Profile updated as of 10th July. 2019
A profile for the FX-8300, a Bulldozer-derivative CPU based on Piledriver architecture. I don't remember these too fondly... lol.
Please click on the gallery to enlarge to see the full die shot.
Brand Name: AMD FX-8300 Eight-Core Processor
Manufacturer: Advanced Micro Devices
Release Date: December 29, 2012
MSRP: $129 USD
CPU Implementation: Full Chip
CPU Type: CPU
Lithography Process: Global Foundries 32nm SOI
Socket: AM3+ (PGA 941)
Approximate die size: 315mm²
SashleyCat's CPU die Size Rating: large
Approximate Transistor Count: 1,200 Million
Approximate Transistor Density: 3.8 Million / Square Milimetre
Simultaneous Multi-Threading: No * (see note)
Out of Order Execution: Yes
Address Space: 64-bit
Integrated Graphics: No
On-die USB/SATA: No
On-die Network/Modem: No
PCI-E Lanes: 32 Gen2
Unlocked for Manual Overclocking: Yes
Turbo Boost: Yes
Automatic Overclocking: No
Instruction Set Architecture: x86
Additional Instructions (potentially non-exhaustive):
SSE, SSE2, SSE3, SSSE3, SSE4, SSE4.1, SSE4.2, SSE4a, AES, AVX, FMA3, AMD-V, XOP, EVP, TBM, F16C, BMI1
CPU Computing Resources
Processor Cores: 8 *
Independant Processor Cores: 4 Modules
Logical Threads: 8
Core Topology: 4 x 2-core Modules
Level 4 Cache: N/A
Level 3 Cache: 8192 KB
Level 2 Cache: 4 x 2048 KB
Level 1 Instruction Cache: 4 x 64 KB
Level 1 Data Cache: 8 x 16 KB
Core Execution Width: 4 (Integer core) 4 (FPU Engine)
Core Seperated INT/FP Engines: Yes
Core Integer Execution Units: 2 ALU, 2 AGU *
Core Floating Point/SIMD Execution Units: 2 x 128-bit FMAC (MUL/ADD) 2 x MMX Pipes
CPU Memory Subsystem
Main Memory Type: DDR3
Maximum Main Memory: 64 GB
Main Memory Composition: dual-channel (2 x 64-bit)
Main Memory Supported Frequency/Data Rate: 933 MHz / 1866 MHz (PC3-14900)
Main Memory Peak Memory Bandwidth: 29,856 MB/s
CPU Frequency and Peak performance
Core Base Clock Speed: 3300 MHz
Core Peak Turbo Speed: 4200 MHz *
Interconnect Clock Speed: Unknown
CPU SIMD Computing Power FP16: 422,400 Million operations per secondwith FMA *
CPU SIMD Computing Power FP32: 211,200 Million operations per secondwith FMA *
CPU SIMD Computing Power FP64: 105,600 Million operations per second with FMA *
CPU Thermal and Power
Standard Cooling Solution: Small top-down cooler with Aluminium heatsink
Thermal Design Power: 95 W
Typical Maximum Power Use (including Turbo): 100W+ *
Maximum Allowed Junction Temperature (TJ Max): 70.5*C
Normal Peak Voltage Range: < 1300 mV
The FX-8300 is a 8-core, 4-module CPU released in late December, 2012, based on AMD's Vishera chip; using the 2nd generation "Bulldozer" derivative known as "Piledriver". This architecture was somewhat lacklustre in performance, especially in single-threaded workloads. Piledriver improved IPC throughput of Buldozer by a small amount but it is still extremely weak when compared to modern cores like Zen or Skylake. This processor's module design is unique in that the chip contains four independant processing elements ("Modules") which each contain two Integer Cores and a single FPU Core. The FPU has four pipes which can be accessed by the two threads on the module, using something similar to SMT, while each Integer Core services just one of those two threads. The result is a relatively higher amount of Integer throughput in multi-threaded workloads, but a single thread may only access half of the module's full ALU throughput; reducing single-threaded performance signficiantly versus a monolithic, wide core design with two threads and SMT: whereby even a single thread can use all of the core's resources if required.
CPU approximate Performance
Sashleycat CPU Performance Rating: (2019): Poor performance overall. Acceptable for well threaded tasks.
The FX-8300 has poor performance in many tasks, including gaming and especially in lightly threaded workloads. It somewhat redeems itself when all 8 cores can be utilised, specially in Integer heavy workloads, but even then its performance is never consistantly greater than an 8-thread Intel CPU from the same time: those based on Sandy and Ivy Bridge using a more traditional monolithic wide core with SMT. The only saving grace for this CPU is the price, it can potentially offer a reasonable amount of multi-threaded grunt for some applications at a price often south of 100 US dollars. Power efficiency is generally poor, however.
CPU SIMD Computing Power
These are based on my understanding of the FPU in the CPU. Piledriver's module has a single FPU with two 128-bit FMAC pipes. This represents a total vector width of 256-bit per module. That is able to work on 8 Single-Precision numbers in one clock, per module. However it can use Fused-Multiply Add to do two ops per clock. I am currently unsure of how this is represented in the block diagram. This is to the best of my understanding.
Core Peak Turbo Speed:
Maximum possible core clock speed is listed. The CPU will run its cores lower due to power/electrical and thermal limitations.
Typical Maximum Power Use (including Turbo):
It probably uses more power but I can't rememeber if I tested it.
Core Integer Execution Units:
Listed is per Integer Core and per single thread. Each independant processing element (module) with two Integer cores contains 4 AGU and 4 ALU, but a single thread can only access half of those.
FPU engine technically employs a form of SMT to address two logical threads.
This bit is for my personal opinion on this CPU
Sashleycat's Awesomeness Rating: Poop