Messages

1

General CPU Discussion / Re: CPU Performance DD2.2 vs DD2.3 (v2)

« on: May 02, 2020, 07:56:20 pm »

Since the stepping is specific to the chip (Nimbus in this case), I think it would make sense to add your findings to the Steppings secton of the POWER9 page; if we start to get too-much Nimbus-specific information, it can always be split off into a separate page.

As an aside, the CP9M## is not actually a part number, it is (as far as I can tell) just a SKU for Raptor to manage their inventory; according to ClassicHasClass you can see the actual part number in lshw. We are actually missing the parts number for the CP9M31 SKU in the decoder table on the Sforza page, if you could add what yours is that would be excellent.

2

General Discussion / Re: The pride of being Power!

« on: April 29, 2020, 09:51:19 am »

Thank you for your confirmation of your CPU model info, I was trying to add these to a list on the Sforza wiki page, but it has been guesswork until now. If others could check via lshw what their CPU models are, that could help flesh out the list.

3

Mod Zone / Re: Convert GPU PCI-E X16 to X8

« on: April 21, 2020, 05:19:32 pm »

Sforza's lanes are the highest so far in the 3 models. Sforza has 48 PCI-e lanes against 42 LaGrange …

Nimbus Module	PCIe 4	OpenCAPI 3
Sforza	48	N/A
LaGrange	42	16
Monza	34	48

I put that into a bbcode table for easier reading, though a more detailed version exists on the POWER9 RCS Wiki page. I would also note that OpenCAPI starts with version 3, as it continues the CAPI numbering. The Sforza modules, and therefore Talos II & Blackbird actually have CAPI 2 support, as it runs on top of PCIe 4 physical interface.

Please also get rid of all SATA ports. SATA is now a thing of the past or replace SATA with SAS (but I do not think it would be a good idea because of cost factor).

I also think the board should come with at least two M2 port. One is for NVMe SSD and the other for Wifi.

Maybe not in future boards, but I am definitely happy to see SATA support on Blackbird. For those of us with limited resources, being able to use cheaper SATA SSDs or even HDDs makes getting into OpenPOWER that much easier; especially when already dealing with the hard requirements of:
• non-commodity mainboard
• non-commodity CPU
• non-commodity heatsink
• more expensive ECC RAM
Having cost savings possible somewhere, such as in storage, is a godsend.

I should note, the reason for SATA rather than SAS is partly due to the SATA controller having open source firmware actually I think it has no firmware at all. I remember reading somewhere that there is little to no hope of the any SAS controller being clean-room reimplemented in open source code, unlike the NIC.

4

Talos II / Re: Imprecise EATX terminology considered harmful - SSI-EEB

« on: April 19, 2020, 11:18:56 pm »

Is that an April Fools joke? It looks like a Mac Pro ripoff.

It appears to be a real product, and it is also definitely a Mac Pro rip-off. Linus Tech Tips made a video with a prototype a while ago, but I have not heard anything recently about it.


Returning to the actual topic, Talos II board size and mounting/standoff locations, it would be helpful to hear something official from Raptor Engineering/Computing before I start making corrections to the wiki.

I am also purely curious about…

There does appear to be an absence of components where the C mounting location would be. I am quite curious if someone from Raptor could answer, why was this non-EEB B/C′ location chosen? Are there trace routing issues that I cannot see preventing the use of location C?

5

Applications and Porting / Re: Google Earth?

« on: April 19, 2020, 10:53:17 pm »

Marble is indeed interesting, but in my experience it has been very unstable; last time I experimented across Mac/Windows/Linux, on each platform half of the basemaps would not work, and the set of non-functioning maps varied between OSs.

I have not looked at it recently, but the USA's NASA had an open source 3D globe program of their own called World Wind; that could potentially be worth investigation.

For everyday users however, GNOME Maps is probably a better solution. While it is 2D only, it does have substantially better satellite imagery than Marble or World Wind due to a partnership with Mapbox.

6

Talos II / Re: Imprecise EATX terminology considered harmful - SSI-EEB

« on: April 03, 2020, 10:44:09 pm »

Looking at the board images on the wiki, it looks like the Talos II mainboard uses the SSI EEB alternate F′ and J′ mounting locations from SSI EEB 2011 v1.0.1, but the C mounting location is missing. Instead we have a location between A and C. Old versions of SSI EEB seem have this location as it seems to date back to the original AT specification; it is labelled “G” in version 1.0, and unlabelled in version 3.0, before being completely removed in version 3.5. This was before SSI relabelled their mounting locations to match ATX; were they to add it back now, as an alternate to location C, it would probably be labelled C′.

The ATX specification does have this missing mounting location, but calls it B, and only uses it for microATX and in earlier versions miniATX. To quote the ATX v2.2 specification, “The hole at B in Figure 2 was previously required along the rear edge of the board. This location is not required for ATX designs.” Older versions of ATX also mention the F′ mounting location, calling it E instead.

