Joseph C. Rhodes, III
***** ***** ****** **. **** 36 Cell: 512-***-****
Austin, TX 78747 **********@*******.***
SUMMARY
Proven strong contributor with 10 years of direct experience in the
semiconductor industry (7 years in VLSI implementation and 3 years in
system and product engineering). Implemented physical designs meeting
timing, power, area and other physical goals for both 2+ GHz x86 CPU cores
(in 65nm and 45nm) and x86 chipsets (in 130nm and 90nm) utilizing both full
custom (manual design) and automated design flow. Understand VLSI CAD
tools and wrote additional scripts, when needed, in order to manage files
in the design database as well as to automate several design specific
tasks. Supported products for the embedded market segment through board
design, system bring up, and benchmarking of custom boards that proved
functionality and optimized CPU performance and customer support.
SKILLS / CORE COMPENTENCIES
Operating environments Unix/LINUX, Windows XP/Office
Programming languages PERL, TCL, C, protected mode x86 assembly
Formal verification Cadence LEC (standard EC, clock domain CDC, low power
CLP, ECO)
Implement design from Either by hand (Verilog or Mentor da), or using tools
Verilog specification (Synopsys dc_shell, Cadence rc, First Encounter), PTSI
timing closure
Interpret and address Timing/noise reports from Primetime or ETS and
qor metrics physical checks form Astro or Calibre, perform ECO to
meet DFM or functional requirements
Documentation Word, PowerPoint, Twiki, Framemaker
Collaboration Livemeeting, provided and coordinated training, setup
TWIKI and forms
Content management MySQL, Designsync (Source control management SCM)
PC architecture X86 ops, system buses (ISA, PCI, and AGP)
System design and use PCB design / FPGA (Xilinix or Lattice)
for bring up /
optimization
EMPLOYMENT
04/08-04/09 Freescale, Austin, CAD Software Engineer
TX
Part of small centralized CAD organization that provided design support
in the form of common tool automation, documentation / collaboration, and
additional support enabling design groups across the entire company
. Worked with designers in various different groups to understand their
needs and methods to accomplish their project needs
. Met with tool vendors (mainly Cadence) to address tool requirements
vs. current capabilities and recommended best practices
. Developed automation scripts in Tool Control Language (TCL) to provide
a common methodology allowing various design teams to efficiently
utilize the design tools. This was managed as part of larger project
through using Design Sync as a SCM system.
. Co-developed new TCL wrapper for formal verification automation using
Cadence Conformal Logical Equivalence Checker (LEC) with one other
engineer in Israel
o Enabled smooth transition from the synthesis phase to formal
verification related phases
o Developed and documented automated ECO flow to make targeted
last minute changes taking advantage of formal verification
tools as opposed to the time-consuming manual process
o Added and documented several features to standard equivalency
checking flow
o Implemented rapid changes to automated procedure allowing for
more flexibility
o Provided several training sessions on the new features that were
created and setup several online collaborative environment for
the design community to make effective use of these new
capabilities
. Responded to Remedy tickets from different design teams requesting
support relating to formal verification (quality checks of the clock
synchronization in HDL, equivalency between HDL and physical netlist,
and ECOs)
06/01-04/08 AMD, Austin, TX Design Engineer
10/02-04/08 (CPU Design)
Hand implemented logic to meet timing, area, and power goals and verify
implementation to model for 2+ GHz x86 microprocessor
. 01/04-04/08 Optimized Floating Point Unit (FPU) for multicore K8 in
both 65nm and 45nm
o Partitioned the Floating Point Unit design into 17 smaller sub
pieces for our team to work with simultaneously
o Worked on 3 sub blocks (SLMs) to meet all timing area metrics
(Some new design, some optimization of existing design)
o Performed more detailed timing and electrical analysis of one
route level module (RLM) consisting of four routed SLMs combined
together and extended out to the rest of the chip to resolve all
issues (timing, electrical, physical, and tool related). Data
from RLM was used to make more accurate SLM designs and
architectural trade offs in subsequent iterations
o Translated several blocks to make use of the newer CAD
environment for the 45nm Shanghai project. Converted 20
schematics from Mentor da to Mentor IC Station, converted
