Isn't Risc-V supposedly supposed to be more secure and isn't open power based on the old risc? I am just wondering, if the level of evils risc-v has done, justifies abandoning usage of it when you could just as easily do some risc-v processors with or without their help and moreover, power, doesn't it require an immense, even crazy amount of watts? I guess my point is, they use more battery power than intel even. And intel is bad enough.
Supposedly though I thought anyone could make risc-v processors as long as they dont use the trademarks.
That being said, I noticed you guys have been quiet for a while. I hope you guys are doing well. :)
I also hope you can reach your goal, who knows, maybe you can do twice what you thought someday with even 1W used only while being run.
If I am misunderstanding something, feel free to let me know, I just think making a line of Risc-V based processor is the way. Libre of course, but yeah... and yes I saw your update.
On 1/2/20, zap zapper@disroot.org wrote:
Isn't Risc-V supposedly supposed to be more secure and isn't open power based on the old risc? I am just wondering, if the level of evils risc-v has done, justifies abandoning usage of it when you could just as easily do some risc-v processors with or without their help
that's the point: we *can't* do a massive libre / open innovative enhancement to the RV ISA without official support.
the consequences - the risk of a clash in the ISA - would be absolutely disastrous.
I guess my point is, they use more battery power than intel even.
what, PowerPC? that's a poor micro-architectural choice that doesn't have anything specifically to do with the ISA, especially when the VLE book (Variable Length Encoding) exists (equivalent to RVC).
l.
On 01/01/2020 11:07 PM, Luke Kenneth Casson Leighton wrote:
On 1/2/20, zap zapper@disroot.org wrote:
Isn't Risc-V supposedly supposed to be more secure and isn't open power based on the old risc? I am just wondering, if the level of evils risc-v has done, justifies abandoning usage of it when you could just as easily do some risc-v processors with or without their help
that's the point: we *can't* do a massive libre / open innovative enhancement to the RV ISA without official support.
the consequences - the risk of a clash in the ISA - would be absolutely disastrous.
Oh really? That's odd... hmm... so you have to abandon risc-v you are saying? Pity that's your only option. I wonder if I should tell others about this.
I guess my point is, they use more battery power than intel even.
what, PowerPC? that's a poor micro-architectural choice that doesn't have anything specifically to do with the ISA, especially when the VLE book (Variable Length Encoding) exists (equivalent to RVC).
l.
Okay, well you did say openpower, what is that?
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On Thursday, January 2, 2020, zap zapper@disroot.org wrote:
.
the consequences - the risk of a clash in the ISA - would be absolutely disastrous.
Oh really? That's odd... hmm... so you have to abandon risc-v you are saying?
abandon *innovating* with RISCV, yes.
Pity that's your only option. I wonder if I should tell others about this.
the circumstances are pretty unique, so it's up to you. they are:
* a Mass Volume commercial product * a Mass Volume LIBRE product * a Mass Volume Libre product where the business opportunity is FULL TRANSPARENCY * a paramount need to innovate BEYOND that which is dictated in a fascist manner by the RISCV Foundation (i.e. they refused to follow Trademark Law's requirements)
very few other products have these requirements.
* a closed secretive company that is developing a proprietary extension may do so as a CUSTOM extension. there is no chance of the extension becoming public.
* an "open source" product simply implenenting EXISTING standards without ISA innovation clearly has no problem.
* an "extension vendor" has no problem because they will be selling to proprietary secretive corporations.
* an academic product also has no problem.
* a NONCOMMERCIAL "open" design which NEVER becomes a commercial product also gaw no problem.
* an UNSUCCESSFUL or niche product, even if it violates Trademark Law by using Custom OPcodes without authotisation, that is not intended to be mass volume also does not cause absolute chaos.
I guess my point is, they use more battery power than intel even.
what, PowerPC? that's a poor micro-architectural choice that doesn't have anything specifically to do with the ISA, especially when the VLE book (Variable Length Encoding) exists (equivalent to RVC).
l.
Okay, well you did say openpower, what is that?
google "open power foundation"
l.
Just to be clear, you cannot legally sell risc-v processors even if you remove the trademarks.
And, OpenPower can be made more secure and lightweight then Risc-V.
Do I have that right?
If so, then I will recommend the Hyperbola devs work within the OpenPower thought process for making libre, lightweight processors.
