AMD Navi "Blockchain" Card Support Being Added To Linux 5.10

[Jumbotron]

I thought they were unusually good questions.



Fun fact... back around 1981 I was working with AMD silicon a lot, designing graphics engines and other specialized processors using their 2900 bit-slice logic (before x86). I thought AMD was the coolest company in the world, particularly after visiting the head office and meeting Jerry Sanders, but they had essentially zero presence in Canada.

Twenty five years later I was working for ATI when the AMD/ATI merger happened. It took a long time to get the CPU and GPU parts of the company really working together, and IMO we only started to see real synergy in shipped products over the last few years, but it still feels like the coolest company in the world.

That alone just made my day!

I was there when Viacom bought Blockbuster. I was there when CSX Railroad bought half of Conrail from the Federal Gov. I was there when Time Warner bought AOL. I was there for a handful of regional media and tech companies who were bought out or merged. Each and everyone ended in disaster. The fact you've been around that long with essentially one company before and after a merger is astonishing and inspirational. Thanks for sharing that !

And Jerry Sanders !! What a character !! IMHO...the only other CEO as impactful to AMD as Jerry Sanders is Lisa Su and she may end up being the more.

[bridgman]

Hey...I'm just pitching the questions. Bridgman is hitting them out of the park. It has connected a lot of dots for me as well !!

I thought they were unusually good questions.

Just to round things out, here is a REALLY good, concise history of AMD I just found. Here is the last portion of it as a teaser.....

Fun fact... back around 1981 I was working with AMD silicon a lot, designing graphics engines and other specialized processors using their 2900 bit-slice logic (before x86). I thought AMD was the coolest company in the world, particularly after visiting the head office and meeting Jerry Sanders, but they had essentially zero presence in Canada.

Twenty five years later I was working for ATI when the AMD/ATI merger happened. It took a long time to get the CPU and GPU parts of the company really working together, and IMO we only started to see real synergy in shipped products over the last few years, but it still feels like the coolest company in the world.

[Jumbotron]

Just to round things out, here is a REALLY good, concise history of AMD I just found. Here is the last portion of it as a teaser.....

" AMD is currently in the strongest position that they've ever been in its 51-year history. With the ambitious Zen project showing no signs of hitting any limits soon, the company's phoenix-like rebirth has been a tremendous success. They're not at the top of the mountain though, and perhaps for the better. It's said that history always repeats itself, but let's hope that this doesn't come to pass. A healthy and competitive AMD, fully able to meet Intel and Nvidia head-on, only brings benefits to users. "

https://www.techspot.com/article/2043-amd-rise-fall-revival-history/

[Jumbotron]

I was never sure exactly where the lines were drawn re: the definition of Freedom Fabric, but IIRC it included:

- a 2D torus interconnect at 1 Gb/s between blades in a box (some reports say this was 2-lane PCIE rather than GbE)
- conventional 10 Gb/s interconnect outside the box (10 GbE)
- management SW that bridged across multiple boxes and did things like bandwidth allocation & QOS using FPGAs on both 1 Gb and 10 Gb NICs

Treat the above as low quality information but it's probably pretty close.



I'm going to have to "no comment" that one.

So...SeaMicro's Freedom Fabric was a sort of rough quasi-Gen Z interconnect before Gen Z became a thing as it simultaneously looked at each blade as a node and then looked at the entire frame as a node to connect to another frame. At least going by the info you provided and with the nod to the fact it may not be the whole correct picture. And here again...we see FPGA's being used for QoS and bandwidth management. And whaddaya know...AMD has those FPGA's now. Kinda back to the future for AMD vis a vis SeaMicro albeit using Infinity Architecture, Xilinx FPGAs and Gen Z to tie boxes or racks together.

Yeah....and I figured you might have no comment on the CCIX/CXL/Infinity Fabric angle concerning Xilinx. That's a bit of the domain of the suits i would imagine. Of course internal discussions for engineers as well. It looks like from what I read and from my personal grok on it is that for everything Intel it will be CXL as their replacement for Omnipath for internal interconnects and instead of Omnipath or CXL being used also for tying together racks that will now be Gen-Z seeing as how there was a lot of vendor concern for fragmentation hence the memorandum of agreement between the CXL and Gen Z consortiums so each interconnect tech works together in their proper or assigned roles.

As far as CCIX I imagine that continues to be the domain of ARM and where-ever an ARM/Xilinx combo is called for. Otherwise...for you guys it's Infinity Architecture for internal interconnects including engineering that for Xilinx FPGAs and using Gen Z to tie boxes and racks together seeing as you guys are in the Gen Z consortium. If I am not mistaken Xilinx was also in the Gen Z consortium along with the CCIX consortium.

And finally...let me apologize to you Bridgman...for saying AMD sucks at software. That was harsh...too harsh. Yeah...we fan bois would like for that to be better and you guys are going balls to wall to make that happen. It shows. It just hit me that I don't care too much for people who bash the Bulldozer arch and its subsequent revisions and improvements all the way to my Bristol Ridge equipped desktop and laptop. I for one think that Mike Clark the main designer of the Bulldozer series and the APUs was a freakin' genius and the Buildozer and subsequent revisions are marvels of engineering. The problems that AMD found themselves in during and after the Financial Crisis of 2008 put them in a precarious spot. With quarter after quarter of mounting losses and Intel hitting all cylinders I think there had to be some tradeoffs in design to meet performance goals and the ability to hit process nodes that frankly couldn't be met the way Intel was hitting them and all under the financial constraints you guys found yourselves in. MY GOD....the fact that you guys could cram what you did into a 28nm process for Bristol Ridge even going so far to use what was essentially GPU Silicon libraries to get even more transistors, reduce power, and up the single and multi-threaded IPC in the same package as Carrizo is really remarkable IMHO. To this day Intel has done NOTHING like the Bulldozer/Steamroller/Piledriver/Excavator APUs. Hell...there would not BE a PS-4 or Xbox One without AMD APUs of that generation.

