ESP32 Forum

Support for external RAM isn't in the current esp-idf yet; we have a branch that does this but because of some cache bugs, it needs a fair amount of workarounds to work on current (v1) silicon.

hi jeroen,
no problem. so i know it.

i will play with the RAM on CMD and SPI in the meantime

best wishes
rudi

Support for external RAM isn't in the current esp-idf yet; we have a branch that does this but because of some cache bugs, it needs a fair amount of workarounds to work on current (v1) silicon.

With workarounds performance/stability is OK?

Stability seems to be okay. Performance is impacted a bit; can't tell you how much (also depends on what you're trying to do) but I'd guess it's in the 10-20% range. We plan to release the branch that supports psram eventually; I'll also make a post (or we'll integrate a document in esp-idf) detailing the ins & outs of what's going on and what to watch out for at that time.

eventually

Sprite is always so conservative in his timeline, the anti elon musk.

Hope it is soon since psram is sampling and pycom OEM supposed to ship this month. Performance reduction not too bad. Is there plan to fix this in rev2 and where is that on roadmap? 2018?

Hehe, as long as it's not going to be Duke Nukem Forever, I think I'm still reasonable It's kinda hard to say when, gotta sanitize the PSRAM branch, write documentation etc in order to make everything public. I estimate it's gonna happen this or next month, but please don't hold me to that.

We will indeed be releasing a new chip in which psram should work without any workarounds, plus some other improvements (it's not a direct new rev of the ESP32): my guess is that that chip will see the market within this year. Again, those are guesstimates, so take with a grain of salt.

We plan to release the branch that supports psram eventually; I'll also make a post (or we'll integrate a document in esp-idf) detailing the ins & outs of what's going on and what to watch out for at that time.

would love to put hand on this jeroen asap

...It's kinda hard to say when, gotta sanitize the PSRAM branch, write documentation etc in order to make everything public. ...

chance for a untested preview for a play? ( link is welcome )

.. We will indeed be releasing a new chip in which psram should work without any workarounds, plus some other improvements (it's not a direct new rev of the ESP32): ...

ESP64 ?



best wishes
rudi

No proper datasheet yet... coming pretty soon though. It's RAM, so if any, you do not have pages as such, and you do not need to erase them; you can just write different data over them. I think I can copy/paste some data here: be warned that's not from a released or finished datasheet yet, so I don't guarantee these figures are correct.

hi Jeroen,

just in time tested CMD READ_ID and CLK out 4 times 0x00 for a test
cause not sure how long is ID.

Can it be ID is only 1 Byte 0x5F ( 1b1011111 )

how looks this to you for READ_ID
for the first try on READ_ID

best wishes
rudi

edit:
Read ID change at each start

btw it looks like a small typo:
From the Table
SPI write command option must be 0x02 right ? until the chapter appears in technical reference,
we continue to play a little bit in the meantime.
play with 1.8 v and 3.3 v
and pullups in the lines.

eventually

Sprite is always so conservative in his timeline, the anti elon musk.

Hope it is soon since psram is sampling and pycom OEM supposed to ship this month. Performance reduction not too bad. Is there plan to fix this in rev2 and where is that on roadmap? 2018?

"WiFive" (FiPy) - you are from preferred pycom team? i hope not.
( personaly i have nothing gen pycom team but for the Art of preferring how they become all infos and we engineers chewing on the fingernails here .. and we have to find everything and roll the dice.

... rev2? 2018?

i am happy after rev0 for get rev1 in the month june 2017 - not july - july is to late - ready WROOM32 and WROVER with functionally pSRAM and ESP-IDF on the market is this unrealistic @jeroen @ivan ?
so we do here then a big pause do 2018 for sure, do not go standby, rattle us out and then concentrate on other things.

We would like to first of all the everything here want to implement in ESP32 (rev0/rev1) of which we are talking since esp31b .
please - no new promise on new chips with recurring rev0 rev1 . these are not good prospects.

(Just see byself my euphoria dwindle)

push!
my last push and then close the theme

i want work this weekend with this!
not later.


ESP-PSRAM32
372016

= produced 37 week in 2016

since 37 KW 2016 the ESP-pSRAM is here
since 50 KW 2016 is it on ESP Wrover and you play oficially with it

now we go into 21 KW 2017

36 weeks - and no datasheet? no working info?
only "wrong" CMD infos - why? why?

bad - very bad guys
this is the badest thing i ever feel in my sympathy last 3 years for espressif