Socket AM3+
(Redirected from Delhi)
AMD Socket AM3+, also known as Socket AM3b or Socket AM3r2, is compatible with a subset of AMD Orochi processors. It is backwards compatible with K10 processors designed for the AM3 socket, but these are not yet supported by coreboot-15h.
Official Documentation
Orochi Processors
Orochi FX (Zembezi and Vishera) and Opteron (Zurich and Delhi) processors are compatible with the AM3+ socket. TDP ratings for these processors range between 25 W and 220 W. Each AM3+ motherboard has a maximum supported TDP for CPUs. For example, the Supermicro H8SML is only designed to support up to 65 W processors.
Official Source Code and Documentation
- AGESA Source Code
- AGESA Interface Specification
- BIOS and Kernel Developer's Guide
- Product Data Sheet (FX)
- Product Data Sheet (Opteron)
- Software Optimization Guide
FX "Zembezi" series (Bulldozer, 32 nm)
- Transistors: ~1.6 billion (real node count)
- Die size: 319 mm2 (real measured up size)
- L1 data cache (per core): 16 kb
- L1 instruction cache (per module): 64 kb
- L2 cache (per module): 2048 kb
- All models support: MMX(+), SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4a, NX bit, AMD64, AMD-V, IOMMU, AES, CLMUL, AVX, XOP, FMA4, F16C, ABM, Turbo Core 2.0, PowerNow!, ECC
| Model | Step. | Cores | Frequency (GHz) | L3 cache |
HT clock (GHz) |
Multi | Core voltage (V) |
TDP (W) | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Base | Full-load turbo |
Half-load turbo | ||||||||
| FX-4100 | B2 | 4 | 3.6 | 3.7 | 3.8 | 8 MB | 2.0 | 4–19× | 0.775–1.425 | 95 |
| FX-4120 | 3.9 | 4.0 | 4.1 | 4–20.5× | ||||||
| FX-4130 | 3.8 | 3.9 | 4.0 | 4 MB | 4–20× | 125 | ||||
| FX-4150 | 8 MB | 2.2 | 0.75–1.40 | 95 125 | ||||||
| FX-4170 | 4.2 | 4.3 | 4.3 | 4–21.5× | 125 | |||||
| FX-4200 | 4 | 3.3 | 3.6 | 3.9 | 7–19.5× | 0.9625–1.4125 | 125 | |||
| FX-6100 | 6 | 3.3 | 3.6 | 3.9 | 2.0 | 4–19.5× | 0.775–1.425 | 95 | ||
| FX-6120 | 3.6 | 3.9 | 4.2 | 4–21× | ||||||
| FX-6130 | 3.8 | 3.9 | 4–19.5× | |||||||
| FX-6200 | 3.8 | 4.0 | 4.1 | 2.2 | 4–20.5× | 0.75–1.40 | 125 | |||
| FX-8100 | 8 | 2.8 | 3.1 | 3.7 | 2.0 | 4–18.5× | 0.775–1.425 | 95 | ||
| FX-8120 | 3.1 | 3.4 | 4.0 | 4–20× | 95 125 | |||||
| FX-8140 | 3.2 | 3.6 | 4.1 | 2.2 | 4–20.5× | 95 | ||||
| FX-8150 | 3.6 | 3.9 | 4.2 | 4–21× | 125 | |||||
| FX-8170 | 3.9 | 4.2 | 4.5 | 4–22.5× | 0.75–1.45 | |||||
FX "Vishera" series (Piledriver, 32 nm)
- L1 data cache (per core): 16 kb
- L1 instruction cache (per module): 64 kb
- L2 cache (per module): 2048 kb
- All models support: MMX(+), SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4a, NX bit, AMD64, AMD-V, IOMMU, AES, CLMUL, AVX, XOP, FMA3, FMA4, F16C, ABM, BMI1, TBM, Turbo Core 3.0, PowerNow!, EVP, ECC
| Model | Step. | Cores | Clock frequency (GHz) | L3 cache |
HT clock (GHz) |
Multiplier | Core voltage (V) |
