I'm out of ideas for news-titles

[Swordfish II]

True story, when I got my Pandora, I saw the "12hour battery life" claim, and thought "yeah when on standby," since normal laptops never meet the battery life they claim with any usage. I was amazed that I can actually use it and get that long. It still amazes me, and I love it.

The Pandora is honest but the mobile and laptop industry is shady

 

[FBnil]

My Ghz has a sweetspot of 800Mhz, while stil on the highest OPP, it consumes less power and is practically powerful enough for all things I do (listening to music, playing frootopia/candycrush, mindmapping).

 

[Linux-SWAT]

I'm out of ideas for news-titles

I have a neat idea for your next news post :
"Mass production finally started"

 

[Rockthesmurf]

On Windows, the most popular these days in Chrome, does this run on Linux? If so, does it somehow use up loads less RAM than on Windows? I currently have 10 tabs open and that alone is using up around 2 GB of memory. Here are the top few entries, from task manager:

I'm just used to Windows, and seeing 16GB of memory being consumed quite easily day to day, and I know Linux can, in some circumstance, use up less memory, but I guess it is a bit more interesting if doing exactly the same task on Linux vs Windows resulted in different memory consumption (i.e. running exactly the same build, i.e. built from same revision of source code, of Chrome on both, and visiting the same 10 websites on both).

 

[Silent-Hunter]

On Windows, the most popular these days in Chrome, does this run on Linux? If so, does it somehow use up loads less RAM than on Windows? I currently have 10 tabs open and that alone is using up around 2 GB of memory. Here are the top few entries, from task manager:

I'm just used to Windows, and seeing 16GB of memory being consumed quite easily day to day, and I know Linux can, in some circumstance, use up less memory, but I guess it is a bit more interesting if doing exactly the same task on Linux vs Windows resulted in different memory consumption (i.e. running exactly the same build, i.e. built from same revision of source code, of Chrome on both, and visiting the same 10 websites on both).

I can test this if you like, and check how much it's using.
[doublepost=1505070129,1505069377][/doublepost]

Here's the two, to compare. I've got the same plugins and the same three sites open on each. Youtube, Shadertoy, and Emby.
[doublepost=1505070507][/doublepost]Oh, the left column of memory info on the Windows one is labeled Private Bytes, and the right column is labeled Working Set. I'm not sure which one should be compared to the Linux one for a fair comparison.

 

[eric]

Don't know though, if it goes in squared - like voltage does in terms of electrical power ( P = U²/R).

Heat transfer is linear with the driving force, i.e. the temperature difference.

Sent from my Nexus 7 using Tapatalk

 

[Bokkertoff]

On Windows, the most popular these days in Chrome, does this run on Linux? If so, does it somehow use up loads less RAM than on Windows? I currently have 10 tabs open and that alone is using up around 2 GB of memory. Here are the top few entries, from task manager:

I'm just used to Windows, and seeing 16GB of memory being consumed quite easily day to day, and I know Linux can, in some circumstance, use up less memory, but I guess it is a bit more interesting if doing exactly the same task on Linux vs Windows resulted in different memory consumption (i.e. running exactly the same build, i.e. built from same revision of source code, of Chrome on both, and visiting the same 10 websites on both).

This is on my Laptop with Ubuntu 16.04, Both chrome and firefox running.
Both with 10+ tabs open:

