<[TheBug]>
if a lower amount of sata devices, then there are m.2 to 5 port sata controllers you can get if you wants to use an option which require m.2 without an pcie adapter
<[TheBug]>
None will have it built in, best options will be rk3399 board with an m.2-> pcie x4 adapter (or similar, rockpro64 has an pcie x4 port built on and is only one I know t hat doesn't need adapter board to put card) or a RPi4 CM4 SoM w/ Carrier board
2021-01-26
<buZz>
its got a chip thats bridging the PCIe
2021-01-24
<KotCzarny>
and you can put in pcie-nvme card, which means 3200/2000MB/s transfers
<Ashleee>
any generic -ish PCIe SATA card will work, only problem would be something that requires large address space
<Ashleee>
but no native SATA nor PCIe
<Ashleee>
Pine64 doesn't have PCIe nor USB3
<Ashleee>
rk3399 has mainline PCIe working fine
<Ashleee>
if you want PCIe then not really
<Ashleee>
and PCIe NIC
<buZz>
you could put a PCIe bridge on the bus
<Ashleee>
Allwinner H6 has PCIe but last time I checked the driver was still not working as AW goofed up something in the design
<Ashleee>
you can get rk3399 -based SoCs with PCIe... RockPro64, Orange Pi 4 etc
2021-01-17
<clementp[m]>
MoeIcenowy: do you know if PCI maintainers agree to add a possible deferred_io to PCIe memory space?
2021-01-10
<smaeul>
IIRC, synquacer developerboard may not require a blob, but it has most of its I/O behind a PCIe x1 link and switch, which is why I ignored it when it was new
2020-12-16
<karlp>
apritzel: cmon, what? have you got evidence that the connector is unsuitable? or jsut "omg, its not a pcie socket" ?
<apritzel>
karlp: so that's not the the kind of carefully designed, tested and certified connector like PCIe
2020-12-07
<KotCzarny>
forget about pcie on allwinner tho
<ullbeking>
KotCzarny: how does such Marvel mini-PCIe SATA get installed on an H3/H5/H6 or Rockchip OPi? I have not looked at SBC's for most of 2020 as i've been looking after my daughter full time, and just now getting back into it.
<KotCzarny>
karlp: how about 'if it only had noncripple pcie port' ?
2020-11-20
<MoeIcenowy>
by trying to make the PCIe controller looks normal to OS
2020-11-19
<karlp>
so, this AW device isn't the only one with a "too small" pcie window then right?
<KotCzarny>
i wonder if writing some virtual pcie host that could act as a proxy might work
<jernej>
I heard that Linux once had accessor methods for PCIE but they removed that and replaced with direct access
<bauen1>
still a mini pcie slot makes a board a lot more useful (but might also be another security nightmare)
<apritzel>
bauen1: that's not the H64's fault, the PCIe controller in the H6 SoC is broken beyond repair
<bauen1>
it's such a shame that the h64 doesn't have a (functional) mini pcie slot
<apritzel>
jernej: only legit with that purple patch cable and the useless (and even wrongly connected) mini PCIe slot
2020-10-26
<buZz>
no usb3 and no pcie, boring
2020-10-22
<diego71>
damex: if you are interested in pcie on a H6, you should read this
<damex>
is there actually any supported (upstream) sunxi board that have pcie (working) ?
<mru>
"decent pcie wifi card" is very much not equal to "horrible realtek usb"
2020-09-14
<jernej>
I know heard that BSDs are using accessor functions for PCIE and Linux doesn't (supposedly it did in the past, but they throw them away)
<sunshavi>
Yes. At that time someone said "That happens with a well designed OS.". Cos FreeBSD reported a working pcie
<sunshavi>
jernej: MoeIcenowy got the hint for solving pcie problem from FreeBSD ;)
2020-07-24
<clementp[m]>
MoeIcenowy: Did you get any answer from an upstrem of the H6 PCIe driver ?
2020-07-08
<diizzy>
Speaking if printing, it like I need to make my own case with pcie card support for the rockpro64 :/
2020-05-28
<mru>
7 pcie 4.0 x16 slots
2020-05-15
<mru>
how about a pcie 4.0 nvme?
