EPYC Datacenter Commitments

[Gideon]

https://www.amd.com/en-us/press-releases/Pages/amd-epyc-momentum-2017aug23.aspx?sf108604957=1

Looks like Epyc has got some business for AMD.

 

[DuronBurgerMan]

Juanrga and Team Blue are probably pissed. IIRC, they said Epyc would be a dud in the server space.

 

[Nathan_P]

I haven't seen them in here much recently, maybe they realised that TR & Eypc are actually pretty good

 

[SighTurtle]

Meh. Its easy for everyone to commit, but real market penetration will be seen in the numbers. If AMD makes more money from their Enterprise division, there we go, AMD is back. If not, then most likely everyone is stringing AMD along. Wait for the financial reports.

 

[Gideon]

I dont think Juanrga can come over here anymore. Dont know for sure, but my posts tend to bring him out and nothing so far. But anyway it seems they were wrong about Epyc having no traction in the server market.

 

[DuronBurgerMan]

Meh. Its easy for everyone to commit, but real market penetration will be seen in the numbers. If AMD makes more money from their Enterprise division, there we go, AMD is back. If not, then most likely everyone is stringing AMD along. Wait for the financial reports.

Saw the Server The Home review of Epyc. It's going to dominate certain areas of the market, if anybody has half a brain cell. But it will still lose to Intel for users of software that relies on per-core licensing (which, IMHO, is a blatant bullshit money grab, but whatever). It has a solid place in the market. Very solid.

 

[SighTurtle]

Saw the Server The Home review of Epyc. It's going to dominate certain areas of the market, if anybody has half a brain cell. But it will still lose to Intel for users of software that relies on per-core licensing (which, IMHO, is a blatant bullshit money grab, but whatever). It has a solid place in the market. Very solid.

I read the review as well, but just because something is good does not mean adoption. I'm just waiting for concrete numbers from AMD.

 

[Gideon]

I read the review as well, but just because something is good does not mean adoption. I'm just waiting for concrete numbers from AMD.

That will take time, but the fact some have committed to it already is a good sign.

 

[Nobu]

Saw the Server The Home review of Epyc. It's going to dominate certain areas of the market, if anybody has half a brain cell. But it will still lose to Intel for users of software that relies on per-core licensing (which, IMHO, is a blatant bullshit money grab, but whatever). It has a solid place in the market. Very solid.

Unfortunately, coding something which uses that many threads for such specialized applications takes a lot of good developers, and then you have to support it. If you're going to go that far, might as well get your money back however you can. Besides, it scales really well with performance in those applications, so it makes good sense for the companies who use the software to choose a reasonable number of cores for their workload and then buy a license for that number of cores, rather than pay more for software that will run on cores they don't have, or less for software that's not optimized for as many cores as they have, or not maintained as well due to lack of funds/developers.

 

[JustReason]

I dont think Juanrga can come over here anymore. Dont know for sure, but my posts tend to bring him out and nothing so far. But anyway it seems they were wrong about Epyc having no traction in the server market.

no. It is just the zen platform is too good to throw stones and getting harder to skew information against it. So now he is doing his anti-AMD thing in the AMD video card sub-forum since Vega has some negative opportunity with less work.

 

[juanrga]

I don't get you guys. When I post something you guys get angry because you don't like the facts and the data in my posts. When I ignore the threads, then you guys are anxiously asking me to post my opinion. Ok, on request, I am giving my opinion on the OP:

No one said that EPYC wouldn't get "some business". Even Bulldozer and Itanium got some sales. What was stated is that the impact of EPYC on the server market will be minimal and in the low single-digit percent.

The link in the OP is not saying anything that obliges me to reconsider my position. Just as remainder I add two former press releases by AMD about older server products

http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-newsArticle&ID=1701991

http://www.amd.com/en-us/press-releases/Pages/new-amd-opteron-processor-2011nov14.aspx

Some excerpts from the pair of links:

“Our industry is at a new juncture; virtualization has provided a new level of reliable consolidation and businesses are now looking to the cloud for even more agility and efficiency. We designed the new AMD Opteron processor for this precise moment,” said Paul Struhsaker, corporate vice president and general manager, Commercial Business, AMD. “The wait for the most anticipated new product and architecture for servers is over. Leading OEMs are now offering cloud, enterprise and HPC customers a full suite of solutions based on the industry’s most comprehensive server processor portfolio, the new AMD Opteron family of processors which deliver an inspired balance of performance, scalability and efficiency.”

