Support NeoGAF

[StevieP]

AMD makes i7s? It was an AMD investor conference. I am not talking about anything but AMD in an AMD thread.

Edit. On a side note, AMD mentioned consoles a bunch as how they see the market developing. Specifically, they see clients demanding partially customized/optimized solutions just like consoles manufacturers do with AMD, Nvidia, and Intel. Take IP and designs from AMD and then play with those building blocks for a fit for purpose solution.

I would like AMD to be competitive in that field as well. Sandy Bridge is an awesome freaking CPU, and I doubt I would trade it for a Bulldozer (as an example) but to have 2 companies pushing eachother is better than only having one. A proud Athlon owner like me will tell you that.

 

[dionysus]

I would like AMD to be competitive in that field as well. Sandy Bridge is an awesome freaking CPU, and I doubt I would trade it for a Bulldozer (as an example) but to have 2 companies pushing eachother is better than only having one. A proud Athlon owner like me will tell you that.

Me too, but in AMD's own words that is not a goal of theirs. It was in the Q&A session where they said there was no high end desktop CPU development, so maybe it was just worded badly.

 

[·feist·]

Keep the stuff coming, feist. It's very nice to have a single place to keep track of all this, thanks. ^^

NP. I appreciate your added perspective. Going by some of the responses, there are some members who don't seem to be reading my own comments, or info I've linked to. Hopefully, contributions like yours prove more successful.

As I mentioned before already (haha) the whole APU/HSA integration is AMD's only feasible way left to really compete, they already overhauled their whole CPU design process to adapt to that. The only reason left for CPUs without GPUs to exist for AMD is as a test bed for new changes to optimize further for HSA. It's a great thing that the new management seems to be fully on board with this already on-going development.

Yup. I think people are mistaking a change in focus as a sign of abandoning desktop CPUs. Currently, AMD can not properly go toe-to-toe with Intel at the upper end of the performance spectrum. For the time being, they're looking to increase profits by focusing on APUs, which are a major area of growth, and happen to be a strength of theirs. OEMs love them, the vast majority of users are well served by them. If not for manufacturing supply shortages, AMD could have added Apple to its client list (apart from GPUs). That's a massive missed opportunity, and money you can not replace.

So, focus largely on APUs, and SoCs of all types (including the potential for specialized server parts), increase profits, have more money to devote to R&D and better products across the board. Most mid-level, and high end APUs will share CPU cores with standard desktop parts. As efficiency and performance increase, that transfers directly over to products like the FX line.

There's very little area for AMD to make up considerable market share with desktop CPUs. There's vast room to make money with APUs. Ultimately, APU advances directly benefit GPU, and CPU development.

 

[·feist·]

Follow up on these FX Bulldozer roadmaps:

http://www.neogaf.com/forum/showpost.php?p=33461566&postcount=953
http://www.neogaf.com/forum/showpost.php?p=33621622&postcount=966
http://www.neogaf.com/forum/showpost.php?p=34822680&postcount=1018
http://www.neogaf.com/forum/showpost.php?p=35048170&postcount=1035

DH Exclusive: Three New 95 Watt TDP AMD FX Processors Are Coming
http://www.donanimhaber.com/islemci...-95-Watt-TDPli-uc-yeni-FX-islemci-geliyor.htm

Three New 95W AMD FX CPUs Will Arrive This Month or the Next
http://news.softpedia.com/news/Three-New-AMD-FX-CPUs-Will-Arrive-This-Month-or-the-Next-253496.shtml

They are called FX-8140, FX-6120 and FX-4150 and will probably be launched in March, 2012 (next month).

FX-4150, as the weakest of the trio, features four cores, each working at a base frequency of 3.9 GHz and a Turbo Core rating of 4.1 GHz.

The second unit, FX-6120, can reach the same Turbo Core level as above, even though its six cores function at 3.5 GHz normally.

As for the FX-8140, it is a strong, 8-core central processor (4 modules) whose base frequency is of 3.2 GHz and whose maximum Turbo Core performance is of, again, 4.1 GHz.

All three newcomers draw 95W of energy, though it may be possible for the 8-core to come in a 125W variant as well.

If anything, this shows that AMD is not about to back out of the processor segment, even if it has stopped bothering with trying to outmatch Intel on the high-end front.

Finally, the FX-4150 boasts 12 MB of cache memory, while the FX-6120 comes with 14 MB and the FX-8140 has 16 MB.

 

[wario]

don't post much but i like reading about the latest with amd.

much respect for your dedication feist.

 

[·feist·]

Thanks. Have some more.


Cyclos Semiconductor Announces First Commercial Implementation of Resonant Clock Mesh Technology

AMD Realizes Significant Reduction in Power Consumption by Implementing Cyclos Resonant Clock Mesh Technology
http://eon.businesswire.com/news/eo...nces-Commercial-Implementation-Resonant-Clock

BERKELEY, Calif.--(EON: Enhanced Online News)--Cyclos Semiconductor, the inventor and only supplier of resonant clock mesh technology for commercial IC designs, today announced at the International Solid State Circuits Conference (ISSCC) in San Francisco, CA that AMD has successfully implemented Cyclos’ low-power semiconductor intellectual property (IP) in the AMD x86 core destined for inclusion in Opteron server processors and client Accelerated Processing Units (APUs). The adoption of the Cyclos resonant clock mesh IP to reduce power consumption demonstrates the commitment AMD has made to provide its customers with not only class-leading APU performance but also with the lowest possible power consumption.

AMD’s 4+ GHz x86-64 core code-named “Piledriver” employs resonant clocking to reduce clock distribution power up to 24% while maintaining the low clock-skew target required by high-performance processors. Fabricated in a 32nm CMOS process, Piledriver represents the first volume production-enabled implementation of resonant clock mesh technology. “We were able to seamlessly integrate the Cyclos IP into our existing clock mesh design process so there was no risk to our development schedule,” said Samuel Naffziger, Corporate Fellow at AMD. “Silicon results met our power reduction expectations, we incurred no increase in silicon area, and we were able to use our standard manufacturing process, so the investment and risk in adopting resonant clock mesh technology was well worth it as all of our customers are clamoring for more energy efficient processor designs.”

"High-performance processors have used clock mesh designs for years, but with growing emphasis on power reduction in both servers and mobile PCs, the traditional approach has become too power hungry," said Linley Gwennap, principal analyst of The Linley Group. "This announcement proves that the Cyclos resonant clock mesh technology provides meaningful power savings in real-world products. We expect other processor designers to adopt the Cyclos technology in applications where power reduction is important."

 

[·feist·]

More Windows 7 Bulldozer patch benchmarks.


AMD FX-8150 - Bulldozers in the detailed test
http://ht4u.net/reviews/2011/amd_bulldozer_fx_prozessoren/index51.php

Update 1/30/2012: Windows fixes for performance improvement

With the update, however, a counting method is applied, as we know them already with Intel processors, hyperthreading. For example, Windows 7 recognizes the FX-8150 is now only 4 cores, but allows each core to execute two threads. This is now only one thread is assigned to each module, and only after all modules are occupied, CMT is used.

However, this does not always bring benefits. Since the maximum allowed AMD turbo mode only up to the use of a maximum of 2 modules, with three active threads is now used only the all-turbo core. This may in turn affect the performance.

Two contrasting examples are the Cinebench and Linpack. Both benchmarks we measured, each with four threads active and Turbo mode. Once the threads have been established here, "pinned" to the first two modules set (approximately equivalent to the original scheduler), the second pass, we then have allowed only one thread per module (adapted scheduler). As the table below indicate the goal of responding to Cinebench almost none during the Linpack benchmark benefits very clearly.

Core configurations/core affinity.

4 threads in 2 CU/modules -vs- 4 threads in 4 CU/modules:

In the case of Cinebench, the slightly better thread performance of the second variant, the only drawback clock (3.9 GHz to 4.2 GHz) to compensate, while the Linpack test significant advantages despite the lack of 300 MHz can be measured.

 

[·feist·]

Cyclos Semiconductor
http://www.cyclos-semi.com

"A test chip, created in 2009, was enough to convince AMD to add Cyclos' resonant clock mesh technology to Piledriver."

AMD packs Cyclos clock tech into Piledriver
http://www.bit-tech.net/news/hardware/2012/02/21/amd-packs-cyclos-piledriver/1

The technology, invented by Cyclos, uses on-chip inductors to create an electric pendulum known as a 'tank circuit,' formed by the large capacitance of the clock mesh in parallel with the inductors. The result is a system which 'recycles' the clock power instead of watching it dissipate on every clock cycle as with traditional systems.

In other words: it's regenerative braking for processors.

To back its claims of power savings up, Cyclos has some real-world figures. Based on Piledriver-based x86 processing cores running at 4GHz and above, the resonant clock mesh technology drops clock distribution power by up to 24 per cent at peak and between five and 10 per cent on average in the company's testing. Clock-skew, a serious issue in high-speed processors, is claimed to be unaffected by the drop in power draw.

