I can now understand why you've chosen this route. It's a shame since I'm one of the people who bought a Sun system, and this is something I've came to expect, with time I expected something to happen. Following Sun's management over OpenSolaris in general has really stuck as very controlled, I don't particularly like how deeply consumed Sun is about controlling anything related to Solaris. It's true that no other system except Linux and Mac OS X has real time support. Linux does not bundle it, and I personally have not actually measured the impact of time drift on software such as VMware which relies on RTC when using RT patches, which are experimental still. Mac OS X being a hybrid Microkernel and Monolithic kernel could be said to have real time capability as well, and in some sense it does, though in the traditional sense compared to Solaris for example, it is much different.
My Ultra 20 M2 was bought September 26, 2007 and so far I have found that it would had been better to go with another vendor. The only saving grace for your hardware is the quality, the components are certainly overpriced, in terms of almost being exponentially more costly without much warrant for being so. The compatibility, the integration are of least of my worries now that such decisions have negatively impacted my work flow by costing me real numbers. I hope to see someone with AMD Power feature experience from another camp step up and contribute, but I won't hold my breath. Intel is certainly your main focus, as well as most the world's. I strongly believe you have placed AMD below Intel as of late. For someone who invested in your company through hardware acquisition, I did so because I wanted to keep a company of your ideals and innovation alive, but doing so for those reasons so far has delivered little or nothing that I can really account to be positive. Taking a look at your stock split, ticker change, and still decreasing share price, is something that will drive Sun into the ground. Before the 700mln infusion from the new board member, your stocks over the years depreciated while other companies thrived in this 1987-2008 recession, but it's still slowly bit by bit crumbling from the inside out. I predicted this would be the result of your tight control, bad communication to consumers, and strange decisions over the years. I invest because I don't want to see it happen, because freely thinking companies are a dead breed. Since none of the real engineers have a say in what the marketing team thinks is what Sun should be, Sun as a whole is sinking. Sun is a company I wished as a young man to work for, but there's no real reason behind such thinking now, given the obvious internal problems. My participation in the OpenSolaris project as a user and free thinking individual is a last effort to help in whatever ways I can to steer out of calamity, but so far I as a contributor feel that nothing good comes out of it, the Sun employees override all that is, all that will become, and what has past. James On Sat, Jun 14, 2008 at 4:44 PM, Garrett D'Amore <gdamore at opensolaris.org> wrote: > James Cornell wrote: > >> I just answered to Simon Zheng about this, but just check your power bill >> after experimenting with a lab full of machines running 12-16 hours a day, >> you'll save a good deal of money by having the frequency change. I was sure >> that you tested against your own current hardware, and that's good news to >> me, but the lack of cpu frequency support is really an oversight. With your >> engineering capability I really am concerned about your focus. As I >> mentioned, the sustainability of Solaris for wide use is in question when >> 90% of ALL systems are unsupported for the feature at hand. Linux, *BSD, >> and Windows (XP->Vista) support power management, with some better than >> others, but all support cpufreq so I'm really disappointed. >> > > I presume "you" means Sun, rather than "Garrett D'Amore". :-) > > We (Sun) are working towards new directions, tickless kernel and support > for lower power at idle (I think Intel calls this deep C states, AMD calls > it OPM), that may ultimately bring *better* power savings (with zero need > for frequency tuning by the end-user) instead of continuing to tweak cpu > frequency for older chips. So we're trying to work looking forward rather > than back. While you may disagree with this business