Sounds like to be certain it will fit the Talos II board, one should look for an SSI EEB chassis (for mounting locations F′/E and J′) with microATX support (for mounting location B). As Gamers Nexus points out, chassis with “EATX” marketing might not support the alternate mounting locations, of which Talos II needs F′ and J′. The only issue with a properly complaint SSI EEB chassis would therefore be the potential absence of the microATX mounting location B.

There does appear to be an absence of components where the C mounting location would be. I am quite curious if someone from Raptor could answer, why was this non-EEB B/C′ location chosen? Are there trace routing issues that I cannot see preventing the use of location C?


Sources for specifications:
SSI-EEB 2011 v1.01 - most recent EEB spec, I think
SSI-EEB 2008 v1.01 - contains revision history, unlike 2011 v1.01
SSI EEB v3.51 (from the Internet Archive)
SSI EEB v3.0 (from the Internet Archive)
SSI EEB v1.0 (from the Internet Archive)
ATX v2.2 (from the Internet Archive)
ATX v2.1 (from the Internet Archive)
ATX v2.03 (from the Internet Archive)

7

General Discussion / Power Architecture enthusiasm & SARS-CoV-2/COVID-19

« on: March 31, 2020, 07:59:14 pm »

…  I have always been a supporter of the X86 alternatives, mine is also a battle against [the popular assumption] that X86, just because it is the most used, means that it is the best.

I can certainly understand that. There are many less-popular specifications, from SAS to DisplayPort to IEEE-1394/FireWire/i.link that I find myself sympathetic to, where much of the opposition seems to be due to existing popularity/install-base. I only hope Power ISA’s future is more like DisplayPort than FireWire.


It looks like the USA have different policies per-state, but Illinois (where I think Raptor is based) does have computer companies marked as exempt from the stay-at-home order, in Section 1 Subsection 12.n.
https://www2.illinois.gov/Pages/Executive-Orders/ExecutiveOrder2020-10.aspx

8

Talos II / Imprecise EATX terminology considered harmful - SSI-EEB

« on: March 31, 2020, 04:53:13 pm »

I was initially quite confused at the number of standoff and case size issues I have heard regarding the Talos II mainboard. For examples, see the RCS Wiki’s Hardware Compatibility List for Talos II, which has an entire section for problematic cases:
https://wiki.raptorcs.com/wiki/Talos_II/Hardware_Compatibility_List#Problematic_Cases

However, Gamers Nexus recently published a lengthy article & video about the ambiguity inherent in the “EATX” marketing, which seems to be the explanation I was looking for:
https://www.gamersnexus.net/guides/3566-e-atx-is-a-lie-vs-xl-atx-eeb-ceb
https://youtube.com/watch?v=54VJwwLRJBk
Apparently, the real underlying specification for elongated ATX boards with dimensions of 305 mm × 330 mm (12” × 13”) mainboards is SSI-EEB, whereas EATX is a marketing term that encompasses a wide variety of board sizes whose front-to-back measurement is greater than 244 mm (9.6”) but less than or equal to 330 mm (13”).

I think it would be extremely helpful to customers to make it clear that Talos II is an SSI-EEB mainboard, with a footnote warning that while some case manufacturers may market EEB cases as EATX or E-ATX, cases incompatible with EEB may also be marketed as EATX/E-ATX.

For the curious, SSI documents appear to have been moved to the SNIA’s website:
https://www.snia.org/standards/industry-reference

9

General Discussion / Re: The pride of being Power!

« on: March 16, 2020, 01:40:46 pm »

Hmm, I suppose to me praising a company or expressing pride in an architecture seems like a very strange thing to do -- I wish I knew a more delicate way to say it, but what you wrote sounds to me somewhat fanboy-ish or like corporate PR.

I think I share your enthusiasm for Power/PowerPC, and it does make me happy to see Power architecture being put to work and proving itself, but I cannot imagine myself describing that as "pride", nor Power as "beloved". Your wanting to "pay homage" to a company/product also sounds... religious? As if you are a worshiper of Power or IBM.

You write that you are in Italy, so I suspect that something is being lost in translation from Italian?

I point this out not to ridicule your English, but I think your writing *might* be a component of why there were no other responses to this thread. My gut instinct upon reading your post was a suspicion that it might be spam or trolling; or sarcastic. Were I in your shoes I think I might have written something more along the lines of the following for the first post (though were I writing for myself I would write in a different tone):

With COVID-19 putting a strain on the whole world, I am proud to see that Summit is being used help resolve the situation; as this is reaching mainstream news, I have been eagerly sharing information here and on Facebook. I am happy to see US DOE and POWER9 being used to defeat COVID-19! In addition to pride at seeing my favourite architecture at work, it makes for a nice comeback in response to those who say x86 is required to do anything of merit.

10

Applications and Porting / Re: Google Earth?

