Slashdot Mirror
Why Is RAM Suddenly So Cheap? It Might Be Windows
jfruh writes: The average price of a 4GB DDR3 memory DIMM at the moment $18.50 — a price that's far lower than at this time last year. Why is it so cheap? The memory business tends to go in boom and bust cycles, but the free availability of Windows 10 means that fewer people are upgrading their PCs, reducing RAM demand. Analyst Avril Wu said, "Notebook shipments in the third quarter fall short of what is expected for a traditional peak season mainly because Windows 10 with its free upgrade plan negatively impacted replaced sales of notebooks to some extent rather than driving the demand for these products." And prices might stay low for another two years.
3 years ago, I bought 2x8 GB desktop DDR3 memory for about $70 CAD. It is now about $100. Where is Moore's law when we need it?
And DDR4 is even more expensive.
I upgraded my Thinkpad X230 from 8GB to 16GB because it was cheap enough, and because I was occasionally getting slowdowns in Chrome on Linux from so many windows and tabs open.
It fixed the slowdown problem, until recently, when Chrome on Linux decided to simply start crashing after so many (not even that many - maybe 40) were open.
Summary: Latest version of Chrome is total shit on Linux.
www.gaiageek.com
"The average price of a 4GB DDR3 memory DIMM at the moment $18.50"
It is? Newegg is all in the $21 - $23 range. Looking at CamelCamelCamel, it's about the same price it was around this time a year ago.
2x8GB DDR3 is still in the $80 - $90 range, same place it's been for months.
If you're careful with the services you setup, you can easily run 7, 8 and 10 on 240MB of ram without a problem. It's the feature creep that starts cutting into ram usage.
Om, nomnomnom...
I have 16GB and my computer frequently warns me about being out of memory. More than half of this RAM is consumed by Chrome, of course I have way more than 20 tabs open.
Different strokes fof different folks, but the point is that web browsers nowadays require an absurd amount of RAM.
The point is that bad web design nowadays require an absurd amount of RAM (and CPU cycles).
It really is quite easy to hand write rich looking designs. But instead of doing this, most web developers tend to use a metric shit ton of linked script libraries, make stupid non-optimized calls from these non-optimized libraries and generally just make bad design decisions. I puke in my mouth a little every time I see a web design that has jQuery (or even multiple versions of it) and several third party libraries linked *just* to produce something equivalent to a newsletter subscription overlay popup. This means there will be a lot of unnecessary HTTP calls for something that could be done in a one single GET and a result that could be produced with maybe five lines of pure JavaScript.
My customers have been amazed at how fast I can make my sites run even though they look "complex" and more often than not have a lot of graphical elements embedded throughout the design. It's just a question of optimization and having a tiny clue about what you're doing.
Although the impact is a little indirect, medical imaging systems are often rate limited by the hard drive. (When they aren't hamstrung by network speed that is) Frame rates are more a measure of how quickly you can scroll through the image stacks - the scanners themselves don't actually give you an 'image', they give you a bunch of instance objects that can potentially contain a few thousand parameters each - a subset of these within each object define how the pixel data will be interpreted to generate image data appropriate for the display depth.
You might have a 3000 image CT because the tech sent the raw acquisition rather than the more pertinent diagnostic sections, the radiologist expects to be able to scroll these very large stacks end to end ideally in one or two mouse movements - and they want to see every single image as that happens too. You don't always have enough RAM to store the entire data set so you have to load it from the hard drive as needed - then parse it out. Even when a study does fit in RAM the rad will usually have one or two series dragged over to the viewports a fraction of a second after the thumbnail has rendered - they are already flicking at the scroll wheel waiting for some business to happen, behind the scenes the image loader is still asking the PACS for a list of instance UID's and the path to the raw data because WADO is too slow :-)
No matter how fast the hardware is, there's always some inefficiency that people notice. Within an emergency room setting these delays can sometimes be costly.