AMD Phenom II X6 1100T versus FX-8120 Performance Guide

[windwithme]

[windwithme]

Re: AMD Phenom II X6 1100T versus FX-8120 Performance Guide

Upper-Left Corner
990FXA-D3 uses 4+1 phases PWM. The upper white connector is 8-PIN DC-In.
Black AM3+ socket supports AMD AM3+ FX/AM3 Phenom II/AMD Athlon II CPU.


IO
1 X PS2 KB/MS
1 X S/PDIF optical output connector
8 X USB 2.0(Red)
2 X USB 3.0(Blue)
1 X RJ-45 LAN
6 X Audio


System Configuration
CPU: AMD Phenom II X6 1100T / FX-8120 8-Core Processor
MB: GIGABYTE 990FXA-D3
DRAM: CORSAIR VENGEANCE CMZ8GX3M2A1600C9
VGA: msi N560GTX-Ti Twin Frozr II
HD: Intel 510 Series 120GB
POWER: be quiet! STRAIGHT POWER GOLDEN 550W
Cooler: CORSAIR Hydro Series H60
OS: Windows7 Ultimate 64bit SP1


Besides CPUs are different, the other spec is all the same.
Performance Test
1100T
CPU 201.4 X 29 => 3323.1MHz(Torbo 3726MHz)
DDR3 1610.8 CL7 9-8-24 1T
FX-8120
CPU 201.4 X 29 => 3222.4MHz(Torbo 4028MHz)
DDR3 1879.2 CL9 10-9-24 1T

As same software test below, the first pic is 1100T and second is FX-8120.
Hyper PI 32M X 6 => 22m 48.628s
CPUMARK 99 => 504


Hyper PI 32M X 8 => 28m 37.477s
CPUMARK 99 => 423


CPUMARK and PI both have long history. I still remember K8 was always leading Pentium4 in CPUMARK and PI.
Till now, CPUMARK is still very good and easy benchmark software in single core test.
Early stage, SUPER PI only can test single core CPU. Now, latest version Hyper PI can support 16 cores. It’s a very common benchmark software.
1100T single core performance is higher than FX-8210 around 19.1% in CPUMARK. In Hyper PI 32M, multi-core full speed is also 26.2% faster.

Nuclearus Multi Core => 2216
Fritz Chess Benchmark => 24.33/11677


Nuclearus Multi Core => 16231
Fritz Chess Benchmark => 20.87/10019


Nuclearus Multi Core hasn’t been update after 2008. It doesn’t support multi-core well.
As FX-8120, it can run Multi Thread Speed. In contrast, 1100T cannot run this option.
Former 6-Cores 980X and recent 3-Cores A6-3500 also cannot finish the test. Implying it cannot support triple murderer cores?!
However, we can refer to the full finished ALU Speed and FPU Speed items, 1100T is still leading obviously.
Fritz Chess Benchmark is a multi-core benchmark by chess. It can support to 8 threads.
1100T 6-core full speed is 24.33 and FX-8120 8-core full speed is 20.87. 1100T is still 16.6% higher.

CrystalMark 2004R3 => 222715


CrystalMark 2004R3 => 219825


CrystalMark 2004R3 test several items performance in full system. 1100T is still a little ahead in total score.
If look into CPU performance, you can refer to the ALU and FPU to see the difference.
For DDR3 performance, FX-8120 has more advantages. You can see detail test below.

[windwithme]

[windwithme]

Re: AMD Phenom II X6 1100T versus FX-8120 Performance Guide

1100T DDR3 1610.8 CL7 9-8-24 1T
ADIA64 Memory Read - 9280 MB/s
Sandra Memory Bandwidth - 12768 MB/s
MaXXMEM Memory-Copy - 11660 MB/s


FX-8120 DDR3 1879.2 CL9 10-9-24 1T
ADIA64 Memory Read - 14699 MB/s
Sandra Memory Bandwidth - 18120 MB/s
MaXXMEM Memory-Copy - 13862 MB/s


DDR3 bandwidth, FX-8120 wins 1100T in big gap.
FX-8120 leads 58.4% in ADIA64, 41.9% in Sandra and 18.9% in MaXXMEM.
FX-8120 bandwidth score is similar to Intel first Gen. Core i platform. It’s still a little behind 2nd Gen. Core i platform.
At least it’s a shot in the AMD arm. In the past 8 years, this improvement is a big step. I hope they can catch up the competitor’s bandwidth.

Temperature – Enable Cool 'n' Quiet power saving technology
1100T Enter to OS Desktop - 34(Room temperature is 20˚C)


FX-8120 Enter to OS Desktop - 30


1100T Full Speed - 60
LinX 0.6.4


FX-8120 Full Speed - 45
LinX 0.6.4


AMD32nm APU and FX series CPUs always show very temperature in most software.
Sometimes, it’s even lower than room temperature. It’s often to see 8~16˚C. However, CPU is not a cooler. The temperature should be higher than room temperature.
It shouldn’t be lower environment temperature. I will ignore AIDA number.
For right side MB tool to compare, we can use this for reference. Even though, I think this number is still lower than FX-8120 real temperature.
In Touch BIOS, 32nm FX-8120 burning temperature is much lower.
It has 2 more cores than 1100T. As we set both in default mode, FX-8120 temperature owns the obvious advantage.

Power Consumption Test
1100T Enter to OS Desktop (Enable CNQ) - 81W


FX-8120 Enter to OS Desktop (Enable CNQ) - 74W


1100T Enter to OS Desktop (Disable CNQ) - 103W


FX-8120 Enter to OS Desktop (Disable CNQ)- 93W

[windwithme]

WIND3C