"With the Solaris team gutted, it looks like the Sun skeleton has finally been picked clean."
https://meshedinsights.com/2017/09/03/oracle-finally-killed-sun/
18 replies (most recent on top)
Then why are you involved in this discussion, if your care index is at 0 Kelvin, as you say
I told you why: because I read your "HAEAR ARE SOME 2010 BENCHMARKSES" comment and thought your insistence that we should all care deeply about that level of nerd archaeology was cute.
What I said I do not care about -- and I don't -- is digging deeper to find out exactly how derpy your fixation on SPECint is.
See this post here? This is me definitely caring about your getting all serious about this. "I DEMAND everyone tell me why the emperor has no clothes, or doesn't publish SPECint! Excelsior! By the sword of Linux and my rms tat!" Purely for the self-righteous hilarity vaiue.
This is also me not caring about SPARC benchmarks. That ship sailed long before Friday. There is no contradiction here.
Carry on, I can make more popcorn.
And as predicted, falling back on per-core performance comparisons, just shows your IBM colors. I wish you had stayed on your El Register rants instead of trying to sell your bull at a site where people write about layoffs. Real people with families instead of Stahl-type marketing bull.
While running a single thing on a single thread is proof of anything, especially today. Come on...
M7 1-socket (2015): https://spec.org/cpu2006/results/res2015q4/cpu2006-20151026-37722.html
That's not SPECint. Thiat's SPECint_rate. Very different benchmark.
It's also meaningless. Running 224 concurrent instances of the same program tells you absolutely nothing about a CPU's per-core integer performance.
Disingenuous? Not that I am surprised in the least.
Whether or not you like the SPEC benchmark is completely besides the point. It's the benchmark that everyone quotes in their marketing fluff.
You're right on one point though: Intel still outperforms SPARC, today.
I appreciate your default fallback setting of insulting those who don't subscribe to the SPARC Kool-Aid marketing BS. If the only thing you've got left is personal insults, I have nothing more to add.
Enjoy your imaginary M8 SPECint numbers.
Not the same guy as 1yfz.
You are dead wrong and the fact that you can't do a simple spec.org search to find newer results says you are indeed stupid, as shown by your use of specint for any practical business workload comparison.
I really don't care.
Then why are you involved in this discussion, if your care index is at 0 Kelvin, as you say?
And no, Oracle hasn't published any Solaris SPARC SPECint numbers more recent than 2011. I checked. SPEC.org is searchable. And the one SPARC SPECint result from 2011 is worse than the M8000 one.
This isn't about working someone up in a tizzy or whatever you want to call it. It's too late for tizzies. Just as your care index is 0K, Solaris SPARC customers' care index is just about the same.
This is about reality. Something Solaris SPARC fanbois always have had problems with.
See the HBO show "Silicon Valley" and the character Keenan Feldspar. "Why do you make me feel bad? I don't like feeling bad! I like to feel good!"
Let's play your stupid game.
M7 1-socket (2015): http://spec.org/cpu2006/results/res2015q4/cpu2006-20151026-37722.html
Intel 1-socket (2016): http://spec.org/cpu2006/results/res2016q2/cpu2006-20160419-40510.html
Intel 1-socket (2017) http://spec.org/cpu2006/results/res2017q3/cpu2006-20170627-47383.html
So Intel finally managed to out-perform a 2 year old processor, just as M8 is being announced, leap-frogging ahead again.
Now feel free to go on around performance per-core instead of per-chip, which shows actual technical leadership. Then I'll be sure you're the usual El Register comment section f---tard.
PS. You do know that SPECint2006 is a sh-- benchmark anyway right? Hope you don't work for one of my customers, as they would be so unlucky to have a zombie who can't differentiate between sh-- and brains.
And I'll answer: I really don't care.
I just think it's funny you're trying to whomp up a tizzy by quoting 7 year old numbers.
Again, my caring index is right about 0 Kelvin, but I would not be surprised to learn that Oracle has published SPARC benchmarks newer than 2010, and you're having an endorphin rush simulating a righteous lather because SPECint is the one they haven't published.
Back when my caring index was measurable, I know that Sun didn't publish TPC-C, because it was a crap benchmark. And there was similar foam from their competitors. SPECint, imma gonna guess, is way far from real world usage, too.
Checks calendar
Show us Solaris SPARC SPECint results more recent than 2011. Do you have an URL at SPEC.org? If you do, please post it here.
I'll ask again: how come Solaris SPARC stopped publishing SPECint results after 2011? It can't be the $500 fee for submitting a SPECint result. I'm sure Oracle can afford $500.