If you're wondering just how the resonant clock mesh technology works, there's a great write-up over on SemiWiki.

edit:

Trinity to boost performance, save power with resonant clock mesh
http://techreport.com/discussions.x/22520

According to this Cyclos whitepaper (PDF), performance can be improved by 5-10% using a clock mesh instead of the clock trees typically employed by microprocessors. This mesh distributes the clock cycle over a uniform grid that covers the entire chip. The mesh reduces the clock skew associated with tree-based designs, allowing the chip to utilize more of its clock cycle.

The problem with meshes is that they tend to consume more power than clock trees. Cyclos' solution is a resonant clock mesh that uses on-chip capacitors and inductors to create a tank circuit that acts as a sort of electronic pendulum. The charge flowing between the capacitors and inductors is largely self-sustaining, generating an effective clock cycle that needs only a "nudge" from an external source to keep the virtual pendulum swinging in time. Cyclos claims this approach can lower total chip power consumption by up to 30% without compromising the performance benefits of the mesh.

 

[·feist·]

Intel and AMD arm wrestle again: this time on power sipping ultrathins
http://vr-zone.com/articles/intel-a...s-time-on-power-sipping-ultrathins/14962.html


edit:

Ultrabook price unlikely to achieve sweet spot until 2013
http://www.digitimes.com/news/a20120213PD207.html

Most vendors expected to adopt hybrid HDDs for Ivy Bridge ultrabooks
http://www.digitimes.com/news/a20120221PD218.html

As Intel will launch Ivy Bridge processors for use in second-generation ultrabooks in May and demands that data read/write speeds attain a certain level through adopting SSDs or hybrid hard disk drives, most vendors are expected to adopt hybrid HDDs to cut costs, according to Taiwan-based supply chain makers.

In order to promote Ivy Bridge, Intel will lower processor prices by US$60-70 on average, the sources indicated. Because storage devices account for 10-15% of total notebook production costs, vendors will try to minimize storage costs, the sources noted. As hybrid HDDs cost less than 128GB and 256GB SSDs by more than 50%, vendors will use hybrid drives to offer ultrabook models at below US$700, the sources said.​

Acer corporate president Jim Wong predicts prices for ultrabooks equipped with hybrid HDDs will fall to US$600-700 in fourth-quarter 2012, the sources noted.

Ultrabooks will account for 20% of total notebook volumes in 2012, lower than Intel's target of 40%, the sources noted.​

 

[·feist·]

Four page article from Forbes. Worth reading.

"AMD CEO Rory Read says the processor industry is at an 'inflection point,' that will reduce rival Intel's importance."


After Years Of Hard Knocks, AMD Honing Its Killer Instinct
http://www.forbes.com/sites/briancaulfield/2012/02/22/the-predator/

 

[·feist·]

Price cuts, and official release of the FX-6200 and FX-4170.


AMD adds a pair of new FX-series chips
http://techreport.com/discussions.x/22546

Let's start with a look at the price cuts. They've affected the eight-core FX-8120, which has slipped from $205 to $185, and the six-core FX-6100, which is down from $155 to $145. These figures apply to CPUs sold in bulk quantities, of course, so retail prices might turn out to be a tad higher.

MSRP

• FX-8120: $205 $185 (U.S.)
  
• FX-6100: $165 $145 (U.S.)

Now, here are the two newcomers:

Compared to their direct predecessors, the new chips seem to be substantially faster. The FX-6200's base clock is 500MHz higher than the FX-6100, and the Turbo peak has been increased by 200MHz. Meanwhile, the FX-4170 adds 600MHz to the FX-4100's base clock speed while pushing the Turbo maximum up by 300MHz. Both of the new entrants have unlocked upper multipliers, as well, so they should be straightforward to overclock, provided AMD has left some headroom there.

edit:

AMD cuts prices of Phenom II and FX CPUs
http://www.cpu-world.com/news_2012/2012022804_AMD_cuts_prices_of_Phenom_II_and_FX_CPUs.html

Overall, the prices were slashed by 2% - 15% for Phenom CPUs, and by 6% - 9% for FX chips. The parts, discounted the most in absolute terms, are FX-8120, and Phenom II X4 965, 975, and 980. They have got $20 cheaper. AMD FX-6100 had modest $10 price drop, however let's not forget that this is the second drop since processor's introduction last October. The biggest relative price cut, almost 15%, was for Phenom II X4 965. Please see the table below for detailed price changes:

 

[·feist·]

http://www.microcenter.com/specials/promotions/AMDbundlePROMO.html

 

[·feist·]

Posting this snippet from a non-AMD-specific article as another means of highlighting common issues that have already been mentioned multiple times:


- Why GlobalFoundries had struggles with its 32nm products (which are now showing increasing improvements in heat, power consumption, and ease/extent of overclock as it matures)

- Why AMD stated they would no longer be heavily focused on moving to the absolute latest, bleeding edge process, with some of their products

- The increasing difficulty, and ever diminishing returns, that ALL companies are dealing with, particularly with fabs

There is no doubt about it, producing bleeding edge silicon is going through a lot of pain at this years' Mobile World Congress. TSMC's even openly admitted that the yields for its state-of-the-art 28nm process are far from satisfactory - the situation is still better than it was for the 40nm process.

Intel is having issues with the 22nm [Tri-Gate] FinFET (so called 3D transistors), while GlobalFoundries had issues in ramping up both 32nm and 28nm process. Samsung is being silent but we've all seen the amount of 3rd party chips the company ordered from NVIDIA and TI to compensate for its own [internally fabricated] products.

On the disparities in R&D, and profits:

Tech's Biggest Spenders in R&D
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/

No. 1: Microsoft
R&D budget: $9.18 billion
Percent of net sales: 12.93
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/11

No. 2: Intel
R&D budget: $7.71 billion
Percent of net sales: 15.07
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/10

No. 3: IBM
R&D budget: $6.28 billion
Percent of net sales: 6.03
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/9

No. 4: Cisco
R&D budget: $5.82 billion
Percent of net sales: 13.47
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/8

No. 5: Google
R&D budget: $4.92 billion
Percent of net sales: 12.83
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/7

No. 6: Oracle
R&D budget: $4.47 billion
Percent of net sales: 12.69
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/6

No. 7: Hewlett-Packard
R&D budget: $3.24 billion
Percent of net sales: 2.35
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/5

No. 8: Qualcomm
R&D budget: $2.82 billion
Percent of net sales: 23.19
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/4

No. 9: Amazon.com
R&D budget: $2.57 billion
Percent of net sales: 5.07
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/3

No. 10: Apple
R&D budget: $2.43 billion
Percent of net sales: 2.24
http://images.businessweek.com/slideshows/20111028/tech-s-biggest-spenders-in-r-d/slides/2


Everyone knows that the likes of Cisco, and Oracle are giants. Notice how Intel dwarfs them.


Also, seeing as how people are still struggling to understand the whats and whys of the Bulldozer arch, with some even mistakenly believing that it is inherently bad, I'll soon be posting a comparison of AMD's CMT Bulldozer, Intel's SMT Sandy Bridge, and AMD's CMP Phenom II CPUs.

 

[Hazaro]

Looking forward to it.

 

[Ryoma-Echizen]

Im really looking forward for the A10 Trinity APU's and overclock the hell out of it's GPU (1Ghz maybe?).

 

[·feist·]

People have hit ~960MHz+ on air with the Llano A8s, so assuming GloFo's process continues to improve, ~1GHz should be possible with Trinity.

Just remains to be seen how resonant clock mesh factors in to performance improvements, consumption, and overclocking.



CPU-Z 1.60 Released, Brings Ivy Bridge and Trinity Support
http://ht4u.net/downloads/212/

Details about CPU-Z (1.60)
Changes in version 1.60:

• Intel Core i5 2550K, 2450P and 2380P with no GPU.
• Intel Core i3/i5/i7 22 nm "Ivy bridge" (37xx, 35xx, 34xx, 33xx, 32xx).
• Intel Xeon E3 (SandyBridge-WS).
• Intel Z77 platform.
• AMD FX-8140, FX-4150 and mobile Llano (socket FS1).
• "Trinity" AMD APU preliminary support.
• Windows 32/64-bit report.

Another sighting of the 95w FX-8140, and FX-4150 that AMD have yet to officially announce.

 

[·feist·]

AMD to Acquire SeaMicro: Accelerates Disruptive Server Strategy
http://www.seamicro.com/node/238


AMD Executes on Promise of Agility, Intends to Acquire SeaMicro for $334M
http://www.anandtech.com/show/5624/...-agility-intends-to-acquire-seamicro-for-334m


AMD Investors React Positively To SeaMicro Deal; Shares Up 2%
http://www.forbes.com/sites/briancaulfield/2012/03/01/amds-seamicro-acquisition-pleases-investors/


Tying It All Together
http://blogs.amd.com/work/2012/02/29/tying-it-all-together/

Fabric – noun; framework, structure. Everyone knows fabric when it comes to cloth, but when it comes to computing, not as many people understand the concepts of fabric computing.

But AMD does. We understand it so much that today, we made a significant investment in fabric computing, announcing that we’ve entered into a definitive agreement to purchase SeaMicro.

