VLSI
ANSWERS

PUBLIC SEMICONDUCTOR KNOWLEDGE DESK

ISSUE 034SEOUL · AUGUST 28, 2026VLSI KOREA ↗

CITED WORKING ANSWER / STA

Why can useful skew improve setup but damage hold?

캡처 클록을 늦춰 setup에 쓸 수 있는 시간을 늘리면 같은 엣지 관계의 최소 지연 경로에서는 hold 여유가 줄 수 있습니다. 실제 영향은 launch·capture endpoint, 타이밍 도구의 skew 부호 규칙, uncertainty와 derate에 따라 달라지므로 두 검사를 함께 확인해야 합니다.

READ SOURCES & CONTEXT →ADD PUBLIC CONTEXT
CAN'T FIND IT?

한 가지 공개 가능한 문제로 좁혀 질문하면 검토 대기열 또는 이 기기의 초안으로 보존됩니다.

DRAFT A PUBLIC QUESTION ↗
OPEN KNOWLEDGE EXPORT Working answers · sources · verification states
CSV EDITION ↓JSON API ↗
DV / 02

What is the boundary between a UVM driver and sequencer?

The sequencer coordinates sequence items; the driver translates them into interface activity.

Accellera UVM resources →
Device / 03

Why does GAA improve electrostatic control?

Wrapping the gate around more of the channel improves gate control over channel potential.

Technical review open · Primary device reference required →
Career / 04

How should a new graduate compare STA and PD roles?

Compare daily problem types, ownership boundaries, tool exposure and evidence you can build publicly.

Experience-based · No employer-specific claim →
Architecture / 05

What roles do CXL.io, CXL.cache, and CXL.mem serve?

protocol roles using the public CXL overview; avoid reproducing normative detail.

Compute Express Link Consortium →
Architecture / 06

Why should CXL device type be stated before comparing devices?

type/usage-model context before any performance claim.

Compute Express Link Consortium →
Packaging / 07

What problem does UCIe address in a chiplet system?

UCIe as package-level die-to-die standardization at a high level.

UCIe Consortium →
Packaging / 08

What changed in UCIe 2.0 for 3D packaging?

Cover announced 3D-package support without copying controlled specification details.

UCIe Consortium →
Interface / 09

Which UCIe 3.0 data rates are publicly announced?

with source-scoped public release values and conditions.

UCIe Consortium →
Memory / 10

Why must HBM bandwidth comparisons name the exact product and conditions?

standards taxonomy versus product-specific metrics.

JEDEC Solid State Technology Association →
Public Question Library33 matching answers
Evidence-aware · public only
01STAOpenROAD STA documentation

Why can useful skew improve setup but damage hold?

Relative launch and capture clock movement changes setup and hold margins in opposite ways. →

02DVAccellera UVM resources

What is the boundary between a UVM driver and sequencer?

The sequencer coordinates sequence items; the driver translates them into interface activity. →

03DeviceTechnical review open · Primary device reference required

Why does GAA improve electrostatic control?

Wrapping the gate around more of the channel improves gate control over channel potential. →

04CareerExperience-based · No employer-specific claim

How should a new graduate compare STA and PD roles?

Compare daily problem types, ownership boundaries, tool exposure and evidence you can build publicly. →

05ArchitectureCompute Express Link Consortium

What roles do CXL.io, CXL.cache, and CXL.mem serve?

protocol roles using the public CXL overview; avoid reproducing normative detail. →

06ArchitectureCompute Express Link Consortium

Why should CXL device type be stated before comparing devices?

type/usage-model context before any performance claim. →

07PackagingUCIe Consortium

What problem does UCIe address in a chiplet system?

UCIe as package-level die-to-die standardization at a high level. →

08PackagingUCIe Consortium

What changed in UCIe 2.0 for 3D packaging?

Cover announced 3D-package support without copying controlled specification details. →

09InterfaceUCIe Consortium

Which UCIe 3.0 data rates are publicly announced?

with source-scoped public release values and conditions. →

10MemoryJEDEC Solid State Technology Association

Why must HBM bandwidth comparisons name the exact product and conditions?

standards taxonomy versus product-specific metrics. →

11ArchitectureRISC-V International

What is the purpose of a RISC-V profile?

why a profile provides a named software/implementation target. →

12ArchitectureRISC-V International

Why is RISC-V vector length not a universal performance number?

implementation-specific vector length and workload dependence. →

13Physical DesignOpenROAD project

What does OpenROAD cover in an RTL-to-GDSII flow?

Map documented tool scope while preserving flow/version boundaries. →

14Physical DesignOpenLane project

How does OpenLane 2 relate to an ASIC implementation flow?

documented flow orchestration without claiming PPA outcomes. →

15Physical DesignSkyWater PDK project

Why must an open-PDK result identify its PDK release?

rule/model/version dependence of implementation results. →

16Physical DesignSkyWater PDK project

What is SKY130 and what should a project record alongside it?

PDK identity, release and toolchain provenance. →

17Physical DesignGlobalFoundries PDK project

What is GF180MCU and what are its documentation boundaries?

source/version-driven use of its public PDK docs. →

18DVIEEE Standards Association

How should a SystemVerilog clocking/event question be sourced?

Point readers to the official IEEE language-standard channel and distinguish tool behavior. →

19DVIEEE Standards Association

When should blocking and nonblocking assignment behavior be checked against IEEE 1800?

Frame a canonical semantics answer with a language-standard source, not folklore. →

20DVAccellera Systems Initiative

What is the UVM sequence-item handshake at a high level?

roles while linking official Accellera UVM materials. →

21DVAccellera Systems Initiative

Why use the UVM factory instead of hard-wiring a component type?

configurability concept with official UVM reference. →

22DVAccellera Systems Initiative

What lifecycle concern do UVM objections address?

phase-completion coordination using UVM source material. →

23DVIEEE Standards Association

What question should functional coverage answer?

Frame coverage as planned scenario observation, distinct from code coverage. →

24DVIEEE Standards Association

What makes an assertion property different from a test stimulus sequence?

temporal checking versus stimulus generation. →

25DFTIEEE Standards Association

What problem is IEEE 1149.1 boundary scan intended to address?

standardized boundary-scan context without reproducing standard text. →

26STAOpenROAD project

Why are timing constraints as important as path delays in STA?

analysis context and constraints with a documented open STA tool source. →

27STAOpenROAD project

What is the conceptual difference between setup and hold checks?

Contrast maximum- and minimum-delay timing obligations, leaving tool convention explicit. →

28STAOpenROAD project

Why should clock uncertainty not be silently folded into a path delay?

auditability benefit of named timing assumptions. →

29Physical DesignOpenROAD project

Why can macro pin placement affect routing congestion?

geometry/demand relationship without asserting a universal fix. →

30Physical DesignNational Institute of Standards and Technology

Which inputs must an IR-drop estimate disclose?

List current, topology, resistance and operating-condition provenance. →

31PowerNational Institute of Standards and Technology

What does the αCV²f dynamic-power model leave out?

explicit model exclusions and unit conversions. →

32AnalogNational Institute of Standards and Technology

Why is ideal ADC LSB not the same as measured ENOB?

Distinguish ideal quantizer step from measured nonideal performance. →

33PackagingNational Institute of Standards and Technology

Why must a junction-to-ambient thermal-resistance number include its conditions?

environment and board dependence in a first-order thermal model. →

03 / KNOWLEDGE LOOP

Question
→ source
→ revision.

  1. 01Search a precise technical question
  2. 02Open the cited public source
  3. 03Add safe context or a correction
  4. 04Follow the reviewed revision