Slashdot Mirror
Intel Says It Will Stop Developing Compute Cards
Intel will not develop new Compute Cards, the company said this week. From a report: Compute Cards were Intel's vision of modular computing that would allow customers to continually update point of sale systems, all-in-one desktops, laptops and other devices. Pull out one card, replace it with another, and you have a new CPU, plus RAM and storage. "We continue to believe modular computing is a market where there are many opportunities for innovation," an Intel spokesperson told Tom's Hardware. "However, as we look at the best way to address this opportunity, we've made the decision that we will not develop new Compute Card products moving forward. We will continue to sell and support the current Compute Card products through 2019 to ensure our customers receive the support they need with their current solutions, and we are thankful for their partnership on this change."
who cares
I tried two different version of these and they sucked bad.
Here are some examples of problems.
The biggest problem is the terrible implementation. Nothing was planned for expansion. The USB ports were limited though understandble, but often had a weird problem. The devices emitted an EMI field that prevented many wireless keyboards and mice from working. Had to get an extension usb cable or dongle to bring them far enough away from the device to function.
No flexible cabling to plug them into devices with limited profiles for their HDMI connectors and really the same for the other ports. They really should have broken out ports to allow for far more flexibility than they did. The NUC was not as bad as the Sticks because they had more surface area to work with but still found many problems with them.
Their designs just kept them down and prevented them from being as useful as they could have been.
On one side of their mouth they say "We continue to believe modular computing is a market where there are many opportunities for innovation." while the other side says they will no longer develop new Compute Cards."
So, which is it Intel?
Why not just be honest and say "We couldn't compete in this market/make enough money, so we're leaving it."
I'd respect you more if you were honest.
So rise up, all ye lost ones, as one, we'll claw the clouds.
...and meanwhile, the raspberry pi lives...
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
Intel doesn't want to compete in low-margin product areas where it can only make a few dollars off each sale. Intel wants to sell expensive desktop and laptop processors where it is the only supplier, and rake in the profits. That is the market that made Intel what it is today, and is also very clearly the type of business that its current management wants to pursue. Managing a low-margin product line must be career suicide at Intel.
This is why Intel abandoned the IoT and the Arduino-style computer markets (e.g. the Galileo). There's no money to be made in them (by their standards). I accepted some donated Galileo boards from Intel, and tried in vain for months to get them to provide some additional parts. Every few months when I emailed an engineer, he had moved on to another department within Intel. The Galileo product line had the stench of death on it from the very beginning.
Remember Netburst- the dreadful Intel x86 architecture all tech sites (including this one) promised readers were the only future - especially since Intel promised to hit 10GHz with its very long pipeline design?
The shilling hid the fact that netburst was a disaster- which Intel could no longer hide when AMD invented x64 and true dual core x64 chips. Intel went back to the Pentium 3 design, crossed it with the best AMD had (legal via cross patent agreements), and made the 'winning' multicore 'Core 2' (now Core) architecture.
Jost ONE thing. This new architecture was always a sack of cheating sh-t. Why? Cos all thread memory access security was missing- by design. Proper multi-core/multi-threading chips must implement "lock and key' where every thread has an id number used to unlock access to appropriate memory resources using an actual physical (transistor) circuit.
But 'lock and key' increases memory latency, power usage, and thus reduces max clocks. By going the NSA friendly route, Intel gained better performance while tech 'journalists' carefully avoided mentioning this fatal flaw for more than a decade.
This flaw cannot be fixed accept by running only ONE thread on the Intel CPU, and flushing on-chip memory resources when the multi-tasking time-slice OS swaps threads. Guess how much slower this would make your Intel chip run?
Meanwhile AMD's Zen fully implements 'lock and key'. No known exploits based on Intel's fault have ever been shown on Zen.
If and when Intel releases a new architecture with proper thread security (not for at least THREE years), Intel would then declare all earlier Intel CPUs as "not fit for purpose" and Microsoft and Linux will patch older (all Intel chips used today) parts to run at between 30-10% of their performance today. Not 10% off. 10% (1 tenth) of todays's performance. Yes, it is that bad. Obviously when Intel has 'fixed' chips, Intel with have no incentive to maintain the lie.
Today it should be illegal for any sensitive corporation or government department to use Intel x86 CPUs. But Intel links to a certain vile nation that has total control over the governments of the West, so Intel considers itself fully 'protected'.
to systems people never upgrade. Seriously, it was for stuff like Point of Sale and cheap laptops. Folks don't upgrade those unless the system breaks, catches fire, burns to ashes and a dog sneezes on the ashes. If it's a cat they might still call in IT to fix it (smaller sneeze).
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
Maybe there ought to be a new website: killedbyintel.com
we Thank for their buying all new systems due to this change.
I'm a consultant - I convert gibberish into cash-flow.
They ceded to the chinese around 10-15 years ago when they dropped all their embedded products departments, including high volume low margin items like i8051s, memory chips, the majority of flash products, etc.
What many people forget is that Intel used to be the FIRST SOURCE for a little big of everything, having rushed to the forefront of process technology and staggered their product lines so as processes became obsolete they could move their older designs onto newer processes while their newest designs were taped out on the bleeding edge processes, first RAM or Flash to validate a new process, then when yields were high enough, CPUs before eventually passing on to consumer/commercial low margin parts, with the industrial editions remaining on larger more mature processes that met the reliability requirements for industrial and military grade customers.
Giving up all that is directly what lead to Intel's current predicament, in exchange for making profits look better to shareholders while effectively slitting their own wrists.
There was all that talk about bringing tech back onshore in America. Intel *HAD* most of that capability in the states up until that embedded divestiture. Most of those older processes were already in US fabs, some of them dating back multiple generations of processes.