Oracle Loses Appeal Against $3 Billion Payment To HPE Over Withdrawal of Itanium Support (theregister.com) 47
The Supreme Court of California has thrown out Oracle's appeal against a decision to award $3 billion damages to HPE in a case which dates back a decade and relates to Big Red's commitment to develop on Itanium hardware. From a report:On Wednesday, the court denied a review of Oracle's appeal against a summary judgement, apparently without comment or any written dissents. The decision follows a ruling made in the California Court of Appeal that affirmed HPE's $3.14bn win for alleged contract violation, stating that an agreement between the firms had created a legal obligation for Oracle to support software on HPE's Itanium server. The case hinged on the companies' statements that they had a "longstanding strategic relationship" and a "mutual desire to continue to support their mutual customers." The agreement stated that Oracle, for its part, "will continue to offer its product suite on HP platforms" while HPE "will continue to support Oracle products (including Oracle Enterprise Linux and Oracle VM) on its hardware." The ruling reads: "We conclude that the second sentence, moreover, does more than declare an aspiration or intent to continue working together, as Oracle claims. It commits the parties to continue the actions specified (Oracle offering its product suite and HP supporting the products)," as it had done previously.
Itanium (Score:4, Interesting)
I always wanted an Itanium box but even now they command ridiculous prices. Who can even say they saw one in a production setting? I just want to have an OpenVMS cluster running across three architectures. VAX, Alpha, and Itanium.
Re:Itanium (Score:5, Insightful)
I remember back when the HP and Oracle people were all about the Itanium boxes, but all I ever wanted to hear about was commodity AMD servers
Oracle simply listened to their customers while HP tried to shove a broken chip architecture down the throats of customers who had been forced off of the Alpha chip line
Sad but true, HP decided to become a toner reseller, while customers wanted to deal with a hardware manufacturer
Re: (Score:2)
Blame's on both HP and Compaq for that disaster.
Re: (Score:2)
Why? HP co-developed Itanium with Intel. Compaq did not. Also, why blame Compaq for alleged failures of HP?
Any discussion of Itanium brings out morons.
Re: (Score:2)
Compaq owned DEC, then buried the Alpha by selling the IP to Intel.
Re: Itanium (Score:2)
Re: (Score:3)
Re: (Score:2)
Upside:
Alpha had a 30 year roadmap
Alpha used an advanced memory bus that Intel could not match for about a decade
Alpha had better multithreading
Alpha had better integer and float (not to mention encryption) performance than Pentium
DEC had a viable lawsuit against Intel for IP theft
Downside:
DEC couldn't market their way out of a wet paper sack
DEC relied on MSo$t for NT version (btw, WNT was increment to VMS, with same architect) that was abandoned by M$oft
Alpha required re-compile and frequently re-coding to
Re: (Score:1)
"...customers who had been forced off of the Alpha chip line"
LOL who were they?
"Sad but true, HP decided to become a toner reseller, while customers wanted to deal with a hardware manufacturer"
Capitalizing on the opportunity to sound even more ignorant?
Re: (Score:2)
You can see the idea behind Itanium wasn't terrible. Modern x86 treats machine code as an intermediate language, compiles and optimizes it on the fly down to microcode. You could potentially get rid of a lot of silicon (and heat/die space) by having the compiler do all the work up-front.
Problem is it turns out that there is no substitute for run-time optimization and hyperthreading.
Re: Itanium (Score:2)
The last batch of Itanium CPUs was shipped at the end of July.
Itanium was a huge win for Intel (Score:1)
As Executive VP at HP, his main accomplishment was killing HPUX and PA-RISC in favor of WinNT-on-Itanium (when Windows NT for Itanium was little more than a pre-announcement press release).
He then went to SGI as president where his main accomplishment was killing IRIX and 64-bit-MIPS in favor of WinNT-on-Itanium (before WinNT-on-Itanium ev
Re: (Score:2)
Re: (Score:2)
A Raspberry Pi 4 can emulate a VAX CPU faster than any VAX CPU that Digital ever produced. There was never a VAX machine with 8GB of RAM, and well a microSD card can hold more than any VAX ever had. They never had a 1Gbps network connection either.
