KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
And this 0.7 watt chip can be powered by a single AA battery.
Read more.
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Wonder how many Transputers you could get on a chip these days.
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
Originally Posted by
DanceswithUnix
Wonder how many Transputers you could get on a chip these days.
A lot.
One could have a play with an FPGA and http://www.opentransputer.org/
Thinking about processors in general: https://www.cs.bris.ac.uk/~dave/benes.pdf
And if you can get hold of James Hanlon's thesis, he looks at some of the physical aspects of cramming cores in, I think: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.652048
As for me, I'm currently working on getting 32 picoRV32 cores into a network-on-chip in an FPGA, with the possibility of expanding that out to get something in the region of a couple of hundred cores. As an ASIC, 1K cores might be feasible.
...and they're all a nightmare to program for, in the conventional sense ;)
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
Originally Posted by
Steve
...and they're all a nightmare to program for, in the conventional sense ;)
Which is the thought I had which made me think "Haven't we been here before" and reminded me of Transputers :D
I think I would want a modern take on the idea though. Proper RISC design, 64 bit, rather than the funky stack based instructions of the Transputer. I don't think the signed address space worked out that well either, certainly not for me as a C programmer. But the idea of a simple core with comms links tied in seems still valid, just these days you could make most of those links into a FIFO to talk to other on chip cores and only go serial when heading off chip.
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
A little bit of misleading writing here - the first paragraph states:
Quote:
this chip contains 621 million transistors, and is capable of executing 1.78 trillion instructions per second, yet is hugely power efficient and can be powered by a single AA battery.
whereas it's later clarified that it must be clocked lower to meet AA battery power levels, and achieves less than 7% of that peak performance.
Also, whilst I'm sure the AA battery is capable of powering it at 0.7W, that doesn't tell us how long the battery will last for! 10 seconds? A couple of minutes? Long enough to actually be useful?
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
Originally Posted by
DanceswithUnix
I think I would want a modern take on the idea though. Proper RISC design, 64 bit, rather than the funky stack based instructions of the Transputer. I don't think the signed address space worked out that well either, certainly not for me as a C programmer. But the idea of a simple core with comms links tied in seems still valid, just these days you could make most of those links into a FIFO to talk to other on chip cores and only go serial when heading off chip.
This is very close to what I'm doing, except the RISC V variant I'm using is 32-bit, because it's small. Alternative, larger implementations are 64-bit though, as the ISA is designed with support for either.
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
Originally Posted by
scaryjim
A little bit of misleading writing here - the first paragraph states:
Quote:
this chip contains 621 million transistors, and is capable of executing 1.78 trillion instructions per second, yet is hugely power efficient and can be powered by a single AA battery.
whereas it's later clarified that it must be clocked lower to meet AA battery power levels, and achieves less than 7% of that peak performance.
Also, whilst I'm sure the AA battery is capable of powering it at 0.7W, that doesn't tell us how long the battery will last for! 10 seconds? A couple of minutes? Long enough to actually be useful?
The instructions per second they're quoting is still around i5 level, which is pretty good for 0.7w. Although there's nothing to say about the efficiency of each instruction so it's a wild guess as to how effective it really is.
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
Originally Posted by
scaryjim
A little bit of misleading writing here - the first paragraph states:
whereas it's later clarified that it must be clocked lower to meet AA battery power levels, and achieves less than 7% of that peak performance.
Also, whilst I'm sure the AA battery is capable of powering it at 0.7W, that doesn't tell us how long the battery will last for! 10 seconds? A couple of minutes? Long enough to actually be useful?
0.7W at 1.5V is about 0.5A, and according to this site you'll get an hour or two out of it assuming the chip doesn't need exactly 1.5V
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Plenty of overclocking headroom then ;) mind you might have to slap an heatsink on the sucker then.
In fact on second looks it has mounts for an heatsink ... shrug.
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
Quote:
Originally Posted by
frozen-monkey
But can it run Doom?
You're doing it wrong
Re: KiloCore 1000-core chip can execute 1.78 trillion IPS
please the tech guys: Xlucine, HW90, DancesWithUnix, Steve, ScaryJim, CaptainAmerica please tell us how the kilcore differentiates to a Tesla or Firepro processing core in terms of architecture, because I do understand Tesla has CUDA cores which are many'cores'