So, what does this acquisition mean for AMD? Clearly, we are moving from a being merely a silicon provider to delivering systems-level integration and capability. When you look at the IP that AMD has at its disposal, large cores like “Bulldozer” and the upcoming “Piledriver”; smaller, energy efficient cores like “Bobcat” and the upcoming “Jaguar”; as well as leading GPU technology, it all means that we have the ability to build a variety of CPU and APU products that can be integrated into servers. And the unique fabric solution from SeaMicro helps AMD tie all of these pieces together.

AMD announces plans to acquire SeaMicro
http://techreport.com/discussions.x/22560

Apparently, AMD's new executive team had some specifics in mind when they were talking about making some unconventional moves in the server market a few weeks back—especially the part about building a cloud server product based on an array of low-power Brazos-derived CPU cores. AMD has just announced its intention to acquire SeaMicro, the innovative maker of high-density servers. This is fresh news, so I'll drop a big chunk of the press release on you for now:

This move does seem to fit with AMD's new direction—and in some intriguing ways, too.

AMD's heavily threaded Bulldozer, APUs, GCN are good fits for Seamicro's compact cloud computing servers
http://www.dailytech.com/AMD+Acquires+Cloud+Server+Maker+SeaMicro+for+334M+USD/article24132c.htm

Advanced Micro Devices, Inc. (AMD) has struggled mightily in the server market in recent years, seeing its market share fall from nearly 15 percent in 2007 to less than half that -- roughly 6.5 percent in 2011.

I. AMD Server Division -- In Need of a Turnaround

AMD can try to write off part of its struggles to rival Intel Corp. (INTC) using anti-competitive techniques to squelch its performance during its strong years in the middle of the last decade, a big part of the troubles have come due to AMD's trailing die shrink timing, which has not improved since it spun off its fabs. While AMD finally dropped a new architecture (Bulldozer) in Sept. 2011, it disappointed in clock speeds and power performance -- something that may be attributable to die shrinks. Difficulty getting to 32 nm may have left AMD with too little time to thoroughly test and refine the new cores.

Approximately 21.89 percent of AMD's market share is tied up in its server sales, so clearly this is a major issue for the company and its shareholders. AMD desperately needed a new tactic. While allowing competitive interplay between Taiwan Semiconductor Manufacturing Comp., Ltd. (TPE:2330) and GlobalFoundries in die shrinks may be a potential long term solution, AMD needed something more immediate.

That's why the news of its acquisition of SeaMicro for $334M USD (a mix of $281M USD cash and stock) is a bit surprising, but a bit unsurprising. The small 80-person Silicon Valley server maker is known as a premium maker of highly dense and power-efficient servers. It sells heavily to large-scale cloud computing businesses.

II. Meet SeaMicro

[...]

III. Folding in Piledriver, APUs, GCN GPUs, ARM into Thread-Shredding Beasts

[...]

"SeaMicro makes its servers in California -- not China. It contracts NBS, a small local manufacturer (pictured)."

"SeaMicro's server boards drastically slash space and power via custom chipsets.

Assuming AMD explores these tracks thoroughly, its new subsidiary could soon be producing some sweet thread-shredding mixed designs.

SeaMicro fabric could be the glue for future AMD chips
http://techreport.com/discussions.x/22561

The reasons for AMD's decision to purchase SeaMicro are coming into focus quickly. AMD's John Fruehe has written a blog post explaining what AMD gets in this acquisition, and it's worth reading. Most notably, Fruehe says the key technology involved is SeaMicro's interconnect fabric:

The basis of the SeaMicro technology is the interconnect fabric. A custom ASIC allows multiple computers to be linked together via their PCI Express bus. This fabric allows a large number of servers, to be tied together in a fabric within a dense chassis, all sharing a high-performance connection to networking and I/O peripherals. Servers go from being a large box to a small card as they are aggregated into this environment.

Fruehe then makes it very clear how this technology ties into the "SoC approach" to chip design that AMD's execs communicated at the firm's Financial Analyst Day. SeaMicro's fabric may be the glue that binds together future AMD chips:

At the most recent AMD Financial Analyst Day, Mark Papermaster and Lisa Su both spoke of the new AMD direction towards SOCs (system on a chip) that will allow us to take a wide range of IP and integrate it into purpose-built SOCs that meet particular computing needs. With SeaMicro we now have the fabric to pull all of these together as well. The SeaMicro IP that helps connect all of these computers together could one day be built directly into these SOCs, providing an easy way to tie systems together with the smallest silicon footprint possible. But not just AMD SOCs, the technology will be available for others as well.

So it seems AMD's current partners will get access to the SeaMicro fabric technology, potentially enabling them to create high-density server solutions, presumably based on AMD microprocessors. Meanwhile, AMD remains largely a technology firm, not a server vendor.

 

[Hazaro]

Looks neat.

 

[·feist·]

Multi Thread Efficiency: AMD's CMT Bulldozer vs. Intel's SMT Sandy Bridge, and AMD's CMP Phenom II CPUs

*NOTE: These benchmarks have all been done without the Windows 7 BD patch (some are even incorrect, such as the chart showing the FX-8150 falling behind the i7 2600K in 7-Zip)


Threading Tricks or Not?
http://www.anandtech.com/show/5279/the-opteron-6276-a-closer-look/6

AMD claimed more than once that Clustered Multi Threading (CMT) is a much more efficient way to crunch through server applications than Simultaneous Multi Threading (SMT), aka Hyper-Threading (HTT). We wanted to check this, so for our next tests we disabled and enabled CMT and HTT.

First, we look at raw throughput (TP in the table). All measurements were done with the "High Performance" power policy.

However, Hyper-Threading is only capable of delivering a 3-8% performance boost.

While it is not earth shattering, CMT does not disappoint: we measure a very solid 24-37% increase in throughput.

In the end, both multi-threading technologies improve performance. CMT seems to be quite a bit more efficient than SMT.

http://www.anandtech.com/show/5279/the-opteron-6276-a-closer-look/7

MS SQL Server 2008 Power Analysis

We'll let power consumption be the final judge:

CMT increases the amount of power consumed by 6-10%, but only at high loads. The extra clusters probably allow the modules (as AMD likes to call the cores) to sleep more frequently at lighter loads, and we measure no increase or even a small decrease in power consumption. The message is clear: there is no reason to disable CMT when running MS SQL Server.

Hyper-Threading seems to increase the power dissipation always. At higher concurrencies, the higher performance must be paid with a 10-14% power increase, so you might consider disabling Hyper-Threading if your want to cap maximum power output for some reason (e.g. getting to close to the maximum amount of amps allowed in your rack).

http://www.anandtech.com/show/5279/the-opteron-6276-a-closer-look/8

Two Bulldozer Modules are capable of slightly outperforming four cores of the Opteron Magny-Cours. The ideas behind Bulldozer are sound: two modules are smaller (157 mm²) and more power efficient than four K10 cores (231 mm²).

http://www.anandtech.com/show/5279/the-opteron-6276-a-closer-look/10

MySQL OLTP

The Intel X5650 gets a 30% boost from SMT, which is more or less equal to adding two extra cores (compare Xeon X5650, which is a hex-core, and the E5640 quad-core). This shows that this benchmark scales well over more cores, threads, or clusters.

The second integer cluster inside the new Opteron offers 40% more performance. So once again, CMT does the job.

http://www.anandtech.com/show/5279/the-opteron-6276-a-closer-look/12

It is clear to us that quite a few things are suboptimal in the first implementation of this new AMD architecture. For example, the second integer cluster (CMT) is doing an excellent job. If you make sure the front end is working at full speed, we measured a solid 70 to 90% increase in performance enabling CMT (we will give more detail in our next article). CMT works superbly and always gives better results than SMT... until you end up with heavy locking contention issues. That indicates that something goes wrong in the front end. The software applications that do not scale well could be served well with low core count "Valencia" Opteron 4200s, but when we write this, the best AMD could offer was a 3.3GHz 6-core. The architecture is clearly capable of reaching very high clockspeeds, but we saw very little performance increase from Turbo Core.

CMT Efficiency
http://www.behardware.com/articles/842-9/amd-fx-8150-and-fx-6100-bulldozer-arrives-on-am3.html

With the aim of getting a clearer picture of things we carried out the following tests on a Bulldozer processor clocked at 3.2 GHz:

- 4-module, 8-core test
- 2-module, 4-core test
- 4-module, 4-core test

Why carry out this test? It allows us to see if AMD’s claim that two CMT cores are equivalent to 80% of two same architecture standard cores is valid and whether the performance improvement from four to eight cores justifies AMD’s 8-core label.

In this first table, we set an index of 100 to the 4-module, 4-core version.

If you exclude the tests that don’t fully load four cores (WinRAR, the 1st x264 pass and games), you get between 71 and 95% of the 4-module / 4-core performance with the 2-module / 4-core configuration. AMD’s claim with respect to CMT efficiency therefore appears to be right. Where an application doesn’t fully load all four cores however, the 4-module / 4-core mode is fastest, even if the gap is often reduced.

