Actually not. It seems they use an I2C extender for GPIO.
Actually not. It seems they use an I2C extender for GPIO.
-- Failure does not prove something is impossible, failure simply indicates you are not using the right tools... nico@nctdevpuntnl (punt=.) --------------------------------------------------------------
400MHz is fine if you don't do video.
But i only want half of a Zynq, and only pay half. Projected price of the X1000 is $20 to $30 vs. $50 to %60 for the Zynq.
but you said you wanted an FPGA, there goes the price difference and zynq has GbE, dual SDIO, CAN, proper IO, ADC
-Lasse
dache of using crap products like Intel, I seriously doubt the Arduino cust omer base will give Intel a second thought.
It might
slow a 400MHz "pentium" is
hy not
ssle
the
Obviously, price is an important factor whether Quark will be successful or not. Based on projected size on 32nm process, the chip should go for $20 to $30. But if Intel is agressive, it could be close to the low end or eve n below $20. FPGA would add $10. So, for a total cost of $30 in major BOM .
Zynq is a much bigger chip and cost around $50 to $60.
Furthermore, Quark has 8M internal flash. Zynq needs some external flash f or it to work.
No sure what is this "proper IO" issue. Quark has 10 GPIO and 6 GPIO_SUSP (work in suspend mode). Have not look into what Zynq has.
It'd be like 1997 all over again.
-- For a good time: install ntp --- news://freenews.netfront.net/ - complaints: news@netfront.net ---
On a sunny day (2 Nov 2013 06:08:25 GMT) it happened Jasen Betts wrote in :
I have been told, and IIRC have read up on it, thar the current 'pentium' is a completely different core than the old one. Not the ssme processor at all. There was and is a lot of in depth info about that architecture on
Ok, 1997 without the arithmetic problems :)
-- For a good time: install ntp
e.
ise.de,
In English, just visit intel.com
I've read the datasheet and many more. I suggest you guys to do so as well . Basically, it's a 486 core with modem bus (SPI, USB, PCIe) in sub volt 3
5nm process. Around 2.2W, 3.3V or 5V. Pentium instructions microcode upgr ade. Interesting things are the 8M FLASH and 11K EEPROM onboard, as well a s 512K SRAM.Without the DDR3 SDRAM, it might be possible to do a two layer board. PCIe would be an interesting expansion path. Unfortunately, i haven't found an y FPGA that can clock at 2.5GHz.
one.
heise.de,
ll. Basically, it's a 486 core with modem bus (SPI, USB, PCIe) in sub volt 35nm process. Around 2.2W, 3.3V or 5V. Pentium instructions microcode up grade. Interesting things are the 8M FLASH and 11K EEPROM onboard, as well as 512K SRAM.
Ie would be an interesting expansion path. Unfortunately, i haven't found any FPGA that can clock at 2.5GHz.
OK, Xilinx Artix-7 and Altera Cyclon IV have GHz transceivers.
I'll build two 486 chips for my old motherboard (somewhere in storage). Ad dress and Data bus should come out of the FPGA. Perhaps a USB port on top of the chip.
Embedded may go to ARM, too.
I think the medium-to-big embedded processor wars in the next decade are going to go to the companies that make the most reliable ARM-cored parts with the best peripherals, possibly with a dash of "lowest power" thrown in there, too.
I'm not sure what's going to happen to the little embedded processor market currently served by 8-bitters: ARM is going to eat down into that one, but if you froze me today and thawed me out 500 years from now, I wouldn't be surprised if the Transporter Room didn't contain at least one
8051.-- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
Absolutely. Rumor has it that Intel may try to do core licenses of x86 architectures so people can embedd them in ASICS. Good luck.
Intel has given up on building chips that have any i/o except bad-latency PC stuff: pcie, usb, cached dram. Embedded apps need a chip with ports and timers and memory busses (PICs or LPC17xx sorts of things) or FPGAs with hard ARM cores. And they have to be low power and cheap.
Intel had an ARM license and gave it up! Companies don't have the guts to kill off their own legacy products, so they let someone else do it.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
But the Quark has legacy I/O bridge and AMBA bridge (stolen from ARM) as well, so this chip can do some good raw I/Os. Alternatively, a high speed interface into an FPGA; namely PCIe.
Quark + Cyclone IV might be a good idea. Cyclone IV GX has hard core Gbps transceiver.
Hey John, I will be in San Francisco shortly. Perhaps we can argue about this over beer and pizza.