<plaes>
buZz: h6 has PCIe but it has design flaw and is not usable by default in Linux
<buZz>
none have PCIe thusfar, right?
2020-04-17
<tllim>
when the icenowy's H6 PCIe driver completed and upstream. PINE64 plan to release the PINE H64 model A SBC again and revised the circuit match up with model B.
<MoeIcenowy>
and A is now suspended because of broken PCIe
2020-03-29
<MoeIcenowy>
clementp[m]: try to port dw_pcie_setup_rc() code to mainline?
<clementp[m]>
I made it start with backporting AW quirks on "pci/setup-res.c" last week on your PCIe "STASH" code. And I was able to load the FW and read the MAC address.
<clementp[m]>
one*. But regarding the Wi-Fi over PCIe it could be the same.
<KotCzarny>
h6-pcie-wrapper ?
<MoeIcenowy>
it's really PCIe?
<clementp[m]>
my beelink gs1 have a realtek pcie card
<MoeIcenowy>
clementp[m]: I have no performance-constraint PCIe card now, waiting for shippment of a SATA card
<MoeIcenowy>
wrote a simple hypervisor that wraps PCIe access on H6
2020-03-24
<mirko>
is there currently any way to get the msata slot running under mainline on the orangepi 3? i know about the pcie mess on h6 - i read that there won't be any driver accepted upstream because of the mapping - does that mean no effort happened on a driver at all (whether upstream compatible or not)?
2020-03-12
<clementp[m]>
I'm a newbie on PCie but what I Understood is that hard coding the PCIe memory device mapping in the driver would make it very board specific dependent but could work. Just want to now if someone test this or if even this couldn't work.
2020-03-11
<clementp[m]>
MoeIcenowy: do you have a dirty patch to support PCIe on H6 ?
<micken>
MoeIcenowy: so you think that the pcie version of 8723 is a completely different chip?
2020-01-12
<tuxd3v>
but you don't have the pcie since the way it works is not compatible
2020-01-02
<MoeIcenowy>
damex: tuxd3v: theortically there's a way to make it work -- make a hypervisor that maps the broken PCIe to a normal one
<tuxd3v>
A kernel driver specific to address the way the pcie works in H6
<tuxd3v>
so a new driver is needed if you want pcie
<tuxd3v>
Linux doesn't suport separated zones form adressing and data of pcie
2019-12-03
<mru>
pcie also has spread spectrum
2019-11-06
<willmore>
I'm okay with it missing (broken) PCIe and (never used by me) CSI, but EMAC? I guess it needs an external GMAC then?
<clementp[m]>
Hi, I saw that there is a new H6 called H6 CV200 OS which is an H6 without EMAC, PCIe and CSI. Maybe it will be something like H6T / sun50iw6p2 ?? Does someone has more info on it ?
2019-10-12
<[TheBug]>
ullbeking: well when someone says PCIe I usually go to thinking of ESPRESSOBin to start with, but that SBC is more so (imo) for network or NAS storage platforms, so not sure that would match you needs. That said, it does have a fully working Mini-PCIe port and I use one in 'production' as a NAS with a 4 port SATA card in that mini-pcie slot. You needs though sound more like the NanoPi
2019-10-11
<[TheBug]>
or you saying this is some prototype pcie device and you want to interface it with the SBC?
<[TheBug]>
ullbeking: I mean you still didn't actually state what you were goingf to use PCIe for or maybe I misunderstood?
<hellsenberg>
pcie on sunxi? I thought H6 had useless PCIe?
<[TheBug]>
ullbeking: whats your intended use fof PCIe
2019-07-31
<KotCzarny>
pcie is a powerhog
2019-06-24
<wens>
KotCzarny: I don't think it would matter, at least until they come out with a compute module where you can use the PCIe lanes yourself
<wens>
w/ PCIe connecting the USB 3.0 hosts
2019-06-11
<tllim>
Overall, H6 has the best A53 performance on this class of SoC. However, get a bad reputation due to PCIe cannot mainlining (not broken, still usable and good performance on PCIe-SATA interfacing)..