Can Bulldozer therefore be thought of as an 8-core architecture or should we rather be talking about a 4-core, 8 thread processor as with Intel’s processors with Hyperthreading. Here’s a breakdown of the gains you get with a) Hyperthreading on Sandy Bridge, b) moving from four to six cores on K10 and c) CMT on Bulldozer in the most multithreaded applications:

On average, Hyperthreading gives a gain of 23.4%, moving from four to six cores on K10 a gain of 42.2% and CMT in Bulldozer 53.1%. We’re therefore well beyond what you get with Hyperthreading and talking about Bulldozer as an 8-core architecture therefore does seem most accurate.

AMD’s Bulldozer CMT Scaling
http://atenra.blog.com/2012/02/01/amd’s-bulldozer-cmt-scaling/

Introduction

[...]


CMP, SMT and CMT

[...]

Testing procedure

In order to measure the Thread scaling performance we used the following three microprocessors from AMD and Intel.

AMD Phenom II X6 1100T, AMD FX 8150 and Intel Core i7 2600K.

Processor frequencies remained constant at base level by disabling Turbo in all three processors.

Phenom II X6 was used as a reference for the CMP processor. We measured with a single core, dual core, quad core and 6-core. Base frequency 3.3GHz

Intel Core i7 2600K was used as the SMT processor since it has Hyper-Threading. We measured with a single core, single core with HT (dual threads SMT), dual core(CMP), dual core + HT (4 threads SMT), quad core(CMP) and quad core + HT (8 Threads SMT). Base frequency 3.4GHz

AMD FX8150 was used for the CMT processor. It was measured with single core, single Module, two cores, two modules, four cores and four modules. Base frequency 3.6GHz

Mod x1 Core 1 = Single thread (Only a single core in a single Module)
Mod x1 Threads x2 = Dual Threads CMT (both cores in a single Module)
Mod x2 Threads x2 = Dual Threads CMP (Only one core per two Modules used)
Mod x2 Threads x4 = Quad Threads CMT (dual Modules with four cores)
Mod x4 Threads x4 = Quad Threads CMP (Four Modules with only one core per Module)
Mod 4 Threads x8 = 8 Threads CMT (Four Modules with 2 cores each)

POV-Ray 3,7 RC

We start with POV-Ray, AMD’s FX single thread performance is 21% lower than last generation Phenom II and 74% lower than Intel’s Core i7.
Scaling from single core to dual Tread in CMT is 77,14% when SMT scaling is at 31,9%. With 8 Threads the FX8150 scales higher than 8 thread SMT.

Cinebench 11,5 (MultiThread)

The single core performance of the FX8150 is very low in this benchmark, but CMT scaling is very high. SMT scales at 24,63% at dual thread and CMP scaling of four threads is at 399,25 for the Intel processor. FX exhibits the higher scaling both in CMP and CMT modes against the Core i7 in this test.

7-zip

The 7zip score is the compination of both Compress and Decompress scores of a 32MB file. FX 8150 single thread performance is on par with Phenom II but still lags 32% of the Core i7 2600K. Again FX-1850 has the higher scaling both in CMP and CMT modes and it manages to catch the performance of an 8 thread SMT Core i7 2600K.

x264 HD v4.0 (Second Pass)

This is the first benchmark that FX8150 single thread performance is faster than Phenom II but Intel Core i7 is still 26,5% faster. Because FX has a strong single thread performance and higher thread scaling it manages to catche Intels Core i7 performance at the Dual thread SMT mode and continues to be in front of it even in quad and 8 thread SMT mode.

TrueCrypt (AES)

Because Phenom II doesn’t have AES, we only tested FX and Core I processors in this benchmark. Again FXs single thread performance is lower by 35% than Intels Core i7 but due to higher scaling it catches the Intel processor at the end.

Conclusion

AMDs CMT design scales much higher than Intels SMT and it is close to the 80% claims AMD have made. It seams that if FXs single thread performance is close to or better than 35% in relationship to Core i7 single thread performance, then both CPUs perform the same at a higher thread level. But if the single core performance is lower than 35% then the higher CMT scaling cannot help the AMD FX processor to catch Intel’s stronger single thread performance.

It will be very interesting to see what will happen if AMD could increase its single thread performance keeping the higher scaling at the same time. CMT is an interesting and new technology for the desktop platform and windows as well software is not yet optimized for it. When software can take advantage of FX SIMD’s instructions and CMT scaling as shown in AES TrueCrypt, the performance is in par with the Intel Core i7.
The AMD CMT architecture in Bulldozer has a lot of potential and performance will only go up in the future with Piledriver and more optimized software.

So, is the FX-8xxx an 8-core? Yes, and no:

• It has 8 integer cores, similar to Intel's 8-core/16-thread Sandy Bridge-E X79 Xeons
• Phenom II X6 has 6 total, Phenom II X4 has 4 total

Is the FX-8xxx a 4-core? Yes, and no:

• It has 4 "Flex" FPUs, similar to Intel's 4-core/8-thread i7 2700K Sandy Bridge, and 4-core/8-thread i7 3770K Ivy Bridge
• Phenom II X6 has 6 total, Phenom II X4 has 4 total

Also, this:

Piledriver core improvements:

http://www.neogaf.com/forum/showpost.php?p=34047016&postcount=974
http://www.neogaf.com/forum/showpost.php?p=34191413&postcount=977
http://www.neogaf.com/forum/showpost.php?p=34197197&postcount=978
http://www.neogaf.com/forum/showpost.php?p=34823039&postcount=1020
http://www.neogaf.com/forum/showpost.php?p=35327427&postcount=1056
http://www.neogaf.com/forum/showpost.php?p=35348336&postcount=1058

 

[Hazaro]

I skimmed most of it, but it just seems like they are saying that it scales better and the cores work more like cores than HT.

That's all true and great and works out for video stuff and the server space I guess.

 

[markot]

Its so sad that AMD doesnt even rate as an option for anyone building a system. How much longer can they keep it up?

 

[Mudkips]

Its so sad that AMD doesnt even rate as an option for anyone building a system. How much longer can they keep it up?

It's so sad that you don't know what you're talking about.
I'm currently speccing out servers and the AMD options are at the top of the list.

I can get twice as many cores as a comparable Intel server, 50% more RAM, and still save money. This means 2 ESXi hosts running 5 and 3 VMs are replaced with 1 ESXi host running 8 VMs.

 

[Tallshortman]

Its so sad that AMD doesnt even rate as an option for anyone building a system. How much longer can they keep it up?

It definitely rates as an option as the low to medium price market they are very competitive and probably sell better on the lower priced CPUs. They're just not as competitive on the higher end for gaming. They still beat Intel on price to performance for everything except the top tiers of gaming performance. It's also competitive for people building machines that need heavy multithreaded threaded applications that aren't gaming. I own a 2600k but to say AMD is no longer an option is ridiculous.

 

[hteng]

this new fabric tech sounds crazy, i bet Intel has something similar brewing.

 

[Postman]

I still have a phenomII x2 550 running 3.1ghz and ddr2 yet I can still run most games in 1080p full settings.

Just saying.

 

[GameplayWhore]

Its so sad that AMD doesnt even rate as an option for anyone building a system. How much longer can they keep it up?

Aye, "anyone building a system" is really broad. What you meant was "any bleeding-edge dedicated gamer" which translates to "a very, very small minority of system builders".

The 48-core AMD system I installed last week and have been playing with since says hi.

 

[·feist·]

As much as I don't care for sourcing sites like SemiAccurate (their skewed, pro-AMD/ATI stance, combined with blatant anti-Intel/Nvidia remarks being among the reasons), I'll post another site reporting on their rumor.

Despite the what the title implies, and the ongoing rumors, there is no confirmation that the PlayStation 4 will run an AMD CPU, GPU, or APU.

Another win for AMD and their non-standard product marketing
http://www.pcper.com/news/General-Tech/Another-win-AMD-and-their-non-standard-product-marketing

AMD has a double win to announce this morning as they are not only going to be providing the silicon for the graphics on the PS4 but will also be providing the GPU. SemiAccurate goes into the details of what this chip ... or chips ... might be like as Sony has a history of designing very unique systems but have definitely soured on the Cell architecture.

"Yes, you heard that right, multiple sources have been telling SemiAccurate for some time that AMD won not just the GPU as many are suggesting, but the CPU as well. Sony will almost assuredly use an x86 CPU for the PS4, and after Cell in the PS3, can you really blame them? While this may point to a very Fusion/Llano-like architecture we hear that is only the beginning."

"If you look at the Bulldozer architecture, it does have some really creepy high level similarities to Cell, doesn’t it? SemiAccurate’s sources won’t spill the beans on the exact generation of CPU and GPU that are in the PS4, but we expect it to be a very customized version of an existing or near future design."

"Sony is going stacking crazy. The leaks all say that there are multiple additions to the core CPU/GPU chip, and they are not on the same die. Actually, given the steady stream of hints surrounding stacking coming from our Japanese speaking moles, the CPU and GPU could very well be on separate, or even stacked dies too."

