( ESNUG 583 Item 4 ) ---------------------------------------------- [06/06/18]
Subject: SCOOP -- Mentor Veloce to announce first HW emulator on the Cloud
SCOOP! -- In his CDNlive keynote, CDNS SVP Tom Beckley accidentally
Leaked on stage that for its customers Cadence is moving its Verilog
simulation tools into the Amazon AWS cloud. "You'll see more at DAC!"
Clearly this announcement wasn't planned because his staff looked
very unhappy when I asked for details about this. (Oops.) - John
P.S. now the question is: "what are Synopsys and Mentor going to do?"
- John Cooley of DeepChip.com (05/02/2018)
From: [ John Cooley of DeepChip.com ]
So I get an anonymous tip from a good source that Mentor was not only doing
a cloud announcement for the upcoming DAC'18 -- which I thought would be all
about running Questa System Verilog or Calibre DRC SW in the cloud -- but
this tip was that Mentor was putting its Veloce2 HW emulator into the Amazon
AWS cloud!??!! ("WTF??!! How does that even work??!!")
I call Jean-Marie Brunet, the Mentor marketing director bigwig for Veloce.
(He's French and noticeably angry with me, so for full accuracy read his
direct quotes here with a grumpy French accent.)
After a few "colorful" Cooley/J-M exchanges, we *finally* get to:
Jean-Marie: "Yes, Mentor will have a press release of Veloce
being on the AWS Cloud on either Thursday or
Friday; depending on approvals."
Apparently, Siemens has been using Amazon AWS so deeply in its many other
(non-EDA) industrial projects, that AWS uses Siemens itself as a customer
case study on the Amazon.com website -- so doing something special like
putting Mentor Veloce2 boxes on the AWS cloud wasn't all that politically
difficult for them to do!
Jean-Marie: "This is real. We've been working on a pilot project
for 8 months on this with Amazon AWS. It works now."
Cooley: "Do you have an actual customer using it?"
Jean-Marie: "Yes, an IP company is using it."
Cooley: "Is that another way of saying ARM is using it?"
Jean-Marie: "I can not tell you who it is."
Then more colorful "exchanges" happen between J-M and I -- and J-M still
will NOT say who the IP customer is -- so I instead move on to the "how does
Veloce on the Cloud work?" questions.
Jean-Marie: "Putting HW like Veloce onto the cloud is not the same
as putting SW onto the cloud."
From the AWS environment the customer will select running Veloce from within
the AWS VPC (Virual Private Cloud). The customer's design db has to be
uploaded onto VPC. After that, using AWS Veloce is flexiable and dynamic.
"It's just yet another option for them in their VPC. You need to simulate
100 M gates for 3 months? You get 100 M gates for 3 months. You need
2 billion gates for 11 months, you get 2 billion gates for 11 months."
No special commands nor scripts for the cloud are needed. It's like you
have a fully loaded top-of-the-line Veloce2 HW box itself inside your own
company as far as the look and feel of AWS Veloce is. It's a transparent
HW Veloce user experience.
And physically, no, it's not a bunch of Veloce2 boxes sitting in the Mentor
Wilsonville HQ that customers access through the web. Instead it's a farm
of Veloce2 boxes that Amazon AWS itself hosts at the Amazon site that the
customer accesses through their own private AWS VPC environment.
Cooley: "Oh, cool. Got any pics of the site installation?"
Jean-Marie: "No, I can't give you a picture of the Veloce AWS
farm, John. But I can give you a generic pic of
one of our Veloce2 boxes if you want. Amazon has
very strict privacy requirements. Very strict."
Cooley: "Not a problem. I can work around it."
The DeepChip artist's interpretation of the super secret
Amazon Veloce2 AWS cloud farm at an unknown location
PRICING MODEL: Using Amazon AWS Veloce follows something like an Uber model.
"How many users for how many gates for how many months?"
2 users for 200 million gates for 3 months == 1200 M user-gate-months
4 users for 50 million gates for 6 months == 1200 M user-gate-months
1 users for 600 million gates for 2 months == 1200 M user-gate-months
1 users for 1200 million gates for 1 months == 1200 M user-gate-months
It automatically grabs as many Veloce CPUs your job needs. Also your run is
automatically partitioned across those CPUs. The user latency and bandwidth
doesn't change. It's the same as if you owned a Veloce box itself. (J-M
was intentionally vague on exactly how Veloce auto partitioning worked.
"That's secret. Just trust that it's fast, it's automatic, and it works.")
---- ---- ---- ---- ---- ---- ----
I asked Jean-Marie the obvious question that would concern anyone selling
HW boxes: "why do this? Doesn't this hurt your Veloce box sales?"
"No, it doesn't hurt Veloce sales. It's an additional option that
customers have been asking for. It's up to them to decide.
Pretending that the cloud is not there is both blind and stupid.
Instead we serve our customers."
A little over 4 weeks ago I reported in DeepChip that the Big 3 were all
working on putting their EDA SW into the cloud. But I must give Mentor
first prize bragging rights for getting the first EDA HW box into the
cloud! I didn't see it coming. Congrats and well done.
- John Cooley
DeepChip.com Holliston, MA
---- ---- ---- ---- ---- ---- ----
Related Articles
Jean-Marie warns Synopsys Zebu needs 9X more "warm bodies" to run
A surly Jean-Marie sasses Cadence Palladium and Synopsys EVE Zebu
SCOOP -- will the new MENT Veloce Crystal 3 chip crush Palladium?
The 14 metrics - plus their gotchas - used to select an emulator
Join
Index
Next->Item
|
