morrowinder: ARM doesn't really compete in the super computer space...

SPARC and Power have wins in the top 10 but the rest is x86 with one lone Opteron system from DOE hanging around from years ago.

SGI UV 3000 Sets New Throughput Records

Today SGI’s Gabriel Broner announced that the company’s SGI UV 3000 system had set two new benchmarking records, demonstrating the machine’s advanced throughput prowess enabled by its shared memory architecture. The test machine — a 128 socket UV 3000 system equipped with 1,280 E5-4627 v3 2.60 GHz Xeon cores — was assessed using the industry-standardized SPECrate metrics (specifically SPECint_rate_base2006 and SPECfp_rate_base2006), which measure the throughput or rate of a machine carrying out a number of simultaneous tasks.

According to results published by the Standard Performance Evaluation Corporation (SPEC), the system achieved a score of 45,100 for SPECint_rate_base2006 (source) and 42,600 for SPECfp_rate_base2006 (source). Broner notes that the new results put SGI ahead of its nearest competitors by more than 25 percent for SPECfp_rate_base2006 and more than 10 percent for SPECint_rate_base2006 — using 20 percent fewer cores. As you can see in the chart below, SGI overtook a ScaleMP machine comprised of 1,536 Xeon E5-2680 v3 2.50 GHz cores.

“SGI has also published performance results of over twice its closest competitors for both the SPECompG_base2012 shared-memory parallel processing benchmark and the STREAM TRIAD memory bandwidth benchmark,” adds Broner.

The SGI UV 3000 test platform incorporates 128 10-core Intel Xeon E5-4627 v3 chips (2.60 GHz with Intel Turbo Boost Technology up to 3.20 GHz) and 16 TB of memory. The operating system is SUSE Linux Enterprise Server 12.




http://www.hpcwire.com/2016/03/25/sgi-posts-new-spec-cpu2006-results/

Intel Broadwell-EP Xeon E5-2600 V4 Chips Launching on 31st March – Broadwell-E For X99 Arriving at Computex 2016


Broadwell-EP/EX series will have a launch schedule as detailed below:

Broadwell-EP E5 V4 family of processors will be landing in the first half of 2016. This will consist of the Broadwell EP series of processors for one. Which include the Xeon E5 1600 v4 and E5 2600 v4 family of processors for both single and dual socket motherboards. The E5 1600 v4 series will have upto 8 cores whieas the 2600v4 series will rock upto 22 cores and have full support of DDR4 2400 memory. The platform will be compatible with the C610 series chipset. The Broadwell-EP 4S platform (upto 4 sockets on one motherboard) consisting of the Xeon E5 4600 v4 will be launched sometime in Q2 2016.
Broadwell-EP E7 V4 family of processors, constituting the Broadwell-EX platform, and the nomenclature range of E7 4800v4 and E7 8800 v4 will be launched in Q2 2016. It will also be compatible with the existing Brickland platform.
The Xeon Phi X200 series of products (Knights Landing) will be available in Q3 2016. Knight’s Landing was built on the 14nm Process and uses modified silvermont cores (x86 ofcourse). It is also one of the first mass produced components developed for this market segment that features stacked DRAM.

Intel’s HEDT Broadwell-E Is Arriving in Q2 2016, Computex
Another exciting news is that Intel will also be releasing their High-End Desktop (HEDT) platform, codenamed Broadwell-E in Q2 2016. To be precise, Intel will have the lineup ready for unveiling at Computex 2016 and is perfect for users who are eyeing a platform update with upcoming generation of enthusiast-class GPUs (Pascal and Polaris). The lineup will feature several SKUs including the flagship Core i7-6950X which will feature a total of 10 cores.


Read more: http://wccftech.com/intel-broadwell-ep-xeon-e5-2600-v4-march-launch/#ixzz43rX2FSI8

flum: I think its better having a variety of opinions

I think Ash leans to the negative side too often as well but if we only have people who agree with each other we get an echo chamber. More speech is good even if people are wrong. Less speech is bad.

mmoy: A thousand is pretty big...

