5 ms·

I don't really have any comment on the project itself (not something I would ever use and I don't know the value of what they're proposing). But on purely geek

I don't really have any comment on the project itself (not something I would ever use and I don't know the value of what they're proposing). But on purely geek terms this thing seems to warrant a "holy shit": http://www.adapteva.com/products/silicon-devices/e64g401/ http://www.adapteva.com/products/silicon-devices/e64g401/ Again I don't know how (un)common that sort of thing is but I wasn't expecting to see 64 cores in that tiny form factor. Does anyone here know how cutting edge this thing is if at all? [Edit] Also does anyone here want to address use cases for this thing?

Swizec 14y ago
Don't modern GPU's have essentially thousands of cores?
- sspiff 14y ago
  Yes, but modern GPUs also use upwards of 100W of power, and their cores operate mostly in lock-step, which means they aren't fast for all kinds of tasks.
- stonemetal 14y ago
  It really depends on what you consider a core. http://www.anandtech.com/show/2918/2 http://www.anandtech.com/show/2918/2 That first picture shows 4 cores made of 4 sub cores with 32 processing elements each. Now Nvidia would claim each of those 32 processing elements is a core, but each of those cores can not act independently. So it is more like a very wide, very hyper threaded 16 core processor.
  - DeepDuh 14y ago
    I think NVIDIAs definition of a 'core' has some merit. First of all, they have some independency in that you can introduce branches over a subset of them, so they're not just SIMD vector units. Secondly, their threaded programming model is pretty well suited for many computational tasks. Executing the same operations over a whole 2D or 3D region of data is a pretty common thing in computing. If you can't parallelize your task that way, chances are it's not even parallelizeable on N x86 cores. If you compare this to x86 however, you'd have to count n Cores times the SSE vector length on each core to be fair. GPUs still come out ahead for most of heavy computational tasks though - which is why Intel is now fighting back with their Xeon Phi stuff (which sounds very promising btw., looking forward to play with our prerelease model that's coming soon ;) ).
wcchandler 14y ago
I haven't looked at the specs, yet, but this is what I've had in mind: Roll one out to help with deep packet inspection of some of my network traffic. Spam filtering might also be offloaded to one of these guys. Dedicated machines to host backend applications -- SQL servers, Apache, nginx, etc.
Geee 14y ago
You'll need this if you want energy-efficiency when solving parallelizable problems. Use one chip in energy-limited systems, like battery-powered robots. Use multiple chips in power/heat-limited systems, like supercomputers.
stonemetal 14y ago
http://en.wikipedia.org/wiki/TILE64 http://en.wikipedia.org/wiki/TILE64 Tilera did a very similar looking 64 cores on a chip in 2007, which is the oldest instance I know of off the top of my head. Their devices cost(or at least they used to) a few grand though. Tilera has bumped it up to around 100 per chip these days. I don't know anything about either architecture so it is hard to say if 64 1Ghz adapteva cores compares with 64 1.5Ghz Tilera cores. So not quite cutting edge just an under explored side channel.
DeepDuh 14y ago
Well, NVIDIAs Kepler GPUs have 1536 cores on something like 320mm^2. I can't really find the die size of that adapteva product but I'd say it comes out at a similar range. Having looked at the data a bit more: I like their specs concerning system balance. 100 GFLOPS over 6.4GB/s gives you a system balance of 15.625 FLOPS per memory access, that's about the same balance as a Westmere Xeon - pretty good for real world algorithms. For comparison: NVIDIA Fermi has a system balance of about 20. Meaning: Fermi is sooner bounded by memory bandwidth, which is very often the limiting factor in real world computations. One thing though: High Performance Computing is all about software / tooling support. If this company comes out with OpenCL in C (even better Fortran 90+) support, then we're talking. Edit: By similar 'range' I meant core per mm^2 ratio.
- compilercreator 14y ago
  Prepare to be surprised. The die size estimate for the Epiphany IV is 10mm-sq according to Adapteva. It is more appropriate to compare it to embedded GPUs than desktop GPUs in die size, power and performance. For example, one particular embedded 40nm GPU design that I know about can deliver about 25 GFlops or so in the same die area.
  - DeepDuh 14y ago
    Ok, so that's 6.4 cores per mm^2 while Kepler has 4.8. Not bad, considering NVIDIA has already shrunk the scheduling ressources, register blocks and cache sizes per core to a bare minimum (something I don't agree with btw.).
    - adapteva 14y ago
      What do you consider a Kepler core? The Epiphany cores are dual issue RISC processors that run independent C/C++ tasks.
      - DeepDuh 14y ago
        Kepler has 8 "SMX" with 196 parallel sp threads each. For me the number of cores = the number of parallel threads, although on GPU they are not as independent, i.e. each "core" of an SMX either executes the same instruction on adjacent data or does nop. With dual issue do you mean a two stage pipeline or two threads in parallel, both performing FLOP?