This is Solaris SPARC's state-of-the-art in 2010:
Solaris SPARC system: SPARC Enterprise M8000, SPARC64 VII+, 64 cores, 16 chips, 4 cores/chip, 2 threads/core, 3.0 GHz. SPEC submission date: October 2010. Running Solaris 10 09/10
https://www.spec.org/cpu2006/results/res2010q4/cpu2006-20101206-13888.html
This is a ho-hum 2-socket ASUS Intel Xeon server from 2010:
Intel Xeon system: ASUS TS500-E6 (Z8NA-D6) server system, Intel Xeon X5670, 12 cores, 2 chips, 6 cores/chip, 2 threads/core, 2.9GHz. SPEC submission date: July 2010. Running SUSE Linux Enterprise Server 11:
https://www.spec.org/cpu2006/results/res2010q3/cpu2006-20100730-12706.html
There are two columns in the SPEC Results Table: Seconds and Ratio.
Seconds is the clock time taken by each benchmark to run.
Ratio is the placement of each benchmark on the totem pole of benchmark speeds.
Seconds: the lower, the better.
Ratio: the higher, the better.
A Solaris SPARC Enterprise M8000 with 16 CPU's and 64 cores is 3 times slower - on average - than an Intel Xeon with 2 CPU's and 12 cores. Both benchmarks are from 2010.
Given these numbers, how can Solaris SPARC compete with Linux on Intel?
How come Solaris SPARC stopped publishing SPECint numbers after 2011?
It would have gone down hard or been acquired by corporate raiders...
I wonder how come there was no corporate raider this time to acquire SPARC. Or at least some activist investor who could have forced Oracle to attempt to sell. Even a couple of hundred million dollars would have been OK. Anything would be OK when compared to spending a couple of hundred million on "restructuring costs"...
So essentially what 1lyg is saying is that Oracle failed. There were still plenty of deployments in 2009 and 2010. The server and workstation lines had not yet been cancelled. It was ORACLE'S decision to turn Solaris into some sort of cloud OS, and tune it up for their new product line and cancel the mid range products. Saying that it was Sun that failed when Oracle had the reins for 5 years is like blaming Obama for not being in NoLo during Hurricane Katrina when he wasn't even President yet.
There was still plenty of goodwill for Sun and their products prior to the Larry era. Once the customers heard that Larry would be taking over, they bailed in droves mostly bc of ORCL's crappy reputation for service and licensing string arm tactics.
After all, when we were Sun, Solaris and Sparc were our bread and butter, the reason we all came to work everyday. Solaris to Oracle is just another tech they offer from a long, long list of products.
Typical Solaris fanboi.
Why don't you take this toxic attitude to your next job interview. If you ever get one.
Better yet: start telling your interviewer how amazing and smart and speshul Sun, Solaris and SPARC were. You invented everything 30 years ago. Been there, done that. It's just everyone else was too stupid to realize how amazing Sun was. You were the only smart ones.
I predict you'll never make it past the first round.
How come Sun doesn't exist anymore? Whose fault was it? Planet alignment? Asteroid? Global warming? Cosmic rays? Belgium? Linus Torvalds?
Do you still have a job? I do. Not at Oracle. Something you should think about.
"Everything @P6Mmftb-1lyg writes is true. I was there when it happened."
Aw. That's precious. I was there 29 years, Sporto -- did you bail after six months of bad reviews, or did you last longer?
Thank you for the insight, Mr. Ellison.
Sun's failure has nothing to do with LE. Sun was dead at least 5 years before Oracle acquired it.
Everything @P6Mmftb-1lyg writes is true. I was there when it happened. I bailed because the writing was on the wall years before Oracle acquired Sun.
OpenSolaris was Sun's last chance to turn things around. Sun blew that too.
Stop deluding yourselves. It won't help with your job search.
Thank you for the insight, Mr. Ellison.
No, Sun killed Sun. It actually started prior to the Dot Bomb. Mafias, empires and turf kingdoms were rampant. Incompetence at high levels festered. Low, zero and negative ROI lab experiments got approved as products. Way too many sacred cows were allowed to exist. Granted, there will always be an element of this in any tech company. But Sun took it to a whole new level. The Dot Bomb ought to have been a wake up call. Sorry, we need to play a tune to ourselves while Rome burns.
Another inexcusable sin was allowing the aforementioned mafias to squelch any and all serious attempts to grapple with the growing Linux - X86 ecosystem. There was enough market info 20 years ago to come up with a good strategy. Nothing was done, and even worse, the few attempts made by good people contra the mafias were destroyed from within before they gained traction.
Sun is lucky it got acquired when it did by Oracle. It would have gone down hard or been acquired by corporate raiders with immediate blood bath to follow, had Oracle not rode in to save what could be saved.
It is what it is and hopefully a use case for all to learn from.