Slashdot Mirror

The feature that comes to mind that some companies have been hammering at for years is raytracing. I remember a project that intel was doing some years ago with the Return to Castle Wolfenstein source (way back when) to make the engine completely ray traced. I also remember it took a good $10k+ computer to render in less than real time (numbers escape me, and the site appears to be down now). While it did create some pretty unbelievable graphics for the time with true reflections on solid surfaces such as glass and metal, it was completely unapproachable for consumers due to cost, and unusable for gamers because it wasn't real-time yet. Intel does seem to be continuing the work though.
There will always be folks using the cards for password cracking and other "simple" massively parallel tasks. My vote goes for increasing the speed that they can generate more realistic imagery. Then that same 100k (guesstimating) render farm over at Pixar or Dreamworks will give us movies that further blur the line between real and fantasy.

The tech said (via a post) that there was no flashing utility for windows. I would have to use Linux. I said that I couldn't just wipe my hard drive and install linux, and the guy laughed at me and told me to buy another hard drive.

Intel did the right thing and deployed their SSD upgrade software as a bootable CD. In my opinion, this is currently the best way to distribute any kind of PC firmware. You can burn the disc from inside any operating system, and when you boot from that medium, you get a nice clean environment to update the device without a full-blown OS interfering with the process.

Better still, learn about the Intel Phi and let the Intel compiler do the parallelization, all without having to learn a different architecture and API.

The Galileo board looks fairly interesting, but have you seen the Intel Software License Agreement that you have to accept to download the Galileo software?

Apparently nobody told Intel's lawyers that this project was open source and open hardware, and that derivatives based on GPL-licensed sources cannot be encumbered by additional restrictions. That license agreement is completely out of step with everything else in the Arduino community, not to mention also being out of step with Intel's very strong support for Linux in all other areas.

(I expect this is just a temporary mistake by Intel, but it's a big one.)

The "someone" mentioning 230 Hz is INTEL, in their Galileo FAQ.

http://www.intel.com/support/galileo/faq.htm

The question is near the end, specifically "What is the maximum rate at which GPIO output pins can be updated?"

The answer, which you'll see if you click that link and expand the question to see the answer, is:

The GPIO output pins on Intel® Galileo are provided by an I2C Port Expander that is running at standard mode (100 kHz). Each I2C request to update a GPIO requires approximately 2ms. In addition to software overhead, this restricts the frequency achievable on the GPIO outputs to approximately 230 Hz.

While I like the idea of having an Arduino compatible board running Linux to do some more advanced projects, I don't understand what drove Intel to force this board to be Arduino compatible. The Quark processor is not designed for this sort of stuff as it has neither a sufficient number of GPIO pins nor any ADCs. It sure has a lot of interfaces (SPI, I2C, PCI-E, SD-Card, Serial etc.), but it lacks the things that are useful for a hacker project.

So they had to include a separate GPIO extender chip (over a slow I2C interconnect) as well as an separate ADC. The Quark SoC has some 15 GPIO Pins, the extender another 40. But of those 55 Pins only 20 GPIO pins are actually available on the Arduino shield pins -- the rest is used for all the Muxes to switch pins between the ADC, the GPIO Extender and the Quark SoC to emulate the flexibility of the Arduino AVR processor.

While I haven't looked at the actual PCB schematic, I think the board layout is also strange. The ADC is on the opposite side from the analog input pins, meaning that all analog signals have to travel a long distance in the vicinity of some high speed digital signals. And the GPIO Extender Chip is on the opposite corner from all the digital output pins.

This, together with the BGA devices (SoC, RAM), seem to indicate that this is at least an 6 layer board which will make it hard to clone this design -- something that IMHO has contributed to the success of the Arduino. The Schematic for this board has 27 pages compared to the single page of the Arduino Uno

It seems that this Board is designed more as a competitor to the Raspberry Pi than to the Arduino, both in price and in features.The Arduino compatibility is just some marketing thing which makes the board overly complex and more expensive than it needs to be.

But hey, it sure must be fun to employ a few million transistors and a full blown operating system to run the Arduino Blink demo :-)

WebGL is fine.