approach, you can > hardly fault Sun for wanting to focus on the features that sell systems. > (At the end of the day, Sun still gains most of its revenue from Solaris > engineering in the form of sales of new systems.) > > Also, at the end of the day, Solaris is Open Source software. Anyone else > is free to volunteer patches to address PowerNow! for older multi-core > Opterons. Heck, if someone can convince me they have a good technical > solution, I'll be willing to sponsor the integration and any ARC case work. > > I suspect that the big thing is that unlike all those other OS' you > mentioned, Solaris offers hard real-time, and requires excellent time > keeping. I'm not terribly close to the problem, but its my understanding > that fixing the synchronization problems is "hard" on those older systems. > It would not surprise me (and this is conjecture on my part) if the problem > is badly exacerbated by our stricter requirements for accurate timing than > the other operating systems you mentioned. > > -- Garrett > >> >> James >> On Jun 13, 2008, at 9:55 PM, Garrett D'Amore wrote: >> >> James Cornell wrote: >>> >>>> Jeff Cai wrote: >>>> >>>> >>>>> A nice GNOME Power Manager flash demo. >>>>> >>>>> Jeff >>>>> >>>>> ------------------------------------------------------------------------ >>>>> >>>>> Subject: >>>>> GNOME Power Manager flash demo >>>>> From: >>>>> "simon.zheng at sun.com <mailto:simon.zheng at sun.com>" >>>>> <Simon.Zheng at Sun.COM <mailto:Simon.Zheng at Sun.COM>> >>>>> Date: >>>>> Fri, 13 Jun 2008 18:37:02 +0800 >>>>> To: >>>>> Desktop C-Team <desktop-cteam at sun.com <mailto:desktop-cteam at >>>>> sun.com>>, >>>>> solaris-battery-team at sun.com <mailto:solaris-battery-team at sun.com>, >>>>> David Chieu <David.Chieu at Sun.COM <mailto:David.Chieu at Sun.COM>>, >>>>> Randy >>>>> Fishel <randy.fishel at sun.com <mailto:randy.fishel at sun.com>> >>>>> >>>>> To: >>>>> Desktop C-Team <desktop-cteam at sun.com <mailto:desktop-cteam at >>>>> sun.com>>, >>>>> solaris-battery-team at sun.com <mailto:solaris-battery-team at sun.com>, >>>>> David Chieu <David.Chieu at Sun.COM <mailto:David.Chieu at Sun.COM>>, >>>>> Randy >>>>> Fishel <randy.fishel at sun.com <mailto:randy.fishel at sun.com>> >>>>> >>>>> >>>>> Hi All, >>>>> >>>>> GNOME Power Manager will be integrated into Neveda build 92. Here's a >>>>> demo to show what's new features we could see. >>>>> >>>>> >>>>> http://www.opensolaris.org/os/community/laptop/windmill/gpm/GPM_demo.html >>>>> Please note suspend to RAM is an experimental functionality on laptop. >>>>> >>>>> Thanks, >>>>> -Simon >>>>> >>>>> >>>>> ------------------------------------------------------------------------ >>>>> >>>>> _______________________________________________ >>>>> laptop-discuss mailing list >>>>> laptop-discuss at opensolaris.org <mailto:laptop-discuss at >>>>> opensolaris.org> >>>>> >>>>> >>>> Awesome, when will all this be integrated? What CPU's will be supported >>>> by CPUFreq (So far it seems only AMD Barcelona, single-core AMD64, >>>> Opteron forced into uniprocessor mode, and modern P4, Core, Core 2, Xeon >>>> 5000) >>>> >>>> >>> >>> The same chips. In other words, older dual core AMD64 parts and even >>> *older* Intel parts are out of luck. >>> >>> That said, there has been some debate about the utility of CPU frequency >>> tweaking. The thought is that for many systems, the CPU may consume less >>> power overall by "rushing" to idle state (i.e. run at full speed), thanks to >>> reduced voltages/power used in idle. (Note that this is not fully >>> substantiated yet, especially since we don't the full benefits possible >>> owing to our lack of deep C-state support, and the fact in a typical system >>> there are often a lot of tasks with periodic timers which tend to keep the >>> system awake. >>> >>> What kind of chipsets has suspend been tested on? nForce 550 Pro, Intel >>>> 965M? >>>> >>>> >>> >>> Those are probably good ones to use. They seem to match well to what we >>> we have in house (though I'm not sure about the precise model numbers -- the >>> stuff we test most is Ultra 20/40 and Toshiba Tecra laptops. We also have >>> been doing a lot of work with other Intel mainboards, such as found in the >>> Ultra 24.) >>> >>> -- Garrett >>> >>> >> > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/laptop-discuss/attachments/20080614/c50604da/attachment.html>
