AMD THREADRIPPER Officially Announced.

[CVNet1]

From the Forbes article:

John Taylor:

I think making Threadripper a separate version of Ryzen was the right idea too, rather than adding extra digits to the existing Ryzen naming scheme.

Threadripper timeline:

2014 – 2015: Work began on Threadripper concept, mostly in the initial team’s spare time and without a dedicated business plan

2016: The team presented the idea to Jim Anderson just before the Computex tradeshow who had joined from Intel the previous year. Loved the idea so much the launch timeline was cut to 18 months to be the same year as Ryzen and EPYC.

May 2017: AMD officially announces Threadripper, which comes mostly as a surprise to the press.

August 2017: Threadripper goes retail

And now for the other storyline

As I have been saying for months, TR didn't appear in the roadmaps a pair of years ago. The only planned sockets were: AM4, SP4, and SP3.

Then the SP4 socket failed to get traction from the industry, which raised doubts about what do with those 16-core CPUs and, in the other hand, Skylake-X was much more powerful than AMD expected, and in a last minute change AMD merged the SP4 and SP3 platforms in the new SP3r2 socket, releasing the 8, 12 and 16 core ThreadRipper, with TR4 being the marketing term for the SP3r2 socket.

The earliest version (non HEDT) of Skylake was released in August 2015.
The earliest version (HEDT) of Skylake-X was released in June 2017.

Now lets disregard the clear timeline discrepancy and pretend that what actually happened was Skylake-X was thrashing AMD's plan and AMD couldn't get traction with the motherboard industry, so AMD instead reacted to Intel inside of 2 months (not to mention getting the motherboard industry that apparently wasn't onboard for the SP4 socket to rush development of the SP3r2 and have boards ready too) and somehow with Threadripper and the 'rushed' motherboards in a desperate Hail Mary, AMD just happened to get lucky and complete Threadripper with startling success.

If the alternate timeline and events being identified by JuanRGA are to be believed, Intel should be very worried about AMD because JuanRGA's timeline suggest AMD team members can see 2+ years into the future and figure out the performance of Intel's products at that future time and make arrangements to counter Intel's 2 years from now product releases even before Intel has taped out said products. Or alternately, it suggests AMD can react to an Intel product release inside of 2 months with a strong competing product (possibly even beating out Intel's original product release).

It is actually rather irrelevant what socket this processor plugs into in relation to the creation of Threadripper. In fact, running Threadripper on the same size socket platform as Epyc is a brilliant cost savings idea since it eliminates substantial R&D costs and costs associated with Retooling production capacity at the Fab for a third dis-similar socket type and the semi-custom silicon production runs.

Threadripper as a design saw it's genesis back in 2014-2015 by a group of clearly talented and visionary people at AMD. It would appear AMD's design teams made very wise choices in how they built this thing from a production perspective as well as from a HEDT platform perspective. It would also appear that Threadripper is going to generate alot of sales from AMD given how much interest it is receiving from customers interested in HEDT platforms with lots of processing power and lots of connectivity (PCIe channels).

 

[juanrga]

I wrote:

TR didn't appear in the roadmaps a pair of years ago. The only planned sockets were: AM4, SP4, and SP3.

"a pair of years ago" = 2015.

AM4 --> single-die; dual-channel; 4, 6, and 8-core (desktop)
SP4 --> MCM2; quad-channel; 8, 12, and 16-core (microserver and workstation)
SP3 --> MCM4; octo-channel; 16, 24, and 32-core (server)

And Forbes confirms that ThreadRipper didn't exist in 2015.

It might surprise you to learn that Ryzen Threadripper wasn’t originally part of AMD’s plans, which back in 2015 only included Ryzen and the EPYC server CPU.

But as you can check Forbes is not telling all the truth. They only mention Ryzen (AM4) and EPYC (SP3) now, and don't mention Snowy Owl (SP4) anymore.

Skylake-X was released in middle 2017, but Skylake-SP has been selling since 2016 to selected customers. And Skylake-X is derived from it. It is not difficult to infer that the people within AMD knew that was coming and had time to prepare a response, specially since ThreadRipper wasn't designed from zero but derived from Snowy Owl.

This Forbes article has a marketing purpose, and neither Forbes nor AMD people interviewed will say you the real history, They don't say why AMD is lately silent about the SP4 platform, why the silence started somewhat in 2016, the same year when TreadRipper was officially incorporated in the rodmap, why the specs for SP4 CPUs agree with the specs for current ThreadRipper models, or why the article is rewriting the history, pretends that only EPYC and Ryzen existed, and doesn't make mention to Snowy Owl.

 

[noko]

From the article:





This is just what I have been saying...



Skylake-X was released in middle 2017, but Skylake-SP has been selling since 2016 to selected customers. And Skylake-X is derived from it. It is not difficult to infer that people within AMD knew that was comming.

You also can check by yourself that AMD is lately silent about the SP4 platform, which did target microservers and workstation... The silence started somewhat in 2016. The socket is not irrelevant, because SP4 was a MCM2 approach with 8, 12, and 16-core CPUs.

Note I have been saying this kind of stuff about ThreadRipper and the original SP4 socket, since before Forbes published the article. Either I used one of your time-machines to travel to the future, read Forbes article and go back in time to write this stuff in forums OR I have had some information about TR and its history before Forbes published the article.

Evidently this Forbes article has a marketing purpose. AMD will not say the real history, just as no one at AMD will say officially the true reason why Keller left AMD.

Jim Keller moves around, left Apple, Dec Alpha, AMD before etc. I wouldn't read anything into it. It is obvious that Jim's work was done at the time and moving to Samsung was probably to his best interest. He probably, my guess gets really bored with regular management, budgets, staffing issues etc. and really is an engineer at heart pushing for the next real leap in design.

 

[juanrga]

Jim Keller moves around, left Apple, Dec Alpha, AMD before etc. I wouldn't read anything into it. It is obvious that Jim's work was done at the time and moving to Samsung was probably to his best interest. He probably, my guess gets really bored with regular management, budgets, staffing issues etc. and really is an engineer at heart pushing for the next real leap in design.

Damm it! You replied before I could edit my post and delete that part. The 'official' explanation that Keller left AMD because finished his work at AMD isn't true. Keller left AMD after having sound dissagrements with Lisa Su about K12 and SkyBridge. she canceled the developments and Keller left AMD. I had edited my post because this thread is about the true history behind ThreadRipper and I don't want to derail it with a discussion about the true history behind Keller second departure from AMD.

 

[noko]

Damm it! You replied before I could edit my post and delete that part. The 'official' explanation that Keller left AMD because finished his work at AMD isn't true. Keller left AMD after having sound dissagrements with Lisa Su about K12 and SkyBridge. she canceled the developments and Keller left AMD. I had edited my post because this thread is about the true history behind ThreadRipper and I don't want to derail it with a discussion about the true history behind Keller second departure from AMD.

Keller departs everyone, lol. I wouldn't read too much into it and besides he worked for AMD not AMD worked for him. Lisa Su had to protect a very fast sinking ship - looks like she came through. To backseat drive and narrate the past is pointless at this point of time.

I would like to see Keller come back to AMD but he maybe a rather hard man to please - he constantly pushes the envelope.