Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
You can run them quite nicely and very fast. The major EDA tools for Opteron (64-bit) have started arriving on Redhat Linux. The tools will continue to arrive right through 2004-2006 which by then probably all the tools would have been ported.
Is anyone running 32 bit executables (say design compiler) under a 64 bit linux release?
My understanding is that they should work if you have 32 bit linux libs installed on the machine, but I'd like to have this confirmed (I'm thinking on building a 64 bit system, but would still have to run some 32 bit tools).
They are supporting 64-bit, calibre and modelsim have just been released for AMD64 LINUX. Modelsim is a very big one IMO, it is the standard and the best for HDL simulation etc.. Once the big packages are ported everything else will follow.
And a QUAD Opteron with 32GB-64GB of RAM as a workstation is bloody good for me, AKA the Appro 4145H.
Even if on AMD64 there is a 32bit execution Unit, to take full advantge of this CPU (i.e. Superscalar Architecture) you need to run binaries compiled for this machine.
I'm really curious to know if there is a speed improvement in comparison with a 32 bit architecture like a P4.
I know that there is a resource gain in terms of memory, but if I use DC with a P4@3GHz and an Opteron@2GHz (assuming that there is enough memory to manage the design), which is the faster architecture?
Please report as many benchmarks as you can, with different applications.
For LINUX 64-bit there is REDHAT WS 3 for upto 2 CPUS, and also the AS 3 for 2 CPU +. Both have natively compiled versions for Opteron. For comparison to the P4 in 32-bit designs Id say the Opteron is better as it has higher FPU performance and shorter pipeline. As for 64-bit performance between nocona and opteron the difference in performance will be greater as the nocona is IDENTICAL to xeon with only 64-bit extensions, for the Opteron NUMA and the low latency, high BW memory architechture will kick in for designs bigger than 4GB. I shouls also note the EMT technique intel is using in the nocona is flawed for accessing the 4GB+ address space, REDHAT allready knows about this and has solved the issue but at a heavy performance penalty in memory performance.
IMO if your looking for a 64-bit workstation, INTEL is out of the game. Mentor has allready released 64-bit version of MODELSIM and CALIBRE and did not even mention or make references to INTELS NOCONA.
We are using single Opteron Linux boxes for all our design workstations, and multiple CPUs for simulatin farm. All 32-bit EDA applications run like charm. For some 64-bit ready applications, they are runnning great.