Swizec 14y ago

Don't modern GPU's have essentially thousands of cores?

sspiff 14y ago
Yes, but modern GPUs also use upwards of 100W of power, and their cores operate mostly in lock-step, which means they aren't fast for all kinds of tasks.
stonemetal 14y ago
It really depends on what you consider a core. http://www.anandtech.com/show/2918/2 http://www.anandtech.com/show/2918/2 That first picture shows 4 cores made of 4 sub cores with 32 processing elements each. Now Nvidia would claim each of those 32 processing elements is a core, but each of those cores can not act independently. So it is more like a very wide, very hyper threaded 16 core processor.
- DeepDuh 14y ago
  I think NVIDIAs definition of a 'core' has some merit. First of all, they have some independency in that you can introduce branches over a subset of them, so they're not just SIMD vector units. Secondly, their threaded programming model is pretty well suited for many computational tasks. Executing the same operations over a whole 2D or 3D region of data is a pretty common thing in computing. If you can't parallelize your task that way, chances are it's not even parallelizeable on N x86 cores. If you compare this to x86 however, you'd have to count n Cores times the SSE vector length on each core to be fair. GPUs still come out ahead for most of heavy computational tasks though - which is why Intel is now fighting back with their Xeon Phi stuff (which sounds very promising btw., looking forward to play with our prerelease model that's coming soon ;) ).

wcchandler 14y ago

I haven't looked at the specs, yet, but this is what I've had in mind: Roll one out to help with deep packet inspection of some of my network traffic. Spam filtering might also be offloaded to one of these guys. Dedicated machines to host backend applications -- SQL servers, Apache, nginx, etc.

Geee 14y ago

You'll need this if you want energy-efficiency when solving parallelizable problems. Use one chip in energy-limited systems, like battery-powered robots. Use multiple chips in power/heat-limited systems, like supercomputers.

stonemetal 14y ago

http://en.wikipedia.org/wiki/TILE64 http://en.wikipedia.org/wiki/TILE64 Tilera did a very similar looking 64 cores on a chip in 2007, which is the oldest instance I know of off the top of my head. Their devices cost(or at least they used to) a few grand though. Tilera has bumped it up to around 100 per chip these days. I don't know anything about either architecture so it is hard to say if 64 1Ghz adapteva cores compares with 64 1.5Ghz Tilera cores. So not quite cutting edge just an under explored side channel.

DeepDuh 14y ago

Well, NVIDIAs Kepler GPUs have 1536 cores on something like 320mm^2. I can't really find the die size of that adapteva product but I'd say it comes out at a similar range. Having looked at the data a bit more: I like their specs concerning system balance. 100 GFLOPS over 6.4GB/s gives you a system balance of 15.625 FLOPS per memory access, that's about the same balance as a Westmere Xeon - pretty good for real world algorithms. For comparison: NVIDIA Fermi has a system balance of about 20. Meaning: Fermi is sooner bounded by memory bandwidth, which is very often the limiting factor in real world computations. One thing though: High Performance Computing is all about software / tooling support. If this company comes out with OpenCL in C (even better Fortran 90+) support, then we're talking. Edit: By similar 'range' I meant core per mm^2 ratio.

compilercreator 14y ago
Prepare to be surprised. The die size estimate for the Epiphany IV is 10mm-sq according to Adapteva. It is more appropriate to compare it to embedded GPUs than desktop GPUs in die size, power and performance. For example, one particular embedded 40nm GPU design that I know about can deliver about 25 GFlops or so in the same die area.
- DeepDuh 14y ago
  Ok, so that's 6.4 cores per mm^2 while Kepler has 4.8. Not bad, considering NVIDIA has already shrunk the scheduling ressources, register blocks and cache sizes per core to a bare minimum (something I don't agree with btw.).
  - adapteva 14y ago
    What do you consider a Kepler core? The Epiphany cores are dual issue RISC processors that run independent C/C++ tasks.
    - DeepDuh 14y ago
      Kepler has 8 "SMX" with 196 parallel sp threads each. For me the number of cores = the number of parallel threads, although on GPU they are not as independent, i.e. each "core" of an SMX either executes the same instruction on adjacent data or does nop. With dual issue do you mean a two stage pipeline or two threads in parallel, both performing FLOP?