Intel at their last IDF revealed an aggressive Atom roadmap with a new Atom design, to come out on 22nm, with Atom on 14nm coming out in 2014.
Intel at their last IDF revealed an aggressive Atom roadmap with a new Atom design, to come out on 22nm, with Atom on 14nm coming out in 2014.If Atom/bobcat are able to progress to 22/20nm it probably won't come until the wafers drop to the point of being very cheap. <----By that time we may see Custom "hand optimized" ARMv8 CPU designs on servers.....with some of those "hand optimized" Server CPU designs being shared across platforms. (eg, when the necessary die shrink makes them appropriate for a smartphone SOC)
Lets put some things straight,
AMD Q1 CPU market share
AMD Q2 CPU market share
In Q1 2011 AMD had ~35% of the Desktop Market with an ASP(Average Selling Price) of $54,16
AMD's Desktop Total Revenue in K$s $598,497
Intel's Desktop Total Revenue in K$s $2,872,036
In Q2 2011 AMD had ~37% of the Desktop Market with an ASP(Average Selling Price) of $51,14
AMD's Desktop Total Revenue in K$s $574,381
Intel's Desktop Total Revenue in K$s $2,717,361
I believe they doing ok in Desktop, they are just not competitive in the Performance Desktop market where high ASP makes more profits.
Their APUs (Llano and Bobcat) seams to be doing great and we already know they will release Trinity in Q1 2012.
Server Market
We can all see that Server market has the Highest ASP and that translates to higher profits. Intel makes more revenue in the Server market alone than AMD does in all three markets combined.
In Q1 2011 AMD had ~7,3% of the Server Market with an ASP(Average Selling Price) of $379.03
AMD's Server Total Revenue in K$s $109,161
Intel's Server Total Revenue in K$s $2,194,026
In Q2 2011 AMD had ~6,23% of the Server Market with an ASP(Average Selling Price) of $336.94
AMD's Server Total Revenue in K$s $82,214
Intel's Server Total Revenue in K$s $2,235,818
Mobile Market
The biggest market is the Mobile, especially for Intel the Mobile market revenue is almost the same as the combined of Server and Desktop.
In Q1 2011 AMD had ~15,22% of the Mobile Market with an ASP(Average Selling Price) of $36.98
AMD's Mobile Total Revenue in K$s $259,688
Intel's Mobile Total Revenue in K$s $4,152,149
In Q2 2011 AMD had ~17,25% of the Mobile Market with an ASP(Average Selling Price) of $40.06
AMD's Mobile Total Revenue in K$s $309,467
Intel's Mobile Total Revenue in K$s $4,341,669
AMD has 33-40% of Desktop market share and i believe they will try to keep this margins but they will try to raise ASP and profits if possible. AMD simple doesnt need to try catch Intel in Desktop High End.
Mobile is the biggest x86 market both in units shipped and in revenue. Server market has the highest ASP and the highest profits. In both of this two markets AMD lags behind by substantial margins against Intel.
It seams that Bobcat and Llano has raised both market share and revenue in the Mobile market and it is only logical for AMD to really focus more in this segment than Desktop. Fusion will be their weapon and they have to continue releasing a new product every year with better availability. Mobile market is shifting to lower power laptops and Tablets and Bobcat architecture with ATIs iGPU can give them an advantage in performance per Watt.
Server has the highest ASPs and thats where higher profits are. I believe Bulldozer architecture will give them an advantage in selected areas. They are targeting specific workloads with bulldozer like Cloud, Virtualization etc.
It seams AMD will not try to compete directly with Intel and they will focus in areas where they will be able to catch a bigger market share. That doesnt mean they will exit the Desktop market in favor of ARM.
Yeah, but those ARM Servers you are referring to wouldn't need to be Tegras.
There are a lot of specialized processors on those smartphone SOCs that would never be used for a server.....that is my point.
Q2
Server:
5.9% units(-0.9%)
3.5% revenue(-1.2%)
Intel at their last IDF revealed an aggressive Atom roadmap with a new Atom design, to come out on 22nm, with Atom on 14nm coming out in 2014.