I would imagine the hardest bit of a VAX cluster on a Pi would be doing the shared storage right.
Re: (Score:2)
Re: (Score:1)
I had to set up several clients at last job with some Itanium boxes (rx8620 and 6600), but they're very power hungry, and with Red Hat they were slow running single threaded code. Did very well with large sys5 shared memory accessed by multiple processes like databases but unless that's what you have in mind I'd really question whether you'd be happy with them or the electric bill.
Re: (Score:2)
I've actually supported quite a few Itanium Superdomes in production in at least one shop. But that was around ten years ago, and they were getting phased out in favor of Linux boxes wherever possible.
Can't say that I've seen them anywhere else since.
Re: (Score:2)
On odd thing that I did not understand immediately, was they they could run PA-RISC binaries. AND at reasonable speeds. Meaning at one point we did not have the Itanium binaries for Netbackup client. So, my team lead said, "just use the latest PA-RISC one". And it worked. Backup times did NOT suffer. Perhaps it might have been a little slower. Nothing we really worried about at the time.
Of course, whenc
Re: (Score:2)
I managed a gaggle of Itanium Superdomes, RX servers and of course plenty of storage to go with it all.
Montecito, Montvale and McKinley, oh my ...
Was good stuff for its day - kicked the Sun E15k's ass when it came to I/O throughput. (But I still love Sun with all my heart.... even still touch modern SPARC M8's on a regular basis.)
Re: (Score:2)
Too bad, Oracle... (Score:5, Funny)
Couldn't have happened to a nicer company.
And to HPE: You lie down with dogs...
Re: (Score:2)
Couldn't have happened to a nicer company.
Agreed.
And to HPE: You lie down with dogs...
Oracle, Itanium...both are dogs.
Re: (Score:3)
Oracle losing a contract case...
ooh the irony
Itanium (Score:1)
Did anyone sue Intel for creating such a bad architecture to cause all these issues in the first place?
Re: (Score:3)
Thank you for pointing out my hubris. You have changed my path in life.
Re: (Score:3)
The people who were most upset by the Itanium shenanigans were ALPHA customers who saw a superior architecture shit-canned in favor of... well crap
It just so happened that AMD-x86-64 came along and ate the itanium market
Re: Itanium (Score:2)
In retrospect, HP should have migrated all of its customers and software to Alpha architecture instead of inflicting Itanic on the industry.
Re: (Score:2)
Yes, instead of creating a new architecture from scratch, focus on the best of the ones they collectively already had. That would have achieved the stated goal of standardising and reducing development costs, while also giving them a head start since Alpha was already a functional platform with hardware available, existing compilers and operating systems etc.
MIPS made a similar mistake. they went after the embedded market after losing their highend customer SGI. The transition to 64bit in the embedded space
Re:Itanium (Score:5, Informative)
Did anyone sue Intel for creating such a bad architecture to cause all these issues in the first place?
Itanium wasn't a *bad* architecture, especially for the time. It just had a lot of things against it. It was 64-bit, but it required a whole lot of recompiling with very specific code optimized for it, and it wouldn't run most x86 software. This meant that only a handful of things ran on it, and fewer optimized for it. Now, when that thing was OracleDB, and Oracle did actually optimize for the architecture, there was a notable performance bump...but it wasn't something that was widespread, making Itanium a bit of a niche.
Then AMD made x86_64 that was a drop-in replacement for x86, and multicore processors pretty much ended any performance improvements Itanium provided.
Like virtually all technology, it gets decommissioned on an asymptote. There aren't many Itanium based Oracle servers out there, but like my old job that was still running Oracle on Sun SPARC, it's all but certain that there are at least a handful still in production use.
Re:Itanium (Score:4, Insightful)
No, it was bad. Even Intel couldn't come up with a compiler that would make it out-perform AMD's x86_64 or even x86_32. It was also fantastically expensive and Intel refused to lower the price to match it's actual value. It certainly never out-performed or even matched the other 64 bit processors that were shoved into the dustbin to make room for it. Intel marketed and strong-armed a bunch of people into scrapping their Porsche to make room in the garage for the new Tercel.
