https://www.anandtech.com/show/16007/hot-chips-2020-live-blog-manticore-4096core-riscv-330pm-pt
Manticore 4096 core RISC-V CPU using older 22mm FDX process @ GloFo beating out more modern CPU’s.
RISC-V is Blowing my mind with these kind of announcements + POWER10.
Were I to guess, I think it might be either an initial lack of low end part for POWER10 (Sforza being this for POWER9), or the mandatory use of OMI, that Raptor is disappointed about.
EDIT: While the above may be a concern, there is credible (in my opinion) suspicion that Raptor may in fact be concerned about part of the firmware stack being closed-source, or in some other way not owner-controllable. If you only wish to read the original post about OMI, please skip to the “Options for implementing OMI” heading.
Talospace and Phoronix suspect that IBM may be requiring some firmware blobs (at least initially) to remain closed-source:
Phoronix
Libre/Open-Source POWER10 Hardware Systems Unlikely Until At Least 2022While Raptor Computing Systems has been making fabulous 100% open-source/libre hardware systems based around POWER9 with the likes of their Talos II and Blackbird systems, don’t hold your breath on quickly seeing fully-open POWER10 systems…
This does seem to better explain the rather harsh comments that Raptor made about IBM “blocking” their products, and that “POWER10 systems from competitors in the interim will not meet the normal Raptor standards due to the causative IBM decisions”.
If we want to be optimistic about the way Raptor talks about the “interim”, I have some speculation: For POWER9, IBM presents itself as having entirely separate firmware for its PowerVM systems (IBM i, AIX, Linux on PowerVM). Maybe IBM is holding back the OpenPOWER firmware until 2022, because the only machines shipping in 2021 will be those running PowerVM.
The following is the rest of the post as originally written:
To detail the OMI issue, as I see it, POWER10 board vendors have three options:
This is akin to how all IBM POWER8, and some IBM POWER9 machines attached memory; IBM would manufacture custom “CDIMMs”, modules with memory chips and Centaur memory-controller/buffer/L4-cache. An article by Johan de Gelas (Anandtech) has photos and diagrams of this.
With POWER10 and OMI, this is less of a proprietary solution, and I believe Micron has said they will make these modules. Using OMI memory modules directly would make the switchover from DDR4 to DDR5 very simple, just buy newer OMI modules; however, unless other widely used architectures (ARM, x86) also adopt this, these modules might not see similar economies of scale, or even if the POWER10 market can drive the necessary volume, resellers of OMI memory may be hard to come by, and less willing to deal with individual end-users.
As I mentioned in the Hot Chips 32 post, if you look at Raptor’s proposal for the POWER8 Talos I, they planned to use normal DDR3 DIMMs by moving the required Centaur chips onto their mainboard, which I imagine adds expense, complexity, and consumes board space. Looking at the Talos I design, these were probably under the two rectangular heatsinks between the DIMM slots:
| A. | 1 × PCI slot | G. | 8 × DDR3 ECC DIMM slots |
| B. | GPIO header | H. | 6 × PCI Express slots |
| C. | mPCIe slot | I. | AST2400 BMC with HDMI video |
| D. | 8 × 6 Gbps internal SATA | J. | Integrated I/O |
| E. | 2-port USB 3.0 header | K. | 1 × socketed POWER8 SCM |
| F. | 2 × internal USB 3.0 |
A more flexible but more expensive alternative could be to have riser cards which themselves use standard DDR4/DDR5 memory. This could let you switch out the riser interface when DDR5 becomes available rather than switching out the entire mainboard.
I am taking inspiration from what the Russian company Yadro did for its Vesnin POWER8 machine (@olddellian mentioned these earlier in this thread).
My guess is that Raptor will take the option B approach, and may be waiting until 2022 for DDR5 to reach mainstream (more or less), rather than release a board that will only work with DDR4. On the off chance that OMI really appears to be taking off in 2021, they could change plans and switch to option A and release an OMI-module board instead.
Raptor-specific speculation aside, Timothy Prickett Morgan at Next Platform tends to be quite knowledgeable about POWER and IBM, so I shall to link his article:
Well I’ve got to say, I’m very impressed by the design of POWER10 overall, but I’m naturally disappointed by the fact that Raptor won’t be able to get a product out to enthusiasts based on the new chip in the near future. I’m really hoping for the best, because I sold my Blackbird in anticipation for either the Condor (which was recently cancelled) or a board using POWER10 instead.
There is credible (in my opinion) suspicion that Raptor may in fact be concerned about part of the firmware stack being closed-source, or in some other way not owner-controllable.
The tweet reeks of an NDA being in place, and the court of public opinion doesn’t look too kindly on those that enforce NDA’s too harshly.
