DAC presentation slides (survey results on RISC V implementations)

[Guy Lemieux]

Hi,

I'm attaching my DAC presentation which includes the survey results on various RISC V implementations that are available in the ecosystem. These slides will be available on the DAC website in a few months (!).

I'll forward the detailed survey results to the RISC V Marketting Committee. I know they are planning their own survey, and these results will help their planning.

A big thank you to all who filled out the survey, and thanks to Andrew Waterman, Krste Asanovic and Yunsup Lee for sharing their slides and providing feedback on my slides.

Ciao,

Guy

[Ouabache Designworks]

Is github.com/tomythorn/yarvi a valid repo?

John Eaton

[Tommy Thorn]

Arrghh.  It's missing an 'm' :(

github.com/tommythorn/yarvi

On Fri, Jun 23, 2017 at 10:22 AM, Ouabache Designworks <z3qm...@gmail.com> wrote:

Is github.com/tomythorn/yarvi a valid repo?

John Eaton

--
You received this message because you are subscribed to the Google Groups "RISC-V HW Dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hw-dev+unsubscribe@groups.riscv.org.
To post to this group, send email to hw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/hw-dev/.
To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/hw-dev/CAOP4iL2EaFrYTsAqaZz3JDFwMw%2B75QsKogfnP-i2FiF-xZ%3DmxQ%40mail.gmail.com.

[高野茂幸]

Hi Guy-San,

I am interested in how implementing an xnor and pop count in your Binary NN Accelerator

Is cross point and inverter select corresponds the xnor?

Or inherently not implementation of xnor and pop count?

Best Regards,

S.Takano

2017年6月24日(土) 1:40 Guy Lemieux <glem...@vectorblox.com>:

--

You received this message because you are subscribed to the Google Groups "RISC-V HW Dev" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hw-dev+un...@groups.riscv.org.

To post to this group, send email to hw-...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/hw-dev/.

To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/hw-dev/CALo5CZyUN%3DmR1FrG4YoRBXnjWNATfTtoTRdAC9%3D8WFtKDZnorw%40mail.gmail.com.

[Samuel Falvo II]

kresetelcomputer.github.io/krestel

My repo is misspelled quite badly; this is the correct URL:

kestrelcomputer.github.io/kestrel

(As a useful mnemonic, just remember it's named after a bird of prey. :) )

> --
> You received this message because you are subscribed to the Google Groups

> "RISC-V ISA Dev" group.

> To unsubscribe from this group and stop receiving emails from it, send an

> email to isa-dev+u...@groups.riscv.org.
> To post to this group, send email to isa...@groups.riscv.org.
> Visit this group at
> https://groups.google.com/a/groups.riscv.org/group/isa-dev/.

> To view this discussion on the web visit

> https://groups.google.com/a/groups.riscv.org/d/msgid/isa-dev/CALo5CZyUN%3DmR1FrG4YoRBXnjWNATfTtoTRdAC9%3D8WFtKDZnorw%40mail.gmail.com.



--
Samuel A. Falvo II

[Guy Lemieux]

My apologies for the project URL typos ... they are now fixed and I am re-posting the final slides here.

Thanks for filling in the surveys!

Guy

On Fri, Jun 23, 2017 at 9:08 PM, Samuel Falvo II <sam....@gmail.com> wrote:

kresetelcomputer.github.io/krestel

My repo is misspelled quite badly; this is the correct URL:

kestrelcomputer.github.io/kestrel

(As a useful mnemonic, just remember it's named after a bird of prey.  :) )


On Fri, Jun 23, 2017 at 9:38 AM, Guy Lemieux <glem...@vectorblox.com> wrote:

> Hi,
>
> I'm attaching my DAC presentation which includes the survey results on
> various RISC V implementations that are available in the ecosystem. These
> slides will be available on the DAC website in a few months (!).
>
> I'll forward the detailed survey results to the RISC V Marketting Committee.
> I know they are planning their own survey, and these results will help their
> planning.
>
> A big thank you to all who filled out the survey, and thanks to Andrew
> Waterman, Krste Asanovic and Yunsup Lee for sharing their slides and
> providing feedback on my slides.
>
> Ciao,
> Guy
>

> --
> You received this message because you are subscribed to the Google Groups
> "RISC-V ISA Dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to isa-dev+unsubscribe@groups.riscv.org.

> To post to this group, send email to isa...@groups.riscv.org.
> Visit this group at
> https://groups.google.com/a/groups.riscv.org/group/isa-dev/.
> To view this discussion on the web visit
> https://groups.google.com/a/groups.riscv.org/d/msgid/isa-dev/CALo5CZyUN%3DmR1FrG4YoRBXnjWNATfTtoTRdAC9%3D8WFtKDZnorw%40mail.gmail.com.

--
Samuel A. Falvo II

--
You received this message because you are subscribed to the Google Groups "RISC-V ISA Dev" group.

To unsubscribe from this group and stop receiving emails from it, send an email to isa-dev+unsubscribe@groups.riscv.org.

To post to this group, send email to isa...@groups.riscv.org.
Visit this group at https://groups.google.com/a/groups.riscv.org/group/isa-dev/.

To view this discussion on the web visit https://groups.google.com/a/groups.riscv.org/d/msgid/isa-dev/CAEz%3Dsonz89MrQFAWwiSOYMcpeaSUyFXM1_b%3DhJcQQ7%3DJjiA2Tg%40mail.gmail.com.

[Guy Lemieux]

It's not based on XNOR/popcount. It's for binary weights with uint8 activations.

Details on the binary-weight accelerator are in the attached paper.
The paper will soon be posted to arxiv.org here:
https://arxiv.org/html/1704.08802

Guy

[高野茂幸]

Hi Guy-san,

Thank you for sharing your paper.

I read several binary CNN papers, then I wonder how interface present API and the accelerators or simply it is not programmable accelerators.

Best Regards,

S.Takano

>> email to hw-dev+unsubscribe@groups.riscv.org.