From this page, Atom N450, Xubuntu 12.04, Xorg 1.11.3, Firefox 24: "Hmm. While your browser seems to support WebGL, it is disabled or unavailable. If possible, please ensure that you are running the latest drivers for your video card." This Mozilla page recommended looking in the "Additional Drivers" utility that came with the operating system, but all it showed me was the driver for a Broadcom NIC that I'm already using. The Mozilla page also referred me to Intel's driver update utility, but that's Windows-only.

"Secret" 3G Intel Chip Gives Snoops Backdoor PC Access

vPro processors allow remote access even when computer is turned off

Paul Joseph Watson | Infowars.com | September 26, 2013

http://www.infowars.com/91497/

Intel Core vPro processors contain a "secret" 3G chip that allows remote disabling and backdoor access to any computer even when it is turned off.

Although the technology has actually been around for a while, the attendant privacy concerns are only just being aired. The "secret" 3G chip that Intel added to its processors in 2011 caused little consternation until the NSA spying issue exploded earlier this year as a result of Edward Snowden's revelations.

In a promotional video for the technology, Intel brags that the chips actually offer enhanced security because they don't require computers to be "powered on" and allow problems to be fixed remotely. The promo also highlights the ability for an administrator to shut down PCs remotely "even if the PC is not connected to the network," as well as the ability to bypass hard drive encryption.

"Intel actually embedded the 3G radio chip in order to enable its Anti Theft 3.0 technology. And since that technology is found on every Core i3/i5/i7 CPU after Sandy Bridge, that means a lot of CPUs, not just new vPro, might have a secret 3G connection nobody knew about until now,"reports Softpedia.

Jeff Marek, director of business client engineering for Intel, acknowledged that the company's Sandy Bridge" microprocessor, which was released in 2011, had "the ability to remotely kill and restore a lost or stolen PC via 3G."

"Core vPro processors contain a second physical processor embedded within the main processor which has it's own operating system embedded on the chip itself," writes Jim Stone. "As long as the power supply is available and and in working condition, it can be woken up by the Core vPro processor, which runs on the system's phantom power and is able to quietly turn individual hardware components on and access anything on them."

Although the technology is being promoted as a convenient way for IT experts to troubleshoot PC issues remotely, it also allows hackers or NSA snoops to view the entire contents of somebody's hard drive, even when the power is off and the computer is not connected to a wi-fi network.

It also allows third parties to remotely disable any computer via the "secret" 3G chip that is built into Intel's Sandy Bridge processors. Webcams could also be remotely accessed.

"This combination of hardware from Intel enables vPro access ports which operate independently of normal user operations," reports TG Daily. "These include out-of-band communications (communications that exist outside of the scope of anything the machine might be doing through an OS or hypervisor), monitoring and altering of incoming and outgoing network traffic. In short, it operates covertly and snoops and potentially manipulates data."

Not only does this represent a privacy nightmare, it also dramatically increases the risk of industrial espionage.

The ability for third parties to have remote 3G access to PCs would also allow unwanted content to be placed on somebody's hard drive, making it easier for intelligence agencies and corrupt law enforcement bodies to frame people.

"The bottom line? The Core vPro processor is the end of any pretend privacy," writes Stone. "If you think encryption, Norton, or anything else is going to ensure your privacy, including never hooking up to the web at all, think again. There is now more than just a ghost in the machine."

Facebook @ https://www.facebook.com/paul.j.watson.71
FOLLOW Paul Joseph Watson @ https://twitter.com/PrisonPlanet
---
http://www.intel.com/content/www/us/en/enterprise-security/

Gloom - even the best only supports 4GB RAM. Not enough for ZFS server.

http://ark.intel.com/products/76760/Intel-Atom-Processor-Z3770-2M-Cache-up-to-2_39-GHz/

Actually it appears that Intel removed the A20 line starting with Haswell.

Check out page 271 of the Intel System Programmers Manual Vol. 3A from June 2013. Notice the following excerpt: "The functionality of A20M# is used primarily by older operating systems and not used by modern operating systems. On newer Intel 64 processors, A20M# may be absent."

Now check out page 368 from the May 2011 version of that same document. In the same paragraph, the statement above is not present.