TDP (W) | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Base | Full-load turbo |
Half-load turbo1 | ||||||||
| FX-4300 | C0 | 4 | 3.8 | 3.9 | 4.0 | 4 MB | 2.0 | 7.0–20.0× | 0.8875–1.4125 | 95 |
| FX-4320 | 4.0 | 4.1 | 4.2 | 7.0–21.0× | ? | |||||
| FX-4350 | 4.2 | ? | 4.3 | 8 MB | 7.0–21.5× | 0.8375–1.4125 | 125 | |||
| FX-6300 | 6 | 3.5 | 3.8 | 4.1 | 7.0–20.5× | 0.9000–1.4250 | 95 | |||
| FX-6330 | 3.6 | 3.9 | 4.2 | 2.2 | 7.0–21.0× | 0.8875-1.4250 | ||||
| FX-6350 | 3.9 | ? | 2.0 | 0.8750–1.4250 | 125 | |||||
| FX-8300 | 8 | 3.3 | 3.6 | 4.2 | 2.2 | 0.8250–1.3750 | 95 | |||
| FX-8310 | 3.4 | 3.7 | 4.3 | 7.0–21.5× | 0.8375–1.4125 | |||||
| FX-8320E | 3.2 | 3.5 | 4.0 | 7.0–20.0× | ||||||
| FX-8320 | 3.5 | 3.7 | 4.0 | 0.9500–1.4250 | 125 | |||||
| FX-8350 | 4.0 | 4.1 | 4.2 | 7.0–21.0× | 0.8750–1.4250 | |||||
| FX-8370E | 3.3 | 3.6 | 4.3 | 7.0–21.5× | 0.8250–1.3750 | 95 | ||||
| FX-8370 | 4.0 | 4.1 | 0.8625–1.4250 | 125 | ||||||
| FX-9370 | 4.4 | 4.7 | 7.0–23.5× | ? | 220 | |||||
| FX-9590 | 4.7 | 5.0 | 7.0–25.0× | 0.8875–1.5375 | ||||||
Opteron "Zurich" series (Bulldozer, 32 nm)
- All models support: MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4a, IOMMU, NX bit, AMD64, AMD-V, AES, CLMUL, AVX, CVT16–F16C, XOP, FMA4.
- All models support single socket configurations
- Memory support: Up to 4 DIMMs per socket
- Memory controller: Two channels of UDIMM up to DDR3-1866
- Die size: 315 mm2
| Model Number |
Step. | Cores | Frequency (GHz) |
Cache | Multi | TDP (W) | |||
|---|---|---|---|---|---|---|---|---|---|
| Base | Full Load turbo |
Half Load turbo |
L2 | L3 (MB) | |||||
| B2, Four core, high-efficiency | |||||||||
| 3250 HE | B2 | 4 | 2.5 | 2.8 | 3.5 | 2 × 2 MB | 4 | 12.5–17.5× | 45 |
| 3260 HE | 2.7 | 3.0 | 3.7 | 13.5–18.5× | |||||
| B2, Eight core, high-efficiency | |||||||||
| 3280 | B2 | 8 | 2.4 | 2.7 | 3.5 | 4 × 2 MB | 8 | 12–17.5× | 65 |
Opteron "Delhi" series (Piledriver, 32 nm)
- All models support: MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4a, NX bit, AMD64, AMD-V, AES, CLMUL, AVX, XOP, FMA3, FMA4, CVT16–F16C, Turbo Core 3.0.
- All models support single socket configurations
- Memory support: Up to 4 DIMMs per socket
- Memory controller: Two channels of UDIMM up to DDR3-1866
- Die size: 315 mm2
| Model Number |
Step. | Cores | Frequency (GHz) |
Cache | Multi | TDP (W) | |||
|---|---|---|---|---|---|---|---|---|---|
| Base | Full Load turbo |
Half Load turbo |
L2 | L3 (MB) | |||||
| C0, Four core, energy-efficient | |||||||||
| 3320 EE | C0 | 4 | 1.9 | 2.1 | 2.5 | 2 × 2 MB | 8 | 9.5–12.5× | 25 |
| C0, Four core, high-efficiency | |||||||||
| 3350 HE | C0 | 4 | 2.8 | 3.1 | 3.8 | 2 × 2 MB | 8 | 14–19× | 45 |
| C0, Eight core, high-efficiency | |||||||||
| 3365 | C0 | 8 | 2.3 | 2.6 | 3.3 | 4 × 2 MB | 8 | 11.5–16.5× | 65 |
| 3380 | 2.6 | 2.9 | 3.6 | 13–18× | |||||