Recommendations+For+Computer+For+SmartSDR?

Jay Nation

Glad I could help you out with your retirement funding, with my retirement, It's all new and shiney, and barely begun. The money I spend now won't be available later. It was all planned for while I was still working and saving for it. No animals were injured, no children went to bed hungry. Make your purchases when its time. Spend as much as you want too. Be bold when your able.

Neal_K3NC

Read and okay Peter's kind words, he has spent many hours with his head in the Windows codebase and is one of the most recognized driver experts in the country. We are not worthy!

Neal_K3NC

If you are using Realtek chipset in your  NIC, see what version of the driver you are using. Look at the end of the driver version number and it ends with the year of release. If its not 2011 you got a bad driver which is full of DPCs.

Peter K1PGV

Thank you Neal.  I'm blushing now... 

K1PGV

Walt - KZ1F

Wow Peter, I am glad you mentioned Windows Kernel mode developer. And just how long have you been doing that? I believe you are referencing to someone's blog post, hardly an architectural design.That's different than being there at the time. I am referring that which was done in the 70's, 80's, and 90's. Should you look at the Windows Performance Monitor, you'll notice they call it page writes and page reads. Maybe it's just your group that made the new designation. IBM, which created the virtual memory model and predates Bill Gates, also doesn't. I've been in that space for 43 years now. I was also heavily involved in OS/2 long before Windows hired Dave Cutler and splintered off NT. I also attended seminars done by Russinovich. And, you are contradicting yourself. This is not the forum to pick fights.

Neal_K3NC

I think Burroughs/Unisys had virtual memory before IBM (at least thats what they told me...)

Peter K1PGV

Mr. Corey: I don't want to argue with you.  I've been writing operating systems for more than 30 years.  I've been working with the Windows OS, at the source code level, for more than 15 years. The guy who writes the Windows Memory Manager is a friend of mine.  Mark Russinovich use to work for me.

Look in Resource Monitor, in Windows 7 or later, and you'll see the hard page faults reported.  Here's a quick screen grab from Windows 7:

In 1978, when both Dave Cutler and I were working at DEC, there was the concept of hard and soft page faults in VMS.  I don't know if this was Dave's doing or **** Hustefedt's.

The link I provided, is a blog post from TechNet. Or, if you prefer, check Windows Internals, Sixth Edition, Part 2 here -- you'll see Russinovich and Solomon describe a hard page fault.  This really isn't debatable.

Your entitled to your own opinion, but you are not entitled to your own facts.  I might not know much about amateur radio, but when it comes to how the Windows OS works, I feel like I have a pretty good grasp.  Because I'm not guessing, or taking somebody else's word for it. I know from seeing the code.  Or writing it.

Now I'm done with this thread. 

Peter K1PGV

Walt - KZ1F

More power to you Jay. I don't want to see prospective users get scared off.

Walt - KZ1F

Walt - KZ1F

It was late '60s IBM had MVS. I, frankly am unfamiliar with what the industry was doing prior to that. At the time I was in Big Blue country. My only point was page frames and mapping virtual addresses to page and offset were around before Microsoft. And my only point to even bringing up paging was it is very fast and efficient over normal disk access. So even if you have a machine that does page slightly, you likely will never know it. I also vaguely remember how IBM did 2 bit error correction but would be hard pressed to articulate how even though that has no bearing on what level of compute power is necessary.  My main point was something like passmark is only good for bragging rights, mine is bigger than yours or how can I make mine bigger. The degree of IO Boundness SSDR is, is a function of processor speed. The faster your processor the more time (wall clock and percentage) you instance of SSDR will be waiting for IO. With the the faster machine you have, the less important the cpu is.  SSDR is 100% dependent on input which comes either from inputs to the GUI or data flowing from the radio itself. So long as your processor can keep up with the data flow from the radio itself, no extra speed or thread count is meaningful. As for memory, according to microsoft perfmon the footprint of SSDR with all slice receivers and panadaptors running only took just over 4GB. The difference between 1 slice receiver and 4 was about 200MB. I suspect the difference between adding 3 more and 7 more is 260MB. So we're still talking under 4.5 GB. One would be hard pressed I think to find a machine with less than 8GB, my phone has 16 and my tablet has 32 64bit memory. Even though my tablet it Android ( I like Microsoft even less now, thanks Steve) it will be running my complete shack shortly. In 12 months most of the rest of you will have Window Surface III tablets and carrying your shack around in your satchel. Price over radio, a few hundred dollars. 