From this, we can infer that between May 2011 and June 2013 some new Intel chip dropped support for A20M#. In that timeframe, Ivybridge and Haswell are the only 2 chips that were released. Since Ivybridge is the same architecture as SandyBridge just manufactured on 22nm and we know that SandyBridge did have A20M#, I think its a fairly safe assumption that Haswell is the first x86 chip that has _finally_ done away with A20M#. That said, it would be nice if Intel actually said in the manual which chip was the first to remove it.

Does someone have a new Haswell system that they can do a quick DOS ASM program on to verify this? Even better, if someone has an Ivybridge system we can narrow this down.

I have read - http://software.intel.com/en-us/articles/intel-digital-random-number-generator-drng-software-implementation-guide

There is no untestable magic there:
1) entropic source
2) digial state algorithm
3) async sampling

"The ES runs asynchronously on a self-timed circuit and uses thermal noise within the silicon to output a random stream of bits at the rate of 3 GHz. The ES needs no dedicated external power supply to run, instead using the same power supply as other core logic. The ES is designed to function properly over a wide range of operating conditions, exceeding the normal operating range of the processor."

The digtal part behind the entropic source is what the article discusses. Digital clouds are fully verifiable with scan, even async ones with proper test logic insertion which breaks loops. Fully analog entropic circuits like the thermal noise source are also verifiable.

What makes it "random" is that it a) has an entropic source and b) runs async to the rest of the design. Both of those are testable with proper test circuits. The pertebations described in the article are commonly tested for faults. If they were nto detected 5-10% of processors would not work at all.

The article talks about it being undetectable because they are only looking at the reduced space random sequence and that is effectively still random. Scan test is able to look at the value of every logic state and how it come about (ok there is not usually 100% coverage, but large portions of the design do get 100% coverage and LFSR type logic is easily covered in scan).

haswell makes full windows with 100% backwards compatibility in a tablet device a desirable thing

Actually, if I'm not mistaken, the Atom line has been the one championing x86 tablets. Also, it is the line Intel feels is their best bet for entry into the tablet and phone market: http://www.anandtech.com/show/7263/intel-teases-baytrail-performance-with-atom-z3770-cinebench-score

The end result is the same, though:
RT is destined for the bin.
ARM SoCs are getting competition from SoCs made by a very potent behemoth.
x86 will rise in the mobile market.

To further support the latter I'd like to note that Intel is also putting effort into getting Android x86 working on the Atom, with success:
http://reviews.cnet.com/tablets/asus-transformer-book-trio/4505-3126_7-35827211.html

Also:
http://www.pcworld.com/article/2044617/new-intel-chief-sees-150-atom-tablets-this-year.html
http://www.intel.com/content/www/us/en/tablets/tablets-atom.html

I think you would download the library from intel. http://software.intel.com/en-us/articles/intel-digital-random-number-generator-drng-software-implementation-guide

Or use one of the many libraries like OpenSSL that use it.

Or read from /dev/random, because the OS does it for you. Well partially.

What makes you think Linux's RNG is 'best practice'. There are many of us in the huge and diverse RNG designer community that think it stinks.

These days, almost every time a story is posted along the lines of "Linux says X" it's frequently framed in such a way as to paint Linus as a frothing madman hurling not just insults but entire furniture factories at his cringing subordinates. It's become such a regular occurence that I half expect them to be followed up with a story on how Steve Ballmer has converted to buddhism and will be using the armpit sweat from his meditations to irrigate the sahara.

Reading the article, of course, usually reveals a different picture, but that gets in the way of attention-grabbing headlines. I'm not really sure how the following post can be construed as "fury"; irritation, indignation, perhaps, but not fury.

Where do I start a petition to raise the IQ and kernel knowledge of people? Guys, go read drivers/char/random.c. Then, learn about cryptography. Finally, come back here and admit to the world that you were wrong. Short answer: we actually know what we are doing. You don't. Long answer: we use rdrand as _one_ of many inputs into the random pool, and we use it as a way to _improve_ that random pool. So even if rdrand were to be back-doored by the NSA, our use of rdrand actually improves the quality of the random numbers you get from /dev/random. Really short answer: you're ignorant.

