Sun Microsystems has canceled its lower-end UltraSparc IIIi+ ‘Serrano’ processor, choosing instead to focus its priorities on alternatives including its more radical ‘Niagara’ chip family. Sun had planned to introduce the UltraSparc IIIi+ chips in three servers, the V215, V245 and V445, models that likely will be announced at a Sun event Sept. 13. But the company chose to scrap the chip, John Fowler, Sun’s executive vice president of systems, said in an interview Thursday.
Looks like Sun has finally got its house in order; purging the SPARC loyalists from the ranks in favour of a more pragmatic approach to using UltraSPARC where it makes sense.
When you’ve got Opteron machines that deliver better performance, out of the box, than the above configurations (UltraSPARC IIIi+), one really has to ask, does it make sense spending millions breathing life into a chip that just isn’t a viable competitor.
Opteron is already being provided by Sun in configurations up to 8 way (16 cores) that beat 12 way (24 core) SPARC monsters, it jsut really didn’t make sense putting good money after bad.
Oh, and just as a side note, Sun as quietly upgraded their worstation like to the new M2 Opterons, which should mean for customers, a longer life in regards to being able to upgrade the processor and so forth.
Opteron is already being provided by Sun in configurations up to 8 way (16 cores) that beat 12 way (24 core) SPARC monsters, it jsut really didn’t make sense putting good money after bad.
At somethings yes. Other things no. I suppose you should refer to a specific benchmark or application before you jump the gun. You buy the platform that best suits your needs rather than buy it because you are a fanboy or you like the colour of the lights.
The UltraSPARC IIIi+ makes less sense now due to intel and AMD following Sun multicore lead. With Niagara, Rock, APL, and Ultra IV+, there are plenty of options to cover Sun’s customers needs into the future.
We use both V210’s and V240’s at work and they perform quite well as small database and backup servers.
It only makes sense not to produce the CPU if there is only a slight increase in performance and increased cost that didn’t justify the performance gain. And considering the performance of the T1, I’m glad that Sun is focusing efforts on Niagara, Niagara 2 and the UltraSparc IV+, and not trying to extend the life of low-end CPU’s.
The US IIIi+ was getting covered up with the big Opteron servers, both price and performance — this makes a lot of sense to me. I would think that getting the Niagara II and III to market sooner would be a far better use of resources. As long as the US line stays above multi-chip Opteron performance, works for me.
Finally, Sun really seems to be doing things intelligently. The USPARC line has sucked for some time. The new IV+ chips might be better, but they still don’t “wow” me or anybody I’ve worked with. CERTAINLY not for the price.
That being said, the Opteron gear has been pretty awesome, and the direction they are going with Niagra is great too. The performance of Niagra (the current revision) hasn’t astounded me – BUT – it shows what will come as they increase the capacity of each core (or add more of them…) It’s a kick-butt proof-of-concept.
I hope Sun keeps doing things right, my stock might actually be WORTH something in the future. Not to mention I won’t be paying 50k for a server that a 5k home-built could outperform. (Obviously Sun wins in the reliability/design department!!) I wouldn’t mind paying 10k for a Sun server that is equiv. in performance to a 5k home-built machine. 2x the cost is worth it for what you get in exchange.
I just wonder which processor will power the next SPARC based workstation? Certainly not Niagara or APL. So the only choise is UltraIV+, which seems to be quite expensive. There is still lot of software, which demands SPARC processor and is not compiled for Solaris x86. Transitive announced a transition software, which allows SPARC binaries to run on x86 processors, but this program isn’t out yet and is not proven. Only if the transition is as smoothly as currently at Apple, then Opteron based workstations will be accepted.
Anton
We need cheap sparcs for our desktops also. SUN must start producing a mass market chipset for all the SPARC fans. With mini-itx form factor making leaps they should have a competitive solution in the low end desktop also.
That wouldn’t fit with Sun’s model. They would prefer you buy an expensive server, and a bunch of thin clients. The key here is – Sun needs to come out with inexpensive thin clients.
It would be a no-brainer to buy a 50k server and 100 200$ thin clients, if it would work well. That would be a HUGE savings in the end, in terms of reduced administrative costs, maintenence, etc. I really hope Sun is listening!
That wouldn’t fit with Sun’s model. They would prefer you buy an expensive server, and a bunch of thin clients. The key here is – Sun needs to come out with inexpensive thin clients.
Sun also love selling workstations, though Sun Ray’s do rock!
Sun has always persuaded this model. If you remember some 8 years ago, the thin clients idea was quite popular in this time, but obviously was too early for that. It is not rare that good ideas don’t find their future because they were born just too early or being killed by a larger market drivers (MS and IBM at that time).
You can see what I mean if you watch this video
http://video.google.com/videoplay?docid=-1763081921280236634&q=scot…
where Charlie Rose interviews Scott McNealy and he explains the vision of thin clients, though in the light of interoperability and networked environment.
The Opteron is a nice chip, but you can’t replace every SPARC solution with Opterons.
In the midrange to highend computing under Solaris, SPARC is the only solution. Maybe this will change if AMD provides 16+ way setups.
Sun could have just used the other archetectures they’ve alreadyt been doing and are focusing on.. since manufacturing isnt perfect im positive some chips will only have one working core, etc.. etc.. and those would be more sensible lower end chips for lower end sparc servers as its cost effective, the chips work great and it just makes sense.