"PS4 might be a ‘two chip fusion’ design with the intent to weld the two when the technology allows.

If you do that, you need an interposer, something that Intel has been talking about for a while, and recently shown off parts of too. Luckily, if you know where to look, you will see that AMD is behind, but not by much. Given what they are showing off, the tech will be more than ready by any realistic PS4 ship date. With an interposer, you can do things like stack memory on it, and stack a large number low wattage chips."

I'll leave it there for now.

this new fabric tech sounds crazy, i bet Intel has something similar brewing.

Between internal developments, and their acquisitions, Intel always has something cooking up.


Intel buys Qlogic InfiniBand business for $125 million
http://www.eetimes.com/electronics-news/4235052/Intel-buys-InfiniBand-business

Intel buys RealNetworks' patents and video coding tech
http://www.bbc.co.uk/news/technology-16746971

Intel to Acquire Fulcrum Microsystems
http://newsroom.intel.com/community...1/07/19/intel-to-acquire-fulcrum-microsystems

I skimmed most of it, but it just seems like they are saying that it scales better and the cores work more like cores than HT.

That's all true and great and works out for video stuff and the server space I guess.

You know, I haven't been in the PC thread in a good while, so I have no idea what discussions have been like lately. Still, it wouldn't entirely surprise me to see comments like the one in the post below yours being made. Even before that one there have been recent remarks made about BD in here that show a basic misunderstanding. People are still pining for the mythical "Phenom II/III X8" that wouldn't have worked.

It's pretty clear that per core efficiency is BD's biggest area of concern, and thankfully they're currently addressing that with tweaks (although big structural changes won't come until Steamroller, or Excavator at the very latest). I'm just highlighting multi thread efficiency as an already established strong point, which will also improve as latency and bandwidth are worked on.

 

[·feist·]

Another brief comparison of Bulldozer in Windows 7 to Windows 8.


Windows 8 - Installation, Basic Overview and AMD Radeon 7750 / FX-6100 Performance
http://www.hardwareheaven.com/revie...on-7750-fx-6100-performance-introduction.html

Test Specification and Methodology

XFX Radeon HD 7750 1GB DD
AMD FX-6100
2x 4GB Corsair Vengeance DDR3-1600

Windows 7 Ultimate 64-bit SP1
Windows 8 Consumer Preview 64-bit
AMD Catalyst February 2012
DirectX 11

Battlefield 3
Skyrim
3DMark 2011 Professional
Cinebench R11.5 x64
Fraps

In our first test, Cinebench we see that the FX-6100 CPU (set to 3.3GHz) scores 4.0 points in Windows 7, this is a little more than Windows 8 where it hits 3.97points. The PC Mark scores however show far more variance with the three most notable results being significant increases, on Windows 8 for Video Transcoding, Image Manipulation and Data Decryption which allow the new OS to score much higher. It also edges ahead on 3DMark but not by much, so let's see about real world gaming...

Gaming Performance

In these two tests we have taken popular, recent games and run them on identical hardware/settings for both operating systems. The only point of note is that AMDs drivers for 7000 series cards are not ready for public download yet however those who want to try those cards with the new OS can use the Catalyst 12.2 pre-certified drivers (they simply lack the WDDM 1.2 functionality of the full release).

Battlefield 3 was tested at 1680x1050 with no AA, 16xAF and all other settings at their maximum.

Skyrim was tested at 1680x1050 with FXAA, 16xAF and all other settings at their maximum.

So, overall we have a situation when gaming on the XFX 7750 with the newest available AMD driver that the two operating systems perform at very similar levels. The highs and lows vary slightly between the two however the averages are near identical. We also noted that both games tested were completely stable on Windows 8 without any image quality issues when compared to Windows 7.

Conclusion

Looking first at performance we have a set of results here that show Windows 8 will perform at very similar levels to Windows 7 on current hardware and drivers. There are a few exceptions, shown by our PCMark results, where the new OS varies in performance and as the drivers mature, especially on the video card front, we may see further improvements. Users should have no issue though with building a system based on the FX-6100 and XFX 7750 combination (combined cost of less than £200) to power their Windows 8 build as each provides a speedy, stable experience.

 

[specialguy]

You know I was actually thinking the newly released FX4170 might be a nice little gamer chip. Priced at 150, quad core, 4.2 ghz. I was vaguely interested.

I had forgotten how absolutely horrible for gaming the FX architecture is. I bet my ancient 2.8 ghz Q6600 might be faster, no joke. I found a site that had benchmarked the 4170 months ago via google

 

[DeFiBkIlLeR]

As much as the AMD marketing department try and tell me they are selling 4,6 and 8 core CPU's , my head just wont stop telling me they are really 2,3 and 4 core CPU's with AMD's version of hyperthreading (albeit a very efficient hyperthreading).

If they had marketed them as such from the beginning, i might have been open to replace my Phenom 2 setup with a Bulldozer rig, but they didn't, so i wont.

Sorry AMD, you had my loyalty for 6yrs, but you blew it with Bulldozer, having a lower IPC than a Phenom II is just unforgivable in 2012, so Ivy Bridge it is for me.

 

[specialguy]

As much as the AMD marketing department try and tell me they are selling 4,6 and 8 core CPU's , my head just wont stop telling me they are really 2,3 and 4 core CPU's with AMD's version of hyperthreading (albeit a very efficient hyperthreading).

Me too! Especially when I see these terrible benchmarks.

Not to anger Feist but, ok lets see for the consumer gamer, these chips really could not have been worse. Also just looked at some Anand gaming benches where the 8150(!) barely keeps pace with the Phenom X4's (usually losing slightly) and is destroyed by the Sandy Bridges.

Maybe it's good for the server market or some crap I dont know about, but for me it's horrendous. They would have been infinitely better off just shrinking their old phenom X4 cores to 32nm unchanged (where they could have probably fit at least 8 on a die, and maybe upped the clock too), and saved a few billion and years in R&D, and it would have performed much better.

It's most frustrating because I've always argued that AMD would be well served by making a chip design that excels in gaming even possibly at the expense of some other things if needed, yet the FX is the exact opposite of that. It's an absolute dog in gaming.

 

[·feist·]

You know I was actually thinking the newly released FX4170 might be a nice little gamer chip. Priced at 150, quad core, 4.2 ghz. I was vaguely interested.

I had forgotten how absolutely horrible for gaming the FX architecture is. I bet my ancient 2.8 ghz Q6600 might be faster, no joke. I found a site that had benchmarked the 4170 months ago via google

You're forgetting that the FX-4xxx are not priced anywhere near any of the AMD, and Intel quad cores people always compare them too. That isn't by chance. Your Q6600 had a street price of ~$800, before price cuts eventually whittled that down. The AMD quads, which fall within the FX-4xxx's launch price range, are only there due to cuts that have seen them drop quite a ways from their own launch MSRPs.

FX-4xxxs have two FPUs, like an i3 2100, and four integers, like an i5 2500K. So, despite some objections, AMD can call it a quad. Even if AMD had hit their initial performance projections, you'd still see it fall behind a four FPU quad in certain workloads.

Again, they're priced similarly to Intel's 2C/4T i3 parts. Of course, someone will then mention that you have 65W i3s, going against 95W FX-4xxxs. Which is true, though that has been the case for years. It's as if everyone forgot how AMD sold 125W 45nm parts, alongside Intel's 95W Core 2 Quads, and quad core LGA 1156 products.

As much as the AMD marketing department try and tell me they are selling 4,6 and 8 core CPU's , my head just wont stop telling me they are really 2,3 and 4 core CPU's with AMD's version of hyperthreading (albeit a very efficient hyperthreading).

If they had marketed them as such from the beginning, i might have been open to replace my Phenom 2 setup with a Bulldozer rig, but they didn't, so i wont.

Well, after the old management was replaced, the new management fired nearly all of their old marketing department, and PR people last fall.

Still, performance aside, the behavior of their CPUs are generally closer to a CPU with the number of integers they each have, as opposed to their number of FPUs. If per core efficiency, had been in the ballpark of, say, LGA 1156/LGA 1366, we'd be having an entirely different discussion.

Not to anger Feist but...

Heh. If it wasn't clear enough already, I have zero allegiance to, or affiliation with, AMD. I'm not bothered by anyone choosing to buy whatever they want. This is merely an info thread, in part meant to help dispel some of the common misunderstandings. I only ask that people take a moment to look over some of the info posted in here (with some main points of interest included in the OP, which has an update section) to see that some of the arguments posed, are flawed, and why. Either way, despite my redundancy on the matter, it's not that serious.

 

[·feist·]

8-Core 65W Socket AM3+ Opteron 3280 SE On Sale
http://ht4u.net/news/25213_bulldozer_mit_einer_tdp_von_65_watt_erreichen_den_handel/

Already in October of last year showed some information about the Opteron-3200 Series, which are to be used in the server segment as well as in the desktop area. Now, the first model achieved quiet trade.

The sighted processor, which has surfaced in the Asian region is the AMD Opteron 3280 SE for socket AM3 +. The processor was HE already known with its specifications as Opteron 3280, which is why there is no really new evidence here. So, the model provides eight cores with a frequency of each 2.4 GHz. The L3 cache is 8 MB and the TDP is located at 65 Watts.

