( DAC 01 Item 28 ) --------------------------------------------- [ 7/31/01 ]
Subject: Magma "BlastFusion"
BETA CODE & NEEDING CASH: On the technical front, Magma's Blast tools only
do a gates-to-P&R-gates flow. It's *not* an RTL-to-GDSII tool yet, so users
must use DC in the Magma flow. All but one of the Magma users distrust
Blast's LVS/DRC capabilities. And BlastFusion also has the same hierarchy
troubles that PKS & PhysOpt have. (That is, BlastFusion only works on 100K
to 200K blocks. Stitching those blocks together is still somewhat of a
headache for users.) Also, if you take a look at my Magma tape-out count in
ESNUG 374 #7, you'll find that most Magma users are 'using' the Blast tools
in taxicab mode -- i.e. the Magma A.E.'s are the ones actually running Blast,
*not* the customers. Overall, Magma's Blast tools are still beta code. But
compared to its rivals, that's not as damning as it sounds.
Synopsys ########################################### 170 tape-outs
Silicon Perspec. ########### 43
Magma #### 13
Cadence PKS ### 12
Cadence non-PKS ###### 23
Mentor TeraPlace ## 7
Sapphire/Sequence # 6
Monterey 0
OK, so PhysOpt's kicking ass here. No news there. But, beyond that, Magma
is pretty much on par with most of its other competitors.
On the business front, Magma is one of the most heavily financed start-ups
in EDA history. Since its founding in April '97 to Dec 2000, Magma has
burned through $87 million for just $8.4 million in revenues. Earlier this
year, Magma had to lay off 20% of its workforce. And it's been widely noted
that Magma had been trying to get itself bought out by one of the Big Four
EDA vendors. (Cadence, Synopsys, and Avanti are all rumored to have said no
right off. Mentor's rumored to have said no twice -- which is interesting
because Mentor is seriously lacking in the physical synthesis department.
The sticking point I've heard is that Magma asking too much for a buyout.)
Magma gave killer DAC parties in prior years. This year, no Magma party.
""This seems like a premature IPO. Magma has been trying to sell
themselves for some time now to no avail."
- Jessica Kourakos of Goldman Sachs (EE Times 6/4/01)
""There's no question that Magma needs the money given their $3 million
a month burn rate and lack of revenues. What can I say? Magma has
an appetite for cash."
- Jay Vleeschhouser of Merrill Lynch (EE Times 6/4/01)
"I'd echo what you've already heard about Magma. I have also heard from
all the major EDA players that Magma was shopped heavily prior to this
filing and that all the majors turned them down. Two reasons cited were
failure to penetrate the market significantly in the past two years and
a product that had trouble managing large designs.
With $8.4 million in sales and this kind of burn rate, I'd say the
investors who buy this thing after the large funds flip risk getting
burned themselves."
- Jennifer Jordan of Wells Fargo Van Kasper (ESNUG 374 #1)
"We did our block level design using Magma (no "real tapeouts") that
were assembled them on the top level using Avanti for 2 reasons:
a) Magma didn't support hierarchical design which we needed due
to our configurable IP
b) Our current internal customer uses our internal plain vanilla
design flow and has to integrate our core in their system
and thus needs the Apollo Database anyway.
So we figure if we can integrate the Magma blocks into our Avanti
top level then our customer will have no problems integrating
this top level in their systems (just to be on the safe side.)"
- [ An Anon Engineer ]
"Magma and Monterey are the only comprehensive alternatives for 0.18 um
and below. We're concerned about Magma's financial viability. Silicon
Perspective is a useful stopgap measure to augment existing tools from
Avanti, Cadence, and Synopsys."
- [ An Anon Engineer ]
"People buy Magma because they can do 10% better than Silicon Ensemble,
or they can do larger blocks than PhysOpt."
- [ An Anon Engineer ]
"Early on, Magma had a unique story. But this year Synopsys and Cadence
have the same story. I must have passed Magma's booth on the floor two
dozen times. Every time the booth was empty. By the way, I passed the
booth two dozen times without stopping."
- Richard Lowry of Starburst Technologies
"Your Magma desperados column was pretty eye opening since we're building
a flow around Blast Fusion here at TI ASIC. And yeah, we're only doing
blocks with it right now."
- Larry Buckman, Texas Instruments ASICs (ESNUG 374 #1)
"I visited the Magma booth and got an overview on their 'flow_tool',
for lack of a better term. Good intentions; but as a user, I'm
uncomfortable with the notion of getting into a methodology that did
not offer a 'way-out': specifically, one that lacked 'hooks' to get
into non-Magma tools. Being a startup company, we didn't have the
wherewithal or the schedule to allow an extended eval, which is what
we'd have needed to gain comfort in their new flow.
The most common usage model for BlastFusion (from friends who are
using it), seems to be one that replaces Apollo (and Saturn) or
SE (and PKS) for P&R. In our case, we weren't comfortable with
(a) having to retrain the team on an entirely new toolset (we have
major deadlines in the not-too-distant-future), and (b) our current
flow is getting us good results, except in the area of design planning.
Seeing that Magma had no 'stand-alone' design planning solution, we
decided to try out Silicon Perspectives instead. We're also using
PhysOpt, which seems to be doing a decent job."
- [ An Anon Engineer ]
"Since I work at Vitesse, I found your recent comments about Magma
interesting. FWIW, it's not all taxicab mode here; I ran Magma a fair
bit earlier this year and never talked to a Magma AE."
- Natalie Overstreet Ramsey, Vitesse Semiconductor
"A number of ASIC manufacturers seem to be impressed with Magma. Most
are supporting Synopsys PhysOpt.
- Carl Wakeland, Creative Advanced Technology Center
"Magma: integrated RTL-to-GDSII flow
-----
1. Hasn't had a lot of good press lately; ESNUG and analysts have been
questioning the capacity and performance of their tools, but
recently some users have come forward with success stories.
2. RTL synthesis front-end (BlastChip) is brand new
- uses a "supercell" concept for fast front-end synthesis
(logic structure), before considering gate-sizing.
- supports arithmetic/datapath synthesis and resource sharing
3. BlastFusion: block building, claim 36 tapeouts
- Will do long-wire buffering, critical path logic restructuring,
gate resizing.
- Clock Tree Synthesis: hierarchical, balanced H tree, supports
gated clocks & "useful skew" concept between hierarchies/trees,
but not within blocks (flop-to-flop).
- Signal Integrity-driven routing, variable width & spacing
- Table-based RC extraction, tables from built-in Magma 3D
field solver
- built-in DRC & LVS, ECO place & route capabilities
- No Scan/DFT capabilities built into the flow -- how do we do
scan insertion and placement-based reordering??? Need to
investigate.
- Supports min/max corner timing analysis
Sun Solaris now, Linux 'soon'."
- Kris Monsen of Mobilygen Corp.
"Monterey Dolphin and Magma BlastFusion are really targeting the
gates-to-placed gates flow (or gates-to-GDSII). I'm interested in
RTL-to-placed gates, so I'm not tracking them."
- [ An Anon Engineer ]
"Magma's pitch sounds great. It seems to be a complete solution. Will
they survive though? We are looking at Magma, but also a DC / Silicon
Perspective flow."
- [ An Anon Engineer ]
|
|
