<ZipCPU>
... and here I thought you had given it that name.
<q3k>
no, this was the typical 'ugh just name it whatever', as this was me experimenting with zfs on my shiny new NVMe
<q3k>
then that pool became production. whoops.
gruetzkoa has joined ##openfpga
<ZipCPU>
I'm about to head to bed here. Can't figure out how it is that you are still up.
gruetzkoa has quit [Excess Flood]
<q3k>
i can't either
<ZipCPU>
I spent my evening applying formal methods to Vivado's self-generated/get-you-started/demo AXI-lite code. Talk about how *not* to build an interface. :D
gruetzko- has joined ##openfpga
<shapr>
I spent my evening trying to figure out how this docker container runs test code from this dir, after I delete the test code.
<shapr>
My evening has not been successful
<q3k>
ZipCPU: yeah, it was somewhat hard to follow when I had to hack on some
<ZipCPU>
It's not just the hard to follow issue. The performance of this thing is just pitiful.
<q3k>
hm
<q3k>
do you have a timing diagram?
<ZipCPU>
Sure ... I just have to switch from "mode bmc" to "mode cover" :)
<q3k>
yep
<q3k>
i'm curious how it compares to my artisanal handcrafted axi lite peripherals in hardflag
<shapr>
I get the feeling that the FPGA tools just try to outdo the competition, not actually perform well
<ZipCPU>
One difference is that this thing can't return two valid requests in a row.
* ZipCPU
doesn't think this thing is outdoing anyone's competition
* ZipCPU
doesn't think highly of AXI in general.
<q3k>
ZipCPU: oh
<awygle>
Helps to have no competition
<q3k>
ZipCPU: are you talking about the interface design, not the implementation?
<ZipCPU>
No, I'm talking about the demo implementation.
<q3k>
oh.
<ZipCPU>
The AXI design itself should be decently fast.
<q3k>
well, save for the fact that axi lite has no bursts
<q3k>
but it's still somewhat pipelineable
<ZipCPU>
It'll burst if you want it too--just feed it fast enough ;)
<q3k>
but that does depend a lot on the implementation
<q3k>
yeah
<awygle>
axi lite and axi stream make way more sense to me than vanilla axi
<ZipCPU>
awygle: I think I've put in between 2wks and a month trying to build a full AXI formal property list. I'm not sure, after all that work, that I'm any closer to having one. AXI-lite I have, full AXI? not so much.
azonenberg_work has quit [Ping timeout: 252 seconds]
<awygle>
I just don't understand the use case. Configuration registers? axi lite. data flow? Axi stream. I guess axi makes sense specifically for talking to memory?
<q3k>
yes.
<q3k>
as a main system bus.
<awygle>
which I basically never do, so *shrug*
<awygle>
just doesn't seem very fpgaish
<ZipCPU>
awygle: Not the term I would use. I would say it doesn't seem well tuned to low logic applications--such as all of the cheap FPGA's I keep working with.
<q3k>
it is a fat pipe.
<awygle>
this all goes back to "I don't get why y'all keep wanting to do CPUs on FPGAs" is all
<q3k>
awygle: it's fun
<ZipCPU>
awygle: Not necessarily CPU's, AXI has a lot of memory features to it as well. Features that could be appropriate for some complex algorithms.
<awygle>
fair enough :-D
<q3k>
awygle: well. a control cpu does make sense for fpga, but that can be pretty slow and doesn't need a fat pipe to a dram controller
<jn__>
as someone who knows software much better than hardware, the ability to write software draws me to CPUs
<q3k>
awygle: nvidia sprinkles a few dozen falcon cores in every chip, so i might as well sprinkle in a few picorv32s and pretend that makes me a professional
<ZipCPU>
The reconfigurability of a CPU within an FPGA makes CPU's easier to work with.