In terms of price the processor in the trade is 27.480 yen, what are converted between 250 and 260 euro [street price is ¥ 27480]. It is unclear, however, when the processor in Europe will be to have. As well, the two smaller variants Opteron 3260 EE and 3250 EE are missing. They have two switched off modules and thus only four cores, the TDP is also at only 45 watts.

AMD Server Processors
http://www.amd.com/us/products/server/processors/Pages/server-processors.aspx

 

[·feist·]

You try to tell folks in advance, and some ignore it, until...


Advanced Micro Devices, Inc. - Special Call
http://seekingalpha.com/article/411671-advanced-micro-devices-inc-special-call

AMD dumps SOI
http://www.electronicsweekly.com/Articles/08/03/2012/53158/amddumps-soi.htm

AMD Ditches SOI: Kaveri goes Bulk at GF, More Details From the New WSA
http://www.brightsideofnews.com/new...lk-at-gf2c-more-details-from-the-new-wsa.aspx

On the conference call regarding the Wafer Supply Agreement (WSA) amendment covering 2012, AMDs CFO Thomas Seifert delivered a few interesting statements that reveal some small details regarding the roadmap going forward. We will also go into a few details on the impact of the new WSA on AMD.

There is one statement made in the Q&A section of the call on the new WSA that caught our attention:

"So with respect to SOI (Silicon On Insulator), we made statements that on 28-nanometer, all of our products will be bulk."​

Also he basically confirmed, that Kaveri, the successor to Trinity is going to be manufactured at GlobalFoundries at the 28nm node:

"If you look at the roadmaps that we have presented at Financial Analyst Day, there is a 28-nanometer successor product to Trinity on the roadmap that we will ramp next year, and that is also manufactured at GLOBALFOUNDRIES."​

Kaveri will integrate 2-4 Steamroller CPU cores (successor to Piledriver cores in Trinity, 3rd Generation Bulldozer core) and a GPU based on the GCN architecture. It will also provide support for HSA, which means integration of third party IP into the same die. The APU is scheduled for a 2013 launch.

Departure from SOI
At the Morgan Stanley Technology, Media and Telecom Conference Thomas Seifert last week gave a bit more detail on the matter:

"We said that at the 28nm node we are going to be on bulk silicon across all products, not only graphics but also CPUs. And We have made no statement beyond that. But for 28[nm] we will be on bulk for all products."​

He also added:

"There are always tradeoff decisions. But the flexibility that we gain moving in that direction... the flexibility across foundry partners, across design tools outhweigh that by far, the benefits of SOI."​

So the reasoning behind moving away from SOI is centered on being more flexible with manufacturing. How big the impact on the performance (or lack thereof) will be, remains to be seen. It shouldn't materialize until high-end CPUs are being manufactured on 28nm bulk silicon, which will only happen in 2013.

While technically he hasn't denied the use of SOI going forward, for example at the 22nm node, at this point it seems a bit unlikely. Once you have transitioned your entire product lineup to a bulk process you are not going to move back to SOI at the next half-node step again, unless there are significant benefits to do so. The statement regarding the increased flexibility at a bulk process are a strong hint that going forward AMD would rather avoid SOI.

Regarding development of 22nm node technology, Seifert answered a bit ambiguously:

"We have very strong and engaging discussions on the roadmap, and we will provide updates on our roadmap as we proceed. But this is not a reflection of disengagement on future nodes."​

It doesn't preclude a 22nm bulk process at GlobalFoundries, but depending on what processes TSMC offers it might be smarter to go straight to 20nm bulk silicon in order to leverage multiple sources of bleeding edge processes.


The details of the amended WSA

[...]

After SeaMicro Deal Goes Awry, Intel Tries to Save Face
http://www.hardwarecanucks.com/news/after-seamicro-deal-goes-awry-intel-tries-to-save-face/

Last week AMD performed an impressive coup-de-data-centre against Intel, acquiring server manufacturer SeaMicro for its intellectual property – of which Intel once sought. Now, in a brave act to save face, Intel has said that it wasn’t even interested in SeaMicro in the first place.

“We weren’t interested,” said Diane Bryant, the general manager of Intel’s data center group, to Wired Enterprise. “We looked at the fabric and we told them thereafter that we weren’t even interested in the fabric.”

The IP behind the fabric is one of the reasons why SeaMicro is an asset to AMD. While the exact technical details are a protected trade secret, this technology apparently is a more power-efficient vehicle for transmitting data though server farms than Ethernet or InfiniBand.

Ms. Bryant’s disinterested sentiment is a notable change from the previously warm relationship that existed between Intel and SeaMicro. In early January the two companies hosted a joint press conference, where SeaMicro announced it was planning on shipping power-efficient servers with Intel’s Xeon chips.

Now, Ms. Bryant claims that Intel has its own fabric plans in the works. Speaking to Wired Enterprise she said, “We believe we have a compelling solution; we believe we have a great road map. We just didn’t feel that the solution that SeaMicro was offering was superior.”

AMD, in contrast, is still confident they have made a strategic acquisition.

“With the acquisition of SeaMicro, we believe we will have the best server IP portfolio in the marketplace,” AMD spokesman Phil Hughes said.

 

[·feist·]

ASRock First Introduces AMD MB Supporting Lucid Virtu Universal MVP™ Technology – A75 Pro4/MVP
http://www.asrock.com/news/index.asp?cat=News&ID=375

This is the first AMD motherboard supporting Lucid Virtu Universal MVP™ technology in the world. It allows gamers to take advantage of this new technique and enjoy the amazing performance boost from HyperFormance™ and Virtual Vsync™.

1. HyperFormance™ : A new technology which combines integrated and discrete graphics for uncompromised performance and improves gaming frame rates 30-70%. Special highlight here is that HyperFormance is also supported with both ATI™ and NVIDIA® graphics cards.

Taken NVIDIA® GTX560 graphics card for testing. When enabling Virtu HyperFormance in ASRock A75 Pro4/MVP motherboard, the graphics scores increase significantly. Compared to the integrated VGA and discrete VGA, it provides up to 41.12% performance boost in 3DMark 11 (Extreme mode) test. HyperFormance gives a pretty good showing in performance term, demonstrating a huge improvement in graphics result.

2. Virtual Vsync™ : A technology designed to solve the ever lasting debate between quality and performance. Bring GPU performance from 60 FPS Vsync limit back to its maximum while keeping Vsync image quality and no tearing.

http://www.asrock.com/news/images/20120308-3.jpg

3. Switchable Graphics : Dynamically assigns tasks to be the best graphics resource regardless of physical cable connections.

PR, and marketing aside, Lucid Virtu Universal MVP is very real. It's platform agnostic, can be retrofitted, and is being used as a big selling point for Ivy Bridge 7 series motherboards.


Lucid's 'smarter vsync' could revolutionize game performance - It's called HyperFormance, unfortunately
http://techreport.com/articles.x/21682

Just after publishing my Inside the Second article, I jetted off to the Intel Developer Forum last week in order to soak up the latest info on industry happenings. Little did I know that I would end up having a very interesting conversation with the folks at Lucid about one of the more intriguing concepts we'd discussed in that article. Yet that's exactly what happened, and we're now able to relay some news about a "better vsync" implementation with the potential to revolutionize the way graphics solutions deliver consistent and responsive in-game motion.

[...]

If the latest Ivy Bridge launch date holds, Trinity desktop APUs, and motherboards should launch roughly 2-3 weeks afterwards.


Socket FM2 A85 FCH motherboards for Trinity coming in June
http://www.kitguru.net/components/s...-fch-motherboards-for-trinity-coming-in-june/

Through discussions with sources in the motherboard industry at CeBIT it’s been determined that AMD’s upcoming A10 and A8 “Trinity” APUs will be launched near the end of Q2 and into Q3 (May – June), with accompanying motherboards to start arriving in June.

AMD’s Trinity APUs are built using the new FM2 package which means the requirement of a new breed of FM2 socket motherboards due to backward incompatibility with the FM1 socket. This will begin with a bit of a sidestep initially where The “Virgo” platform will migrate the existing A75 FCH chipset with Trinity’s FM2-accommodating socket.

The following run of FM2 boards said to be arriving in June will be based around AMD’s next-gen A85 FCH chipset which we still know very little about at this time other than picking up USB-IF certification, which indicates the possibility of incorporated native USB 3.0.

CeBIT also saw the launch of various Trinity-based mobile products from laptop vendors.

Along with that, AMD had some Trinity mobile and desktop demos (3-screen Eyefinity, running off a development board, powered by a single desktop APU's iGPU), and the usual raft of marketing slides.