IT customers of all sizes can today take advantage of new servers based on the engineering and business advantages of AMD’s new “Bulldozer” architecture, which has been engineered for the leading datacenter workloads. Systems from Acer, Cray, Dell, HP, IBM and many additional channel and motherboard partners are expected on the market in the coming days and weeks.

The AMD Opteron processor family has seen significant momentum since its new "Bulldozer" core-based families of server CPUs were launched in November 2011. This includes multiple major customer wins featuring a supercomputer deployment with the National Science Foundation's Blue Waters project, Canada's TELUS, which launched a virtual private cloud; and an upgrade to the "Jaguar" supercomputer at Oak Ridge National Labs.

"Since the launch of our latest AMD Opteron processors based on our 'Bulldozer' core, we have seen steadily increasing demand for our high-end processors," said Lisa Su, senior vice president and general manager, AMD Global Business Units. "The new additions to the AMD Opteron processor family, along with the latest offerings from HP and Dell, further strengthen our ability to offer greater choices to an agile and data-hungry base of enterprise and cloud customers."

Key Facts, Performance and Technical Detail

* 89 percent greater performance than the most popular server processor from the competition at the same pricev
* An extremely comprehensive processor portfolio scaling from 4- to 16-cores
* Unmatched power efficiency with power envelopes that are consistent with previous generation platforms and as low as 4.375W per core
* Up to 24 percent to 84 percent better performance on key cloud, virtualization and HPC workloads
* Highest 2P server TPCC score
* Lowest cost per virtual machine (VM)
* Up to 4 memory channels with up to 1600 MHz memory
* The only x86 processor to support ultra-low voltage 1.25v memory
* Supports up to 12 DIMMs per CPU for up to 384GB memory per CPU
* Up to Four x16 HyperTransport™ technology (HT3) Links at up to 6.4GT/s per link

Together, these new features allow AMD to deliver unparalleled performance, scalability and efficiency for highly threaded workloads like HPC, database, virtualization, and especially, the emerging web and cloud market.

As all of you can check the language used then is very similar to the language used today.

Recall my claims about products are always based in data and technological aspects, not press releases, hype, and so. For instance, in the thread about Zen markestshare I wrote that the 31% hype was all plain nonsense and that

I expect further corrections to the graph in next days with a final value around 21--22%.

It is 22.20% today.

 

[JustReason]

I don't get you guys. When I post something you guys get angry because you don't like the facts and the data in my posts. When I ignore the threads, then you guys are anxiously asking me to post my opinion. Ok, on request, I am giving my opinion on the OP:

No one said that EPYC wouldn't get "some business". Even Bulldozer and Itanium got some sales. What was stated is that the impact of EPYC on the server market will be minimal and in the low single-digit percent.

The link in the OP is not saying anything that obliges me to reconsider my position. Just as remainder I add two former press releases by AMD about older server products

http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-newsArticle&ID=1701991

http://www.amd.com/en-us/press-releases/Pages/new-amd-opteron-processor-2011nov14.aspx

Some excerpts from the pair of links:










As all of you can check the language used then is very similar to the language used today.

Recall my claims about products are always based in data and technological aspects, not press releases, hype, and so. For instance, in the thread about Zen markestshare I wrote that the 31% hype was all plain nonsense and that



It is 22.20% today.

No. You skew facts and your word connotation is to incite unrest, not relay facts. Like in this case, using percents hide the real facts. What do you think 5% of the server market is worth? It isn't thousands. And sorry the bulk of the community also said that 30% share for Ryzen was in no way accurate or indicative of the real marketshare.

 

[juanrga]

Stop the BS. Percents don't hide anything. Not only percents are routinely used in analysts reports on server market, but are also used by AMD in their own predictions of how EPYC will do in the server market.