On 1/3/20, zap zapper@disroot.org wrote:
Just to be clear, you cannot legally sell risc-v processors even if you remove the trademarks.
like any trademark, if you make no mention of the trademark, or any claims of "compliance", you're probably ok.
from the time i worked on samba-tng, you can claim *compatibility* with something that is a pun or the *inversion* of a trademark. "arcfour-compatible" rather than "RC4 compliant".
etnaviv.
v-sirc.
if you say "v-sirc compatible" and you're ok.
And, OpenPower can be made more secure and lightweight then Risc-V.
that's very difficult to say. you start having to delve into what "secure" means at both the architectural, ISA *and* design level. "lightweight" is much easier to compare however would still take a significant amount of time.
Do I have that right?
If so, then I will recommend the Hyperbola devs work within the OpenPower thought process for making libre, lightweight processors.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk
On 01/03/2020 07:25 PM, Luke Kenneth Casson Leighton wrote:
On 1/3/20, zap zapper@disroot.org wrote:
Just to be clear, you cannot legally sell risc-v processors even if you remove the trademarks.
like any trademark, if you make no mention of the trademark, or any claims of "compliance", you're probably ok.
from the time i worked on samba-tng, you can claim *compatibility* with something that is a pun or the *inversion* of a trademark. "arcfour-compatible" rather than "RC4 compliant".
etnaviv.
v-sirc.
if you say "v-sirc compatible" and you're ok.
So to be clear, is it because it could be very dangerous to work on risc-v without their help.
And, OpenPower can be made more secure and lightweight then Risc-V.
that's very difficult to say. you start having to delve into what "secure" means at both the architectural, ISA *and* design level. "lightweight" is much easier to compare however would still take a significant amount of time.
Well said then. Lightweight is what I concern over more to be fair. I am sure they are both equally or close to equally secure though - the meltdown spectre crap. ;)
On 1/4/20, zap zapper@disroot.org wrote:
On 01/03/2020 07:25 PM, Luke Kenneth Casson Leighton wrote:
On 1/3/20, zap zapper@disroot.org wrote:
Just to be clear, you cannot legally sell risc-v processors even if you remove the trademarks.
like any trademark, if you make no mention of the trademark, or any claims of "compliance", you're probably ok.
from the time i worked on samba-tng, you can claim *compatibility* with something that is a pun or the *inversion* of a trademark. "arcfour-compatible" rather than "RC4 compliant".
etnaviv.
v-sirc.
if you say "v-sirc compatible" and you're ok.
So to be clear, is it because it could be very dangerous to work on risc-v without their help.
no, not at all. there's no need for "help". and it's not "dangerous", either.
* RISC-V is Trademarked. * therefore if you want to use the Trademark, you *must* respect the requirements set by the Trademark Holder. * if you do not want to use the Trademark in connection with your product, you do *NOT* have to meet the requirements.
there is no "danger" here, nor a "need for help".
in addition:
* if the Trademark Holder acts in a persistently UNREASONABLE WAY, they LOSE the Trademark.
And, OpenPower can be made more secure and lightweight then Risc-V.
that's very difficult to say. you start having to delve into what "secure" means at both the architectural, ISA *and* design level. "lightweight" is much easier to compare however would still take a significant amount of time.
Well said then. Lightweight is what I concern over more to be fair. I am sure they are both equally or close to equally secure though - the meltdown spectre crap. ;)
that's a micro-architectural design decision, not a fault of the ISA itself.
l.
like any trademark, if you make no mention of the trademark, or any claims of "compliance", you're probably ok.
from the time i worked on samba-tng, you can claim *compatibility* with something that is a pun or the *inversion* of a trademark. "arcfour-compatible" rather than "RC4 compliant".
etnaviv.
v-sirc.
if you say "v-sirc compatible" and you're ok.
So to be clear, is it because it could be very dangerous to work on risc-v without their help.
no, not at all. there's no need for "help". and it's not "dangerous", either.
- RISC-V is Trademarked.
- therefore if you want to use the Trademark, you *must* respect the
requirements set by the Trademark Holder.
- if you do not want to use the Trademark in connection with your
product, you do *NOT* have to meet the requirements.
there is no "danger" here, nor a "need for help".
in addition:
- if the Trademark Holder acts in a persistently UNREASONABLE WAY,
they LOSE the Trademark.
hmm... okay. I must've gotten confused then. I will wait to see your plans and whether you plan to make your own risc-v or openpower processor then.
