( ESNUG 416 Item 3 ) -------------------------------------------- [07/30/03]
From: Maynard Hammond <maynard.hammond=user domain=sciatl spot calm>
Subject: Why Won't Synopsys Support 64-bit Linux On The AMD Opteron Boxes?
Hi, John,
On our last chip we required 64-bit Solaris machines with 12 Gig of memory
to do our static timing analysis (PrimeTime). Several of our physical
blocks required more than 4 Gig of memory to compile in PhysOpt. Our new
chip is almost double the size. I believe we will need boxes that support
memory configurations over 16 Gig to build our next chip. We have also
migrated to Linux. Our sever farms are predominately Linux now and we would
like to buy a Linux solution.
As I look at the cheaper 64-bit Linux machines we can buy to build this next
chip, I find I can get Intel Itanium 2 or AMD Opteron platforms. If I look
at 8 Gig configurations (probably OK for PhysOpt, but still not enough for
PrimeTime) I find:
1 - Dual AMD Opteron 244 CPUs,
8 Gigs of PC2700 ECC DDR,
40 GB EIDE,
CD,
Dual nics
Price: ~$6,500
above with 16 Gigs of PC2700 memory
becomes ~$16,500
2- Dual Itanium 2 1.4 GHz CPUs,
4 Meg L3 Cache,
8 Gig memory
Price: ~$17,000
above with 6 Meg L3 cache, 16 Gig PC2100 memory
becomes ~$30,000
In eWeek, they state that for SPECint tests:
score
2 GHz AMD Opteron -------
w/ PC2700 memory 1,202
1 GHz Intel Itanium 2 810
2.8 GHz Intel Xeon (32-bit only) 957
The Itanium 2 did better on floating point (SPECfp)
Intel Itanium 2 1,370
AMD Opteron 1,170
Intel Xeon (32-bit) 887
It's hard for me to compare as I don't have either machine now. From cost
and SPECint performance I would like to buy AMD Opteron servers. Even
though the Itanium 2 has better SPECfp performace, I believe I would prefer
two AMD Opteron machines to the one Intel Itanium machine I could buy for
roughly the same amount of money. I don't know what the difference in cache
sizes and other differences between processors buys me. I'm sure this is
not an apples-to-apples comparison. However, until the EDA community ports
to both, I don't know how I'll know for sure.
I will also mention that I have been able to get quotes for AMD Opteron
machines with delivery times within the month. For the Intel Itanium 2
servers, I have been able to get quotes, but NO ONE has been able to give me
a delivery schedule. I know the Intel Itanium 2 severs exist. I saw them
at DAC. I just don't know how I would get one in my hands.
What frustrates me, is that the EDA community in general is porting
everything to the Intel Itanium 2 machines but not to the AMD Opteron. This
doesn't make any sense to me. Mentor has officially announced that they
have Calibre running on the Opteron. I have asked Synopsys when I can get
an AMD Opteron port for PrimeTime and PhysOpt. (They do support VCS if you
ask the right person.) They tell me the AMD Opteron is not officially being
supported at this time.
Does anyone else have the same needs and concerns as I do? Does someone
have better analysis that would make me want to buy and Intel Itanium 2
solution? Has someone who has run the same application on both platforms
have any results they could share? Has someone other than HP been able to
get an Intel Itanium 2 solution? I would appreciate any help or comments.
- Maynard Hammond
Scientific-Atlanta, Inc Lawrenceville, GA
|
|
