Synopsys IC Compiler vs. Magma Talus vs. Cadence Encounter GXL http://www.deepchip.com/items/snug07-08.html NOT 1/3, 1/3, 1/3 -- I used to believe that the digital backend P&R market was, in rough terms, divided equally between Cadence, Synopsys, and Magma. This is "used to"; because it's not any more from what I'm seeing. 5. Because this is a weird 1/3, 1/3, 1/3 niche, I have to ask that you please select the right question for your company below: a.) In the next 12 months as a Synopsys P&R user, for our production flow my project will most likely (choose) keep using PhysOpt/Astro: ########## 20% upgrade/keep using IC Compiler: ################### 37% switching away from Synopsys P&R: ### 5% b.) In the next 12 months as a Magma P&R user, for our production flow my project will most likely (choose) keep using old Blast Fusion: ##### 9% upgrade/keep using Talus: ######## 16% switching away from Magma P&R: # 2% c.) In the next 12 months as a Cadence P&R user, for our production flow my project will most likely (choose) keep old SoC Encounter/NanoRoute: ####### 14% upgrade/keep new Encounter L/XL/GXL: ########## 19% switching away from Cadence P&R: 0% What's surprizing here are the overall numbers. If you had 100 digital P&R users sitting in a room, this survey finds that they're: 2007 - using Synopsys P&R tools: ############################ 56% using Magma P&R tools: ############# 25% using Cadence P&R tools: ################ 31% Huh? That 56-25-31 ain't the 1/3, 1/3, 1/3 which I was expecting!!! "Relax, John, my data agrees with yours. I saw the same Synopsys first, Cadence second, Magma third ranking," said Gary calmly. "I think that this is from the reorg Synopsys did 2 years ago. They brought in some good talent from Cadence and they focused on what Synopsys does best: building best-of-breed tools." "I think Cadence is in trouble and they'd best be starting to introduce some new tools soon," Gary added. "What I'm seeing from them is that the their winning tools are coming from their RTL verification people, not the Cadence P&R people." "Magma had a tool hick-up with Talus, but from I hear they're focused on rebuilding their internal R&D group to fix this," said Gary. "For example, Patrick Groeneveld is back, who was the orginal architect of Blast Fusion. I've also heard other former R&D stars are coming back to Magma. That will whip them back into shape." Another trend to notice is that P&R customers aren't switching backend P&R tools as much. Here's the old 2005 users: 2005 - switching away from Synopsys: ########### 21% switching away from Magma: ## 3% switching away from Cadence: ########## 19% Now here's the new 2007 users: 2007 - switching away from Synopsys: ### 5% switching away from Magma: # 2% switching away from Cadence: 0% Suddenly the usual customer churn in backend sales has died off?!? WTF?! "Again, my data agrees with yours. We're getting power users in Asia and India for the first time," explained Gary. "And those guys are tending to go Synopsys and Magma. (People use Synopsys and Magma together a lot.)" "The mainstream users in the US and Europe are tending to die off," said Gary. "The power users make up 57% of the EDA sales while the mainstream users have dropped to 38% of sales. Cadence's strategy has been to sell to the mainstream, which hurts them now because the big money is now with the power users. It used to be with the mainstream." ---- ---- ---- ---- ---- ---- ---- Answer: keep using PhysOpt/Astro, we tried Magma - it was a complete bug infested disaster and nearly killed my last project. I'll stay with what I know works for a while "PhysOpt/Astro". - [ An Anon Engineer ] Our project will continue using Blast Fusion. Magma gave us a pitch about why we should switch to Talus. It was the same old EDA-vendor pitch: incrementally better timing/area on a small set of secret internal test cases, and incrementally faster run time some of the time on some cases. We felt it was off the mark for our needs. The improvements we want to see from Magma are mostly in the area of usability. We want to be able to get designs from kickoff to tapeout faster, and the execution time of the Mantle tool and it's timing/area are already good enough to not be the major barrier to taping out faster. Instead of saving us 5 minutes of run time, how about saving us the days at a time that our design engineers spend tearing their hair out figuring out which Mantle command to use, applying trial-and-error to fill in for the gaps in the documentation, and getting back cut-&-paste-from-the-manual answers from Magma support? - [ An Anon Engineer ] We tried upgrading to IC Compiler but went back to Astro due to bugs in IC Compiler. - [ An Anon Engineer ] X upgrade/keep using IC Compiler. Even with ALL THE BUGS! - [ An Anon Engineer ] I've always done netlist hand off. We are switching to COT and are now shopping for this tool. I like the video demos that I've seen on Magma. - Jared Bytheway of Cirque Corp. Definitely this---> upgrade/keep using IC Compiler. - [ An Anon Engineer ] Continue to use a mix of Astro and ICC - [ An Anon Engineer ] Will keep using new Encounter. - [ An Anon Engineer ] We are contracted with Synopsys Design Services for P&R. - [ An Anon Engineer ] We will keep using Blast Fusion for most of the blocks. We would upgrade to Talus on certain critical blocks. We did see Talus producing better routing (quality of routing) compared to Blast. I should rather say BLAST was pretty bad in certain routing cases. - Jay Pragasam of PLX Tech. Switching from Talus to ICC. Corporate policy. - [ An Anon Engineer ] We will be switching to Cadence (due to no fault of Magma's; their tool so far as been awesome) due to resource constraints (financial) - [ An Anon Engineer ] Don't use any of them - [ An Anon Engineer ] Keep using old SoC Encounter/NanoRoute. - Gautham Kamath of Cirrus Logic Keep using ICCy(with Astro for some non-critical blocks). After 12 months, looking at Magma. - [ An Anon Engineer ] Combination of continuing to use PhysOpt and switching to Magma - [ An Anon Engineer ] Dual flow supported; will likely use upgrade to both ICC & Talus. - [ An Anon Engineer ] Keep using Encounter XL - David Schwan of Sirenza Microdevices upgrade/keep using new Encounter L/XL/GXL. - Tom Mannos of Sandia National Laboratories We're only larger geometries, so using dc_shell to Astro. Doubt we'll move to IC Compiler for awhile. - [ An Anon Engineer ] Keep using Encounter GXL - but we will take a look at Mentor/Sierra. If better we may switch. - [ An Anon Engineer ] Astro, Jupiter, IC Compiler - Wolfgang Welser of Qimonda Keep using Astro for parts of the chip. Keep using Cadence for another part of the chip. - [ An Anon Engineer ] Upgrade from Astro in progress. Currently using IC Compiler. Used Blast Fusion because it was signoff kit proven. Used SoC Encounter w/NanoRoute. - [ An Anon Engineer ] Started project with IC Compiler, will keep using IC Compiler because already signed 3 yr contract. - [ An Anon Engineer ] ICC - John Stiles of Silicon Logic Engineering keep using old SoC Encounter/NanoRoute - Manfred Kaiser of Atmel keep using old SoC Encounter/NanoRoute. upgrade/keep using new Encounter L/XL/GXL. - [ An Anon Engineer ] Upgrade/keep using IC Compiler. If we had a license, then upgrade/keep using Talus. - [ An Anon Engineer ] Using Magma and Synopsys in-house, no decision to go solely with one or the other, but leaning towards Synopsys. - [ An Anon Engineer ] Using SNPS. Will likely switch to Magma. - [ An Anon Engineer ] --

※ 發信站: 批踢踢實業坊(ptt.cc) ◆ From: 220.138.132.40