Design space exploration

For CXL Attach Modes, senior architects map options against cache/memory latency, coherency traffic, attach-mode bandwidth on the product workload — not on a single directed test.

Option A — conservative

• Minimal / simple: helps timing, area, verification

• Risk: bandwidth and latency tails

• Validate with: control paths and low-rate peripherals

Option B — buffered / outstanding

• Buffered / outstanding: helps throughput under latency

• Risk: deadlock and debug complexity

• Validate with: DMA and memory-class traffic

Option C — QoS / arbitration

• QoS / arbitration: helps product-critical traffic wins

• Risk: verification state explosion

• Validate with: mixed CPU/GPU/DMA SoCs

Option D — software-first

• Software contract: helps predictable programming model

• Risk: portability and driver cost

• Validate with: platforms with long SW lifetime

DESIGN SPACE — CXL Attach Modes

        performance
            ^
            |     [C] QoS-heavy
            |        *
            |   [B] buffered *
            |              *
            | [A] simple *
            +--------------------> complexity
                      [D] SW-first

Pick the smallest option that moves cache/memory latency, coherency traffic, attach-mode bandwidth on the product workload.

Design pitfalls

• Sizing for peak headline bandwidth instead of payload efficiency

• Adding outstanding depth without ordering analysis

• Choosing aggressive hardware before a reduced sequence proves the mechanism

Tradeoff curve

BEFORE / AFTER — CXL Attach Modes

           failing        target
metric  |    ●              ┄┄┄┄┄┄┄
        |     \
        |      \___ ● bounded fix
        |           \
        |            ● validated
        +-------------------------------> change set
Prove the mechanism moved the metric; one good dot is not proof.

Protocol deep dive

PCIe is reliable packet delivery over unreliable links; debug flows PHY -> DLL -> TLP -> firmware.

Concept diagram

PCIe DEBUG TOP-DOWN

L0 link healthy?  -> credits OK?  -> TLP completes?  -> driver happy?

Skip a layer and you will mis-own the bug.

Metric graph

LINK DEGRADE EXAMPLE

target x4 Gen4  ---- ---- ---- ----
actual   x4 Gen4  ---- ---- ---- ----   (eval board)
actual   x1 Gen3  -                   (product board)

Package/SI often shows up as width downgrade, not hard fail.

Metrics and artifacts to collect

• link width/speed

• replay count

• completion timeout

• AER error log

• LTSSM history

Mini case study

Endpoint enumerated but DMA timed out: completion credits exhausted because a switch port was misconfigured in firmware, not because the endpoint was broken.

Debug branches

• If degrade at width/speed, PHY/SI before driver.

• If replay storm, link layer before transaction layer.

• If CXL coherency bug, separate .io vs .cache vs .mem traffic.

Senior review question

Ask: what is the first transaction that deviates, and which spec rule does it test?

Key takeaways

• Connect every protocol claim to a transaction identity and measurable metric.

• Store the artifact (waveform, log, counter) next to every signoff decision.

Common pitfalls

• Debugging timeouts without finding the first bad transaction.

• Quoting peak bus width without payload efficiency and retry overhead.

• Treating VIP compliance as a substitute for system integration replay.

Principal review addendum

Re-read CXL Attach Modes against one concrete product workload, not a synthetic directed test.

CXL.io, CXL.cache, and CXL.mem layer coherent and memory semantics on PCIe infrastructure.