6540 paul 8.90M 50.78M 7.93M 0.00k chrome
6552 paul 8.97M 50.85M 8.00M 0.00k chrome
6570 paul 8.99M 50.88M 8.02M 0.00k chrome
6653 paul 8.99M 50.84M 8.02M 0.00k chrome
6457 paul 10.42M 53.94M 9.41M 0.00k chrome
6644 paul 11.21M 52.85M 10.23M 0.00k chrome
2144 paul 12.84M 25.47M 12.28M 2.04M nm-applet
6203 paul 13.17M 57.82M 12.09M 0.00k chrome
6483 paul 19.13M 63.41M 18.11M 0.00k chrome
6493 paul 20.58M 64.96M 19.52M 0.00k chrome
6423 paul 21.57M 66.22M 20.52M 0.00k chrome
6438 paul 22.74M 67.71M 21.66M 0.00k chrome
6378 paul 23.40M 68.89M 22.33M 0.00k chrome
6216 paul 23.57M 69.48M 22.43M 0.00k chrome
1904 paul 23.66M 34.58M 23.18M 1.17M bamfdaemon
6152 paul 23.68M 69.47M 22.55M 0.00k chrome
6229 paul 23.81M 69.69M 22.68M 0.00k chrome
6944 paul 25.13M 34.20M 24.68M 12.84M python mainwindow.py
8992 paul 25.44M 39.18M 24.41M 1.90M gnome-terminal-server
6608 paul 25.58M 56.49M 18.33M 0.00k chrome
6295 paul 26.88M 71.47M 25.84M 0.00k chrome
4688 paul 27.14M 35.00M 26.59M 44.00k gvfsd-google
6112 paul 29.98M 75.65M 28.85M 0.00k chrome
6527 paul 38.12M 84.41M 36.98M 0.00k chrome
2230 paul 38.85M 48.25M 38.20M 3.93M goa-daemon
6448 paul 40.46M 86.38M 39.31M 0.00k chrome
6307 paul 45.16M 90.89M 44.05M 0.00k chrome
6357 paul 45.62M 91.42M 44.52M 0.00k chrome
6244 paul 46.20M 91.91M 45.09M 0.00k chrome
6458 paul 46.90M 92.76M 45.80M 0.00k chrome
6347 paul 48.14M 94.03M 47.02M 0.00k chrome
2148 paul 48.66M 64.57M 45.55M 1.89M nautilus -n
6584 paul 52.76M 98.16M 51.69M 0.00k chrome
6511 paul 55.86M 102.65M 54.53M 0.00k chrome
6256 paul 56.78M 103.04M 55.57M 0.00k chrome
6614 paul 56.79M 102.20M 55.71M 0.00k chrome
6192 paul 57.63M 102.98M 56.56M 0.00k chrome
6590 paul 58.61M 104.01M 57.54M 0.00k chrome
1975 paul 63.62M 70.21M 63.01M 1.12M hud-service
2038 paul 69.00M 73.50M 68.93M 5.96M indicator-power-service
6171 paul 82.60M 130.65M 81.14M 0.00k chrome
6052 paul 90.58M 139.17M 88.38M 0.00k chrome
6555 paul 92.58M 152.39M 77.49M 0.00k chrome
6271 paul 94.11M 141.55M 92.81M 0.00k chrome
1992 paul 101.32M 113.80M 100.69M 2.90M unity-panel-service
6319 paul 119.79M 167.11M 118.51M 0.00k chrome
6130 paul 122.76M 170.16M 121.36M 0.00k chrome
6088 paul 153.19M 203.27M 149.46M 0.00k chrome
5879 paul 184.09M 225.64M 178.21M 0.00k chrome
2143 paul 184.91M 197.79M 183.70M 16.16M gnome-software
5956 paul 211.91M 252.24M 191.22M 0.00k chrome
2049 paul 245.16M 261.69M 243.02M 33.44M compiz
5975 paul 328.79M 380.73M 324.04M 0.00k chrome
17937 paul 1.04G 1.07G 1.03G 0.00k firefox
--------------------------------------------------------------------
pid owner smap-pss smap-rss private swap Name
--------------------------------------------------------------------
Total 4.41G 6.79G 4.28G 257.82M 129

BTW This is with everything open (2 browsers with many tabs) that I use on a daily basis and still only use 4.4GiB

Although I ordered a 4G units, I was in 2 minds whether I should. But in the end it wasnt much more for extra 2G, so I went for that.
Its not a case that I absolutely wanted 4G RAM.

 

[WizardStan]

On Windows, the most popular these days in Chrome, does this run on Linux? If so, does it somehow use up loads less RAM than on Windows?

It wasn't uncommon for my gmail or facebook tabs to exceed half a gig, and would even occasionally get over 1GB. Something happened in an update a few weeks ago and suddenly they barely brush 300MB. Most normal tabs are around 50-80MB (which is still way too much in my opinion. Back in my day we used to browse the internet one page at a time on computers with 4MB of RAM and they worked just fine)

 

[Caine]

If you have 4GB of ram then it is silly to just utilize 4MB of RAM and suffer a penalty in responsiveness.

 

[WizardStan]

Hey! Get off my lawn!

 

[Pyramancer]

Keep in mind that the Pyra uses DDR and not an LPDDR type. Nature of the SoC.

Point of information:

The Pyra RAM chips (in the 4 GB units at least) are low power; DDR3L rather than DDR3; 1.35 V rather than 1.5 V. Or does this not count as LPDDR?

