(ICU 04 Item 1 ) ------------------------------------------------ [10/07/04]
Subject: Customers Rate The Best & The Worst Cadence Tools
READING TEA LEAVES: Whenever I do a survey it's very telling to see how
customers rate specific tools in an EDA company's product line. (I've been
doing this with SNUG surveys for years. Usually the product marketing
marketing managers with "winning" products love me and those with "losing"
products hate me. That's life.)
1.) Which specific Cadence product(s) are you MOST satisfied with?
(write in):
Which specific Cadence product(s) are you LEAST satisfied with?
(write in):
The big thing to track here is the "ratio" for each tool; that is the ratio
of "MOST satisfied" to "LEAST satisfied" answers. For example, in the table
below you'll see:
MOST satisfied LEAST satisfied
----------------------------- -----------------
Ambit BuildGates: ######## 4 ###### 3
NC-Verilog: ###################### 11 ###### 3
Silicon Ensemble: #### 2 #################### 10
This means Ambit has a ratio of 4 to 3 -- or roughly 1:1 -- this means Ambit
has approximately an equal number of people who strongly love it and people
who strongly hate it. As a rule of thumb, I've found a ratio of 1:1 means a
tool is OK in most user's eyes.
A ratio of 2:1 or greater means a tool is doing really well. NC-Verilog
has a ratio of 11 to 3, which is roughly 4:1, which says NC-Verilog is a
really successful tool as far as its user see it.
A ratio of 1:2 or less means a tool is in trouble. For example, the ratio
for Silicon Ensemble is 2 to 10 which equals 1:5. For every one user who
loves SE, there are 5 users who hate it. Not good.
It's important to note that this only gives general trends and that a small
number of responses can throw ratio analysis off. For example, NC-SystemC
has a ratio of 1:0 -- it literally got 1 "best" vote and 0 "worst" votes,
so does this say NC-SystemC is taking over the EDA world or that it's an
obscure tool that seems to do well in its little niche with its small number
of users? (I'd say the latter is probably closer to the truth here...)
MOST satisfied LEAST satisfied
----------------------------- -----------------
Ambit BuildGates: ######## 4 ###### 3
Get2Chip: #### 2 0
Verilog-XL: ########## 5 ## 1
NC-Verilog: ###################### 11 ###### 3
NC-VHDL: ## 1 0
NC-Sim: #################### 10 ###### 3
NC-SystemC: ## 1 0
SignalScan: ## 1 0
SimVision: ## 1 ## 1
NC-COV: 0 ## 1
HDLScore: 0 #### 2
HAL: 0 ###### 3
Palladium: ## 1 0
Test Encounter: ## 1 0
scan support: 0 ## 1
Verplex Conformal is looking good here. No one hated it and 7 loved it.
The Tuxedo and Blacktie vote numbers are too small to tell anything. I was
surprised to see the 1:1 ratios on the Simplex tools, though. I figured
they'd do better than "average", but they didn't.
Verplex Conformal: ############## 7 0
Verplex Tuxedo: ## 1 0
Verplex Blacktie: 0 ## 1
Perl: 0 ## 1
VoltageStorm: #### 2 ###### 3
Simplex QX: ## 1 0
Fire & Ice: 0 ## 1
CeltIC: ############ 6 ######## 4
PacifIC: 0 ## 1
VSOC: ## 1 0
HyperExtract: 0 ## 1
Cadence kicks ass in the Custom IC market. If you add up all the Virtuoso
"best" votes 13 + 6 + 5 == 24 to its "worst" votes 2 + 0 + 0 == 2, you get
a kick ass ratio of 12:1. Wow! That might be a record for positive ratios.
Spectre's looking good, too, with a 7:1 here.
Virtuoso: ########################## 13 #### 2
Virtuoso-LE: ############ 6 0
Virtuoso-XL: ########## 5 0
Spectre: ############## 7 ## 1
Spectre-RF: ## 1 0
Composer: ############ 6 ####### 3
ICFB/DFII: #### 2 #### 2
Analog Artist: ## 1 ## 1
UltraSim: ## 1 0
ICC/CCT: ## 1 ## 1
TDM: 0 ## 1
Aptivia: ## 1 0
AMS Designer: 0 ###### 3
ADE: 0 ## 1
TLF Generator: 0 ## 1
Abstract: 0 ###### 3
The acquired tools are looking great, too. FE gets 3:1, SoC E 4:1, and
Plato -- err -- "NanoRoute" gets 7:1. PKS gets a mediocre 1:1, but at least
it beasts native Cadence tools like SE and CTgen. And I've never seen such
hatred for a tool line until I saw these Assura numbers -- 0 to 25 -- which
is probably a record for a this-tool-sucks ratio, too.