Had no idea the server side was looking so dreadful
I think Windows 8 has given AMD and Intel some wind in their tablet x86 sails.
I realize Hindsight is 20/20 but maybe AMD could have done something better than Bulldozer?
Could this line-up have worked better?
1. Phenom II IPC increased to a higher level. Call this new part AMD Phenom III+ for the sake of argument. Sure this would have killed performance per watt (due to Pollack's rule) but it would have helped the ASPs for desktop and maybe even server. Why Server? Over at AMD Zone, a criticism of Bulldozer is the high licensing costs of "Oracle" and "Microsoft SQL 2012" due the fact these companies charge fees on a "per cpu core" basis. Therefore any performance per watt decreases of Phenom III+ based servers might have been offset by lower Server software licensing costs.
2. Bobcat for mobile and various server applications where "per core licensing" isn't a roadblock to achieving low cost.
In all fairness to AMD, maybe some of these factors like "per core" server licensing were not an issue back at the time the BD concept was being developed? Is this a new type of server licensing? Maybe someone in the tech industry can shed some light on the history?
1) If you post revenue numbers, as you originally did, and then refer to "market share" without making the distinction between unit or revenue market share, then pretty much everyone is going to assume you mean to imply that you are quoting revenue market share numbers.
Otherwise why post revenue numbers but not revenue market share, only to then post unit market share without posting the unit numbers?
2) Your math is still in error
11,050 + 31,807 = 42,857
11,050/42,857 x 100% = 25.8% (not ~35%)
In all fairness to AMD, maybe some of these factors like "per core" server licensing were not an issue back at the time the BD concept was being developed? Is this a new type of server licensing? Maybe someone in the tech industry can shed some light on the history?
I dont understand your math. Looking at desktops, which is my main area of interest:
Total sales = 3,470,533
AMD sales = 598,497
598,471/3,470,533 = 0.172 or 17.2 percent. Where do you get 37% from those numbers?
AMD didnt even have much more desktop market share, if even that much, during the heyday of the Athlon chips.
They did #1, right? Llano. They just didn't package it for the market you have in mind, but they could have. They already paid the piper so to speak in terms of the R&D of shrinking the core and debugging it.
Per core licensing has been around for a decade or more.
Refer back to the BD threads and you will see I brought up this exact issue with JFAMD. You will find that it was too inconvenient for him to address.
Per core licensing has been around for a decade or more.
Refer back to the BD threads and you will see I brought up this exact issue with JFAMD. You will find that it was too inconvenient for him to address.
Thanks for shedding light on the issue.
Based your information my guess is that the Bulldozer design was a massive compromise in an attempt to address too many markets simultaneously.
Maybe a new direction for AMD could be:
1. Enhancing Llano/Phenom II cores for desktop/Server. (focusing on absolute performance with no iGPU attached)
2. Keep Bulldozer confined to APU designs only.
Intel has worked very hard to earn themselves every bit of skepticism and doubt that can be heaped upon these sorts of "roadmaps" when it comes to their non-mainstream x86 consumer parts.
If its larrabee, atom, itanium, or the xeon/extreme variants of x86 then you can count on timeline slips, project cancellations, and outright dead-silence for lengthy periods of time.
Yeah Intel said "yaddi yaddi" about atom...and yet where is 32nm atom? I'll believe the so-called "plan" for atom when it actually materializes.
AMD believes the per core licensing is a non issue. They believe less than 10% of the total market using it due to most of the enterprise workloads are licensed in a variety of different ways.
Some examples are, License by Server, License by Socket, by user name, Enterprise wide license etc.
That is the official AMD answer.
Microsoft Unveils SQL Server 2012 Licensing and Pricing
By Kurt Mackie
11/05/2011
Microsoft rolled out a description of its SQL Server 2012 licensing model on Thursday.
Microsoft's up-and-coming relational database management system (formerly code-named "Denali") promises self-service business intelligence features and other new capabilities when commercially launched. However, organizations still will have to figure out complicated licensing considerations and costs. Microsoft attempted to kick-start that effort by publishing its "SQL Server 2012 Licensing Datasheet" document this week, which can be downloaded here.
