• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

New High Performance, x86 Compatible Microprocessors from Centaur / VIA

Very likely. It's a little more fine-grained than just lumping the Atom in with one series or another, but at their core, Atom had a lot more in common with the Pentium3/PentiumM than with any of the 486 line.
(BTW, thanks for pointing that out, I read up on more detail aspects of the Atom line and learned a few more things I didn't know before.)

Speculative execution is a huge core design fundemental decision, not sure I can agree with that but it's down to perspective I suppose.
 
Speculative execution is a huge core design fundemental decision, not sure I can agree with that but it's down to perspective I suppose.
Speculative execution is also very power hungry, which is why Intel cut it from the feature set. Atom was intended to be very power efficient, so power intensive features were removed leaving only essential functionality. However, what remained was still directly derived from Pentium M. From there Atom evolved in directions that followed the idea's of power efficiency. As each iteration was produced the line became more effective. I personally think that Intel did well to make a power efficient CPU that also performed in a competent way...
Yes, they were.
...despite what many may think. I still have two Atom powered devices, both still running strong and both still performing well.
 
Last edited:
Speculative execution is also very power hungry, which is why Intel cut it from the feature set.

I was talking about in terms of being similar, not in terms of whether or not it was a good design decision (it was).
 
A few retail boards beside the known board from Cjoyin:


attachment.php
attachment.php


attachment.php
attachment.php


source: https://www.forum-3dcenter.org/vbulletin/showpost.php?p=12268260&postcount=492
 
Zhaoxin KX-U6780A x86 CPU Tested: The Rise of China's Chips
source: https://www.tomshardware.com/features/zhaoxin-kx-u6780a-x86-cpu-tested


______________________________


I underclocked to 2.7GHz my Eight Core processor AMD FX-8300 3.3GHz

Zhaoxin KaiXian KX-U6780A 2.7GHz vs AMD FX-8300 3.3GHz @ 2.7GHz


Cinebench R20

Multi-threaded
Zhaoxin KX-U6780A 2.7GHz: 982 (+11,72%)
AMD FX-8300 @ 2.7GHz: 879 (89,51%)

Single-threaded
Zhaoxin KX-U6780A 2.7GHz: 127 (94,07%)
AMD FX-8300 @ 2.7GHz: 135 (+6,30%)



Geekbench 4

Multi-threaded
Zhaoxin KX-U6780A 2.7GHz: 9 128 (+8,99%)
AMD FX-8300 @ 2.7GHz: 8 375 (91,75%)

Single-threaded
Zhaoxin KX-U6780A 2.7GHz: 1 780 (94,53%)
AMD FX-8300 @ 2.7GHz: 1 883 (+5,79%)
 
Last edited:
Steve tested older Zhaoxin CPU very simular as my VIA QuadCore C4650 2.0GHz (4C/4T) from 2015:

Steve: 28nm Zhaoxin Kaixian ZX-C+ C4701 2.08GHz (4C/4T)
Tomshardware: 16nm Zhaoxin KaiXian KX-U6780A 2.7GHz (8C/8T) SoC

______________________


Nice desktop with Zhaoxin Eight Core processor

c06537734.png


HP 268 Pro G1 MT
ZhaoXin KaiXian KX-U6780A Processor
2.7 GHz
8 MB L2 cache, 8 cores
8 threads
ZhaoXin C960 UHD Graphics
70 W
Supports DDR4 memory up to 2666 MT/s

source: https://support.hp.com/in-en/product/hp-268-pro-g1-mt/32591011/document/c06537715#AbT2
source: http://h10032.www1.hp.com/ctg/Manual/c06558314
 
Last edited:
Steve tested older Zhaoxin CPU very simular as my VIA QuadCore C4650 2.0GHz (4C/4T) from 2015:

Steve: 28nm Zhaoxin Kaixian ZX-C+ C4701 2.08GHz (4C/4T)
Tomshardware: 16nm Zhaoxin KaiXian KX-U6780A 2.7GHz (8C/8T) SoC

______________________


Nice desktop with Zhaoxin Eight Core processor

c06537734.png


HP 268 Pro G1 MT
ZhaoXin KaiXian KX-U6780A Processor
2.7 GHz
8 MB L2 cache, 8 cores
8 threads
ZhaoXin C960 UHD Graphics
70 W
Supports DDR4 memory up to 2666 MT/s

source: https://support.hp.com/in-en/product/hp-268-pro-g1-mt/32591011/document/c06537715#AbT2
source: http://h10032.www1.hp.com/ctg/Manual/c06558314
That's different! Very interesting. Don't think it'll sell any better than the pet rock...
 
i miss cyrix.

I don't, it was a HOT! pile of SHIT, days after getting the 150+ i wished i had saved a little more for AMD's chip.

