Home /Modules/cpu/SPECIFICATION.md

SPECIFICATION.md

lib/cpu/amd-threadripper/SPECIFICATION.md · 105 line(s) · markdown

AMD Ryzen Threadripper 9000 (Zen 5 x86-64) Specification

1. Identity

Model: AMD Ryzen Threadripper 9000 ("Shimada Peak", Zen 5 x86-64) Architecture class: 64-bit many-core HEDT / workstation microprocessor Data path: 64-bit Addressing: 48-bit external address space (256 TiB) Register model: CISC named/split register set Process: 4 nm "Zen 5" Socket: sTR5 (TRX50 for HEDT, WRX90 for PRO WX)

The profile exposes a programmer model while using the shared SLeeLa CPU objects (clock, register file, ALU, decoder, bus) to sequence each instruction.

2. SKU roster (Zen 5 "Shimada Peak", launched July 2025)

High-End Desktop (HEDT) โ€” 4-channel DDR5-6400, sTR5/TRX50

ModelCores / ThreadsBase clockMax boostL3 cacheTDPEvidence
Threadripper 9980X64 / 128~3.2 GHz5.4 GHz256 MB350 Wdocumented
Threadripper 9970X32 / 644.0 GHz5.4 GHz128 MB350 Wdocumented
Threadripper 9960X24 / 484.2 GHz5.4 GHz128 MB350 Wdocumented

HEDT PCIe: up to 88 PCIe 5.0 lanes (platform-dependent).

Workstation PRO WX-Series โ€” 8-channel DDR5-6400, sTR5/WRX90, up to 128 PCIe 5.0 lanes

ModelCores / ThreadsMax boostL3 cacheTDPEvidence
Threadripper PRO 9995WX96 / 1925.4 GHz384 MB350 Wdocumented
Threadripper PRO 9985WX64 / 1285.4 GHz256 MB350 Wdocumented
Threadripper PRO 9975WX32 / 645.4 GHz128 MB350 Wdocumented
Threadripper PRO 9965WX24 / 485.4 GHz128 MB350 Wdocumented
Threadripper PRO 9955WX16 / 325.4 GHz64 MB350 Wdocumented
Threadripper PRO 9945WX12 / 245.4 GHz64 MB350 Wdocumented

Each SKU above is its own runnable class (SLAMDThreadripper<model> / SLAMDThreadripperPro<model>) extending SLAMDThreadripperCPU; they share the Zen 5 x86-64 programmer model and decode() and differ in rated clock and identity label. All rate their spec clock at the 5.4 GHz family max boost; a specific SKU's base clock is a configuration of the same model via setSpecClock.

3. Programmer-visible state

x86-64 register model (Zen 5 microarchitecture) with per-core L1/L2 and shared L3 per CCD, aggregated across multiple CCDs on an I/O die.

The SLeeLa model stores these in SLRegisterFile and exposes 16 addressable register slots to the runnable model.

4. Address and data path

5. Clock

Rated model clock: about 5400 MHz max boost (overclockable in the model via setSpecClock).

The model records nominal frequency, cycle number, bus direction, and completion phase. Clock frequency is not instruction latency: one instruction may consume multiple cycles.

6. Execution model

The SLeeLa model sequences each instruction as:

  1. instruction fetch;
  2. opcode decode;
  3. operand/address acquisition;
  4. ALU/register operation;
  5. memory read/write;
  6. next-instruction transition.

7. Cache

cache coherence protocol: modeled where multi-core

8. DMA

The base CPU has no integrated DMA controller; DMA is a system-level bus-master concern and SLDMAController is left uninstantiated for the base profile.

9. Interrupts

The model records interrupt request, recognition, vector/entry and timing through SLInterruptController without inventing undocumented internal behavior.

10. MMU/TLB

The processor integrates an MMU/TLB; address translation is modeled as virtual -> physical through that unit.

11. Evidence labels

Sources