AMD Trinity Notebook
http://www.youtube.com/watch?v=mRaA-k0oRII

AMD Trinity Eyefinity
http://www.youtube.com/watch?v=eb4uEvaUTjA

AMD’s 2nd Generation APU, Codenamed “Trinity,” Will Enable Superior Multimedia Experience for Our “Connected” Generation
http://blogs.amd.com/fusion/2012/03...ia-experience-for-our-“connected”-generation/

• In 2012 we will introduce our “Trinity” ultra-low voltage (ULV) APU, ideal for premium ultrathin notebooks and other innovative form factors. This 17W BGA APU is designed to match the experience enabled by the AMD 2011 APU (35W) at half the power.
• “Trinity” uses the next-generation “Piledriver” modular CPU core for improved notebook compute performance of up to 29% (as compared to our “Llano” first generation AMD A-series APU).
  [*]That “Piledriver” CPU core combined with AMD Radeon™ HD 7000 discrete-class graphic cores provides up to 56% more graphics performance boost in notebook games and the increasing number of applications that utilize massively parallel computing.
  

[...]

http://prohardver.hu/hir/cebit_2012_amd_trinity_apu_szolgaltatas.html

"AMD HD Media Accelerator, Quick Stream Technology and Perfect Picture HD"

"MotionDSP vReveal, Polycom VuRoom and ArcSoft Link"

 

[Refreshment.01]

HyperFormance™ : A new technology which combines integrated and discrete graphics for uncompromised performance and improves gaming frame rates 30-70%. Special highlight here is that HyperFormance is also supported with both ATI™ and NVIDIA® graphics cards.

Been thinking about this since the integrated GPU was announced. I thought that AMD could use the APU as an answer to Nvidia Physx. Thta would've been a compeling reason to have both an AMD procesor with IGP and a discrete card.

"Hyperformance" sounds better if true. But i dont understand something, will this technology also be available for Intel CPUs?

 

[Zombie James]

What? I don't get how an APU or a video driver is supposed to give me lower pings or more bandwidth.

 

[Rufus]

What? I don't get how an APU or a video driver is supposed to give me lower pings or more bandwidth.

I assume they are comparing equal connection qualities here.

 

[Lactose_Intolerant]

lol at the guy trying to drive on the notebook video.

 

[Veal]

Lucid's 'smarter vsync' could revolutionize game performance - It's called HyperFormance, unfortunately
http://techreport.com/articles.x/21682

Please tell me this is going to be software that could be used with older IGP's!! If I could use the HD3200 IGP in machine to get rid of input lag, while maintaining vsynch, that would be a dream come true!! I HATE screen tearing. I'd rather have a slower framerate than screen tearing!

 

[·feist·]

Been thinking about this since the integrated GPU was announced. I thought that AMD could use the APU as an answer to Nvidia Physx. Thta would've been a compeling reason to have both an AMD procesor with IGP and a discrete card.

"Hyperformance" sounds better if true. But i dont understand something, will this technology also be available for Intel CPUs?

Yes. You may have overlooked the link in that post which details its use on an Intel platform.

I'd like to see how it responds across different iGPUs from both camps.

http://i.imgur.com/2zyWB.jpg

What? I don't get how an APU or a video driver is supposed to give me lower pings or more bandwidth.

Software based network/packet control. It's not unlike what you can do with some download managers, firewalls, or other programs. OEMs can set it however they want, such as prioritizing gaming specific use. You should be able to easily disable it, or, at least, select which programs you'd like to have priority.

Asus has something similar with their ROG GameFirst.

Please tell me this is going to be software that could be used with older IGP's!! If I could use the HD3200 IGP in machine to get rid of input lag, while maintaining vsynch, that would be a dream come true!! I HATE screen tearing. I'd rather have a slower framerate than screen tearing!

It can be retrofitted to Intels, and AMDs that didn't leave the factory with the feature. I'm not at all certain of this, but I believe Lucid handles their licenses on a per motherboard model basis, as opposed to a per motherboard vendor basis. If so, you'll likely see it on a very limited number of older Intel 6 series, and Llano boards. It's going to be a major selling point for getting you to buy a new board, so vendors would much rather you buy an Ivy Bridge 7 series, or Trinity board - leaving Virtu Universal MVP as an exclusive feature of sorts.

 

[·feist·]

Well, already indirectly sourced the likes of SemiAccurate, so why not go ahead and add some Fud...

AMD FX 8350 Vishera CPU production in Q3
http://www.fudzilla.com/home/item/26253-amd-fx-8350-vishera-cpu-production-in-q3

Eight cores and better scores

AMD flagship FX 8350, the one that will replace the current flagship in the AMD world known as FX 8150, will enter in production in Q3 2012. We don’t know if it can make it for a launch in the same quarter, but it looks like that is the plan.

Currently the new FX processor is expected as an engineering sample, and AMD tells its partners that this is expected to happen in early Q2 2012 and by early Q3 2012 it should get to production candidate sample stage.

It all goes well in Q3 2012 AMD enters production and launching this production whenever it feels like it has enough of them. AMD FX 8350 is based on eight Vishera 32nm cores that will support Enhanced AMD turbo core technology, DDR3 1866 and fit the same AM3+ socket. At this time AMD doesn’t mention any clocks.

It replaces FX 8150 clocked at 3.6GHz with 8 cores, 16MB cache and turbo ability to jump to 4.2GHz in needed. This one is based on 32nm Zambezi desktop bulldozer core. AMD FX8350 should upgrade the performance of AMD top category and Vishera cores are promising some better performance at the same clock.

Trinity A85X chipset ready
http://www.fudzilla.com/home/item/26382-trinity-a85x-chipset-ready

FM2 support

Some of you are aware that currently available AMD A75 FCH and AMD A68 FCH support for both FM1 and FM2 sockets. Despite that there will be a new chipset that is shipping right now to customers and is meant for Trinity CPU and the new Virgo FM2 platform.

It comes with 8 SATA 6Gb/s ports as well as 4 USB 3.0, 10 USB 2.0 and two USB 1.1 ports if anyone still needs that. It also has RAID 0/1/5 and 10 support and it does support Crossfire. It has a TDP of 7.8W and ships in FCBGA 656 packaging.

AMD has been shipping A85X FCH since mid-Q1 2012 and it should launch some motherboards based on A85X together with Trinity CPUs in Q2 2012. The price of these boards should not get too high, and we expect to see them retail in the €60 to €100 range, depending on the spec.

It sure looks like a decent chipset for the new king of Fusion graphics, the Trinity APU.

Dual core Trinity in Q3 2012
http://www.fudzilla.com/home/item/26383-dual-core-trinity-in-q3-2012

Two 65W SKUs

AMD is not ready to talk about the clock speeds of dual-core Trinity cores but at least we have managed to learn their names.

AMD plans to introduce the faster and unlocked A6 5400K, a dual-core Trinity with HD7560D DirectX 11 graphics. We don’t know the clock speed but the new CPU supports socket FM2 and DDR3 1866 memory.

The second SKU, the slower of two is A4 5300 and this one has HD 7580D graphics that at least sounds a bit faster than the 7560D on the A6 5400K. Both CPUs are 65W parts and both are obviously meant for desktop computers.

AMD still keeps their clocks quiet but we can tell you that slowest of quad core Trinity desktop parts the A8 5500 runs at 3.2GHz and with AMD turbo implementation can reach even 3.6 GHz.

We would not be surprised to see some even higher clock speeds as Dual core Trinity parts will only have two cores enabled and a lot of room for nice clock speeds under 65W TDP.

Kaveri 2013 Fusion supports DDR3 2133
http://www.fudzilla.com/home/item/26280-kaveri-2013-fusion-supports-ddr3-2133

28nm Quad-core Steamroller

AMD plans to replace Trinity at some point in 2013. Trinity is a part of the Virgo platform and features up to four Piledriver CPU cores, supports AMD Turbo Core 3.0, DirectX 11 GPU, socket FM2 as well as DDR3 1866.

The third generation Fusion, or the third generation APU, has up to four Steamroller CPU cores, new DirectX 11 GPU, FS2L socket and DDR3 2133 support. This is the highest DDR3 spec we’ve seen to date and even DDR3 1866 for Vishera and Trinity sounded quite nice.

The codename that AMD wants to talk about is wrapped around Kaveri APU. AMD has revealed in its financial roadmap update that Kaveri APU is a 28nm mainstream part, while Kabini essential level APUs replaces Bobcat-based Brazos 2.0 and they both should be manufactured by GlobalFoundries.

2012 CPU/APU leaks, and official info, for reference:

http://www.neogaf.com/forum/showpost.php?p=35092499&postcount=1039
http://www.neogaf.com/forum/showpost.php?p=35048170&postcount=1035
http://www.neogaf.com/forum/showpost.php?p=34823039&postcount=1020
http://www.neogaf.com/forum/showpost.php?p=34597461&postcount=1005
http://www.neogaf.com/forum/showpost.php?p=34316065&postcount=992

 

[·feist·]

Worldwide PC Microprocessor Revenues in 2011 Rise 13.2% Compared to 2010, According to IDC
http://www.idc.com/getdoc.jsp?containerId=prUS23376112

Intel Dominated, AMD Gained in Chip Market in 2011: IDC
http://www.eweek.com/c/a/IT-Infrast...AMD-Gained-in-Chip-Market-in-2011-IDC-148600/

Intel Continues to Dominate, But AMD Manages to Slightly Gain CPU Market Share in 2011 - IDC.
http://www.xbitlabs.com/news/cpu/di...ightly_Gain_CPU_Market_Share_in_2011_IDC.html

Via loses more ground in x86 chip market to Intel and AMD
http://www.computerworlduk.com/news...ses-more-ground-in-x86-chip-market-intel-amd/

Via fights Intel, AMD even as its x86 chip market share shrinks
http://www.computerworld.com/s/arti...AMD_even_as_its_x86_chip_market_share_shrinks

edit:

 

[·feist·]

Benchmark rumor time. You know the drill.