Some people has predicted/expected huge success for EPYC in the server market, I can quote people predicting "16% server marketshare for EPYC in Q4" and similar nonsense. This people was also making nonsensical claims about RyZen marketshare.

Everything what I have done is to mention how wrong all their claims are and I am getting attacks for saying the obvious, just as I also got attacks in the RyZen thread when I said that the 31% marketshare reported by certain press was a raw cheating, and today marketshare of 22.20% is only 0.20% above where I said it would be.

 

[DuronBurgerMan]

This gentleman has a somewhat more optimistic view on AMD's future in the server market. It is cautious, of course, but he sees a 20%-30% AMD server market share by 2020 as being quite possible. And unlike you (so far as I know, anyway), he has skin in the game with substantial AMD holdings.

https://seekingalpha.com/article/4086345-evaluating-amds-server-market-potential

 

[Shintai]

This gentleman has a somewhat more optimistic view on AMD's future in the server market. It is cautious, of course, but he sees a 20%-30% AMD server market share by 2020 as being quite possible. And unlike you (so far as I know, anyway), he has skin in the game with substantial AMD holdings.

https://seekingalpha.com/article/4086345-evaluating-amds-server-market-potential

Same was said with Bulldozer, same was said with Barcelona. 2020, ye funny one.

 

[kac77]

no. It is just the zen platform is too good to throw stones and getting harder to skew information against it. So now he is doing his anti-AMD thing in the AMD video card sub-forum since Vega has some negative opportunity with less work.

Now he's back and it's your fault...lol.

 

[DuronBurgerMan]

Same was said with Bulldozer, same was said with Barcelona. 2020, ye funny one.

Not by me, and so far as I can remember, not by him either. So that is a spurious argument. Different products. Different time.

My man, everybody knew Bulldozer sucked. Well... most everybody, anyway. It had a few defenders. But hey, some people like Kim Kardashian too. There's no accounting for taste.

 

[juanrga]

This gentleman has a somewhat more optimistic view on AMD's future in the server market. It is cautious, of course, but he sees a 20%-30% AMD server market share by 2020 as being quite possible. And unlike you (so far as I know, anyway), he has skin in the game with substantial AMD holdings.

https://seekingalpha.com/article/4086345-evaluating-amds-server-market-potential

That article is from 7 Jul 2017. and the author was so confident on that optimistic vision about AMD's future that sold all the stock 19 days latter: "I sold my remaining shares of AMD at $15.51 on July 26, 2017."

 

[DuronBurgerMan]

That article is from 7 Jul 2017. and the author was so confident on that optimistic vision about AMD's future that sold all the stock 19 days latter: "I sold my remaining shares of AMD at $15.51 on July 26, 2017."

Where is the link to him selling his shares of AMD? I am unable to locate this.

Edit: And since you complained about the date of the article, here's a more recent one from the same site (different author): https://seekingalpha.com/article/4102480-epic-server-market-amd-cutting-intels-grass

 

[juanrga]

Where is the link to him selling his shares of AMD? I am unable to locate this.

Edit: And since you complained about the date of the article, here's a more recent one from the same site (different author): https://seekingalpha.com/article/4102480-epic-server-market-amd-cutting-intels-grass

The quote is from his blog. I didn't complain about the date. I simply mentioned how days after publishing that "optimistic view on AMD's future in the server market" He sold all his shares of AMD.

About the new article that you bring. Well I can confirm this part did me laugh: "Given AMD’s collaboration with Microsoft on their new console, one could speculate the former will swiftly follow if Tencent’s experiment is successful."

 

[pendragon1]

then you guys are anxiously asking me to post my opinion.

nobody asked for your opinion. they stated they were surprised you weren't here spouting off but youre here now so...

 

[juanrga]

nobody asked for your opinion. they stated they were surprised you weren't here spouting off but youre here now so...

I guess that my wording "anxiously" was too subtle. I would have added a big sarcastic icon or something.

 

[DuronBurgerMan]

nobody asked for your opinion. they stated they were surprised you weren't here spouting off but youre here now so...

It's my fault. Speak of the Devil, and all that. *sigh* Lesson learned.

 

[DuronBurgerMan]

The quote is from his blog. I didn't complain about the date. I simply mentioned how days after publishing that "optimistic view on AMD's future in the server market" He sold all his shares of AMD.