Re: (Score:2)
Intel had form for this with the i860 too. Performance as a general purpose processor was poor largely due to compiler technology, but with properly optimized code the i860 actually performed very well.
The i860 ended up being used for devices where it would be running specific hand tuned code rather than user supplied code, being used as geometry calculators on some highend graphics boards for instance.
Re: (Score:2)
That's actually a great mystery to me. They KNEW that getting the compiler to produce good code for Itanic would be difficult to impossible because they tried and failed with the i860, but they went full steam ahead anyway, right into the wall.
Given the HUGE success of the 80x86 line, it's easy to forget that Intel has had a few big fails in their day. The 8086 was supposed to be the I/O co-processor for iAPX 432 systems, but the iAPX 432 was dog slow and their internal teams noticed it was faster to do com
Re: (Score:2)
No, Itanium made with assumptions about compiler optimization that couldn't be realized in the real world.
Donald Knuth: "The Itanium approach was supposed to be so terrific until it turned out that the wished-for compilers were basically impossible to write."
It also was a performance dog and its vaunted x86 compatbility mode was as slow as Pentium I.
Customers bought it as only way to run VMS or HP/UX with a future, but that meant its market was tiny.
The thing was a farce. Look at the
Re: (Score:1)
Correct answer. Itanium was predicated on the existence of sophisticated compiler technology that never developed.
Re: (Score:2)
Itanium was built on the notion that x86_32 would become unusable as demand for memory forced a migration to 64bit across the board. Intel assumed that users would have no choice but to migrate to Itanium irrespective of how badly it performed. They didn't count on AMD coming up with a 64bit extension to x86.
It's quite telling that the first generation Itanium developed by Intel performed extremely poorly, while the second generation developed by HP was a big improvement.
What few Itanium customers there wer
Re: (Score:1)
Yea well, the fact that Intel didn't make the Itaniums also capable of running 32-bit code was purely a strategic blunder based on their unfettered greed and decades of escaping direct prosecution for their low-effort lies. The market didn't buy their excuses, and neither did AMD.
Re: (Score:1)
Honestly they should be sued by the entire planet for that or EOLing the Itanium hardware. The fact I was able to predict this entire scenario the moment the creation of the Itanium product line was initially announced is all the proof I need this entire fiasco was orchestrated to cause maximum financial and logistical damage to all players, exactly in this fashion. If I could see it coming then the people in the inner circle sure as fuck did too.
Re:The binds that tie. (Score:4, Informative)
interesting read on this:
https://www.theregister.com/20... [theregister.com]
Upgrading as suppport: OFFTOPIC? ya rt. (Score:2)
To mark a comment about premature and toxic upgrade and support behaviors as OffTopic?: How quickly manufacturers can drop and shut out support for products seems on topic. Maybe requirements to open source components specs including SW requirements & interfaces that don't allow user diagnosis, repair and upgrade would make the applicability of the topic more clear? Doesn't fit in a '/.' subject field.
Seems especially apropos for closed-source or proprietary HW & SW.
...affirmed HPE's $3.14bn win (Score:2)
3.14? Nicely done to whoever made the fine be pi.
Lifetime sales? (Score:1)
Isn't that like 10 times the entire lifetime sales of itanium?
Was Mark Hurd worth $3.1 Billion? (Score:3)
What neither the article nor the linked story mention is that it all started when Mark Hurd, CEO at HP, left the company and was immediately hired by Oracle. HP sued regarding some non-compete agreement, trade secrets, the usual. But then HP and Oracle settled, with the settlement agreement including the language:
"Oracle will continue to offer its product suite on HP platforms ... in a manner consistent with that partnership as it existed prior to Oracle’s hiring of Hurd."
It is that language (combined with fluff around it) that is at the core of the $3.1 billion bill it now must pay. So tell us, Larry Ellison - did Mark Hurd bring you more than $3.1 billion in value?