placement of 6 blocks to new placement language
o Member of six person team that performed full chip false path
timing review before 65nm Barcelona tapeout
o Preformed cross module hand edits for timing, noise, and other
manufacturing issues
. 10/02-01/04 - Designed four issue x86 instruction decode from RTL
description (New core)
o Designed single cycle four issue instruction cracker (all
previous AMD processors were three issue or less) to a timing
goal of 450ps
o Designed 32 byte instruction skid buffer (selected up to four
instructions, depending on length, from buffer and stored off
any remaining bytes for next cycle processing)
o Synthesized and Auto Place and Route (APR) large decode table
(between 90K and 100K gates) providing a quick first pass
reasonable placeholder
06/01-10/02 (Chipset Design)
Provided chipset solutions to enable the AMD CPU platform
. Backend implementation of Hyper Transport to PCI / PCI-X bridge (over
700k gates) in an ASIC flow
o One of three authors and one of two maintainers for entire
backend compile script
o Ran biweekly full chip builds using backend scripts, reviewed
the quality of results (qor), made improvements to help close on
design goals (revised floorplan, changed tool optimizations,
updated timing budgets, and feedback of other required design
changes)
. Developed netlist parser written in PERL and preformed several
optimizations on the physical netlist (scan chain balancing and a
large number of last minute performance fixes)
. Reviewed all physical design checks before passing design on to
physical layout
. Supported effort to be first to market with an AGP 3.0 platform by
producing a test chip to mimic a graphics card (since there were no
AGP 3.0 cards available to bring up the platform)
o Developed PERL script to generate random AGP bus traffic test
patterns and validated 96%+ logical test coverage using Verisity
Surecov resulting in error free tape out of test chip that was
used to bring up first K8 platforms
ADDITIONAL EXPRIENCE
01/97-06/00 (every other AMD, Austin, TX Co-op
semester)
. Member of in-house design team that was quickly reconstituted to
enhance chipset features for the K7 system platform that were lacking
relative to the competition
o Wrote PERL script to migrate two processor K7 Northbridge
chipset to new cell library allowing leverage of old reference
design as a starting point
o Worked on synthesis and timing analysis / optimization on bus
synchronization unit
. Supported embedded customers that used established parts which were no
longer supported by a design team
o Solved a customer system level JTAG issue by running processor
simulations on Zycad (a software simulation environment) and
ultimately preformed an ECO to fix bugs in AMD Elan SC520 (486
system on a chip)
o Characterized running K6 processor at a lower voltage and single-
handedly implemented a system level test allowing the Embedded
Processor Division to verify K6 in system stability at that
lower voltage to delivered critical customer samples stable and
on time (received Vice President's Spotlight Award)
o Created interactive Web based tools using PERL Common Gateway
Interface (CGI) scripts that found recurring sleep mode problem
on 486 wafers at certain chip (die) locations improving yield
fallout substantially
. Provided real platforms to test processors and other enablement
technology to guarantee reliable and optimal real world performance
o Designed and debugged K6 processor compare printed circuit board
with several other engineers that debugged a bus cycle
discrepancy in some K6 processors
o Designed K7 processor voltage simulation board, using a Xilinx
4036 FPGA, for a fast reliable voltage level conversion for the
K7 processor, which used different voltage level, HSTL (1.5V),
from the standard TTL (3.0V) logic on most motherboards
o Verified that several OSes (Windows and Unix based) booted on a
simulated K5 processor
o Performed K6 Fast PC benchmarking, using Winbench 97, that
determined the optimal bus speed used in the definition of the
AMD Super 7 design (100 MHz Socket 7 bus)
05/96-07/96 Compaq, Houston, TX Summer Intern
. Supported marketing by providing stable custom configured prototypes
for large customers by configuring, stability testing, shipping
workstations, and rebuilding returned units after the preview
. Reworked system boards to replace a backward capacitor on one lot
(about 30 systems) rather than send the units back for rework saving
two weeks
EDUCATION AND TRAINING
1995-2001 Southern Methodist University, Dallas, TX, BS in Electrical
Engineering with a minor in Computer Engineering and Mathematics