As far as I can tell, no-one's found any evidence for rdrand being backdoored, and even if it were, there's bigger issues at foot with things like microcode. Linus explains how the kernel implementation uses random data from several different sources to guard against this kind of stuff. Plus, as other people have pointed out, you can disable rdrand with a kernel parameter. Linus is primarily a pragmatist, so it doesn't really make much sense to excise the code from the kernel - throwing out the baby with the bathwater if you will. Surely if there were any hardware to worry about, it'd be the hardware providing AES-NI? Why isn't there a petition to have that removed...?

Shouldn't we be welcoming RdRand with open arms? It's a mathematically proven high-quality random number generator that lets chips like Ivy Bridge & Haswell produce large amounts of true random data (not a simple PRNG data) at multi-gigabit speeds.

There are some excellent slides describing RdRand here: http://software.intel.com/en-us/tags/20757

I would strongly recommend using it wherever feasible as it is a great boon to security in Linux.

So is some AMD/ARM fanboy saying that it's not fair that AMD/ARM haven't bothered to implement RdRand yet so therefore nobody should be allowed to use it? How about we extend that logic to other pieces of hardware? Say, when AMD comes out with an improved GPU, let's say that Linux shouldn't support it because Intel doesn't have the same hardware.. fair is fair right?

I also wonder how deliberate is the confusion. There are MANY areas inside Intel where there is confusion. The confusion is visible even when visiting the Intel campus in Oregon.

Funny story: I visited the Intel web site and was asked to complete a survey. I gave a few of the reasons why Intel CEO Paul Otellini should be fired, like paying $6 Billion for McAfee when Microsoft is giving away its Microsoft Security Essentials anti-virus software. A few months later Otellini left Intel; they didn't say why. I'm not saying my survey answers had an influence, I'm only making the point that the perception of Intel is widespread.

Intel has a long record of failure with consumer products. Now a completely separate division plans a TV product (???): Intel Media aims to remake TV with its own technology. This paragraph indicates some confusion and lack of competent direction: "Intel Media is run by Erik Huggers, an Intel vice president who worked previously at Microsoft and the BBC. He's assembled a team from such high-tech and media heavyweights as Apple, Netflix, Microsoft, Sky TV and Sony. Intel engineers in Oregon are participating, too, providing technical support for the project."

Oh... The Intel people are providing "technical support". Everyone else came from outside Intel??? And they don't know enough about technology to do their own support? There are many, many issues like that inside Intel.

We are having problems with Intel RAID. Intel technical support is poorly organized.

Apparently only the CPU and chipset division of the company is well-run. All other parts of Intel seem to have little competent supervision.

I also wonder how deliberate is the confusion. There are MANY areas inside Intel where there is confusion. The confusion is visible even when visiting the Intel campus in Oregon.

Funny story: I visited the Intel web site and was asked to complete a survey. I gave a few of the reasons why Intel CEO Paul Otellini should be fired, like paying $6 Billion for McAfee when Microsoft is giving away its Microsoft Security Essentials anti-virus software. A few months later Otellini left Intel; they didn't say why. I'm not saying my survey answers had an influence, I'm only making the point that the perception of Intel is widespread.

Intel has a long record of failure with consumer products. Now a completely separate division plans a TV product (???): Intel Media aims to remake TV with its own technology. This paragraph indicates some confusion and lack of competent direction: "Intel Media is run by Erik Huggers, an Intel vice president who worked previously at Microsoft and the BBC. He's assembled a team from such high-tech and media heavyweights as Apple, Netflix, Microsoft, Sky TV and Sony. Intel engineers in Oregon are participating, too, providing technical support for the project."

Oh... The Intel people are providing "technical support". Everyone else came from outside Intel??? And they don't know enough about technology to do their own support? There are many, many issues like that inside Intel.

We are having problems with Intel RAID. Intel technical support is poorly organized.

Apparently only the CPU and chipset division of the company is well-run. All other parts of Intel seem to have little competent supervision.

Well, at least Intel thinks it owns Pentium as a trademark. From: http://www.intel.com/content/www/us/en/trademarks/pentium.html

Usage Guidelines for the Pentium® or Intel® Pentium® Trademark

Whenever the Pentium® name appears, the following footnote must also appear: "Pentium is a trademark of Intel Corporation in the U.S. and/or other countries."