Advertisement
The new SQL Server 2012 licensing model is based on an organization's computing power, number of users and use of virtualization. Beyond that, the devil lurks in the details. The bottom line appears to be that licensing costs apparently won't substantially change that much compared with SQL Server 2008 R2 licensing, except for Client Access Licensing (CAL) costs, which will be higher.
Microsoft expects to release SQL Server 2012 in the first half of 2012. Rob Horwitz, research chair at the Directions on Microsoft independent consultancy, thinks the product may appear sometime in the second quarter of that year.
Edition Changes
SQL Server 2012 will be available in three editions: Enterprise, Business Intelligence and Standard. The Enterprise edition is an all-inclusive product in terms of its features, and Microsoft is positioning it for "mission critical applications and large scale data warehousing" uses. The Business Intelligence edition is a new product offering. It adds BI features while also including all of the features in the Standard edition. Microsoft recommends the Standard edition for "basic database, reporting and analytics capabilities," according to its white paper.
Microsoft rolled much of the SQL Server 2008 R2 Datacenter edition licensing rights into the SQL Server 2012 Enterprise edition, so the old Datacenter edition will disappear as a top product-line offering. Microsoft will offer a Web edition of SQL Server 2012, but only to organizations signing a Service Provider License Agreement. Developer, Express and Compact editions will still be available after the SQL Server 2012 product is released, Microsoft indicated.
Licensing Changes
The biggest licensing change for SQL Server 2012 is Microsoft's shift from counting processors to counting cores (see table). The licensing describes four cores per physical processor as being the minimum licensing basis.
SQL Server 2012 Licensing Options. Source: Microsoft. Footnote: "*Requires CALs, which are sold separately."
Those organizations using virtualization with SQL Server 2012 have two licensing options. Organizations can license virtual machines based on core licenses or they can license virtual machines based on server plus CALs. Four cores per virtual machine is the minimum requirement on licensing. Maximum virtualization (that is, no limits on the number of virtual machines) is only available only with the Enterprise edition of SQL Server 2012, with Software Assurance being required.
Licensing Costs
The licensing costs stayed the same, decreased or increased. It all depends on how you look at it. Horwitz shared his views in an e-mail, where he laid out the changes in bullet points.
"The price of the SQL Server CAL does go up, about 25%.
"The per-server license for Standard Edition remains the same price as before.
"The per-server license for BI server is the same price as the server license for SQL Server 2008 R2 Enterprise…though this isn't an apples to apples comparison given the difference in SKU features.
"The per-core price for SQL 2012 Standard and Enterprise edition is one quarter the price of per-proc licenses for equivalent editions of SQL 2008 R2. So effectively, if you have more than 4 cores per physical processor in the server, your licensing fee goes up."
Paul DeGroot, another Microsoft software licensing expert who now serves as principal consultant of the independent consultancy Pica Communications after working for Directions on Microsoft, offered other insights into Microsoft SQL Server 2012 licensing costs. DeGroot noted that the CAL price increased substantially from $164 to $209 and speculated that Microsoft felt that raising the price of the CALs would have less of an impact on customers than raising server licensing costs. Still, other price changes were somewhat neutral, he contended.
"Overall, I'd say they [the prices] stayed the same or went down, with the reservation that the change from per proc to per core is significant, but may not have a huge impact on a lot of customers, since quad-core procs are probably a common choice for running high-end editions of SQL Server," DeGroot said in an e-mail. He estimated that the price would remain much the same for organizations "so as long as you're using quad-core procs."
Cost considerations largely killed the Datacenter edition of SQL Server 2008 R2, DeGroot contended. "That cost $54,990 per proc, or twice the per proc price of SQL 2008 R2 Enterprise," DeGroot said, adding that "reading between the lines, I'd say that SQL Server 2008 R2 Datacenter sold poorly, and that's not surprising." With SQL Server 2012 Enterprise edition "customers will get Datacenter power at half the price that Datacenter was," he explained.
Similarly, costs considerations were tricky for those running virtual machines on SQL Server 2008 R2, relative to SQL Server 2008. To have unlimited virtual machine capability, organizations either had to buy the Enterprise edition of SQL Server 2008 R2 with Software Assurance or they had to buy the Datacenter edition of that product. The latter option doubled SQL Server 2008 R2 licensing costs relative to SQL Server 2008 licensing costs for virtualization to "about $110,000," DeGroot said.
