Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)

Page 365 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Vattila

Senior member
Oct 22, 2004
821
1,457
136
Except for the details about the improvements in the microarchitecture, we now know pretty well what to expect with Zen 3.

The leaked presentation by AMD Senior Manager Martin Hilgeman shows that EPYC 3 "Milan" will, as promised and expected, reuse the current platform (SP3), and the system architecture and packaging looks to be the same, with the same 9-die chiplet design and the same maximum core and thread-count (no SMT-4, contrary to rumour). The biggest change revealed so far is the enlargement of the compute complex from 4 cores to 8 cores, all sharing a larger L3 cache ("32+ MB", likely to double to 64 MB, I think).

Hilgeman's slides did also show that EPYC 4 "Genoa" is in the definition phase (or was at the time of the presentation in September, at least), and will come with a new platform (SP5), with new memory support (likely DDR5).

Untitled2.png


What else do you think we will see with Zen 4? PCI-Express 5 support? Increased core-count? 4-way SMT? New packaging (interposer, 2.5D, 3D)? Integrated memory on package (HBM)?

Vote in the poll and share your thoughts! :)
 
Last edited:
  • Like
Reactions: richardllewis_01

Tuna-Fish

Golden Member
Mar 4, 2011
1,667
2,537
136

How is this done? 2 from the dedicated link, and then 2 split off from the GPU link, leaving 8x for the gpu?

For real, though, I really wish someone would make a good PCIe 5.0 switch that's cheap enough to use for consumer motherboards. Something like 4x 5.0 -> 3* 4x 5.0 would be ideal. No consumer really needs all the PCI slots to work at the same time, they just need very fast access to any one of the devices at one time. The entire PCIe switch market seems to now be catering to server customers, with higher-capacity but also much higher cost products.
 

nicalandia

Diamond Member
Jan 10, 2019
3,331
5,282
136
Okay, took my time and did a better Graphic Chart, comparing the 7900X vs the 5900X, 12900K and 5950X...
1663182203037.png


That beast is an Integer Monster.... It matches the 5950X in FP, but its much superior Integer Operations.


Thread for Thread Ryzen 7000 is stronger than Alder Lake(24T 7900X vs 27T 12900K) in Sisoftsandra
 
Last edited:

TESKATLIPOKA

Platinum Member
May 1, 2020
2,696
3,260
136
Okay, took my time and did a better Graphic Chart, comparing the 7900X vs the 5900X, 12900K and 5950X...

View attachment 67556

That beast is an Integer Monster.... It matches the 5950X in FP, but its much superior Integer Operations.


Thread for Thread Ryzen 7000 is stronger than Alder Lake(24T 7900X vs 27T 12900K) in Sisoftsandra
I find comparing It to 5900x the most interesting.
R9 5900xR9 7900xDifference in %
Whetstone Double32442932.4 %
Whetstone Single38851733.2 %
Dhrystone Long59491954.7 %
Dhrystone Int58988650.4 %

edit: added %
 
Last edited:

Kaluan

Senior member
Jan 4, 2022
515
1,092
106
Okay, took my time and did a better Graphic Chart, comparing the 7900X vs the 5900X, 12900K and 5950X...

View attachment 67556

That beast is an Integer Monster.... It matches the 5950X in FP, but its much superior Integer Operations.


Thread for Thread Ryzen 7000 is stronger than Alder Lake(24T 7900X vs 27T 12900K) in Sisoftsandra
I find comparing It to 5900x the most interesting.
R9 5900xR9 7900xDifference
Whetstone Double32442932.4
Whetstone Single38851733.2
Dhrystone Long59491954.7
Dhrystone Int58988650.4
Based on this data, asuming 7900X to 7950X performance scaling (same TDPs on both Ryzen 9s within their generations) is similar to 5900X to 5950X (29,5-34% faster), the 7950X should score roughly:

- 570 in Whetstone Double
- 690 in Whetstone Single
- 1220 in Dhrystone Long
- 1150 in Dhrystone Int

I could be very off mark, or maybe right on it, IDK if I'm missing something. Was about to ask if someone has a chart with both 5900X AND 5950X on it for this purpose then @nicalandia delivered even before I had a chance to ask haha
 

maddogmcgee

Senior member
Apr 20, 2015
410
421
136
I know this is a CPU thread and not a GPU one, but i thought I would post this as it looks like User Benchmark has applied the CPU scoring system for the GPU section as well.
 

Attachments

  • user.jpg
    user.jpg
    357.1 KB · Views: 92
Jul 27, 2020
28,038
19,139
146
Imagine RAID-0 with PCIE gen 5 drives. You know someone will do it, because why not?
Yeah. Someone will do it. Someone who has too much money to burn or someone who is dumb enough not to realize that the faster they can write to the SSD, the faster it degrades and approaches EOL.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,777
6,791
136
Someone who has too much money to burn or someone who is dumb enough not to realize that the faster they can write to the SSD, the faster it degrades and approaches EOL.
So in other words the average Linux kernel developer or distro maintainer or Android System Developer (e.g. GKH/Linus Torvards, Android folks) is dumb and stupid and read and writes millions of files to the SSD day in day out. They should go back to spinning disks.
 
Jul 27, 2020
28,038
19,139
146
So in other words the average Linux kernel developer or distro maintainer or Android System Developer (e.g. GKH/Linus Torvards, Android folks) is dumb and stupid and read and writes millions of files to the SSD day in day out. They should go back to spinning disks.
I was replying about PCIe 5.0 RAID-0. Shouldn't developer workloads involve more random I/O than sequential I/O (which is what PCIe 5.0 mainly improves)? Please correct me if I'm wrong.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,777
6,791
136