But your IT department is definitely unconventional. Most won't let users have that much control. I know that last time I checked at Intel they had specific laptops you could order and they have a software store to customize your laptop to your needs as they evolve.. They would pre image them. They even have vending machines where you can enter in your employee ID number and get say a new charger for your laptop. And the only way you can do stuff like that is standardization. You get in big trouble for plugging in a rogue wi fi station or non sanctioned PC on Intels network.

mmoy: it's not just about the end user with Windows pro...

I am sure you have all the hooks you need to log in to your corporate environment just fine with OS X. So you yeah it will work. But Total Cost of Ownership in enterprise is about a lot more than just plugging in to the network. It is about managing the entire server, client and network infrastructure. It's about making a series of complicated software work to gether on thousands or even tens of thousands of PCs. In a small environment you really don't need much management. So it's about giving IT the tools to effectively manage a large installed base cost effectively. Think of software packages like the old LANDESK which Intel used to own. OS X itself is fine if you are sort of loosey goosey about adminstering things(although I bet your IT guys hate the people who bring in their own Macs :P). Intel has a whole driver environment called Active Management Technology to help basically administer the entire network remotely. You can literally use Wake on Lan to turn on a computer and then reimage it from an engineering build to an admin setup and manage the software licenses etc. from a thousand miles away. And they can quickly deploy software patches across the entire network and or get rid of something that is a problem child. Thats where the big difference is and why Macs have never really taken off in enterprise.

Dmcq: Not that Windows Pro :P

This windows 10 Pro:

http://www.microsoftstore.com/store/msusa/en_US/pdp/Windows-10-Pro/productID.319935900?VID=320028200&s_kwcid=AL!4249!3!82561420373!b!!g!!buy%20%2Bwindows%20%2B10%20%2Bpro&WT.mc_id=pointitsem+Google+Adwords+5+-+Windows+10&ef_id=VTmN2QAABBHHux7f:20160322183841:s

Dmcq: Apple has been claiming that for decades...

And yet their enterprise business is still anemic for Macs. They may get fewer calls but there are some serious manageability deficits on the Mac software side and you have to factor in total cost of ownership. There is no OS X equivalent of Windows pro. And yeah Macs are a lot more money than PCs. Intel has been working on enterprise manageabilty forever. They even have deployed Macs interally but they will never be anything but a minority option because of TCO. Apples strength is with the consumer and maybe prosumers in areas like photography and video but not enterprise.

Apple is gunning for PC users, but Microsoft has little to fear

"But it seems unlikely that Apple will manage to commandeer massive swaths of the Windows user base. Businesses appear to be doubling down on Windows with massive, earlier-than-expected rollouts of Windows 10, while consumer interest in iPads appears to be waning, not growing.

“So far, we haven’t seen any real impact from the iPad Pro on Windows 10 uptake with enterprises,” Gartner analyst Steve Kleynhans said. “I’d say generally, enterprises are moving very rapidly towards Windows 10, and nothing from Apple has slowed that down. I don’t think a smaller version of the same device is going to really change that.”

Enterprises, which are in a position to make large-scale purchases of devices like iPads and Windows tablets, are usually slower to adopt things that they aren’t familiar with. IDC Vice President Lauren Loverde said it’s still easier for businesses to manage Windows devices than ones running iOS. That may lead companies to stick to the platform they’re comfortable with."

http://www.pcworld.com/article/3046844/tablets/apple-is-gunning-for-pc-users-but-microsoft-has-little-to-fear.html

borusa: that was constructive confrontation

And for better or worse, that is pretty much dead at Intel. Political correctness seems to be the order of the day these days. Granted I was reacting a little emotionally and Elmer is right, you weren't really saying anything that bad. So I apologize for what it is worth.

Maybe but I am very sad...