« on: March 14, 2020, 08:51:27 pm »

I am confused, is not Google Earth Pro a proprietary, x86-only application? Is the idea behind a google-earth-pro package to just load the x86 binary for Linux in a prec-onfigured emulator?

11

Operating Systems and Porting / Re: Possibility to install Amiga OS 4

« on: March 14, 2020, 08:42:11 pm »

It doesn't know how to deal with the architecture; it's written only for PowerPC-based Amigas, which despite being the same family of CPU, have very different system facilities.

Are you specifically referring to the firmware? At a glance, AmigaOS seems to call its firmware Kickstart, but according to Wikipedia since version 4 (when Amiga switched to PowerPC) it is run off of the disk rather than ROM. While looking around, I came across an AmigaOS4.1 boot process blog post, which would suggest that AmigaOS 4's Kickstart is in fact more of a bootloader, and a portable one at that, seeing as it has already been made to run on top of IEEE 1275-1994 (Open Firmware), CFE, and UBoot.

Maybe porting AmigaOS 4.1's Kickstart to OpenPOWER firmware really is not that unfathomable...

12

General Discussion / Re: The pride of being Power!

« on: March 13, 2020, 10:55:35 pm »

I am not entirely sure to reply to such such a post; I intend no offense, but your writing is difficult for me to read and respond to in a more conversational manner; therefore I will jump straight to saying:

I suspect it may interest you to hear that Sierra, Summit's smaller CORAL sibling, has also been tasked with SARS-CoV-2/COVID-19 related work as well:
Coronavirus: Lawrence Livermore Lab researchers examine virus in hopes of blocking, treating it - ABC7 News (San Francisco news station)

Also slightly on topic, Folding@Home has been working on SARS-CoV-2 proteins as well:
Folding@home update on SARS-CoV-2 (10 Mar 2020)
and FoldIt is also working to a similar goal:
Special update on coronavirus puzzles - round 3 announcement; check homepage (not blog) if round 3 has ended

I do not know if either FoldIt or Folding@Home is available for Power[PC], but if someone has time for puzzles or too many unused threads on their CPU, might be worth taking a look.

13

General CPU Discussion / Ultravisor state and/or FlexVer as a substitute for confidential/trustworthy rem

« on: January 11, 2020, 01:53:02 pm »

I have been occasionally reading the Signal blog,
https://signal.org/blog/secure-value-recovery/
and now that they are once again bringing up SGX as a possible solution to confidential/trustworthy remote processing, I am curious if POWER9’s Ultravisor mode along with Flexver could act as a replacement for SGX.
https://wiki.raptorcs.com/wiki/Power_ISA/Privilege_States#Ultravisor_State

To be honest, I do not fully understand the three (SGX, Flexver, or P9’s Ultravisor), but what Integricloud seems to be claiming to do with FlexVer in terms of allowing a user to verify code running remotely, seems awfully similar to how Signal is trying to use SGX to run code on the users behalf remotely without allowing the SGX server to see the inputs or outputs of the code being executed. From what I have read about Ultravisor state, IBM is certainly positioning it as an alternative to SGX and AMD’S Secure Processor memory encryption.
https://developer.ibm.com/articles/l-support-protected-computing/
https://www.kernel.org/doc/html/latest/powerpc/ultravisor.html

As Signal is pretty much the only messaging program I have significant trust in, part of my interest here is that I would like to see Signal using something other than a DRM mechanism to do private remote processing.

Though regardless, the promise of being able to perform confidential processing remotely is intriguing by itself, especially when done without placing absolute and irrevocable trust in the manufacturer.

14

General OpenPOWER Discussion / Re: Ahoy shipmates!

« on: January 08, 2020, 12:16:24 am »

Sorry JollyRoger; not so much a shipmate yet, as an x86 landlubber who dreams of the open sea, but has only gone as far as exploring the wreck of G4 dinghy found beached on the shore.

Howdy! Glad to see the party move outside of Twitter.

I do hope more people will be signing on here. I only stumbled on the link on Twitter, where I seldom visit.

Outside of Twitter, there is an official OpenPOWER (and PPC) community mailing list, though it sees little activity,
http://lists.mailinglist.openpowerfoundation.org/mailman/listinfo/openpower-community-dev

and discussion in a Level1Techs thread (hi q66), including a post about this very same topic of people/discussions being scattered across different platforms,
https://forum.level1techs.com/t/the-power-and-powerpc-general-discussion-news-thread/122515/245

Also, if I read the sidebar correctly, ClassicHasClass’s Talospace blog is willing to take guest posts too, right? So it, with its comments sections, can be a watering hole of sorts as well,
https://www.talospace.com/


Maybe there should probably be a dedicated community page on the RCS Wiki, rather than just linking to a few places in the main page’s footer?
https://wiki.raptorcs.com/wiki/Main_Page

Edit: Sorry if that everyone already knew about those, just writing my silly ahoy response and leaving it at that felt a bit empty.