First Encounter: ############################## 15 ######## 4
SoC Encounter: ################ 8 #### 2
PKS: ###### 3 ######## 4
NanoRoute: ############## 7 ## 1
Nano Encounter: ## 1 0
Silicon Ensemble: #### 2 #################### 10
CTgen: 0 ###### 3
Diva: ## 1 ## 1
Dracula: 0 ## 1
Assura: 0 ########################## 13
Assura LVS: 0 ######## 4
Assura DRC: 0 ######## 4
Assura RCX: 0 ###### 3
Assura PL: 0 ## 1
common timing
engine: 0 ## 1
I'll freely admit that these PCB and board design tools are a blind spot for
me. I don't use them and I'm fairly certain that 99% of the 21,500 ESNUG
subscribers are NOT users of PCB tools. I was very surprised to see 11 board
designers responding to the survey!
SpecctraQuest: ## 1 ## 1
Specctra: ## 1 0
OrCAD Capture: ## 1 0
Constraint Manager: 0 ###### 3
Library Manager: 0 ## 1
ConceptHDL: ## 1 ###### 3
Allegro: ## 1 #### 2
PowerMeter: 0 ## 1
PAS: 0 ## 1
CDL: 0 ## 1
PSD: 0 ## 1
Now if you add up all the "best" tool votes and all the "worst" tool votes,
you'll get a good picture of how customers rate Cadence tools overall in
one number. That comes out to a 142 total "best" votes to 115 total "worst"
votes giving Cadence a ratio of 142:115 == 1.2 to 1. What's interesting
is comparing the Cadence overall ratio to the Synopsys overall ratios.
Synopsys
"Best" "Worst"
tool votes tool votes ratio
------------ ------------ -------
2001 306 144 2.13X
2002 344 131 2.63X
2003 371 172 2.16X
2004 351 155 2.26X
Cadence
2004 142 115 1.23X
This mildly surprised me. I thought Cadence's ratio would be roughly in
line with the Synopsys ratio -- but 1.2 is noticeably off from the SNPS
2.3 average. I don't think there's any skew to my polling technique; I
really did send out the survey to all 21,500 ESNUG subscribers.
The only responses I excluded where the ones that came from EDA vendors
or from untraceable @yahoo.com, @hotmail.com, etc. accounts. And the
study I did in ESNUG 422 #7 clearly showed me that in 2003 a surprizing 78%
of ESNUG discussion was non-Synopsys oriented. (In fact, Magma posts were
3 of the Top 10 ESNUG Items in 2003!) Also, in my DVcon'04 Trip Report,
in http://www.deepchip.com/items/dvcon04-02.html Cadence simulators
took 51% of response with ModelTech getting 41% and Synopsys VCS getting
a paltry 34%. Cadence Verplex and Palladium also came out on top in that
same DVcon'04 Trip Report -- and that Trip Report was put together using
the exact same methodology that this one was -- I sent out the survey
on the 21,500 member ESNUG mailing list.
There's no exclusion of who can subscribe to the ESNUG mailing list.
Anybody anywhere can get it. Sheesh -- over Labor Day weekend I did a
study and found 257 Wall Street financial guys on ESNUG! Yup, 257.
In addition, there's that fundamental problem that there's no such thing
as a pure Cadence customer (as in "we Cadence and ONLY Cadence tools in
our entire chip design flow"), nor a pure Synopsys customer, instead 98%
of the projects worldwide use a mix of Cadence and Synopsys and Mentor
and (sometimes) Magma tools.
So putting this all together it does appear that Cadence really does have
a ratio of 1.2 while in contrast Synopsys has a 2.3 average ratio.
MOST satisfied user comments:
First Encounter, not the whole 9 yards, just the floorplanner.
---- ---- ---- ---- ---- ---- ----
I like PKS. I've taped out several chips with it and feel that it made
my life easier than it would have been with other tools.
---- ---- ---- ---- ---- ---- ----
SoC Encounter. The integration seems better and the technology good
for DSM.
---- ---- ---- ---- ---- ---- ----
My answer is sort of like saying "The best person in the Guantanamo Bay
prison system is..."
---- ---- ---- ---- ---- ---- ----
Verplex Conformal - excellent and easy to use logic equivalency checker
---- ---- ---- ---- ---- ---- ----
Composer, Virtuoso Layout Editor. MOST satisfied in this case mean
least problems. I don't think I can use "satisfied" and "Cadence" in
the same sentence.
LEAST satisfied user comments:
NanoRoute is really fast to generate bad routes.
---- ---- ---- ---- ---- ---- ----
CeltIC - don't think it is proven for STA signoff
---- ---- ---- ---- ---- ---- ----
I don't like SE. I'm happy to see it leaving official support. It has
made my life more difficult for the last many years.
---- ---- ---- ---- ---- ---- ----
LEAST satisfied with VoltageStorm-PE, flat out doesn't work.
---- ---- ---- ---- ---- ---- ----
Assura. We invested many years of development in this and it never
worked satisfactorily.
---- ---- ---- ---- ---- ---- ----
Verplex Blacktie (formal property checker) We evaluated 2 years ago and
found it hard to use and understand. I was told the new one coming out
will be much better and easier to use.
---- ---- ---- ---- ---- ---- ----
Simplex tools -- in spite of the name, they need to simplify the user
interface.
|
|