And I have even greater appreciation for them now after you revealed that you guys are just now getting your dGPU/CPU gear up to that point of where APUs were in 2014 and really it's going to take the Zen 4 / Infinity Architecture to do it. Mike Clark and crew.....my admiration for you guys is eternal !!

As well as for you Bridgman. This has been a very enlightening conversation and I for one am indebted to you for taking the time. I look forward with renewed enthusiasm to what you guys are doing at AMD. Thanks again !

[Jumbotron]

A big thanks to Jumbotron and Bridgman for really bringing it today. I am blown away by how many dots got connected for me in this thread.

Going back to some of the early comments about AMD not investing in software. AMD is crushing it on the CPU side right now and I can't imagine them no continuing to take market share from Intel. At the same time NVidia brough great graphics cards but can't deliver them. If AMDs new graphics cards turn out to be as good as claimed and they can actually supply them AMD is going to have to hire people just the shovel the money. That solves part of the problem, the other side is graphics engineers don't exactly grow on trees.

Hey...I'm just pitching the questions. Bridgman is hitting them out of the park. It has connected a lot of dots for me as well !!

[MadeUpName]

A big thanks to Jumbotron and Bridgman for really bringing it today. I am blown away by how many dots got connected for me in this thread.

Going back to some of the early comments about AMD not investing in software. AMD is crushing it on the CPU side right now and I can't imagine them no continuing to take market share from Intel. At the same time NVidia brough great graphics cards but can't deliver them. If AMDs new graphics cards turn out to be as good as claimed and they can actually supply them AMD is going to have to hire people just the shovel the money. That solves part of the problem, the other side is graphics engineers don't exactly grow on trees.

[bridgman]

So am I reading you right in mentioning gigabit ethernet and Freedom Fabric that ...

A: Freedom Fabric was an ethernet based interconnect to connect the various blades to one another inside the box
Or B: It was the Interconnect to multiple SeaMicro boxes....roughly analogous to Gen-Z ?
Or C: Both

I was never sure exactly where the lines were drawn re: the definition of Freedom Fabric, but IIRC it included:

- a 2D torus interconnect at 1 Gb/s between blades in a box (some reports say this was 2-lane PCIE rather than GbE)
- conventional 10 Gb/s interconnect outside the box (10 GbE)
- management SW that bridged across multiple boxes and did things like bandwidth allocation & QOS using FPGAs on both 1 Gb and 10 Gb NICs

Treat the above as low quality information but it's probably pretty close.

And while I'm at it talking about interconnects. Things were getting very crowded after it took so long to get PCIe4 off the ground. We had CXL from Intel, OpenCAPI from IBM, CCIX from AMD, ARM and Xilinx and Gen-Z. Not to mention Infinity Architecture from you guys.

Now it looks like it's going to be CXL as a superset of PCIe 5 for internals and Gen-Z to tie together racks. That leaves CCIX and Infinity Fabric. What happens to CCIX now that Xilinx is in the house of AMD and Infinity Fabric? And what about CXL and Infinity Fabric ?

I'm going to have to "no comment" that one.

[Jumbotron]

Not much at this point, unfortunately. Freedom Fabric was a large scale interconnect based on 1 and 10 gigabit Ethernet, while Infinity Fabric is more like an advanced version of Hyper Transport, connecting a smaller number of nodes but with much higher data rates and cache coherency.

So am I reading you right in mentioning gigabit ethernet and Freedom Fabric that ...

A: Freedom Fabric was an ethernet based interconnect to connect the various blades to one another inside the box
Or B: It was the Interconnect to multiple SeaMicro boxes....roughly analogous to Gen-Z ?
Or C: Both

And while I'm at it talking about interconnects. Things were getting very crowded after it took so long to get PCIe4 off the ground. We had CXL from Intel, OpenCAPI from IBM, CCIX from AMD, ARM and Xilinx and Gen-Z. Not to mention Infinity Architecture from you guys.

Now it looks like it's going to be CXL as a superset of PCIe 5 for internals and Gen-Z to tie together racks. That leaves CCIX and Infinity Fabric. What happens to CCIX now that Xilinx is in the house of AMD and Infinity Fabric? And what about CXL and Infinity Fabric ?

[bridgman]

how much of SeaMicro's "Freedom Fabric is engineered into Infinity Architecture ??!

Not much at this point, unfortunately. Freedom Fabric was a large scale interconnect based on 1 and 10 gigabit Ethernet, while Infinity Fabric is more like an advanced version of Hyper Transport, connecting a smaller number of nodes but with much higher data rates and cache coherency.

[Jumbotron]

I just want to thanks Michael for Phoronix exists and we have access to developers ( in this case, bridgman ) so we can ask questions direct to them and have awesome answers

Truer words have rarely been written. It is a blessing. All the more reason to support Michael and his endeavors. With a tip, or better yet a subscription. I need to renew mine...but right now...Covid and a months long lay off have squeezed me dry. But I am trying to scrape up some dough for a sizable tip in the mean time when I can go ahead and make a lifetime sub payment. Hopefully after next years tax return + additional Stimulus + a new job.