<KotCzarny>
i dont have pcie card reader to confirm
<MoeIcenowy>
he says that with PCie card reader it doesn't work
2019-04-10
<mru>
thunderbolt is more or less pcie
2019-04-05
<MoeIcenowy>
it's possible for a USB3 controller to utilize 2 PHYs, one for USB2, one for USB3 (and USB3 one may shared with PCIe)
2019-03-31
<[TheBug]>
<tuxd3v> the PCIe shouldnt have support for what I have read.. <- Not in Linux but several have said that PCIe will work in a BSD implementation which will likely show up
<tuxd3v>
the PCIe shouldnt have support for what I have read..
2019-03-25
<KotCzarny>
and pcie
2019-03-14
<jernej>
tbh, I don't really care much about PCIe
<MoeIcenowy>
but PCIe is cursed
<MoeIcenowy>
USB3 and PCIe is new ;-)
<MoeIcenowy>
the main problem is PCIe
2019-03-13
<MoeIcenowy>
and Allwinner do have a lot of free space around PCIe memory space
<anarsoul|2>
anyway, it was pretty weird decision for allwinner to implement pcie like this... My guess is that their SoC is 32-bit inside so they didn't have a choice
<tllim>
however, BSD folk made the PCIe driver works.
2019-03-12
<tllim>
since no developer interest on the H6 PCIe driver implementation due to the limitation, PINE64 drop the miniPCIe slot and in favor on upcoming small RPi size board implementation
<tllim>
@willmore, the H6 PCIe can use for Wireless card, Ethernet, and SATA
2019-02-18
<kilobyte>
not a problem compared to A64 (no PCIe at all), but people tend to use what they get
<kilobyte>
note that PCIe on H6 is broken
2018-12-29
<BenG83>
NetBSD has kind of working PCIe
<jernej>
plaes: IIRC someone said that H6 PCIe issue is not really an issue on one of the BSD variant. So I guess it depends how kernel is designed.
<plaes>
and sometimes BSP does lots of hacks, for example H6 and PCIe
2018-09-06
<MoeIcenowy>
buZz: but on SBCs w/o PCIe only ath9k-htc can be used
2018-07-23
<KotCzarny>
knowing allwinner they will just reuse H6's pcie broken design
<montjoie>
removing PCIE ... why ? :D
<MoeIcenowy>
comparing with model A (the big ver) it removed PCIe and one LED
2018-07-04
<wens>
so that's what the sunxi_pcie_bus_cutpage_* functions in the BSP are for
<MoeIcenowy>
but in virtual memory space of PCIe
<wens>
IO and MEM space are mapped directly to the CPU address space. that's what the ranges property in the pcie controller node describe
<wens>
it seems we are not the only platform that has such a small ATU window for PCIe
<wens>
do you have any patches that simply get the pcie controller to probe
<MoeIcenowy>
the root node of PCIe
<wens>
MoeIcenowy: do you have any preliminary work for pcie?
<wens>
MoeIcenowy: so my guess is that this pcie controller is limited to serving only one device, no IO space support
<MoeIcenowy>
(The description above is only applied to the broken implementation of DWC PCIe on H6
<MoeIcenowy>
the PCIe space is only 0x20000
<MoeIcenowy>
theortically it might be still possible to use 64k PCIe prefetchable memory, use an unstable mux similar to the one in pcie-tango.c
<MoeIcenowy>
it's not the PCIe memory space is only 64k
<wens>
MoeIcenowy: what did you say the PCIe window size for H6 was? 64k?
2018-04-17
<blu-boards>
I also wanted to make more NIC's using PCIe in one design...
<blu-boards>
@ixnus do you have any knowledge about the H6 PCIe?
<Ixnus>
wait, blu-boards: what you need PCIe for ?
<blu-boards>
also per icenowy "I have thought a way to hack it -- use EL2 MMU to wrap the PCIe part; however implementing such a hack is beyond my knowledge level"
<blu-boards>
per icenowy "I have thought to use EL2 to workaround H6 PCIe"
<blu-boards>
so it looks like icenowy is most knowledgable about PCIe...
<blu-boards>
would anyone in sunxi know more info about the H6 and PCIe?