I wish you the best on this,
And, OpenPower can be made more secure and lightweight then Risc-V.
that's very difficult to say. you start having to delve into what "secure" means at both the architectural, ISA *and* design level. "lightweight" is much easier to compare however would still take a significant amount of time.
Well said then. Lightweight is what I concern over more to be fair. I am sure they are both equally or close to equally secure though - the meltdown spectre crap. ;)
that's a micro-architectural design decision, not a fault of the ISA itself.
Oh, okay. I wonder how much the softcore's use in watts.
https://openpowerfoundation.org/openpower-summit-north-america-2019-introduc...
On 1/4/20, zap zapper@disroot.org wrote:
hmm... okay. I must've gotten confused then.
yes :) it's what's *not* done that matters, not what *is* done.
I will wait to see your plans and whether you plan to make your own risc-v or openpower processor then.
both.
absolute to-the-letter RV64GC compliance in *userspace*... *ONLY*. with full POWER compliance in ***BOTH*** userspace ****AND**** kernelspace.
Oh, okay. I wonder how much the softcore's use in watts.
https://openpowerfoundation.org/openpower-summit-north-america-2019-introduc...
a lot, compared to an ASIC - this is to be expected.
l.
I will wait to see your plans and whether you plan to make your own risc-v or openpower processor then.
both.
absolute to-the-letter RV64GC compliance in *userspace*... *ONLY*. with full POWER compliance in ***BOTH*** userspace ****AND**** kernelspace.
Hmm, okay sounds good, thanks for clearing up that mystery. Still I do love that risc-v is supposedly ultra lightweight. Too bad about their dracoian methods... ;/ They really are like the mozilla of processors.
Oh, okay. I wonder how much the softcore's use in watts.
https://openpowerfoundation.org/openpower-summit-north-america-2019-introduc...
a lot, compared to an ASIC - this is to be expected.
At least until the power developers make some ultra lightweight processors right? Whenever that happens...
On 1/4/20, zap zapper@disroot.org wrote:
Hmm, okay sounds good, thanks for clearing up that mystery. Still I do love that risc-v is supposedly ultra lightweight. Too bad about their dracoian methods... ;/
not draconian: just blatantly arrogant.
They really are like the mozilla of processors.
indeed.
Oh, okay. I wonder how much the softcore's use in watts.
https://openpowerfoundation.org/openpower-summit-north-america-2019-introduc...
a lot, compared to an ASIC - this is to be expected.
At least until the power developers make some ultra lightweight processors right? Whenever that happens...
FPGAs automatically come with a power consumption penalty compared to an ASIC, for a given clockrate in both. therefore, if you drastically cut the performance and expectations back far enough, such that power consumption is "acceptable"... then yes.
this does not have anything to do with the ISA (per se).
l.
On Wed, 1 Jan 2020 20:59:47 -0500 zap zapper@disroot.org wrote:
Isn't Risc-V supposedly supposed to be more secure and isn't open power based on the old risc? I am just wondering, if the level of evils risc-v has done, justifies abandoning usage of it when you could just as easily do some risc-v processors with or without their help
The ISA has nothing to do with security. All the meltdown/spectre stuff is implementation, hence why AMD is unaffected by much of it.
and moreover, power, doesn't it require an immense, even crazy amount of watts? I guess my point is, they use more battery power than intel even. And intel is bad enough.
The old Mac laptops basically tried to use a desktop cpu on mobile, kinda what Intel P4 did. It's about design choices, not the instructions. There are a number of lower-wattage Power cpus; using that does not mean the chip will burn lots.
The POWER9 chips IBM offers, 90W for 4-core, 160W for 8-core and 190W for more, they are server chips. Compare to Epycs and Xeons, not to mobile 15W ones. The older Power core in the Wii U uses around 20W. All about choices.
Okay, well you did say openpower, what is that?
IBM released the ISA under open terms, and started the OpenPower foundation to govern it.
Oh really? That's odd... hmm... so you have to abandon risc-v you are saying? Pity that's your only option. I wonder if I should tell others about this.
Just making sure it's clear, the ISA choice is about licensing and extensibility. Risc-v, mips, power are about equal in other ways; in some ways, power and mips are better, like existing compiler support. IOW dropping risc-v is not a loss in ways most users would care about.
- Lauri
arm-netbook@lists.phcomp.co.uk
-
Lauri Kasanen -
Luke Kenneth Casson Leighton -
zap