Andy was a great man. Certain people here try to diminish him. I WILL NOT let that happen. I think you interacted with him. He was an amazing person. I am very sad.

borusa: obviously you never met the man....

He was brilliant, fucking amazing and succesful beyond your dreams. Fuck off and just fade away you TWAT

Elmer: Agreed :(

Andy was a most formidable intellect...

Remarkable manager, amazing CEO, GOOD man. He will be missed and the company diminishes with his passing sadly.

The new iphone SE supports 150Mbps over LTE...

Is that possibly a win for the 7160 wireless modem from Intel? Apple was typically coy. It is a so called slim modem which fits for a smaller handset. No mention of specific manufacturers. One of the reasons they might have won this business is that you can't switch from ATT to Verizon for example. Which sucks for the customer but the wireless provider loves. Qualcomm would have to be in Verizon/unlocked phones.

http://www.intel.com/content/www/us/en/wireless-products/mobile-communications/mobile-xmm-7160-1-brief.html

TCE: there are flaws in your argument...

I agree that Intel has largely failed in mobile. But server is far more profitable. And I am skeptical about PCs dying. Q1s numbers are important here. Honest question: do you think they will guide down? I don't. They didn't warn. So they enter the quiet period soon. And I WAS right that high end Cores were selling out and your 14nm fail theory was not the case.

Still you look very favorably on AMDs and ARMs attempts to enter there. Certainly one should be paranoid but: Servers are REALLY HARD(TM). It took intel a decade to build up its current position which it has held for a long time. One does not just walk in to the server market and take share. It takes years and years. And in ARMS case they have no obvious advantage. Atom is pretty good for low power. Core dominates every where else.

I also think high end mobile is stalling out. Midrange phones like the Nexus 5x are more than capable and hundreds less than iShiny objects. And 2 year contracts are dying out. Sooooo now people have to shell out 800 big ones for the latest shiny thing with incomes stalled out for 30 years? Yeah you and I have the dinero but most people don't.

And even low end piece of shit phones aren't all that bad. Phones will hit commodity status earlier and faster than PCs imho.

As to Intel's win or not win vs. Qualcomm in modems. It is about establishing a beach head. I don't think it changes the world but its a start if they get some of Apple's business. And carriers have to account for Intel silicon. That is a barrier right now. Certainly Qualcomm appears to be ahead in the specs race. But noone has shown me that 300Mb is even common anywhere. And GigaLTE? I gotta bridge to sell you if you buy that thing happening soon. 5G will be the next big step. I really am tempted to pick up a Nexus 5x and compare it to my shitty Motorola Ultra numbers. I suspect it will get 30Mb too. The average Apple 6plus numbers are telling. So "being ahead" in modem virtual specs doesn't really matter in the real world does it? I also would remind you that the vast majority of broadband users in the US do NOT get faster than 50Mb. I happen to have negotiated 200Mb service from Time Warner Cable. Honestly, even with 4k content on Netflix, I am not really sure it is worth it. Yeah Windows updates are lightning fast but I really do think we are hitting some limits on the upper end of broadband. 30Mb on phones is well pretty great for the vast majority:) Well except for the chronic pirate bay users downloading free illegal movies but they aren't doing that on phones are they...

Hands-on with Intel's Skull Canyon NUC, the most powerful game-ready mini-PC

"Besides packing four of Intel’s latest generation Skylake cores, the chip also features Intel’s most powerful graphics core, the Iris Pro 580.

A typical Core i7 or Core i5 CPU with, say, HD 520 graphics, might feature 24 execution units. The Iris Pro 580 features 72 execution units. It also gets a big boost in performance from an ultrafast 128MB of embedded DRAM on the chip.