The actual chips being used (according to [1] (which was posted just after the recent temporary confusion about the chips was resolved)) are the Samsung K4B8G1646D-MMK0 8 Gb DDR3-1600 low power chips -- four of them, presumably.

[1] https://pyra-handheld.com/pipermail/kernel/2017-August/001954.html

 

[Letalis Sonus]

If you have 4GB of ram then it is silly to just utilize 4MB of RAM and suffer a penalty in responsiveness.

I'd like you to see even getting a negative effect on any original HTML page just by limiting the available RAM, even when fully running off the swap those websites will still be faster and more responsive than any of that modern bullshit.

Actually, one of the most popular German blogs is still using only 2nd gen HTML - you might know that guy as the author of dietlibc. The website's not even caching anything, each page is build from scratch every time - yet the only guys ending up DDoS'd were the victims of his very own Slashdot-effect.

In 2009 he posted a comment about the hardware: a 1.6GHz Sempron single core, 512MB RAM and a slow IDE drive - which handled a peak of 2.1 million page impressions like a breeze. The reason for the comment was the news of another blog that exploded with only 21k PIs - it was upgraded to two double quad core systems with 16GB RAM and a quad 15k RPM SAS RAID each and a load balancer afterwards... The client side is not the only thing that's getting all bloat-y nowadays.

 

[Grench]

Point of information:

The Pyra RAM chips (in the 4 GB units at least) are low power; DDR3L rather than DDR3; 1.35 V rather than 1.5 V. Or does this not count as LPDDR?

The actual chips being used (according to [1] (which was posted just after the recent temporary confusion about the chips was resolved)) are the Samsung K4B8G1646D-MMK0 8 Gb DDR3-1600 low power chips -- four of them, presumably.

[1] https://pyra-handheld.com/pipermail/kernel/2017-August/001954.html

Interesting... 1.35 V is something in between DDR (3.5V) and LPDDR (1.2V) - and closer to the LPDDR side of it. If it works - fantastic!

Utilizing LPDDR2 Vs DDR3 was a primary distinction between the OMAP 5430 and OMAP 5432.
http://www.ti.com/pdfs/wtbu/SWCT010.pdf

It sounds like a solution may have been found to 'split the difference'?

Can someone post up a brief synopsis of the "recent temporary confusion about the chips"?

 

[Swordfish II]

@Grench

Current ram for 4GB wasn't working. They uped voltage over spec and it worked. They have been exploring reasons why this is the case. ED mistakenly passed the wrong part number so they thought the problem was discovered, then found out it was the right part. Trials continue.

 

[Grench]

@Grench

Current ram for 4GB wasn't working. They uped voltage over spec and it worked. They have been exploring reasons why this is the case. ED mistakenly passed the wrong part number so they thought the problem was discovered, then found out it was the right part. Trials continue.

Are the not working 4GB chips just a doubled density version of the (working?) 2GB chips using the same pin count & pattern or are they a different architecture/design? I seem to recall something mentioned where the boards had pads for two different types of RAM pin configurations. If someone has the part number for the chips used in the 2GB configuration, it could be looked up.

Curiosity only - filling me in on details isn't likely to accelerate solutions. Maybe there is a lurker who might?

 

[Swordfish II]

Are the not working 4GB chips just a doubled density version of the (working?) 2GB chips using the same pin count & pattern or are they a different architecture/design? I seem to recall something mentioned where the boards had pads for two different types of RAM pin configurations. If someone has the part number for the chips used in the 2GB configuration, it could be looked up.

Curiosity only - filling me in on details isn't likely to accelerate solutions. Maybe there is a lurker who might?

Sorry, I didn't really follow the tech specs. If I remember right there were two ram slots, originally both were to be filled with 1gb for a total of 2. Once the 4gb plan arrived ED picked 2gb sticks.... For both 2gb orders and 4gb orders (one slot unfilled for 2gb). No issues with 2gb but 4 is in progress

 

[Letalis Sonus]

Or does this not count as LPDDR?

DDRxL and LPDDR are entirely different standards - LPDDR uses a custom protocol.

 

[rygD]

It wasn't uncommon for my gmail or facebook tabs to exceed half a gig, and would even occasionally get over 1GB. Something happened in an update a few weeks ago and suddenly they barely brush 300MB. Most normal tabs are around 50-80MB (which is still way too much in my opinion. Back in my day we used to browse the internet one page at a time on computers with 4MB of RAM and they worked just fine)

I tend to have some problems on lower specced boxes with gmail (I think I have 6 or so accounts for various purposes...yet still get spam in my education and professional ones) if I am doing other stuff. I made the mistake of trying to log in on my Pandora a few times. It works, but really not fun.

