( ESNUG 546 Item 5 ) -------------------------------------------- [01/15/15]
Subject: Spectre XPS out too late for 16 nm memories; too early for 10 nm
>
> "Do you have any good user Spectre XPS benchmarks, John?"
>
From: [ An American Werewolf in London ]
Hi, John,
Saw your reader request for a user Spectre XPS benchmark.
It's too early for those.
Spectre XPS is Cadence's new fast SPICE for memory. It's only 9 months old.
It's to replace their failed UltraSim which Cadence got from Celestry back
in 2003. UltraSim went nowhere even though it ran 4X of Spectre. UltraSim
lost too many benchmarks against HSIM back then.
 
Our group does embedded memory. We use mostly HSIM and FineSim. For better
prices, our CEO likes the idea of a SPICE second source but at present XPS
is too new with respect to our design cycle to benchmark. Magma used to be
our SPICE second source -- but after FineSim beat HSIM in benchmarks, Aart
bought LAVA to put a stop to that.
The Cadence rep claims Spectre XPS is 3X faster than HSIM and it silicon
correlates better than Synopsys because XPS supposedly has better newer
convergence algorithms. The rep also claims Spectre XPS is 100% Spectre
compatible -- meaning no netlist edits (like no tweaks of .MEASURE needed)
nor special syntaxes needed.
 
Everyone is finishing up their 16 nm memories. In a month or two, when we
move to 10 nm, then we'll benchmark XPS, BDA AFS Mega, and HSIM/FineSim.
We've heard Silvaco is making moves in fast memory SPICE, so we might
consider them, too. After that, people will have Spectre XPS benchmarks.
But currently it is too early.
- [ An American Werewolf in London ]
---- ---- ---- ---- ---- ---- ----
Related Articles:
User benchmarks BDA AFS Mega vs. Finesim Pro and HSIM for SRAMs
User dirt on Synopsys/Magma SPICE, BDA, Spectre, Solido, MunEDA
Reader details how SNPS-LAVA dramatically raises SPICE prices
Join
Index
Next->Item
|
|