How much performance does that add up to? On one Skull Canyon NUC we played the new game Just Cause 3 at 1920x1080 resolution, with frame rates in the low- to mid-30s. For PC gamers that’s, well, not great, but for integrated graphics, that’s actually pretty impressive"

http://www.pcworld.com/article/3045374/hardware/hands-on-with-intels-skull-canyon-nuc-the-most-powerful-game-ready-mini-pc.html

ibc: we will know for sure at the Q1 conference call next month

And you selectively choose the negative bs. I rarely see you post anything positive.

TCE: I usually only skim IBCs constant stream of negative BS...

So I didn't read the end. Typically, Q1 is down from Q4 in consumer laptop sales too so I am not sure there is a lull in laptop or if it is normal seasonality.

TCE: Well that is even lamer.

Server shipments don't slow down like consumer pcs do with a new launch. Why defend a bad analyst?

ibc: ummmm Broadwell was 5th gen core...

We are on 6th gen. Kaby lake is next unless he somehow meant Broadwell E. But that is a relatively small volume product and has nothing to do with laptops.

TCE: Not really that close...

From Anandtechs article: " Overall Skylake sees some decent performance improvements relative to the Broadwell CPU in our MacBook – with the exact gains depending on the test – allowing it to widen the gap compared to the A9X. Overall A9X is still competitive in specific scenarios, but on average it definitely trails the Skylake Core m3." And that is in lower power mobile. Generally desktop has more frequency room as Chipguy has pointed out previously.

And if you compare to the current Celeron G3900 at 2.8 GHZ. And since you want to use the iPAD version which is way more than a pc desktop, it is a thousand dollar product with any reasonable amount of storage versus a 2 to 300 dollar desktop. Add in $69 for a monitor and it still is way less money. Hell of a lot cheaper to get on the internet and VASTLY more screen space. Also a hell of a lot cheaper to get a decent smart phone that just makes good phone calls AND a decent desktop :P

http://ark.intel.com/products/90741/Intel-Celeron-Processor-G3900-2M-Cache-2_80-GHz

TCE: There are plenty of core based Celeron systems that are cheap...

And I would bet that 1007 ivy bridge cpu beats the iphone handily on a real benchmark. You have an overly inflated opinion of iCPUs based on Geekbench, the worlds worst benchmark for comparing mobile to PCs.

TCE: there smartphone processors out there than run circles...

Those phones cost 800 bucks. And if you are in the know you can pick up a low cost CORE based Celeron system for as little as $149.

http://www.bestbuy.com/site/asus-vivopc-desktop-intel-celeron-4gb-memory-500gb-hard-drive-silver/6215037.p?id=1219201612121&skuId=6215037

TCE: interesting. Thanks for the mini review :)

TCE: IBM is a bad example for volume process technology...

They went with a Power speedracer design(a la Pentium 4/Bulldozer) and tried to play the GHZ wars and added tons of cache. They were never built for volume, only small volume extreme high end gigantic die unix server configs which are dying. Intel DOES real volume on their mainstream x86 parts and their server parts.

TCE: True but let's see what happens this year.

If as you say Intel is done in mobile, we will see nothing. On the other hand there could be some positive developments. Meanwhile, the fabless folks have been pretty late on 14/16nm high transistor/high frequency GPUs and CPUs. We shall see how that shakes out this year also or if the fabless people are overly biased to low power phone chips where they have been pretty good.

TCE: I see now Huawei uses Kirin not Snapdragon...

Still overall very strong reviews. If only Samsung would ditch TouchWiz :P

TCE: Hows the Samsung Galaxy S7?

I was reading the anandtech review and I was kind of suprised how much of a mixed bag it was performanced wise. CPU is definitely slower than the 6s and a mixed bag as it is behind the S6 on things like Google Octane. And why is the Huawei Mate 8 so far ahead on many benchmarks? If it is a good usable phone it begs the question, how much CPU is good enough for a phone. I was also very interested in the fact that Apples NAND is so much faster than the rest on 256k writes but in the range on 4k.