"To put it simply, the new pricing [of SQL Server 2012] brings the cost of a server licensed for unlimited SQL VMs back to where it was with SQL 2008, before the Datacenter edition doubled the price," DeGroot said.
Transition Caveats
Enterprise edition licensing for SQL Server 2012 will move from being processor based to being core based. Microsoft's document indicated a June 30, 2012 deadline for those organizations transitioning to SQL Server 2012 Enterprise edition.
"New Server licenses for EE will only be available for purchase through 6/30/2012," the white paper states on page 8. "Additional EE licenses in the Server and CAL license model will not be sold thereafter."
There's also a core limitation. Newly purchased Enterprise edition licenses and those upgraded through Software Assurance will be "limited to server deployments with 20 cores or less." Those organizations running more than 20 physical cores with SQL Server 2008 R2 Enterprise edition and the server plus CALs model should contact their Microsoft representative, the document states.
Microsoft is advising that IT pros should run the Microsoft Assessment and Planning Toolkit in their computing environments at the end of their organization's agreement term to take an inventory. They can use that tool to document how many processors were covered under Software Assurance.
"At the end of the SA term, processor licenses will be exchanged for core licenses and customers can renew their SA on core licenses," Microsoft's document explains.
Assessments
The move to core-based licensing was received rather positively, both by Horwitz and DeGroot. The move may kill two-core designs, or at least IT pros should look to buy four-core processors at minimum to support Microsoft's licensing, DeGroot suggested.
DeGroot took issue with the editions being sold sometimes on a per server and CAL basis (Business Intelligence edition) and at other times on a per-core basis (Enterprise edition). The cost calculations can "get whacked" based on the number of users, he explained.
"If you're already licensed per CAL, the BI server is an easy choice and the cost of the Enterprise Edition is way out of line with every other SQL Server you buy," DeGroot stated. "If you're not licensing per CAL, then you bypass BI and go with Enterprise for BI, since it has a superset of the BI server's features. The hierarchy of features -- Standard to BI to Enterprise -- may be clean, but the licensing hierarchy is disjointed."
Customers with existing SQL Server installations will face transition issues, "especially for customers with licenses covered by Software Assurance, including customers who buy SQL Server under the Enrollment for Application Platform (EAP) -- a special option within Enterprise Agreements," Horwitz explained. Despite that complexity, he found the licensing changes to be largely positive.
"I think the adjustments make a lot of sense for a number of reasons, including: (a) competitive pressures, especially from Oracle and how they handle packaging/pricing; (b) evolution of chip architectures (more and core Cores per chip); (c) continued expanded use of virtualization within the datacenter; and (d) pressures to monetize improvements in the technology."
Those wanting more insights might consider signing up for Directions on Microsoft's licensing boot camps.
DeGroot of Pica Communications also conducts workshops on Microsoft licensing
Even if it was, it wasn't widespread, and it's still pretty rare. The vast majority of software is per system, or per physical package like Windows OS-es e.g.
AMD believes the per core licensing is a non issue. They believe less than 10% of the total market using it due to most of the enterprise workloads are licensed in a variety of different ways.
Some examples are, License by Server, License by Socket, by user name, Enterprise wide license etc.
That is the official AMD answer.
Well, I would think something even faster than that would be needed.
The way I look at things if AMD cannot win performance per watt...they should focus on winning with outright single threaded performance.
Higher IPC than Intel Ivy bridge cores (especially with AMD's process tech disadvantage) would have been essential for achieving that.
Maybe four jumbo/extra wide cores without iGPU at a reasonable die size would have kept a lot desktop enthusiasts happy? Sure Pollack's rule would be invoked, but maybe the increase in sales (even server sales due to reduced licensing costs) would have warranted the move?
With regard to Bulldozer, Maybe that CPU can still work nicely for Mobile APUs? I can see the point of saving die size on FPU if they plan on moving more loads to GPU. (although with that said I am disappointed in the die size of a bulldozer module compared to two Llano cores w/ L2 cache.)