Maybe the "unun" in front means that ununpentium doesn't infringe on it. So maybe AMD could name a processor, "ThisIsNotaPentium", and be OK.

No need to take my word when citation is so easy.

http://ark.intel.com/products/71072/Intel-Celeron-Processor-G1610-2M-Cache-2_60-GHz

Its not on the spec sheet, but it has a 6 EU (execution Unit) Intel GPU, roughly equal to Intel HD2500. Not spectacular, but i played Bioshock:Infinite on it at 720p/low and got 33 fps in the benchmark.

Non-free hardware documentation is generally a problem but in this case all of the USB specs are available for free.

Itanium is not a RISC or CISC CPU. It is EPIC (Explicity Parallel Instruction Computing). Sheesh.

But, the i960 is and wasn't even mentioned.

Don't buy a TPM module? Just because a motherboard supports it doesn't mean you have to turn it on... or am I missing something?

It's pretty much impossible to get a new system with any reasonable compute ability without at least some form of back doored TPM-like facility these days. For example, the new Intel Ivy Bridge Chipsets have vPro, which gives similar capabilities. Likewis, the new AMD systems currently being planned have the ability to run TZones in the on-board ARM processor to implement a software TPM, as long as they aren't exposed out directly.

http://www.hardwaresecrets.com/news/Intel-Launches-Ivy-Bridge-CPUs-with-vPro-Technology/6464
http://newsroom.intel.com/community/intel_newsroom/blog/2012/05/15/intel-strengthens-security-boosts-performance-for-business-with-3rd-generation-intel-core-vpro-platforms

The problem is that PHP and web programmers are quite common. Even so, places like Facebook are looking for PHP developers and SQL engineers. Trying to find decent C programmers, especially those capable of working on embedded systems or the Linux kernel or device drivers are much harder to find. As for college, good luck getting started in the industry without a degree unless you've managed to make a name for yourself without it on some well known project.

For example:

(Facebook) https://www.facebook.com/careers/search?q=&location=menlo-park
(Google) https://www.google.com/about/jobs/search/
(Apple) http://www.apple.com/jobs/us/corporate.html
(Tesla) http://tbe.taleo.net/CH07/ats/careers/jobSearch.jsp?org=TESLA&cws=1
(Cavium) http://www.cavium.com/careers.html
(Amazon Lab 126) http://www.lab126.com/careers.htm
(Yahoo) http://us.careers.yahoo.com/
(Xilinx) https://xapps9.xilinx.com/OA_HTML/RF.jsp?function_id=12325&resp_id=23350&resp_appl_id=800&security_group_id=0&lang_code=US&params=mCsTre-AToe2wnIXflPtqsZZTnVM9.N1OyhNnBv5KuqbLKT.chxR3de6DRGMEkZb&oas=suuh5UdozJuyoXGEIHQclw..
(Altera) http://ch.tbe.taleo.net/CH03/ats/careers/jobSearch.jsp?org=ALTERA&cws=1
(Intel) http://jobs.intel.com/
(Qualcomm) https://jobs.qualcomm.com/public/jobSearch.xhtml#messages

I am certainly not lying nor a shill. These are just off the top of my head. Many of these sites have pages of openings as well as openings for new college graduates.

Look into Intel Xeon Phi. It is Intels version of an nivedia tesla. It does not require any special language and is made to program in like a normal intel processor. http://www.intel.com/content/www/us/en/processors/xeon/xeon-phi-detail.html

Yeah. More and more stuff is software nowadays. So much so that nowadays, Software is stuff you configure. Hardware is stuff other people configure ;).

To nongeeks a lot of PC stuff is hardware. To us less so. To some elite hacker most of it is software.

CPUs can be patched after release ( https://downloadcenter.intel.com/Detail_Desc.aspx?lang=eng&DwnldID=14303 ). Same goes for drives and even some mice.

A compute rate that varies with temperature would seem to be a bug, rather than a feature. I don't want a GPU that does that. I need repeatable Gazebo simulations.

I think they're talking about the opposite (a temperature that depends on load), which your CPU has probably been doing for a long, long time.