"Overall, the Snapdragon 820 appears to provide a pretty healthy bump in performance over almost every SoC seen in 2015, although it's hard to declare a clear winner when comparing it to Apple's A9 or Huawei's Kirin 950. If you glanced at the battery life graphs and the performance graphs above it's pretty obvious that Qualcomm has made some enormous strides here. While not quite going from zero to hero, Qualcomm has come close, and that definitely deserves some credit."

http://www.anandtech.com/show/10120/the-samsung-galaxy-s7-review/3

TCE: Why did you return yours again?

TCE: What is weird is...

The 6 plus consistently underperforms on the upload. I ran it a few more times and always 5-6Mbps. Download is far more important of course but it is weird.

TCE: Don't have ATT but...

I did run the following on Verizon:

Motorola Droid Ultra on LTE: Ookla Speedtest

Test 1 21.81Mbps down 25.40 Up
Test 2 40Mbps up down 25.88 Up

Apple iPhone 6 plus:
Test 1 Down 25.40Mbps Up: 4.13Mbps
Test 2 Down 15.74Mbps Up: 5.37Mbps

The Apple iPhone 6 plus uses this modem:
Model A1549 (CDMA)*
Model A1522 (CDMA)

The motorola Ultra I am having trouble finding out what the Modem actually is. But it does use the Snapdragon S4 Pro processor. Much slower than the 6 plus.

Odd results to say the least.

Open Question: Is Intel really behind in modem technology if carrier aggregation is as rare as it looks to be?

Carrier aggregation seems to be the key to getting really fast LTE speeds. I have been trying to research it and it appears that it is branded as either LTE-A and LTE Advanced. Sprint had a press release last year about how its deploying it in a so called tri-band configuration that acheives up to 60Mbps and with carrier aggregation up to 120Mbps. No where near the 450Gbps Intel modems are capable of.

http://www.fiercewireless.com/story/report-sprint-starts-widespread-carrier-aggregation-deployment-25-ghz-band/2015-07-17

So the first problem I see is that you have to be an expert to know about this. Second is that the branding is vague, so consumers don't know which phones to get to take advantage of higher speeds. Third, I have not been able to see any easy way to find out where exactly you can take advantage of this technology. There are no coverage maps. Sprint refers to 100 or so cities but does not have a map telling you where you can get this. Fourth, now companies are looking at 5G. Will 5G be up and running before carrier aggregation capable of 1Gbps is ever available at least in the US? Remember that most carriers just finished switching to plain old LTE recently(Tmobile is still finishing up according to the below).

http://www.fiercewireless.com/tech/story/verizon-sprint-others-turn-carrier-aggregation-small-cells-remainder-2015-c/2015-08-05

Ash has argued that Qualcomm is way ahead of intel on modem technlogy based on these advanced features. And they are definitely ahead with respect to CDMA, since they own that. But it is not clear to me when the infrastructure will be available to support tri band and carrier aggregation technology. The tests I have seen peg current LTE speeds at 19 to 25Mbps in the LA/OC area in California. And it is pretty easy to deploy cell phone networks here. So the question becomes: when will 450Mbps and or 1Gbps be available. And are there any actual tests that show Qualcomm vs. Intel in a real world situation? Thanks for anyone with any thoughts. Edit: for reference, I should point out that at home I get 200Mbps on my Cable modem. This is rare enough. And I am not completely sure I need it other than the fact that I have 8 plus devices hooked up to my wi fi network. So another question is, at what point will consumers stop caring about more bandwidth?

TCE: Isn't available spectrum kind of limiting?

Qualcomm certainly wins the checklist and press release battle but does it actually make a difference in the real world?

I was looking at this Rootmetrics report and the top average speed in Orange County CA where I live is 25.5. Now maybe they aren't implementing the latest LTE features but is anyone? We actually have really good cellphone infrastructure at least on Verizon. Do you know anyone implementing these theoretical 450Mbps or 1Gbps speeds or a date when that would even be available? It involves 4X carrier aggregation just to get to 450 and it seems to me that is problematic. None of the advanced modems matter at 25Mbps. And 25Mbps is actually pretty good. LA gets more like 19.5Mbps.