<blu-boards>
And if PCIe will not work, then I will probably skip that SoC
<ElBarto>
I haven't looked at all at PCIe (even H6) so I don't know about the issue
<blu-boards>
@elbarto ok. the wiki states: Allwinner H6 has a quirky PCIe controller that doesn't map the PCIe address space properly to CPU, and accessing the PCIe config space, IO space or memory space will need to be wrapped. As Linux doesn't wrap PCIe memory space access, it's not possible to do a proper PCIe controller driver for H6. The BSP kernel modifies the driver to wrap the access, so it's
<ElBarto>
blu-boards: I haven't quickly followed but PCIe in mainline linux will not be added as a lot of hacks are needed (or something like that)
<ElBarto>
blu-boards: H6 is the only Allwinner SoC with PCIe yes
<blu-boards>
Additionally I did not see any other SoC in the list that has PCIe, but the list is not exhaustive. Is there another one that contains PCIe?
<BenG83>
maybe the guy/girl working on PCIe left :P
2018-03-28
<mripard>
icenowy[m]: can we get the amount of space a device would require when it's detected on the PCIe bus ?
<icenowy[m]>
and for H6 PCIe I will make it black now
2018-03-19
<icenowy[m]>
The ap6356s module which is used on the Allwinner H6 STD design is also capable of both sdio and pcie
<icenowy[m]>
wens: I think most AC card will have PCIe intf
<tkaiser>
This thing has even a PCIe 1.x interface
2018-03-18
<ullbeking>
16:38 <smaeul> and H6 has broken PCIe, so your best bet is to get an H6 board with USB3 and do USB3 to SATA
2018-03-17
<icenowy[m]>
smaeul: I have thought to use EL2 to workaround H6 PCIe
<smaeul>
and H6 has broken PCIe, so your best bet is to get an H6 board with USB3 and do USB3 to SATA
2018-03-16
<icenowy[m]>
montjoie: I have thought a way to hack it -- use EL2 MMU to wrap the PCIe part; however implementing such a hack is beyond my knowledge level
<KotCzarny>
h6 without pcie is just.. meh
<montjoie>
lots of people could want H6 for PCIe, they must know that it is buggy and probably not(never?) supported by mainline
<icenowy[m]>
mripard, montjoie: I'm thinking whether to fill the H6 PCIe box at the mainline status matrix with black
2018-03-14
<tingoose>
No, I am not aware about PCIe
<tingoose>
No, I am not aware about PCIe?
<adj_>
PCIe v2 4 lanes?
<adj_>
I didn't even know that allwinner had a chip with pcie
<adj_>
BenG83_, PCIe in A13?
<BenG83_>
tingoose, you are aware of the PCIe issues?
2018-03-09
<icenowy[m]>
H6 PCIe seems to be too broken
2018-03-08
<tllim>
if you have interest on PCIe, then go to check out the BSP SDK ver 1.1 that I just release on PINE64 wiki two days ago
<KotCzarny>
tllim: pcie, and make it for icenowy
<BenG83>
fixed as in non-broken PCIe implementation
<icenowy[m]>
wens: however all PCIe drivers currently expect PCIe memory space to be directly accessible via {read,write}{b,w,l}
<wens>
icenowy[m]: the dwc pcie driver library uses callbacks for non-dbi read/writes
<icenowy[m]>
I think... maybe PCIe?
2018-03-05
<tl_lim>
actually I get in the process on getting SDK ver 1.1 for icenowy, which have the PCIe driver implementation.
<jernej>
tl_lim: is there any chance that AW would release new revision of H6 with fixed PCIe? :)
<mripard>
maybe you should send an email to the PCIe maintainers
<mripard>
from what my colleagues who worked on PCIe are telling me, it breaks a few assumptions in how PCIe works
<icenowy[m]>
the PCIe hardware on H6 is quirky as only the DBI (root complex) space is always accessible
<icenowy[m]>
mripard: BTW for the H6 PCIe problem do you have any idea to implement it?