* ZipCPU
steps away
<awygle>
gnight
esden_cloud_ has joined ##openfpga
m4gul0__ has joined ##openfpga
oni_ has joined ##openfpga
scrts has joined ##openfpga
forrestv_ has joined ##openfpga
moho2 has joined ##openfpga
AlexDaniel-old[m has quit [*.net *.split]
m4gul0_ has quit [*.net *.split]
forrestv has quit [*.net *.split]
esden_cloud has quit [*.net *.split]
Adluc has quit [*.net *.split]
oni has quit [*.net *.split]
finsternis has quit [*.net *.split]
moho1 has quit [*.net *.split]
cyrozap has quit [*.net *.split]
Ekho has quit [*.net *.split]
forrestv_ is now known as forrestv
m4gul0__ is now known as m4gul0_
cyrozap has joined ##openfpga
finsternis has joined ##openfpga
<rqou>
offtopic: awygle: be glad you _don't_ get the eecs-grads-misc mailing list right now :P
Ekho- has joined ##openfpga
<awygle>
lol I'm always glad about that
Adluc has joined ##openfpga
<rqou>
my .edu email has been bombarded by a giant pile of unionization/contract discussion for about a week
<rqou>
hey, getting dissertation talk announcements can be cool (although i don't really have time to attend them since i got a "real job")
AlexDaniel-old[m has joined ##openfpga
<awygle>
oh that's actually cool, good for them
<rqou>
ugh afaict the relationship between the university of california, the UAW, and the GSIs/GSRs seems to be an absolute shitshow
scrts has quit [Ping timeout: 265 seconds]
<Ultrasauce>
I'm pretty sure that's the case at most institutions
scrts has joined ##openfpga
rohitksingh_work has joined ##openfpga
wpwrak has quit [Read error: Connection reset by peer]
wpwrak has joined ##openfpga
azonenberg_work has joined ##openfpga
s1dev has joined ##openfpga
massi has joined ##openfpga
s1dev has quit [Ping timeout: 256 seconds]
rqou has quit [Quit: ZNC 1.7.x-git-709-1bb0199 - http://znc.in]
rqou has joined ##openfpga
futarisIRCcloud has joined ##openfpga
scrts has quit [Ping timeout: 240 seconds]
_whitelogger has joined ##openfpga
scrts has joined ##openfpga
Sinclair2 has joined ##openfpga
Sinclair2 is now known as Sinclair
Sinclair is now known as Sinclair2
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
sunxi_fan has joined ##openfpga
Hamilton has joined ##openfpga
scrts2 has quit [Quit: Page closed]
Hamilton has quit [Remote host closed the connection]
rohitksingh_work has quit [Read error: Connection reset by peer]
sunxi_fan has quit [Ping timeout: 272 seconds]
Sinclair2 has quit [Quit: Bye Bye]
CoffeeFlux has quit [Ping timeout: 265 seconds]
CoffeeFlux has joined ##openfpga
CoffeeFlux has quit [Changing host]
CoffeeFlux has joined ##openfpga
CoffeeFlux has quit [Ping timeout: 240 seconds]
CoffeeFlux has joined ##openfpga
CoffeeFlux has quit [Changing host]
CoffeeFlux has joined ##openfpga
sunxi_fan has joined ##openfpga
rohitksingh has joined ##openfpga
azonenberg_work has quit [Ping timeout: 264 seconds]
sunxi_fan has quit [Quit: Leaving.]
azonenberg_work has joined ##openfpga
massi has quit [Remote host closed the connection]
oni_ is now known as oni
oni is now known as Guest45708
Guest45708 is now known as oni
gruetzko- is now known as gruetzkopf
digshadow has quit [Ping timeout: 240 seconds]
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined ##openfpga
ym has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
digshadow has joined ##openfpga
m_t has joined ##openfpga
X-Scale has quit [Ping timeout: 272 seconds]
gnufan1 has joined ##openfpga
X-Scale has joined ##openfpga
<awygle>
Security People - if i get a Yubikey or whatever and use U2F or whatever, what happens if i lose it? what's the recovery scenario like?
<sorear>
You want to buy more than one, attach all of them to your accounts, and keep at least one in a deposit box or similar
<awygle>
hm, i spose that makes sense
<awygle>
there must be a low-tech recovery path though right? (yes, this is also an obvious security hole, but it still seems like something that must exist)
<zkms>
awygle: depends on the service in question, some will let you download some kind of "recovery code" that you are to keep somewhere very secure or w/e
<zkms>
some will straight-up allow reset via SMS which is lulzy and shitty
* awygle
weeps for a lack of standards
<awygle>
okay well those seem reasonable. thanks sorear, zkms
<zkms>
(2FA and/or password reset via SMS is extraordinarily bad because anyone can go to a cell provider store and be like "hi i'm awygle, my phone got stolen and I bought a new phone but I need a replacement SIM card" and they will straight up deactivate your SIM card and give the impostor a replacement)
xdeller has joined ##openfpga
<awygle>
dang the yubikey usb-c thing is _tiny_
<sorear>
On the Versa board docs, does 64 Mb/x16 mean 128MB total?
