible chips to put in the processors.

AMD Next-Gen RDNA 2 ‘Navi 21’ GPU Powered Radeon RX & RDNA 1 ‘Navi 10 Refresh’ GPU Powered Radeon RX Graphics Cards Leaked

“AMD EPYC continues to redefine the modern data center, and with the addition of three powerful new processors we are enabling our customers to unlock even better outcomes at the heart of the enterprise market,” said Dan McNamara, senior vice president and general manager, server business unit, AMD. “With our trusted partners, together we are pushing the limits of per-core performance and value in hyper-converged infrastructure, commercial HPC and relational database workloads.” - AMD

The CPUs do come with a higher TDP bump, pushing them up to 240W and while this may impact the thermal output of the processors, AMD says that it won't be a significant jump from its current offerings and shouldn't be a major concern for its customers. AMD says all of this leads to reduced up-front costs and a 50% reduction in total operating costs for the target markets, which justifies the increased price of the chips.

The new variants making up the 7Fx2 lineup include the 8 core EPYC 7F32, the 16 core EPYC 7F52, and the 24 core EPYC 7F72. AMD provided estimated SPEC2017_int_base benchmarks that demonstrate the increase in performance in performance-per-core and performance-per-dollar in the lightly-threaded workloads, which has been in the Xeon's wheelhouse for quite some time. The vendor provided these benchmarks, so this must be taken with a slight pinch of salt, but it is already looking to be an impressive processor. AMD has also taken the opportunity to up the memory to 8 channel memory with a maximum capacity of 4TB of memory at DDR4-3200 and 128 lanes of PCIe 4.0, which makes Intel's 2TB of memory at DDR4-2933 and 48 PCIe 3.0 look weak.

Double The L3 Cache On The 16-Core Variant With A Boost Frequency Increase That Puts Intel To Shame

AMD's model slot in below the new VMware licensing model that doubles pricing for processors with more than 32 cores. The benefits to this are the increased per core performance that outweighs the per-core cost, but is a better option than to buy a server. This is all achieved through increased performance and specifications in all aspects. The boost frequencies have been increased to 3.7 GHz for its 24-core model, and 3.9 GHz for its 16- and 8-core models. This adheres to AMD's standard practice of rating for one core only, and it is complemented by the thread-targeting to give the fastest core the most of the single-threaded workload.

The increased L3 is no slouch with up to double the cache on some of the processors. The 16-core model has been increased to 256MB of L3 Cache, which is a 128MB increase, so it was doubled. The 24-core model has been increased to 192MB of L3 cache, which is a 64MB increase. All this cache improves the per-core performance and benefits the target applications. To achieve a 500 MHz increase to the base frequencies, the sacrifices made were a TDP increase of 60W in TDP for the 24-core 48-thread EPYC 7F72, a 90W increase for the 16-core 32-thread EPYC 7F52, and a 25W increase for the EPYC 7F32.

The price increases are significant but justified with the specifications the processors boast, and it is not above Intel's offerings, which could put AMD at the forefront of the server market. The 24-core model has an increase in the price of about $667 over a comparable offering from AMD, putting it at $2,450, which is still below Intel's offering, which is the Gold 6248R at $2,700. The 16-core model is $3,100, a $2,122 markup over the 7302, which is quite significant, but in comparison to Intel, it is still below the price of $3,286, which is what the Gold 6246R is currently priced. These prices aren't scaring off any buyers with Dell, IBM, HPE, Lenovo, Microsoft, Nutanix, Supermicro, and VMware, all wanting a piece of the action with these processors.

Processor Cores
/Threads
TDP
(Watts)
Base Freq/
Max Boost Freqvi
Total L3
Cache
L3 Cache
per Core
Price
(1Ku)
7F32 8/16 180W 3.7 GHz/~3.9 GHz 128MB 16MB $ 2,100
7F52 16/32 240W 3.5 GHz/~3.9 GHz 256MB 16MB $ 3,100
7F72 24/48 240W 3.2 GHz/~3.7 GHz 192MB 8MB $ 2,450

All these performance increases are all amazing, but they need some motherboards to come out that can handle this. AMD currenty1 has 10 platforms in the market and plans to expand to 140+ by the end of 2020, and this processor will certainly help AMD grow.

Submit