<icenowy[m]>
I need to set the PCIe MMIO page register when accessing PCIe MMIO space
<icenowy[m]>
I think I will need this to implement proper H6 PCIe support
2018-03-03
<icenowy[m]>
if it's not a big change to PCIe framework
<icenowy[m]>
wens: but it might needs big change to DW PCIe driver... (for H6 PCIe workaround)
2018-03-02
<icenowy[m]>
wink says that H6 PCIE has bug
2018-02-19
<icenowy[m]>
the dw pcie used by aw seems to be strange
<icenowy[m]>
BenG83: PCIe enumerates
2018-02-18
<icenowy[m]>
oh my pcie-sunxi driver passes build
2018-02-14
<BenG83_>
I have read through the BSP PCIe driver code a bit
<BenG83_>
making the glue layer for PCIe is outside of my skill range :P
<buZz>
there are allwinner socs with PCIe now?
<jernej>
what about pcie? should it offer even better performance?
2018-02-12
<maz>
icenowy[m]: I suspect that the rockchip pcie HW is really a DW IP with some ugly clue on the side...
<icenowy[m]>
it's different to rockchip -- rockchip pcie is a dedicate driver, but aw pcie is dw pcie with glue
<BenG83>
is the H6 PCIe controller/phy likely to work with the vanilla synopsis driver?
2018-02-11
<icenowy[m]>
MoeIcenowy: pcie never works
<BenG83>
t3st3r, I think it depends if the USB3 and PCIe are bandwith limited
<BenG83>
PCIe?
2018-01-29
<BenG83>
willmore, the mPCIE has PCIE, USB and I2C
<ElBarto>
willmore: H6 have native PCIe
2018-01-27
<lurchi_>
IIRC crossing of PCIe pairs is allowed
<BenG83>
not sure if a single PCIe 2.0 lane with SATA will be much faster than USB3
<embed-3d>
I see more chances in the PCIE. Pcie with a sata adapter/transreciever can speed this thing up!
<icenowy[m]>
I think the first two batches are all buggy on PCIe
<icenowy[m]>
P.S. some batches of Pine H64 sample boards have PCIE TX and RX reverted, which will make it not work at all
<BenG83>
looking into the PCIe stack on ARM64 right now
<icenowy[m]>
smaeul: I think PCIe-equipped models will need a longer waiting time
2017-12-31
<skiboy>
Even though pcie x1 wouldn't really run any modern graphics cards, it can drive a few sata connections and make a low-cost RAID NAS
<KotCzarny>
who knows, but keep in mind it will be probably pcie x1 link only
<skiboy>
If it has pcie... Could you actually wire up a way to get an external GPU to work with it?
<KotCzarny>
with usb3 and pcie
2017-12-28
<KotCzarny>
no usb3, no pcie, sucky :/
2017-11-24
<icenowy[m]>
but they're not so suprising -- no PCIE ;-)
2017-11-22
<tkaiser>
beeble: With your RK3399 design... did you ever test PCIe 'performance' (asking since I wonder how to test this -- my only approach would be to use a controller known to saturate PCIe lanes and measure a layer above, eg. SATA)
2017-11-15
<icenowy[m]>
USB3 + PCIe on H6
<tllim>
@willmore, how about PCIe?
<icenowy[m]>
so if PCIE WWAN is used one USB Type-A will be not usable
<icenowy[m]>
tkaiser: when talking to tllim he says one USB controller is reused between USB Type-A and PCIE USB pins
<icenowy[m]>
montjoie: another dw part for pcie :-)
<montjoie>
the PCIE seems really low documented
2017-11-14
<icenowy[m]>
the two small variants have no pcie
2017-10-12
<buZz>
H6 has usb3 and PCIe ?
2017-09-20
<MoeIcenowy>
and as steven claims it's real PCIE
2017-08-11
<montjoie>
pcie!!!
2017-08-05
<BenG83>
H6 will have PCIe
<Seppoz>
or has pcie
2017-07-19
<jernej>
but at least it is nice to see USB3, PCIE, HDMI2.0 with HDR, VP9 HW decoder and more
2017-04-12
<tkaiser>
If performance looks ok with a disks, do a 'Pro' board using Marvell 88SE9215 (4 SATA ports, FIS based PM support, stupid anyway behind a single PCIe 2.x lane)
<tkaiser>
That _might_ change with H6 in Allwinner land. Add an ASM1061, check performance. If it's worse than now with A20/R40 then Allwinner's PCIe implementation is broken.