<awygle>
i think x16 is the interface width? not sure tho. there's definitely only one physical chip.
<azonenberg_work>
zkms: do they seriously not at least check for ID?
<awygle>
yeah the user guide says "includes a 16-bit wide memory controller interface" so i'm pretty sure that's what it is
<zkms>
azonenberg_work: they're /supposed to/, but there's no technical factors preventing that
<azonenberg_work>
i thought it normally was done over the phone and required some level of social engineering
<azonenberg_work>
Rather than just walking into a store
<zkms>
azonenberg_work: so you either 1) find an insider (which there are, this is well established phenomenon) or 2) social engineer someone into not doing the ID check (you can give excuses like..."my wallet got stolen along with the phone")
<azonenberg_work>
zkms: in any case, yeah sms 2fa is awful
<azonenberg_work>
at least its better than no 2fa at all
<azonenberg_work>
it requires some minimal level of targeting rather than just scooping up password hashes and throwing them at a couple of gpus
<rqou>
yeah after the recently published news articles about sim swapping people at $WORK were pretty pissed
<rqou>
since The Phone Company (the other one, not the one with the globe logo) are the corporate overlords
<sorear>
The sphere with the missing piece company or the other sphere with a missing piece company
<rqou>
lol i actually don't know which companies you're referring to
<rqou>
my corporate overlords are the red checkmark phone company
Miyu has quit [Ping timeout: 252 seconds]
<awygle>
you're allowed to say the word "verizon"
<rqou>
except Verizon is rather "not popular" at this moment
<awygle>
or, um, any moment?
<gruetzkopf>
is any telco company popular?
<awygle>
no
<gruetzkopf>
Deutsche Telekom (who are not using the tmobile brand in germany these days) is definitly unpopular
* cpresser
started #reworkctf
<cpresser>
i had to try placing the 01005 resistor
<cpresser>
actually not that hard under the microscope. same technique as with 0603, just touch both pads at the same time with the soldering iron-tip
<gruetzkopf>
am i that weird for doing 0402 pad after pad?!
<gruetzkopf>
(without a microscope and with chinesium-grade tweezers)
<sorear>
rqou: Death Star in relief company sells phone plans, Death Star in profile company sells phones and apps
<rqou>
ah i usually hear that second one referred to as the fruit company
<q3k>
gruetzkopf: you mean 0402 imperial? yeah, i also do it pad-after-pad
<q3k>
gruetzkopf: same for 0201, but then I do use my shitty microscope
<q3k>
haven't joined the 01005 club yet
<gruetzkopf>
i've done about a roll of 0402 imperial by hand..
<q3k>
0402 is my go-to size for SMD these days, so yeah, did quite a bit of these by hand myself too
<q3k>
although not realy 4k :P
<q3k>
*not nearly
Bike has joined ##openfpga
gnufan1 has quit [Quit: Leaving.]
<azonenberg_work>
q3k: yeah i default to 0402
<azonenberg_work>
unless spatial constraints force me smaller or capacitance/power constraints force me larger
<azonenberg_work>
cpresser: are you doing the whole reworkctf live? are you tweeting etc pics?
<azonenberg_work>
gruetzkopf: my most 1337 soldering was probably putting 30ga kynar wire onto adjacent pins of a 0.5mm TQFP to probe it
<azonenberg_work>
... with a radioshack cigarette-lighter grade soldering iron purchased out of some cheap shop on apliu street
<azonenberg_work>
In a back room of some random office in hong kong
<gruetzkopf>
ouch
m_t has quit [Quit: Leaving]
<cpresser>
azonenberg_work: I will do the whole board next week on emf-camp. and some other people will give it a try as well. and yes, some pics might show up on twitter
<gruetzkopf>
(reading datasheets) someone at ti has the same miniusb cable i have