Trinity A10-5800K Engineering Sample -vs- Llano A8-3850


"AMD" Trinity first burst of APU A10-5800K performance
http://bbs.pceva.com.cn/thread-39867-1-1.html

AMD will officially launch in June this year Trinity APU and supporting the A85X chipset, forming new Virgo Virgo platform. Road map shows the new generation of AMD Trinity relative in Llano APU in terms of CPU performance will raise 26%, while the GPU portion is 56% to upgrade.

CPU-Z screenshot: recognition errors, the correct frequency should be 3.8GHz

A few days ago, China PCEVA Forum users of exposure for us the next generation of Trinity APU A10-5800K test results ， A10-5800K has two pile-modules (four physical cores), the default frequency is 3.8GHz, dynamic acceleration 4.2GHz, 4MB the second-level cache, an integrated HD display 7660D graphics core, with 384 stream processors, the default frequency is 800MHz. From the test results, A8-5800K complete Super Pi 1MB test 23.775 seconds, 3DMark 06 9,396, CPU 4304,SM 2 3285,HDR/SM 3 4,067.

Contrast A8-3850 complete Super Pi test took 26.039 seconds 1MB, 3DMark 06 6,223, CPU 3814,SM 2 2139,HDR/SM 3 results of 2,552.

From the test results, Trinity APU Llano APU on the CPU performance is only a 10% not to, while the GPU performance upgrade over 50%, the so-called piling too much so.

Of course, this exposure is only a prototype, testing is only limited to the 3DMark 06 and Super Pi, officially released in June when the Trinity of results may be more slowly.

Rumored Trinity line up & specs:
http://www.neogaf.com/forum/showpost.php?p=35048170&postcount=1035
http://www.neogaf.com/forum/showpost.php?p=35092499&postcount=1039


edit:

 

[·feist·]

DH Special: Two new FX processors from AMD
http://www.donanimhaber.com/islemci/haberleri/DH-Ozel-AMDden-iki-yeni-FX-islemci-geliyor.htm


・FX-6130 (3.6-3.9GHz/6C/L2 3x2MB/L3 8MB/125W)
・FX-4130 (3.8-3.9GHz/4C/L2 2x2MB/L3 4MB/125W)

To launch in the second quarter, AMD is working on two new FX processor model. We receive in the details, according to representatives of the new Bulldozer architecture, the FX family of processors and quad-core FX-4130 will be the six-core FX-6130 models. Both are 32nm process technology in the form of Socket AM3 + processors, the FX-4130, as a standard 3.8GHz at 3.9GHz with 2.0 technology that serves and TurboCore 'can go up automatically. Similar to the processor, 4MB of Level 2 Level 3 4MB memory capacity, memory capacity. The price of the processor's thermal design power of 125 watts for the moment unknown.

AMD plans to offer other new model as the standard 3.6GHz FX-6130 works at 3.9GHz and with TurboCore 2.0 technology 'can go up automatically. With a capacity of 8MB of Level 2 and Level 3 6MB of memory from the processor's thermal design power of 125 watts. FX-4130 model of information such as the price is not yet clear about the processor will have taken place on the shelves in the second quarter. FX-4130 and FX-6130, the core design of AMD's Piledriver (updated Bulldozer) based FX Next generation FX processor family, code-named ahead of the new model seems to be the last in the series available.

 

[·feist·]

http://www.neogaf.com/forum/showpost.php?p=35757996&postcount=1083

http://i.imgur.com/Vla7B.jpg
http://i.imgur.com/MXQls.jpg

(PR) AMD Launches New Platform for Dedicated Web Hosting Providers
http://www.amd.com/us/press-releases/Pages/amd-launches-new-platform-2012mar20.aspx

New AMD Opteron™ 3200 Series Processor Family Changes Industry Economics for Single-Socket, Dedicated Hosting and Cloud customers

- Great value with up to 38% better price performance and up to 19% less power per core than the competition;
- ECC memory and server reliability features at a low price-point;
- Fast hardware payback -- in as few as 7 months hosting fees can cover hardware costs -- up to 14% quicker than with the competition;
- Efficient economics for the Cloud with twice the core density per rack.

Key Facts, Performance and Technical Detail

- 45 W to 65 W TDP
- 2.7 GHz base frequency, up to 3.7 GHz frequency using AMD Turbo CORE technology
- 4- and 8-core options
- 2 DDR3 memory channels supporting ECC UDIMM
- 1333, 1600, 1866 MHz memory speed
- Supports 1.5V, 2Rank
- Up to 32 GB memory capacity
- Supports up to 2 DIMMs per memory channel
- Total Cache: 16 MB for 8-core, 8 MB for 4-core
- L2 Cache: up to 8 MB total
- L3 Cache: up to 8MB total

edit:

A Host of Choices
http://blogs.amd.com/work/2012/03/23/a-host-of-choices/

The AMD Opteron 3200 Series processors are designed to deliver the best economics for the hosting community, a part of the business world that often is neglected by traditional offerings. Hosters have a very specific set of needs that includes quick ROI and power efficiency. The AMD Opteron 3200 Series processors enable this through a platform that is designed to deliver all of the benefits of a true server, yet do so with the economics of a desktop platform.

The processors themselves come in 3 models:

• AMD Opteron™ 3250 processor – Priced at only $99 (in 1Ku quantities), this processor is designed to meet the needs for low entry price servers. At less than $25 per core, it makes for an impressive entry point for the market.
• AMD Opteron™ 3260 processor – This model has four cores and a 45W power envelope. The 2.7GHz clock speed and up to 3.7GHz with AMD Turbo CORE technology gives customers a great balance of performance and still delivers a power efficient platform.
• AMD Opteron™ 3280 processor – With 8 cores and a 65W power envelope, this model gives customers a great thread density and low power per core for those workloads that take advantage of a larger number of cores.

Compared to the competition, we see a variety of advantages including:

• Up to 100% more cores [ii]
• Up to 48% lower cost [iii]
• Up to 19% lower watts/core [iv]
• Up to 66% lower cost/core [v]
• Up to 38% better price / performance [vi]

 

[kagete]

The A10-5800K gpu bump looks hawt for entry-level graphics. I wonder if this will eventually make its way onto $500-600 budget laptops that can run diablo at 60fps medium settings 768p.

 

[·feist·]

95W FX-8140s are looking like another step in the right direction. Improving GloFo 32nm process has lead to them being less leaky, reduced TDP/heat, and they overclock well with less voltage needed.

The A10-5800K gpu bump looks hawt for entry-level graphics. I wonder if this will eventually make its way onto $500-600 budget laptops that can run diablo at 60fps medium settings 768p.

Bear in mind the A10-5800K is the highest end desktop APU, with a 100W TDP (although the 65W shouldn't be far off the mark, performance wise). Most of the ~$5-600 mobiles will have 17W, and 35W APUs. Higher performing 45W, and 55W-powered laptops will be above that, unless they're somewhat de-contented, or on sale.

 

[·feist·]

AMD FX-4200 CPU spotted in Gigabyte support list
http://www.cpu-world.com/news_2012/2012032801_AMD_FX-4200_spotted_in_Gigabyte_support_list.html

In the third quarter of the year AMD is going to launch new FX processors with "Piledriver" core, that should improve efficiency of current Bulldozer cores. Before the launch, the company was planning to release a couple of FX chips based on older Bulldozer micro-architecture. In February Donanimhaber published a slide with roadmap for the first three quarters of 2012. If the roadmap is correct, we should see FX-4120 and FX-6120 in the second quarter. It is possible that in addition to these microprocessors, AMD will also release FX-4200, that was spotted today on Gigabyte website by 3DNews.ru:

According to specifications in Gigabyte GA-990XA-UD3 CPU support list, AMD FX-4200 will operate at 3.3 GHz, that is lower than FX-4100, and at the same time have higher 125 Watt TDP. There is one interesting tidbit about this model: it has 8 MB level 2 cache, or twice as much as FX-4100/FX-4170 CPUs. Gigabyte CPU support list is the only source of information on this SKU, therefore we can't tell if it's a mistake in the specs, or the processor does have larger L2 cache, or AMD decided not to follow their naming convention and release 8-core CPU as FX-4200 (which is very unlikely). In any case, the size of L3 cache on the 4200 chip stays the same, as on current FX-4000 series microprocessors. The processor will have core stepping B2.

http://www.gigabyte.com/products/product-page.aspx?pid=3880#bios

 

[Zombie James]

What's going on here? Same specs, same stepping, different TDP?