2017-04-04
<beeble>
by the way, offtopic any pcie2 pcie nic recommendations? since i had only some realtek ones laying around during testing
2017-04-03
<tkaiser>
beeble: Sure, but there is already one PCIe controller on the board and there's also a SIM card slot so I would assume the mPCIe should be used for 4G stuff.
<beeble>
and the pcie based ssds are m-key
<tkaiser>
beeble: Yes, it's not M.2 but since there seem to be no PCIe cards available for this slot (only SSDs) this is IMO the better choice. So without a converter...
<beeble>
i know nobody cares about mini-itx boards here. but our evk baseboard has a full size pcie connector if ever needed
<beeble>
tkaiser: yes, one x4 pcie 2.1
2017-04-02
<tkaiser>
beeble: IIRC you said RK3399 has 4 PCIe lanes?
2017-04-01
<MoeIcenowy>
so let's hope they can keep PCIE and USB3
2017-03-31
<KotCzarny>
ahm, rk3399 has only one pcie
<MoeIcenowy>
but a SATA (I remember RK3399 doesn't have native SATA -- maybe they converted PCIE to SATA
<hojnikb>
sata bridge on pcie
<hojnikb>
hmm i wonder how many pcie lanes are there
<hojnikb>
usb3 and pcie ?
<KotCzarny>
i wonder how having usb3 and pcie would affect max/idle power draw
<MoeIcenowy>
but also pcie!
<KotCzarny>
pcie?
2017-03-29
<TheLinuxBug>
+ pcie
2017-03-23
<KotCzarny>
pcie bw is quite a lot
<tkaiser>
TheLinuxBug: Nope, PCIe controllers can implement as many SATA ports as they want, no PM involved
<KotCzarny>
what port multipliers? in comments someone notices it's 3 pcie lanes (2 drives on each)
2017-03-21
<Ke>
or attach rpi to PCIe and use as gpu!!!!!
<Ke>
there was that one Asus server board that people were interested in that had full pcie
<apritzel>
Ke: take the RK3399 and connect a PCIe graphics card :-D
2017-03-13
<TheLinuxBug>
currently trying to see if I can get it to recognize a pcie wifi card
2017-02-24
<BenG83>
and if the PCIe card reader is not mounted
2017-02-06
<Ke>
but the pcie is only 4x, so you will need physical modifications
2017-01-20
<tkaiser>
beeble: Off-topic: Apple's Ethernet Thunderbolt adapter is not even using TB but PCIe instead ;)
2017-01-08
<MoeIcenowy>
it recommends to use PCIE SSD or USB3.0 UAS
<KotCzarny>
i would like using pcie for sata
<MoeIcenowy>
just use a PCIE channel to wire a new network card
2017-01-04
<beeble>
tuxillo: yes,i use them during development to test minipci cards. just be sure to check you minipcie wan card is actually a usb card and not a pcie
2016-12-14
<tkaiser>
KotCzarny: 3 MacBook Pro lying around, none of them has PCIe slots ;) And the 'PCs' are Mac Mini ;)
2016-12-08
<tkaiser>
KotCzarny: better ask google. I also don't know whether the MediaTek SoC can saturate a PCIe lane or cope with high IO bandwidth. And since it's the worst vendor ever we will not know until someone bought this stuff and tested it.
<tkaiser>
KotCzarny: The SATA ports on this BPi-R2 are most probably provided by one ASM1061 hanging off a single PCIe 2 lane
<tkaiser>
KotCzarny: Which tablet or OTT Box needs PCIe?
2016-11-18
<tkaiser>
KotCzarny: Nope, unlike ARMADA 38x you can't do SERDES voodoo there and choose between SATA and PCIe. It's PCIe. So not one SATA but up to two SATA ports using the right controller in the slot
<tkaiser>
KotCzarny: the MiniPCIe there is really PCIe only. So no simple switch but mPCIe cards with 2 SATA ports and an own controller exist
2016-11-16
<MoeIcenowy>
I don't think it will depend on pcie
<silviop>
in menuconfig is not configured , but it depends of toshiba vebdor that depend from pcie , and my tablet does not have that
2016-11-15
<hojnikb>
you don't need pcie for nvme :)
<tkaiser>
hojnikb: USB3 and PCIe on a $7 device? Really?