I’m really curious to see what RCS has to say about what anti-features their current product lineup lacks.
I sold my Blackbird in anticipation for either the Condor (which was recently cancelled)
Source on this, please? I’ve been waiting for this release, which I assumed was in the near future. If RCS has cancelled the Condor and isn’t rolling out Power10 based system to fill that niche, then it’s looking a lot like there are no compelling Power-based systems on the consumer front for the next few years.
I can’t paste direct links here, but there is a story about the Condor’s cancelation on talospace, Cameron Kaiser’s POWER blog. I’d suggest giving that a read.
[T]here is a story about the Condor’s cancelation on talospace, Cameron Kaiser’s POWER blog.
Thanks!
It looks like the cancellation post was already linked in the thread, I just missed it.
Timothy Prickett Morgan has an article that speculates on how the POWER10 SCM vs DCM (single vs dual chip modules) may fit into IBM’s future line up:
An overview of I/O changes and possibilities in POWER10, also by Timothy Prickett Morgan,
Note that this article seems to think the POWER9′/AIO/Axone chip may still be released, but from my discussions at the weekly OpenPOWER Virtual Coffee Calls (previously mentioned by @Nawrocki) I am quite certain the Axone chip has been cancelled.
At this point, I am really debating on picking up a Power 9 or Epyc for a home server. Power 10 seems enticing, but I don’t know if I can wait that long. I have been waiting on Zen 3 to figure out if I want to build a workstation around Zen 3 or save some money and got to Zen 2.5 as I am coming from Poor Dozer on AM3+
I have not seen the power consumption numbers on Power9. Are they anywhere comparable to Epyc on Zen 1.5 at least? That has always been the factor that has made me choose Intel Compatible in the past for home servers. Old Power/PowerPC was most efficient running at full tilt which meant space heater and high electricity usage.
What do you intend to run on it, that you need Epyc or Power for home use? Just curious.
Phoronix has done benchmarking on Talos II devices, which I think are about as accessible to average users as Epyc systems.
https://www.phoronix.com/scan.php?page=article&item=power9-epyc-xeon&num=1
tl;dr - Performance generally falls between Epyc 7601 and Epyc 7551.
Unfortunately, he only had remote access for testing, so could not pull raw performance-per-watt numbers. 
Would like to run a router in VM with fiber add in cards since I am currently wiring my house for 10Gb fiber. Plan to attach that to some layer 3 switches with VLANs so that I can separate my wireless home automation stuff from my regular wireless and hard wired stuff.
From there, a couple of build servers for ARM and AMD64 (and possibly Power)
And some home lab VMs for some certifications that I plan to acquire in the near future.
Currenly the home automation stuff is running on a RPI4 and Odroid N2 with their own network, but I would like to be able to access everything on one network instead of having to connect to a different wireless network and all that jazz.
For the most part, I need the PCIe lanes more than anything for add in cards
Go Epyc.
Buy Power because you’re explicitly interested in Power, and for no other reason. You won’t see Power specialties listed in the majority of the jobs that care about those certifications.
Nothing you’ve mentioned requires the security of a fully open platform, and the price-per-performance-per-watt isn’t going to come out in favor of Power.
Fair enough. I am a fan of Power and I would like to get back into the platform if it is viable. The only real issue is power consumption for me. I used to administer power servers and some MacOS Servers back in the day as well. Plus I was big in the iMac scene with G3s and G4s.
**Edit
Also going with power will force me to give back to the community to help with software optimizations. I have seen that Power is not doing to well with ffmpeg and media codecs
For this, a Threadripper would probably be enough IMO, but I’m guessing you aren’t stopping just at virtualizing 1 router only. But you did your research and know your requirements better than I can guess them, just pointing out for others.
Why are you using a L3 switch with a router instead of a simple (managed) L2 switch (unless you already had it or you got a nice offer for them)? Currently doing CCNA3, so I have no idea how a serious infrastructure looks, aside from router-on-a-stick (which we have at my work) and a simple network with only L3 switches. Sorry for deviating from the POWER thread.
I may be getting a good deal at DRMO on the layer 3. Otherwise, I am looking at some ubiquity or microtik layer 2 switches with SFP+ powers to allow me to divide the house into 4 major sections with ethernet connectivity as well.
Larabel, Michael — 2020-09-15
IBM Contributing A2O Processor Core To OpenPOWER Community
IBM announced today at the OpenPOWER Summit 2020 that they are contributing the A2O POWER processor core and Open Cognitive Environment…
I do need to double check what this A2O core comes from, the previous core IBM open-sourced came from their BlueGene supercomputer, and this A2O does seem related. The ISA version (Power 2.07) is the same as used by POWER8, but this is clearly different, for instance, it is SMT2 rather than SMT8.
Will do!