Also, as I understand it, certain things don't necessarily benefit all that much from new process technologies as much. Do we have any evidence that 14nm does much with a wireless radio? I know that chipsets often use trailing edge technology because it makes less sense to be cutting edge on those.

This seems to be similar to bragging about 20Mp camera sensors, when we all know damn well that there are significant other factors limiting picture quality including low light performance and the actual lens.

http://www.rootmetrics.com/en-US/rootscore/map/metro/mission-viejo-ca/2015/2H

TCE: Is the Samsung phone really that much better?

You were very enthusiastic over the Apple 6s and it appears to be the one to beat even now. Is the screen really that much better? Do you notice any performance differences?

flumoxed: Good rule of thumb: never wrong to take a profit...

I still have significant holdings in Intel but they are down from my high. Whenever we approach the mid 30s or even the 30s sometimes I sell some. 2000 has scarred me for life :P

I recognize there are some headwinds for Intel. I have been fairly frustrated with the lack of new products this year so far. But I also recognize that saying: Intel is done in mobile after one stupid mobile conference is premature. My personal theory is that Intel screwed up going with Wimax and they are behind on LTE. And they seem to regularly underestimate the power of the carriers: ATT and Verizon. They really dictate who wins the merchant market at least in the US. And it should be clear by Qualcomms crappy earnings that Samsung and Apple own the high end.

One of the other things that bugs me is that Atom while not successful in phones has done fairly well in 2 in 1s and lowend PCs. Intel has learned how to lower costs from mobile and can now successfully compete in the $200 pc market. While Ash regularly reminds us how much Atom "sucks", he fails to account for its good success in lowend and low power PCs. Intel is learning a lot there.

And I also recognize that Intel appears to be in cruise control on 4g but they appear to have embraced 5G and they are not "done" with mobile yet. So they may have their next opening there. Too early to tell. I do think mobile application processors appear to be in decline. So yes Intel failed to dent that market but it is already becoming very commoditized. And while Ash beats Intel with a stick over that he fails to recognize that Qualcomm is getting screwed and faces a horde of cheap chinese companies. And anti trust bullshit. But he regularly has nice things to say about them, excusing the rest of it while beating Intel down repeatedly. He also is apparently out of love with Apple and bought the latest Samsung phone after declaring it incredibly boring on twitter :P Oh well consistency is the hobgoblin :)

TCE: as to Qualcomm being the owner of the majority of merchant Smartphone market...

How's that working out again?

"Qualcomm Inc. said its earnings fell 24% in the most recent quarter, the latest sign that slowing sales of high-end smartphones from Apple Inc. and others are hurting component suppliers."

http://www.nasdaq.com/article/qualcomm-posts-24-profit-drop--update-20160127-01301

TCE: you pretend to be an analyst which implies neutrality...

When in fact you exhibit the worst of the sensationalist trends of the old Inquirer/Register set. Even using the word "chipzilla" erodes your credibility. Mike Magee invented that. Do you even know who he is? Yeah. Take a good look at Mike and Charlie D, its all been done before.

TCE: I am an Intel investor and biased towards Intel success...

What exactly is your excuse? You lack objectivity and are helping people like Nenni pump the fabless sector and Apple and Qualcomm. Qualcomm
hasn't quite been as successful as you were predicting aside from modems. Your credibility is eroding everytime you talk smack. Charlie D did a lot of that. Look where it got him...

TCE: Have I ever had anything nice to say about Windows 10 for mobile?

Nope. MS failed bigtime. The only way Intel could maybe win is if Win 10 for PCs ran on a phone. And even then I would still be skeptical. Apple and Android have won mobile. Period. And I certainly wouldn't want the failure that is HP having anything to do with a broxton win. They have zero phone share and that is unlikely to change.