But you've lost this one, anyway; modern Intel processors have Turbo Boost, meaning the performance does indeed depend on temperature. I was scared, too, from a worst-case provisioning perspective in an environment where I can't predict what will be running on other cores. But I've had a couple years to come to terms with it, and in practice, it doesn't actually seem any worse than other factors like last-level cache contention.

Can you support your allegation that Apple requires payment? Perhaps you are unaware that Apple transferred the Thunderbolt trademark to Intel. Thunderbolt is not at all proprietary to either Apple or Intel; at least since a couple of years ago. It is Intel who controls Thunderbolt now, and they furnish royalty-free licenses. The allegation is that Intel's policies are holding back Thunderbolt, but Intel denies this fairly convincingly.

And in case somebody out there doesn't believe this actually happened, INTEL: Usage Guidelines for the Thunderbolt(TM) Trademark.

To be fair, when Apple DOESN'T fuck up, they don't fuck up.

Thunderbolt and the Thunderbolt logo are trademarks of Intel Corporation in the U.S and other countries

ispc, OpenCL, and LLVM on the way. Failing that, you could of course use C++ and AVX intrinsics (which would be a good choice if you already have a load of SSE4/AVX optimised code lying about).

Using Intel Math Kernel Library on Intel Xeon Phi Coprocessors

No it doesn't. Look up the difference between VT and VT-d. The i5-3570K does not have VT-d (I was aware of that when I bought mine). This feature is only used by Xen and HyperV (I can't speak for ESX) for very specific functions.

Comparison for you (scroll down so you can see VT-d, VPro, and Trusted Execution):

Sandy Bridge:
i5-2500K: http://ark.intel.com/products/52210
i5-2500: http://ark.intel.com/products/52209

Ivy Bridge:
i5-3570K: http://ark.intel.com/products/65520
i5-3570: http://ark.intel.com/products/65702

No it doesn't. Look up the difference between VT and VT-d. The i5-3570K does not have VT-d (I was aware of that when I bought mine). This feature is only used by Xen and HyperV (I can't speak for ESX) for very specific functions.

Comparison for you (scroll down so you can see VT-d, VPro, and Trusted Execution):

Sandy Bridge:
i5-2500K: http://ark.intel.com/products/52210
i5-2500: http://ark.intel.com/products/52209

Ivy Bridge:
i5-3570K: http://ark.intel.com/products/65520
i5-3570: http://ark.intel.com/products/65702

No it doesn't. Look up the difference between VT and VT-d. The i5-3570K does not have VT-d (I was aware of that when I bought mine). This feature is only used by Xen and HyperV (I can't speak for ESX) for very specific functions.

Comparison for you (scroll down so you can see VT-d, VPro, and Trusted Execution):

Sandy Bridge:
i5-2500K: http://ark.intel.com/products/52210
i5-2500: http://ark.intel.com/products/52209

Ivy Bridge:
i5-3570K: http://ark.intel.com/products/65520
i5-3570: http://ark.intel.com/products/65702

No it doesn't. Look up the difference between VT and VT-d. The i5-3570K does not have VT-d (I was aware of that when I bought mine). This feature is only used by Xen and HyperV (I can't speak for ESX) for very specific functions.

Comparison for you (scroll down so you can see VT-d, VPro, and Trusted Execution):

Sandy Bridge:
i5-2500K: http://ark.intel.com/products/52210
i5-2500: http://ark.intel.com/products/52209

Ivy Bridge:
i5-3570K: http://ark.intel.com/products/65520
i5-3570: http://ark.intel.com/products/65702

The i5-3570K does not support VT-d according to Intel ARK.

Go look up the spec sheets for Sandy CPUs. Or better yet Google 3570K and VT-d. Surprise! I found out the hard way myself when I built an ESX server and couldn't install, I found the feature greyed out in the BIOS. A quick Google on that model and I realized I'd been had too.

http://ark.intel.com/products/65520

http://www.tomshardware.com/forum/356118-28-purchased-3570k-virtualization

Obvious sales pitch is obvious:

AMD also has overclocking-friendly K-series parts, but it offers more models at lower prices, and it doesn't remove features available on standard CPUs."

Feature #1 TSE: http://en.wikipedia.org/wiki/Transactional_Synchronization_Extensions I'd imagine nobody codes for this.