P.S. Does anyone see the irony of Bulldozer here? Does a CPU design that seems like a good concept for mobile (small integer cores, reduced FPU so a larger iGPU can be attached) only end up hurting its server potential?
Is that what AMD believes, or is that what the marketing drones are telling you?
Either way, in terms of enterprise application software and middleware, you and they are incorrect.
New SQL Server 2012 per core licensing – Thank you Microsoft
Many of us have probably seen the new SQL Server 2012 per core licensing, with Enterprise Edition at $6,874 per core super ceding the $27,495 per socket of SQL Server 2008 R2 (discounted to $19,188 for 4-way and $23,370 for 2-way in TPC benchmark reports) with Software Assurance at $6,874 per processor? Datacenter was $57,498 per processor, so the new per-core licensing puts 2012 EE on par with 2008R2 DC, at 8-cores per socket.
This is a significant increase for EE licensing on Intel Xeon 5600 6-core systems (6x$6,874 = $41,244 per socket) and a huge increase for Xeon E7 10-cores systems, now $68,740 per socket. I do not intend to discuss justification of the new model. I will say that SQL Server licensing had gotten out of balance with the growing performance capability of server systems over time. So perhaps the more correct perspective is that SQL Server had become underpriced in recent years. (Consider that there was a 30%+ increase in the hardware cost structure in the transition from Core 2 architectures systems to Nehalem systems for both 2-way and 4-way to accommodate the vastly increased memory and IO channels.)
Previously, I had discussed that the default choice for SQL Server used to be a 4-way system. In the really old days, server sizing and capacity planning was an important job category. From 1995/6 on, the better strategy for most people was to buy the 4-way Intel standard high-volume platform rather than risk the temperamental nature of big-iron NUMA systems (and even worse, the consultant to get SQL Server to run correctly by steering the execution plan around operations that were broken on NUMA). With the compute, memory and IO capabilities of Intel Xeon 5500 (Nehalem-EP), the 2-way became the better default system choice from mid-2009 on.
By “default choice”, I mean in the absence of detailed technical sizing analysis. I am not suggesting that ignorance is good policy (in addition to bliss), but rather the cost of knowledge was typically more than the value of said knowledge. Recall that in the past, there were companies that made load testing tools. I think they are mostly gone now. An unrestricted license for the load test product might be $100K. The effort to build scripts might equal or exceed that. All to find out whether a $25K or $50K server is the correct choice?
So now there will also be a huge incentive on software licensing to step down from a 4-way 10-core system with 40 cores total to a 2-way system with perhaps 8-12 cores total (going forward, this cost structure essentially kills the new AMD Bulldozer 16-core processor, which had just recently achieved price performance competitiveness with the Intel 6-core Westmere-EP in 2-way systems).
In the world of database performance consulting, for several years I had been advocating a careful balance between performance tuning effort (billed at consultant rates) with hardware. The price difference between a fully configured 2-way and 4-way system might be $25,000. For a two-node cluster, this is $50K difference in hardware, with perhaps another $50K in SQL Server licensing cost, with consideration that blindly stepping up to bigger hardware does not necessarily improve the critical aspect of performance proportionately, sometimes not at all, and may even have negative impact.
With performance tuning, it is frequently possible to achieve significant performance gains in the first few weeks. But after that, additional gains become either progressively smaller, limited in scope, or involve major re-architecture. In the long ago past, when hardware was so very expensive, not mention the hard upper limits on performance, it was not uncommon for a consultant to get a long term contract to do performance work exclusively.
More recently, performance consulting work tended to be shorter-term. Just clean up the long hanging fruit, and crush moderate inefficiencies with cheap powerful hardware. While this is perfectly viable work, it also precludes the justification for the deep skills necessary to resolve complex problems, which also calls into question the need to endure an intolerably arrogant, exorbitantly expensive consultant.
It had gotten to the point that I had given thought to retiring, and go fishing in some remote corner of the world. But now with the new SQL Server per core licensing, Microsoft has restored the indispensable (though still intolerable) status to arrogant, exorbitantly expensive, performance consultant. So, thank you Microsoft.