I'm looking on his blog right now, and I don't see this. Again, do you have a link for this? I've been digging pretty deep to find it...

 

[KazeoHin]

Bulldozer was a fail, and everyone knew it from day 0. Ryzen is a completely different kettle of fish. People like Ryzen, people are recommending Ryzen, people are buying Ryzen.

No matter what facts the detractors want to cherry pick, Epyc looks to be a compelling product.

 

[Shintai]

When people can show the increased money and profit from the datacenter segment for AMD we can talk, until then its just more of the same PR BS

Even 10% would make it the biggest division in AMD and increase their margins quite a bit. Yet so far margins have only decreased after the first Zen chips.

 

[OrangeKhrush]

At this point the data point is to small sample to probably formulate a pattern of whether or not Naples/Epyc have had an effect on server market, without any substantial data at this point talking about market share is moot. Of course the resident know it all's will tell you otherwise, probably based on their own opinion, when I entered practice I learnt that "In my own opinion" is the equivalent of "my own subjective facts that could not have any direct relevance to fact but sounds good nonetheless".

I don't do opinions, especially not from quasi experts that tout credentials they don't poses. I will wait until those with the qualifications give their "expert opinions".

 

[EchoWars]

It is 22.20% today.

Feel free to correct me, but isn't counting the number of users running Passmark on their Windows system a rather specious method of determining market share? Passmark themselves brag "800,000 Benchmarks", which when viewed globally is a tiny fraction of the PC systems out there (in 2017 alone, it is expected that nearly 200,000,000 systems will be sold, which makes that 800K figure look rather worthless).

Certainly, Intel has a huge lead on AMD, but quoting specific percentages from such a small subset of systems (Windows only, decided to run Passmark, PC's only, no consoles) makes for a poor estimate. I tried to do my own research here but keep running into that Passmark graph that so many seem to take as gospel. Frankly, I'd take that graph at face value: percentage of Intel vs AMD Windows users who decided they wanted to piddle with Passmark.

And WTF does that have to do with EPYC, anyway?

Just curious...

 

[OrangeKhrush]

Feel free to correct me, but isn't counting the number of users running Passmark on their Windows system a rather specious method of determining market share? Passmark themselves brag "800,000 Benchmarks", which when viewed globally is a tiny fraction of the PC systems out there (in 2017 alone, it is expected that nearly 200,000,000 systems will be sold, which makes that 800K figure look rather worthless).

Certainly, Intel has a huge lead on AMD, but quoting specific percentages from such a small subset of systems (Windows only, decided to run Passmark, PC's only, no consoles) makes for a poor estimate. I tried to do my own research here but keep running into that Passmark graph that so many seem to take as gospel. Frankly, I'd take that graph at face value: percentage of Intel vs AMD Windows users who decided they wanted to piddle with Passmark.

And WTF does that have to do with EPYC, anyway?

Just curious...

It is borne from his hatred of AMD, illogical one at best and that his only desire is to see Intel just flat out control the market without competition. That is really why this is here. The next is the quest to take half beaten numbers and try turn them in to a fact.

 

[kirbyrj]

The best thing you can do is place both Shintai and juanrga on your ignore list. Save yourself the aggravation of even seeing anything they post. I have never once gleaned anything useful out of either of them, and even got a week ban for calling one of them an "Intel shill."

 

[OrangeKhrush]

Stock/shares have nothing to do with a companies financials so if a financial analyst says something then sells his shares at top value, that is smart investment, he may very well have re invested some gains and tried again. I find it comical that "El Genius" couldn't figure that one out.

Juan, let me give you some advice, your prediction blog http://www.juanrga.com/2017/02/amd-ryzen-predictions.html that my friend is plagiarism, you have copied sources without citation, if I was the opponents Attorney I would have a field day followed by a pay day. Don't claim that to be yours.

 

[OrangeKhrush]

The best thing you can do is place both Shintai and juanrga on your ignore list. Save yourself the aggravation of even seeing anything they post. I have never once gleaned anything useful out of either of them, and even got a week ban for calling one of them an "Intel shill."