    The only reason I am investing in this type of discussion is I know of several potential buyers of the 6000 series that went elsewhere (more than those who expressed that publicly on here) because reading the mail on here led them to believe the software required too large of a computer investment and was too buggy at this point. People are already whining about the v1x to v2x upgrade cost. That revenue will not sustain FRS long. They need net new buyers. Allowing people to believe they'll need Dell PowrEdge dual Xeon sockets with 20GB of memory is long term counter productive, if we want our investment realized.

And with that I end my involvement with this thread.

Walt Corey - kz1f

Walt - KZ1F

the above is a screen shot of the section of Microsoft Performance Monitor referring to and capturing 'page read' and page write' statistics. I guess Steve's group didn't write that particular code or proof the doc on it.

Steve, people who resort to gratuitous name dropping are generally compensating for some other larger deficiency. And you clearly have no idea of my credentials. I'll leave that as an exercise for you.

And the truly sad part Steve is your little diatribe added zero value, as is generally the case. Paging had the most marginal of relevance to the discussion, beyond the fact it is very fast and efficient compared to std disk and page ins are faster than page writes. As for what was substantitive I couldn't help but note you had no issue with that.So it really was nothing beyond a personal attack.

 I am surprised Tim let that one slide.

KY6LA_Howard

I Used Virtual Memory on a Univac 1107 back in 1963.

donato cardarelli

Passmark could be use to get more data on your pc and gathering data how to upgrade it You need a starting point considering that we do not a full knoledge how our softwae are using our system We Have the windows service information and we know f that mart sdr is a video based application, skimmer is cpu depended and so on, but they are indication n we have to benchmsrkcand to improve. WE know also othat we need to have a balanced system to make sll component wok at better (do not put a super video card in a low cpu base system So passmar is not intended on my side as a click and answer application but an improving system

K1FR

Thanks Neal!  Your post got me to go back one more time to the NIC driver.  Though not a Realtek mine was May 2014.  Latest I could find.  So, decided to take a stab in the dark and roll it back to last driver just to see what would happen.  Totally took the NIC DPC off the list with Latency Monitor.  Machine is really solid now.  Not sure what the issue was with the newer driver, but have had similar things happened with graphics drivers in the past.  Still some tweaking to do, I think but what an improvement!

And, thanks to all that made helpful posts.  Great community.

73, Tom
K1FR

Bob G W1GLV

**** the automatic updates also.

Neal_K3NC

From the wikipedia:
" In 1961, the Burroughs Corporation independently released the first commercial computer with virtual memory, the B5000, with segmentation rather than paging".

http://en.wikipedia.org/wiki/Virtual_memory

DrTeeth

Page faults are normal - http://answers.microsoft.com/en-us/windows/forum/windows_7-performance/in-windows-7-what-is-a-memory...

Plus, a page file is always used no matter how much memory is installed. The best thing one can do for a pagefile is to give it a fixed size (same min-max) so it does not get fragmanted.

DrTeeth

Page faults are normal - http://answers.microsoft.com/en-us/windows/forum/windows_7-performance/in-windows-7-what-is-a-memory...

DrTeeth

Walt, you have nailed your problem "AccuWeather (junkware) and all the other garbage that comes with a Dell". The first thing I do with a new PC is a ****-ware cleanout.

Richard McClelland, AA5S

For those who are interested in setting up fast machines, this article about Intel's new PCIe 750 line of SSDs is relevant:  http://semiaccurate.com/2015/04/02/intel-releases-750-pcie-ssd-line/  I thought I should bury this in a related thread rather than start something new, since this is so far off topic.

Peter K1PGV

Regarding the PCIe 750 NVMe Drives AA5S mentioned:  In my professional life, I'm very involved NVMe technology. I would not put one of these drives in my personal system today.  There is great promise in NVMe , and I'm a strong supporter of it.  However, these products and the software device drivers to support them need some time to "settle down" and mature before they can be used in applications where people need them to just work as part of their everyday systems.

These drives are very cool for people who want to be on the "cutting edge" -- They are not yet good for people who just want to get stuff done using their PC.

Richard McClelland, AA5S

Thank you for the useful information.  The article referenced the 'newness' of NVMe as well.