Overclock.net - An Overclocking Community - View Single Post - [Technopat] Ryzen R9 3950X Breaks WR's Cinebench15/20 & Geekbench4

View Single Post
post #78 of (permalink) Old 06-14-2019, 06:34 AM
KyadCK
Hey I get one of these!
 
KyadCK's Avatar
 
Join Date: Aug 2011
Location: Chicago
Posts: 7,277
Rep: 306 (Unique: 216)
Quote: Originally Posted by m4fox90 View Post
The nerve of moving of to 7nm and putting all their 14/12nm specialized engineers and production lines out of work!


Sent from my iPhone using Tapatalk
The funny thing is they didn't put them out of work, the IO dies, x570 chipset, and Zen+ APUs are still all made on 12/14nm.

Quote: Originally Posted by SwitchFX View Post
I'm not one to nitpick at others representations, but using percent here feels a bit disingenuous to most consumers even though this is an enthusiast site. As you pointed out, there is a large, significant difference in score from the TR1 chip, however, for the Intel processor, there's over a 400 Mhz difference which shows just how much of an IPC improvement there is with Zen 2 over Intel. I'm not sure if this comes down to using the TSMC process or because 12nm and 14 nm on GloFlo was a highly modified mobile platform on licensed tech from Samsung. Unfortunately, that stuff is way out of the breadth of knowledge I know so I'm open to anyone who knows it to explain why apart from that and the modification of how IF works that AMD managed to best Intel, obviously based on what we know and not third party benches.
He used all percentages correctly, taking both clock speed and score into account for all three comparisons. Nothing about it is disingenuous, it is fact.

AMD covered the changes made in Ryzen 3. Anand goes into depth. Skip to page 6 for arch overview;
https://www.anandtech.com/show/14525...-and-epyc-rome


tl;dr, cores are made up of lots of parts. AMD shored up on some of those parts weak points so the core can operate more efficiently and get fed/process data faster, such as expanding cache to reduce cache misses.

It's like fixing that one intersection that causes a traffic jam, just... for data. Yes, now another intersection that didn't used to be a problem is now a bottleneck, but overall traffic is moving faster. They can address those new problems next time and make it even faster. Intel has been doing this for a long time, that is where you see "3% IPC increase", etc. The difference is AMD hit 10-15% (in this case, +18% in Cinebench over Ryzen 1), not the 3-5% Intel has been doing.

Forge
(18 items)
Forge-LT
(7 items)
CPU
AMD Threadripper 1950X
Motherboard
Gigabyte X399 Designare
GPU
EVGA 1080ti SC2 Hybrid
GPU
EVGA 1080ti SC2 Hybrid
RAM
32GB G.Skill TridentZ RGB (4x8GB 3200Mhz 14-14-14)
Hard Drive
Intel 900P 480GB
Hard Drive
Samsung 950 Pro 512GB
Power Supply
Corsair AX1200
Cooling
EK Predator 240
Case
Corsair Graphite 780T
Operating System
Windows 10 Enterprise x64
Monitor
2x Acer XR341CK
Keyboard
Corsair Vengeance K70 RGB
Mouse
Corsair Vengeance M65 RGB
Audio
Sennheiser HD700
Audio
Sound Blaster AE-5
Audio
Audio Technica AT4040
Audio
30ART Mic Tube Amp
CPU
i7-4720HQ
Motherboard
UX501JW-UB71T
GPU
GTX 960m
RAM
16GB 1600 9-9-9-27
Hard Drive
512GB PCI-e SSD
Operating System
Windows 10 Pro
Monitor
4k IPS
▲ hide details ▲
KyadCK is offline