Blocking people is a cop out, I strangely feel that life is hollow without El Genius and the Great Dane.

 

[kirbyrj]

Blocking people is a cop out, I strangely feel that life is hollow without El Genius and the Great Dane.

I used to think so too, but then I got a week ban and missed out on something I was buying in FS/FT and realized it wasn't worth it arguing with people who love the "report" button.

 

[OrangeKhrush]

I used to think so too, but then I got a week ban and missed out on something I was buying in FS/FT and realized it wasn't worth it arguing with people who love the "report" button.

I have had 3 weeks of bans, all asking for proof of claims that didn't exist. apparently asking is a breach of rules. I think one was for blatant sarcasm calling them geniuses, well obviously it is sarcasm though it doesn't seem to be personal. We live in a soft generation.

 

[DuronBurgerMan]

I got you all beat. My fart jokes were probably single-handedly responsible for the restoration of the paywall on the General Mayhem forum.

Then again, I probably deserved it.

 

[thesmokingman]

Blocking people is a cop out, I strangely feel that life is hollow without El Genius and the Great Dane.

There's not much point to engaging with shills.

 

[juanrga]

Feel free to correct me, but isn't counting the number of users running Passmark on their Windows system a rather specious method of determining market share?

It is one metric of those available to us. I have mentioned it here because certain people took the headlines about a magic 31% share increase measured by Passmark as gospel. It resulted to be incorrect.

And WTF does that have to do with EPYC, anyway?

It was a reminder for the people is mentioning me either explicitly or implicitly.

 

[Pieter3dnow]

There's not much point to engaging with shills.

You correct them , don't have to engage anything there not able to do anything but post more nonsense which people who are new to the forum or just not aware of the "factual" content in what they are posting.

 

[BrainEater]

Hey guys, enough of the personal attacks.
I'm an Intel user currently(but not always),and I've learned more about BOTH architectures from Shintai and Juanrga, than from the rest of you , so stfu !
--
I hope EPYC gets some market share, AMD has done well.Seems to me , the current market suggests lower cost server components are long overdue.
I think most 'press releases' are PR garbage.....It'll take at least a year to see how EPYC is doing.

Epyc seems like the new opteron to me...I owned a few of those....great chips.

 

[JustReason]

Hey guys, enough of the personal attacks.
I'm an Intel user currently(but not always),and I've learned more about BOTH architectures from Shintai and Juanrga, than from the rest of you , so stfu !
--
I hope EPYC gets some market share, AMD has done well.Seems to me , the current market suggests lower cost server components are long overdue.
I think most 'press releases' are PR garbage.....It'll take at least a year to see how EPYC is doing.

Epyc seems like the new opteron to me...I owned a few of those....great chips.

You have learned nothing then. Also I have great disdain for those who do not own the CPUs, of any particular kind, nor have experience using them, telling me or others with how they operate. (last part was following a reference to said members but such thin skins around figured best just leave the meat and potatoes)

For example: They both parrot that the latency involved in the CCX and between dies is catastrophic/disasterous (Last one ala-Shintai) and yet when asked how it directly affects the outcome of any real task, neither could give an answer, nor actually have, just skirted the question.

CONTEXT IS KING.

In this thread as example, marketshare in terms of percentages is VERY misleading. 5% of the server market sounds small, and in the totality it is. But wait now lets add context. What percent of the server market will be upgrades this year, or any given year? That would depend on a servers lifecycle and growth. For arguments sake lets say 20%. so that meager looking 5% is actually 25% of this years sales. Which is a far better looking outcome and yet says the exact same thing. But that wasn't the original point of percents, it is more about the value of that meager 5%. That 5% is likely in the millions minimum and of course that doesn't speak to AMDs share of that sale but a better overall picture of penetration.

Realistically AMD will gain some amount of the market, how large is still in question. For tasks that require rock hard stability Intel will likely hold those till AMD has been around just a bit longer to prove they can bring the mustard. EPYC thus far in EVERY reviews looks solid and shows it is very much a performance to dollar king. And keeping in mind this is a new architecture and new showing in that market for the better part of 5 years, there is a lot of potential to be gained as software gets patched and drivers tweaked.