As for this thread i will believe it when they proove it and not before
 
China only... Loeschzwerg boards arrived to Germany (EU) 5.5.2020 ;)


Lochzerg KX6000 mainboard.jpg


source: https://www.forum-3dcenter.org/vbulletin/showpost.php?p=12300213&postcount=520



I compared the results with Loeschzwerg at a distance with my AMD FX-8300 at 2.7GHz.


Zhaoxin KX-U6780A 2.7GHz

pc5ryy.png


CPU-Z 1.92 (Benchmark Version 17.01.64)
Single-Thread: 181 (+14,70%)
Multi-Thread (8T): 1442 (+36,36%)
source: https://valid.x86.fr/pc5ryy



Geekbench 4.4.2 Tryout for Windows x86 (64-bit)
Single-Core Score: 1820 (96,86%)
Multi-Core Score: 9093 (+8,06%)
source: https://browser.geekbench.com/v4/cpu/compare/15466947?baseline=15466403



Geekbench 5.1.1 Tryout for Windows x86 (64-bit)
Single-Core Score: 362 (93,06%)
Multi-Core Score: 2387 (+8,50%)
source: https://browser.geekbench.com/v5/cpu/compare/2070351?baseline=20673433


vs


AMD FX-8300 3.3GHz @ 2.7GHz

UPF3fQb.jpg



CPU-Z 1.92 (Benchmark Version 17.01.64)
Single-Thread: 158,1 (87,18%)
Multi-Thread (8T): 1057,2 (73,34%)


Geekbench 4.4.2 Tryout for Windows x86 (64-bit)
Single-Core Score: 1879 (+3,24%)
Multi-Core Score: 8415 (92,54%)
source: https://browser.geekbench.com/v4/cpu/compare/15466947?baseline=15466403



Geekbench 5.1.1 Tryout for Windows x86 (64-bit)
Single-Core Score: 389 (+7,46%)
Multi-Core Score: 2200 (92,17%)
source: https://browser.geekbench.com/v5/cpu/compare/2070351?baseline=2067343
 
Last edited:
Relative performance of Zhaoxin's processors in SPECint_2006

Single Thread: ZX-A (1) vs ZX-C (1,4x) vs KX-5000 (1,8x) vs KX-6000 (2,7x) vs KX-7000 (4,4x)

Multi Thread: ZX-A (1) vs ZX-C (2,3x) vs KX-5000 (5,8x) vs KX-6000 (8,1x) vs KX-7000 (31x)

source: https://read01.com/J8j2MyQ.html#.Xr7Eu8DgqUl


A large increase in the performance of the new microarchitecture.
ZX-A / ZX-C / KX-5000 / KX-6000 = small cores
KX-7000 = BIG cores
 
New Geekbench 5 results
CentaurHauls 2000 MHz (8 cores) vs ZHAOXIN KaiSheng KH-37800D 2700 MHz (16 cores)
CentaurHauls Family 6 Model 71 Stepping 2 vs 2x CentaurHauls Family 7 Model 11 Stepping 14
8x New BIG Cores vs 16x small Cores (2x 8 small Cores)


Multi-Core Score 3330 (+8,82%) vs 3060
(91,89%)
Single-Core Score 482 (+33,15%) vs 362 (75,10%)

source:
https://browser.geekbench.com/v5/cpu/compare/3358994?baseline=3361909

The new KaiSheng KH-4000O series will have up to 32 of these new BIG cores while in dual socket up to 64 cores...
"Houston, we have a problem"
 
Last edited:
First impression - Rise of the small cores

Cinebench R23 CPU (Multi Core)
VIA QuadCore C4650 2.0GHz (28nm, 4C/4T, 2MB L2 cache): 733 pts (34,49%)
Zhaoxin KaiXian KX-U6580 2.5GHz (16nm, 8C/8T, 8MB L2 cache), : 2 125 pts (+189,90%)


And here is my older game results with VIA QuadCore compare with Zhaoxin OctaCore

World of Tanks enCORE - ULTRA
VIA QuadCore C4650 2.0GHz (28nm, 4C/4T, 2MB L2 cache) + NVIDIA GeForce GTX1050Ti 4GB OC: 5 648 (50,10%)
Zhaoxin KaiXian KX-U6580 2.5GHz (16nm, 8C/8T, 8MB L2 cache) + NVIDIA GeForce GTX1050Ti 4GB OC:
11 273 (+99,59%)

World of Tanks enCORE - MEDIUM
VIA QuadCore C4650 2.0GHz (28nm, 4C/4T, 2MB L2 cache) + NVIDIA GeForce GTX1050Ti 4GB OC: 8 159 (30,56%)
Zhaoxin KaiXian KX-U6580 2.5GHz (16nm, 8C/8T, 8MB L2 cache) + NVIDIA GeForce GTX1050Ti 4GB OC: 26 702 (+227,27%)


And the current game @ 2021

Sniper Ghost Warrior Contracts 2

1080p HIGH Graphics Settings = perfectly playable and huge graphics

 
Last edited:
Back
Top