( DVcon 05 Item 1 ) --------------------------------------------- [ 10/25/05 ]
Subject: Mindshare vs. Marketshare
Q: What is this "Verification Census" measuring?
A: This census is measuring what users actually *use* for verification
EDA tools. It's measuring the "mindshare" of each specific EDA tool.
Q: Isn't this the same as the "marketshare" reports from EDAC/Dataquest?
A: Nope. Marketshare numbers measure the dollars that EDA *vendors*
individually report as their sales for a specific EDA tool. For example
Dataquest FY 2002 ATPG Market (in $ Millions)
Synopsys TetraMAX : ############# $10.6 (45%)
Mentor FastScan : ############## $11.3 (48%)
others : ## $1.7 (7%)
Notice that marketshare numbers are always in sales dollars. Here
Synopsys reported to Dataquest that it had $10.6 million in TetraMAX
sales, Mentor reported $11.3 million in FastScan sales, and the others
reported $1.7 million in ATPG sales.
Marketshare percentages are always in terms of a specific tool's piece of
total sales pie. Marketshare percentages always add up to 100%.
All marketshare data comes directly from the few EDA *vendors*, not from
the hundreds of EDA *users*.
Below is an example of MINDshare numbers:
2004 - "What do you think of assertion languages and assertion
libraries such as SVA, PSL, OVL, 0-In CheckerWare?"
don't use : ############################################### 47%
IBM Sugar/PSL : ################################## 34%
0-In CheckerWare : ################## 18%
System Verilog SVA : ####### 7%
Synopsys Vera OVA : ####### 7%
Verplex OVL : ################# 17%
Mindshare numbers are always percentages. Mindshare does not measure
sales nor even the specific licenses engineers have. Mindshare only
measures what engineers report that they are *actually* using now. In
this example, if you had 100 engineers in a room and asked them what
assertions they're using: 47 would say "I don't use any assertions",
34 would say "PSL", 18 would say "0-in", 7 would say "SVA", etc...
All mindshare data comes directly from hundreds of EDA *users*, not from
the few EDA *vendors*.
Q: Hey! These "mindshare" numbers add up to more than 100%! What gives?
A: That's because because people often use more than one approach for the
same function. ("We use both PSL and OVL assertions.") This means
mindshare numbers almost always add up to *more* than 100%.
The one sure time mindshare numbers will always add up to 100% is when
you compare "use" and "don't use". In the assertions example, 47% of
engineers say they "don't use" assertions. This means 53% of engineers
do "use" assertions of some type. 47% + 53% will always equal that 100%.
One way you can calculate the multiplicity of use is by summing the "use"
subcategory numbers (34% + 18% + 7% + 7% + 17% = 83%) and divide it by
the actual "use" (53%) number. Thus 83%/53% = 1.57. This means the
average assertion user reported using 1.57 different types of assertions.
(Yea, I know there's no such thing as .57 of an assertion, but if you're
an engineer you know what I'm talking about here...)
Q: How were these Verification Census mindshare numbers gathered?
A: The verification census survey was sent out on March 7th and again on
March 30th, 2005 to the 22,000 member ESNUG mailing list. A total of
338 users responded back to the 11 complex question survey, generating
an average ~4 pages of comments/responses per person. This meant that
there were 4 x 338 = ~1,350 pages of detailed data I had to personally
comb through to tally the stats in this census. (Ugh!)
Q: Is this a scientific survey?
A: No, it's a survey of engineers. No scientists that I know of were
involved. Scientists don't use EDA tools.
Q: Isn't this ESNUG mailing list skewed towards Synopsys users?
A: No. At 22,000 subscribers, ESNUG reaches a general EDA user population.
I've done two studies just to check this.
"Out of the top 50 ESNUG Items of 2003, a surprizing 39 (78%) of them
were primarily non-Synopsys discussions."
- from http://www.deepchip.com/items/0422-07.html
"Out of the top 50 ESNUG Items of 2004, a surprizing 36 (72%) of them
were primarily non-Synopsys discussions."
- from http://www.deepchip.com/items/0437-02.html
Also there's that simple fact in EDA that's it's virtually impossible to
get Cadence/Mentor/Magma users who aren't also Synopsys users. Many in
the financial world don't know this, but talk to any chip designer and
it becomes blatently obvious.
Q: What does comparing the 2004 to 2005 mindshare results get you?
A: I was amazed at how many of the big questions remain unchanged over the
two surveys. It's eery when you see almost the *exact* same percentages
of "don't use" vs. "use" showing up across the two reports in the EC,
Assertions, System Verilog, Accelerator/Emulator, Bug Hunter, Specma/Vera,
Co-sim, and Verification IP sections.
That creepy consistency across 8 sections makes the story of where the two
reports are *different* even more interesting.
Q: What is the air-speed velocity of an unladen swallow?
A: What do you mean? An African or a European swallow? You have to know
these things when you're a king, you know.
Index
Next->Item
|
|