Back in the day I would have 20+ instances of the browser open, and would read our type stuff while I waited for other pages to load. This was mostly at schools or libraries. Tabs were a huge improvement for me. (The first thing I got when I was out first out on my own was a phone line so I could have internet access. I didn't even own a phone.)
[doublepost=1505164654,1505163899][/doublepost]

I'd like you to see even getting a negative effect on any original HTML page just by limiting the available RAM, even when fully running off the swap those websites will still be faster and more responsive than any of that modern bullshit.

Actually, one of the most popular German blogs is still using only 2nd gen HTML - you might know that guy as the author of dietlibc. The website's not even caching anything, each page is build from scratch every time - yet the only guys ending up DDoS'd were the victims of his very own Slashdot-effect.

In 2009 he posted a comment about the hardware: a 1.6GHz Sempron single core, 512MB RAM and a slow IDE drive - which handled a peak of 2.1 million page impressions like a breeze. The reason for the comment was the news of another blog that exploded with only 21k PIs - it was upgraded to two double quad core systems with 16GB RAM and a quad 15k RPM SAS RAID each and a load balancer afterwards... The client side is not the only thing that's getting all bloat-y nowadays.

I used to hand write my pages, which I intentionally made as hideous and irritating as possible. I no longer have a site, but if I did I would probably keep it very simple, probably with a bunch of random unlinked pages with various things I am working on. I an reminded of an editor/translator that has a site that has been around since the 90s, and it looks like it (which is a great thing, so simple to navigate, quick to load, etc.). He has transcriptions of books/manuscripts, including some he is currently working on, plus his translations of a few of them. And notes. His notes on his website alone are a huge resource for many people, and are to notch quality, just like the footnotes and appendices in his published books. Such a nice experience compared to most sites and pages I come across today.

Edit: On second thought, maybe I should just go with text files for the "nonpublic" stuff (still public, but not prettied up or intended for mass consumption), as that is how I work on it.

 

[Grench]

Sorry, I didn't really follow the tech specs. If I remember right there were two ram slots, originally both were to be filled with 1gb for a total of 2. Once the 4gb plan arrived ED picked 2gb sticks.... For both 2gb orders and 4gb orders (one slot unfilled for 2gb). No issues with 2gb but 4 is in progress

That doesn't sound right.

I recall that the original 2GB spec called for four 512MB chips - two on each RAM controller on the SoC so that it could get the full data path.

I may be imagining this - but I thought that the RAM had to go in as pairs on each controller and that if only one controller were populated, it would half the data rate - no interleave.

When ED was first talking about the 4GB option, the 1GB per chip equivalent of the 512MB chips were prohibitively expensive, but there was this other kind of RAM that required a different pin-out, but that was OK because the boards were built with the contacts/pads to take either format. I'm really fuzzy about my recollection of this - does anyone else recall those posts from ~2 years ago about this?

But - maybe there are 3 or 4 types of RAM configuration to consider?
A) 4 * 512MB = 2GB
B) 2 * 1GB = 2GB (no interleave)
C) 4 * 1GB = 4GB (chips like the 512MB ones only too expensive?)
D) 4 * 1GB = 4GB (chips that are currently being used in prototypes (Samsung K4B8G1646D-MMK0 8 Gb DDR3-1600?))

I'm pretty sure the pre-Pyra 2GB development boards use method A).

I don't know if B) or C) have been attempted.

Units configured as D) might be able to be initialized as B) to give 2GB active on a 4GB device using 1/2 controllers - I'm wondering if this is the mode where 2GB was rumored to be working on 4GB units? If so, does it work regardless of which of the 2 controllers is active?

 

[rygD]

@Grench, I don't recall the RAM with a different pinout. I also don't remember ED saying the 1GB RAM being more expensive, although I do remember him saying he would look into it, then coming back with numbers for us, which is when serious talk about the 4GB units started.

I don't know who mentioned it, but I do remember some talk of using the 1GB chips for the 2GB units that Swordfish II mentioned. I thought ED shot that idea down, or wanted to avoid it, and there was some discussion of using the sample of the 512MB chips to fulfill 2GB offers if they stayed low enough. I don't know what that number was (plus the prototypes I am sure you are constantly thinking about), or what the plan was after that.

Last I heard the 2GB units/prototypes were using A in your list.