<miasma>
i guess emmc is still cheaper than the pcie ssd modules
<hojnikb>
you can run other stuff via pcie and usb just as well
<hojnikb>
use the same formfactor but something like pcie or usb3 instead
2016-11-11
<apritzel>
beeble: the SoC has 4 PCIe lanes, right?
2016-11-10
<KotCzarny>
beeble, marvel boards feature minipcie ports, i wonder if pcie card would work in arm env
<beeble>
hard for me to think about a soultion that will be competitive against a x86 solutions with standard gpus attached via pcie
2016-11-04
<tkaiser>
But according to rumours he's working on a non-sunxi board with native SATA and IIRC also PCIe
2016-09-26
<apritzel>
willmore: btw, there are some schematics for the Armada board, and the mini-PCIe slot has both proper PCIe and USB connected
2016-09-23
<apritzel>
tkaiser: best thing: seems like there is a dev board .dts in the kernel already, with USB3, PCIe and SATA enabled
2016-08-31
<lvrp16>
you mean lack of pcie and usb3?
2016-08-13
<MoeIcenowy>
for example rtl8723au is usb, rtl8723bs is sdio, rtl8192ee is pcie\
<MoeIcenowy>
"s" is SDIO and "e" is PCIE
2016-08-09
<MY123>
has a full sized PCIe 16x slot
2016-08-08
<tkaiser>
Amit_T: Multiple SERDES lines that can be configured in u-boot to be either PCIe or SATA
<Amit_T>
apritzel: Does it ARMADA 38x SoC has pcie controller built-in ?
2016-07-20
<KotCzarny>
as fast pcie cam be
<vpeter>
pcie can act as sata too - I have 2.5" drive connected there.
2016-06-24
<fire219>
a cheap SBC with an M.2 NVMe (PCIe) slot on it... i'd like to see that
2016-06-21
<maz>
lvrp16: hopefully you'll route the PCIe lines to proper connectors? because that's what is really interesting about the rk3399.
2016-06-01
<apritzel>
but you get DIMM sockets, x16 PCIe, SATA ...
2016-03-29
<lvrp16>
tkaiser: the 10GBe and PCIE is the biggest power consumption, i'd imagine ~5-10W idle though
2016-01-27
<tkaiser>
H3 has no PCIe, just SDIO and RGMII
2015-09-14
<swiftgeek>
so if you find interesting platform with (lots of) PCIe that isn't arm or x86 ... and if you like soldering
2015-02-11
<quitte>
mripard_: true. but Steven did not specify allwinner. i.MX6 has PCIe
<mripard_>
quitte: there's no PCIe on the Allwinner SoCs.
<quitte>
of course it is. SPI is not exactly slow. But PCIe is probably the most sensible choice
2014-12-22
<Turl>
but we don't have PCIe so it's moot :)
2014-11-10
<hramrach_>
similar for general peripherial connection - clear win in long run is USB3 or PCIe but some current SoCs don't support either
2014-11-03
<topi`>
wens: A31 doesn't have PCIe support. The slot just houses a 3G modem and supplies it with the USB signaling
<wens>
iirc, that isn't true m-pcie
2014-10-14
<libv>
last i heard, amd still has broken pcie
<hramrach_>
nVidia was making PCIe stuff before it started in the ARM business so it does not count
<ssvb>
hramrach_: the ARM chip makers are already including PCIe
<hramrach_>
hmm, I wonder how long before ARM chip makers coming from mobile start including PCIe
<hramrach_>
TB is at least clear on including everything and the kitchensink but then you cannot implement fully compliant TB on anything less than Tegra. it's supposed to have at least DP and PCIe multiplexed on the connector
2014-07-12
<CaptHindsight>
did anyone ever come across an allwinner or any ARM tablet with either ethernet or PCIe?
2014-06-30
<cubear>
JohnDoe_71Rus: I don't know about the sata, probably you could do something with the mini-PCIe expansion slots