Feature #2 : http://software.intel.com/en-us/articles/intel-virtualization-technology-for-directed-io-vt-d-enhancing-intel-platforms-for-efficient-virtualization-of-io-devices

It can still do virtualizion just fine: http://forums.anandtech.com/archive/index.php/t-2133898.html

Not an Intel fanboy or anything, but they're not as arrogant as people are making them sound.

The 4770k's maximum Tcase is 73C.
http://www.intel.com/content/www/us/en/processors/core/4th-gen-core-family-desktop-vol-1-datasheet.html

Intel provides rather extensive technical documentation of all their products. http://www.intel.com/content/www/us/en/processors/core/CoreTechnicalResources.html [intel.com] is the page with basic datasheets (basic in this case meaning a couple hundred pages, their more detailed ones are a thousand). If you truly are as interested in the technical details as you pretend, then go look them up.

I've had a look through, but apart from saying "it has 20 execution units", it doesn't really mention any specific figures (for the actually useful information). It does however state that it's OpenGL4.0, which is a little disappointing (a step up from 3.2, but it's still lagging behind AMD & NVidia).

However if you are just throwing out technical shit in an attempt to deflect the argument then knock it off. Particularly since much of what you are asking for are the kind of the things that would be of concern for high end dedicated GPUs for particular applications, not for an integrated controller for general use.

Well, I'm a graphics engineer in the games industry by trade, so I guess you could say I have a passing interest. The things I am asking for, are things that can help improve the performance of the products I work on. Now you might not find this stuff particularly interesting, however I do. So as a very simple example, I have an order-independent-transparency pass to handle pixel perfect transparency. On the current integrated AMD GPU, I can basically pick between any number of algorithms to achieve this (weighted average, dual depth peeling, etc, etc). Now, which one I choose, is going to be largely affected by what GPU resources I need to use for other things, and this includes: memory, the max number of shader attribs, the max number of bindable texture units, etc; but in general, I have resources to spare, so I am free to pick and choose.
The problem with Intel APUs in the past, is that whilst the last generation may have implemented OpenGL 3.2 to the letter, the max attrib counts and shader instructions were significantly lower than the AMD/Nvidia equivalents. This means you typically have to insert an Intel only codepath, where you will either just rip out the nice stuff, or you'll end up using a much slower multipass technique. As a result, making frame-rate comparisons in any game is most likely to be meaningless (since there is a good chance they are running a simplified codepath for intel).

It's all well and good, and matters for certain markets and applications, but those markets are generally not the ones using an integrated GPU. Most people just care how fast it runs their stuff.

Yes, and No. It's very true that most people just want their stuff to run quickly. However, to say that the legions of people out there running low powered ultrabooks and cheap generic laptops don't care about this stuff, is complete and total bullshit. You might imagine that all gamers have £3000 desktop rigs with all the trimmings, but the reality is infact very different. If I can spend a few months optimising the graphics routines to run a game smoothly at 720p on an Intel APU, then the market sector into which we can sell our product, has more or less tripled. Even if you don't go to the effort, you will probably be forced into making those optimisations anyway. Honestly, you would be surprised at just how many people ignore the minimum system requirements on a game, and simply assume their "i3 Dell laptop is brand new, so it should play the latest games". What are you going to do? Refund half of your sales? Or fix it? If you see sense, you'll fix it, and then most of your users will have the luxury of being able to ask how quickly it runs....

Intel provides rather extensive technical documentation of all their products. http://www.intel.com/content/www/us/en/processors/core/CoreTechnicalResources.html is the page with basic datasheets (basic in this case meaning a couple hundred pages, their more detailed ones are a thousand). If you truly are as interested in the technical details as you pretend, then go look them up.

However if you are just throwing out technical shit in an attempt to deflect the argument then knock it off. Particularly since much of what you are asking for are the kind of the things that would be of concern for high end dedicated GPUs for particular applications, not for an integrated controller for general use.

For most people, what matters is how fast it is at running the programs they want to use, like games. All the other stuff is for, as Tam McGleish would say "Specy wanks who get excited about fuckin' GPU clock speeds and hardware tessellation and all that shite folk who are actually interested in playing games dunnie give a stuff about." It's all well and good, and matters for certain markets and applications, but those markets are generally not the ones using an integrated GPU. Most people just care how fast it runs their stuff.

