umtalelab.com
NAVIGATION
  • Writing
  • LabWIP
  • Archive
TOPICS
  • Blogs
  • News
  • Reviews
  • Tests
CONTACT
  • [email protected]
  • Telegram
  • YouTube
  • Patreon
© 2019–2026 UMTALE LABGLORY TO UKRAINEPrivacy Policy
Writing ▾
LabWIP
About
Contact
// UMTALE.LAB/MENUESC — close · ↵ — go
// UMTALE.LABMENU
MAIN// MENU
LabWIP→
UMTALE LAB · BENCH ONLINE
UMTALE.LAB/WRITING/HARDWARE/// AMD-EPYC-MILAN-ZEN-3-CHIPS-ARE-20-FASTER-THAN-EPYC-ROME-ZEN-2
AMD EPYC Milan [Zen 3] chips are 20% faster than EPYC Rome [Zen 2]
NEWS · HARDWARE
08/28/2020
PUBLISHEDAugust 28, 2020
READING2 min

AMD EPYC Milan [Zen 3] chips are 20% faster than EPYC Rome [Zen 2]

Internal AMD data suggests EPYC Milan processors, based on Zen 3, are 20% faster than Zen 2-powered EPYC Rome chips at the same clock speed. However, these are still just rumors.

UmTale
AMD·ZEN 3·AMD EPYC·ZEN 2·ЧУТКИ
READ ↓
РЕАКЦІЇ

According to slides from AMD for internal use (demonstrating performance gains within the company), AMD EPYC Milan processors, built on the new Zen 3 architecture, are poised to deliver a 20% advantage over current Zen 2-based EPYC Rome chips, all while running at the same frequency.

Reliable tech publication Hardwareluxx was the source of this information. However, it's still just a rumor, so take these claims with a grain of salt.

CPU codename

Naples

Rome

Milan

Genoa

Socket

SP3

SP3

SP3

SP5

Core count/Architecture

32x Zen

64x Zen 2

64x Zen 3

>64x Zen 4

CCD / L3 Cache

8 cores per CCD 2x 16 MB L3-Cache

8 cores per CCD 2x 16 MB L3-Cache

8 cores per CCD 1x 32 MB L3-Cache

-

Process technology

14 nm

7 nm

7 nm

5 nm

RAM support

DDR4

DDR4

DDR4

DDR5

PCIe version

PCIe 3.0

PCIe 4.0

PCIe 4.0

PCIe 5.0

TDP

120 - 180 W

120 - 225 W

120 - 225 W

120 - 240 W

The report indicates AMD's Zen 3 architecture should deliver a 15% IPC uplift. However, AMD seems to be taking a differentiated approach with EPYC Milan processors. For instance, 64-core SKUs might see a 15% performance boost, while 32-core variants could gain a more substantial 20% over their current-generation equivalents.

⤢ ВІДКРИТИ

Furthermore, while earlier each of the two CCXs featured its own 16MB L3 cache, it will now become unified, expanding to 32MB. (Essentially, the cache size per CCD won't change, but it will be faster and more accessible). This alteration should significantly impact overall CPU performance, as cores will no longer need to retrieve data from another CCX's cache.

TAGSAMDZen 3AMD EPYCZen 2Чутки
← PREVIOUSAge of Empires III: Definitive Edition gets system requirementsNEXT →First non-reference NVIDIA GeForce RTX 3090s spotted online: Colorful RTX 3090 Vulcan-X OC and ASUS ROG STRIX RTX 3090 OC
CONTINUE READING

Related articles

Intel Arc A380's latest BIOS update boosts GPU clock by 7.5%
NEWS · HARDWARE

Intel Arc A380's latest BIOS update boosts GPU clock by 7.5%

1 min08/21/2023
AMD Inception patch could slash performance by up to 50%
NEWS · HARDWARE

AMD Inception patch could slash performance by up to 50%

2 min08/17/2023
Rumor: Radeon RX 8000 lineup to skip a high-end graphics card
NEWS · HARDWARE

Rumor: Radeon RX 8000 lineup to skip a high-end graphics card

2 min08/06/2023