PCIe is a nightmare. It's amazingly complex, and latency is terrible. I still like the idea of doing a MOV instruction to load a parallel port. Old fashioned, I guess.
We're planning to use a microZED/Zynq on a couple of upcoming projects. Zynq is an FPGA with two 800 MHz hardcore ARMs inside. It powers up running Linux.
(nice pic, huh?)
Absolutely. Unless it's beer and burgers.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
Embedded is already going ARM, if it isn't already there.
Already happened. Many suppliers.
It already is. An M0 with some nice peripherals (several SPI/I2C/Async, touch, etc.) goes for $.50 in quantity now.
Den fredag den 1. november 2013 05.39.47 UTC+1 skrev snipped-for-privacy@gmail.com:
eadache of using crap products like Intel, I seriously doubt the Arduino cu stomer base will give Intel a second thought.
It might
ow slow a 400MHz "pentium" is
why not
hassle
f the
or not. Based on projected size on 32nm process, the chip should go for $2
0 to $30. But if Intel is agressive, it could be close to the low end or e ven below $20. FPGA would add $10. So, for a total cost of $30 in major B OM.you can a 1GHz ARM for less than $6
it is also twice as fast, dual core, has many more peripherals including an ADC, and nearly unlimited bandwidth to the FPGA
for it to work.
but what will you do with 8M ? not nearly enough for windows, and if you do n't want windows what does an x86 do that isn't done faster and cheaper wit h an ARM?
P (work in suspend mode). Have not look into what Zynq has.
Zynq has +100 depending on package and what peripherals you use
-Lasse
You can still do that. 16 GPIOs (6 w/ suspend) are available on direct memory bus. PCIe is good if you can offload all the real time stuff to the FPGA.
Yes, if you really need more than 8M code space. But for us, we can fit everything into 8M and forget about SDRAM. So, Quark will be smaller than Zynq.
OK, i'll look you up when i am in town next month.
l or not. Based on projected size on 32nm process, the chip should go for $20 to $30. But if Intel is agressive, it could be close to the low end or even below $20. FPGA would add $10. So, for a total cost of $30 in major BOM.
But not with 8M FLASH and 11K EEPROM and 512K SRAM.
an ADC, and nearly unlimited bandwidth to the FPGA
I have to look up how fast is "unlimited bandwidth". Will get back to you later.
don't want windows what does an x86 do that isn't done faster and cheaper w ith an ARM?
Have not seen any other chip with 8M FLASH, which is good enough for us. X
86 or ARM does not matter in this case. OTOH, i am spending more than half of my time on an x86 app (CAM350) for layout. I need at least one x86 com puter for as long as i live. Old habit die hard, i can't change just becau se every body else is running ARM.USP (work in suspend mode). Have not look into what Zynq has.
You lose much of this +100 pins as soon as you do external memories, and Zy nq can't run without external memories.
ST are claiming $.32 for their M0 "value line"
-- John Devereux
Den søndag den 3. november 2013 00.59.04 UTC+1 skrev snipped-for-privacy@gmail.com :
ful or not. Based on projected size on 32nm process, the chip should go fo r $20 to $30. But if Intel is agressive, it could be close to the low end or even below $20. FPGA would add $10. So, for a total cost of $30 in maj or BOM.
the flash is external via spi it has to be copied to ram to run code
g an ADC, and nearly unlimited bandwidth to the FPGA
u later.
on-chip directly on the cpu bus is going to be faster and lower latency than any off-chip interface
u don't want windows what does an x86 do that isn't done faster and cheaper with an ARM?
X86 or ARM does not matter in this case. OTOH, i am spending more than ha lf of my time on an x86 app (CAM350) for layout. I need at least one x86 c omputer for as long as i live. Old habit die hard, i can't change just bec ause every body else is running ARM.
the 8M is not on chip it is an external spi device
x86 is fine as a desktop workstation and it runs windows but for pretty muc h everything else I don't see why you would use it
_SUSP (work in suspend mode). Have not look into what Zynq has.
Zynq can't run without external memories.
so you are down to ~50, and it will run just as well as a Quark without ext ernal memory, both will need at least an external flash
-Lasse
Not surprising. The $.50 above was an early budgetary quote for an automotive qualified part with a different mix of peripherals, though a QFP32 or QFN32 package.
The point is that ARM *has* taken over the embedded market, too.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.