Most precompiled software is optimized for a pretty old common target. I would be highly surprised if default code emitted by e.g. GCC is slower on a Core now, than it was five years ago.

Most software is not 1/2 of these benchmarks. I agree you build a generic binary for x86_64 with gcc its basically running the same code on all CPUs. But, povray, cinebench and x264 are all highly optimized. It wouldn't surprise me if all of them are using icc, and have tweaked in on way or the other by intel.

PovRay an open source project has "intel attention" http://software.intel.com/en-us/blogs/2009/03/26/some-tips-for-povray-1

Frankly, this is nothing new. The CPU/GPU/machine vendors have traditionally been willing to help high profile or applications selected for benchmarks run well on their hardware. The problem is that these benchmarks aren't representative of a lot of actual work being done on PCs. If you choose a different video encoding library, you may find completely different results.

Basically, IMHO a significant percentage of the generic performance improvement on intel CPU's in the last 5 years have been benchmarking tricks. Nearly every in house application i've been benchmarking with generic x86_64 gcc targets disagrees with the conclusions drawn on places like tom's hardware. Part of that is gcc (its still not as good as icc, at optimizing for a specific target) part of it is that the basic integer/move/compare/etc instructions that comprise most code haven't gotten much faster. In my tests the only thing intel has done since core2 that affects that kind of code in any significant way has been fixing rep stosd so that it runs at a reasonable rate with respect to the dozen alternative ways to copy memory. This is also why there is a lot of talk that the tables are turned in the AMD/Intel land if your running linux/gcc. Its because AMD looks good and even beats intel in many cases on generic semi optimized linux/gcc code paths that don't leverage SSE4.

http://ark.intel.com/

For me, this is by far the biggest architectural improvement I see in these line of processors (check http://en.wikipedia.org/wiki/Transactional_Synchronization_Extensions and http://software.intel.com/sites/default/files/m/9/2/3/41604 for more information). If it sticks, it will help solving a lot of multi-core shared memory software development issues.

Right, Moore's original piece was rather agnostic to the processes which would be required to increase the density.

http://download.intel.com/museum/Moores_Law/Articles-Press_Releases/Gordon_Moore_1965_Article.pdf

The complexity for minimum component costs has increased at a rate of roughly a factor of two per year (see graph on next page). Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000.

http://download.intel.com/museum/Moores_Law/Articles-Press_Releases/Gordon_Moore_1965_Article.pdf

re:

Wow, thanks for the very informative post. It makes sense that being able to deal with full thermal stress would be useful. I've had my quad-core shut down on me once at 20-30 seconds into the boot-up sequence, and then I realized that the heat-sink was not fully applied to or in contact with the CPU surface. I reseated everything and the boot-up sequence went just fine. So I'm really glad that there was an on-die thermal danger detector that shut down my quad-core as the thermal level went to dangerous failure levels. The fact that it's alsodesigned and tested to be actually able to withstand a full power-load draw for a fixed interval is even more impressive.

Here's a link to the wikipedia page about a "Power Virus". And there's an Intel link about "Thermal performance challenges from Silicon to systems" there on the wikipedia page which is a dead link: http://download.intel.com/technology/itj/q32000/pdf/thermal_perf.pdf .

The most notable example being SATA on Intel chipsets:

http://communities.intel.com/message/133881

If Intel wanted to, they could probably have a new driver that enables support for port multipliers before WD releases the disk.

On the internet, 640 terabytes of data is transferred every minute. That means in a year, you have 330 exabytes of data. Not only that, you need the infrastructure to transfer it. You can deduplicate and stuff, but even deduplicating that much data is not exactly an afternoon hack.

Think of that: you're adding 640 terabytes to your database every minute.

Intel still supports and releases binary drivers for the Poulsbo platform.

http://www.intel.com/content/www/us/en/intelligent-systems/intel-embedded-media-and-graphics-driver/emgd-for-intel-atom-systems.html

It has never been a one-shot or abandoned platform. However, you need to understand that this device was specifically designed for embedded applications and not general purpose computing products.

It's unfortunate in my mind that several manufacturers released it as a consumer product.