5 ms·
> Another way to make a chip faster is to add special circuits that only do one thing, but do it extremely quickly. Roughly 25 percent of the E5’s circuits are
by turbohedgehog 10y ago
> Another way to make a chip faster is to add special circuits that only do one thing, but do it extremely quickly. Roughly 25 percent of the E5’s circuits are specialized for, among other tasks, compressing video and encrypting data. There are other special circuits on the E5, but Intel can’t talk about those because they’re created for its largest customers, the so-called Super 7: Google, Amazon, Facebook, Microsoft, Baidu, Alibaba, and Tencent. ... If you buy an off-the-shelf Xeon server from Dell or HP, the Xeon inside will contain technology that’s off-limits to you. “We’ll integrate [a cloud customer’s] unique feature into the product, as long as it doesn’t make the die so much bigger that it becomes a cost burden for everyone else,” says Bryant. “When we ship it to Customer A, he’ll see it. Customer B has no idea that feature is there.”
This is something I've never heard of before. Anyone have some insight into this? Is it a relatively recent phenomenon?
- cushychicken 10y agoThis is basically the same principle as a GPU - specialized hardware that does one task really well. (In the GPU's case, it's lots of math, very quickly.) The most common example I can think of is BCH coprocessing, which most modern application processors use for verifying data stored in embedded Flash.
- hart_russell 10y agoIt's called an ASIC: https://en.wikipedia.org/wiki/Application-specific_integrated_circuit https://en.wikipedia.org/wiki/Application-specific_integrate...
- turbohedgehog 10y agoYeah, I know what ASICs are, I just hadn't heard of large companies getting their own IP inserted into Intel's chips for their own use.
- nickpsecurity 10y agoLook up AMD Semi-Custom Design. Should get some news articles on where they got this started in x86. Became big business that Intel got into, too.
- aab0 10y agoNormally when you think of an ASIC, you think of a single chip custom-designed for a single customer/buyer. You don't think of a single chip with multiple totally different functions and designs consolidated onto it, which get enabled only if they're shipped to the particular customer who requested that particular function. This is like the CPU/GPU core stuff but on steroids - not just 1 or 2 cores disabled to do yield management and price discrimination, but entire processors.
- bravo22 10y agoThis is not uncommon. Usually a product "family" is all from the same wafer. During packaging and testing either certain pads are not bonded out or certain fuses are blown out (via laser cutting, or as an efuse) to disable certain features that are not available on the lower end version. In this case Intel is building certain accelerators for some of its customers, and enabling it only for them before shipping it out. I don't think any of these cores contain trade secrets, rather specific functions they've requested which helps accelerate their stack.
- uudecode 10y agoThis is common not just in Intel chips but in all electronics. Instead of making several versions, each with with more features, only one version is made that contains all possible features. Whether they are turned on or off is what correlates with the "model" and the price. Turn features off and sell as "basic" model. Turn features on and sell as "pro" model. In this story, turning on features correlates with whether the customer has deep pockets and consistently buys in large quantities.
- akiselev 10y agoThis is also important for the economy of making chips at scale. Since silicon fab is very precise a single wafer can have from dozens to thousands of errors, ranging from failed doping to entire scratches on the surface. Making monolithic all in one chips allows you work around those errors instead of throwing out that part of the wafer. Say you're making an i7 and half the chips on the wafer have scratches on one or two of the cores. Instead of throwing out half the wafer (doubling the per CPU price) you just disable the damaged cores and sell them as i5s instead.
- kabouseng 10y agoYeah sure but there is a tradeoff, making larger (more fully featured) chips also reduces your yield as the chance a chip has a defect is dependant on it's size / surface area.
- williadc 10y agoNehalem is the oldest chip I ever saw a floorplan for, and it had a logic block called "misfit" that had special logic requested by a company whose stock ticker you might be able to guess from the block name.
- amelius 10y agoPerhaps they tell this story to cover up that those extra square millimeters are really for the NSA (?)
- PeCaN 10y agoThey don't try particularly hard to cover that up. It's well-enough known that RDRAND is highly likely to be NSA-backdoored. They could always just shove NSA stuff into the ME anyway (currently there's no reason to believe that they do this, however).
- nn3 10y agoI find it highly unlikely just based on basic economics. Imagine the potential conversation between Intel and the NSA: NSA: Please backdoor your RNG for us. Intel: We're making xx billions of this chip. What happens if someone finds out and we have to recall the chip. Are you paying us the xx billions back then? NSA: ... It is fairly unlikely the NSA could afford Intel.
- nickpsecurity 10y agoYou should look at the Snowden leaks sometime. They had cooperation, including backdoors, in a large number of companies' products and networks. RSA cost them about $30 mil to risk whole market. For carriers, usually around $100 mil or so. Also, remember that it's easier to do if you disguise it as a coding flaw. It wasn't a backdoor: just one among many flaws accidentally hitting our systems. Wouldn't cost any market share.
- CyberDildonics 10y agoI think it would be something more like: NSA: Please backdoor all sorts of stuf Intel: ok Security Researcher: these chips have backdoors Intel: Maybe they do, what are you going to do about it?
- PeCaN 10y agoIntel has plausible deniability. A backdoor in RDRAND would be some very minor statistical flaw. It could just be something tiny that makes private keys susceptible to some known or unknown attack.
- twotwotwo 10y agoSo curious what the other hardware is/does. Dan Luu wrote that Intel's cache allocation tech particularly helped Google, so they could run multiple workloads on one machine without totally trashing the larger caches at every context switch: http://danluu.com/intel-cat/ http://danluu.com/intel-cat/ The Xeon D chips, which provide cheap low-clocked big cores and integrated NICs for small servers, were requested by Facebook: https://code.facebook.com/posts/1711485769063510/facebook-s-new-front-end-server-design-delivers-on-performance-without-sucking-up-power/ https://code.facebook.com/posts/1711485769063510/facebook-s-... Someone who follows server chips (maybe AnandTech?) called Xeon D's one of Intel's coolest product lines to come out recently. They're also arguably a smart way for Intel to seal off the lower end of the market against approaches with tons of cheap ARM or AMD cores.
- iheartmemcache 10y agoThe Zeptobars[1] team and/or Chris Tarnovsky/Travis Goodspeed/Bunnie Huang/et al are probably better equipped for reverse engineering hidden functionality. Anandtech's analyses is great, don't get me wrong, when provided with the proper ISA/datasheets and then giving great critique on what implication feature foo chip bar when it's publically available, but I don't think they're the type of people who take red fuming nitric acid to get the epoxy off chips or are experienced enough in microscopy at the um,nm level to perform such analysis. [1] zeptobars.com/en/read/open-microchip-asic-what-inside-II-msp430-pic-z80
- johnflan 10y agoAny other big buyer of compute not mentioned.... No Such Agency?
- gravypod 10y agoAfter reading some of the other replies that more directly address your question you might want to read up on binning [0]. A lot of Intel's chips are recycled binned Xeons. They've almost completely stopped making desktop processors. The I7s you buy are usually failed Xeon chips. To my knowledge Intel currently only fabs laptop and server chips. [0] - https://en.wikipedia.org/wiki/Product_binning https://en.wikipedia.org/wiki/Product_binning
- colejohnson66 10y agoCitation for the binned Xeons resulting in i3 to i7?
- dogma1138 10y agoThat's not true at all in some cases it's actually the other way around entry level xeons are failed i5/7's that can't turbo boost as high. There were only a few xeons that were made into desktop processors and they ended up as i3's there were a few of those that supported ECC. The 'E' edition CPU's can potentially be xeons but then they aren't failed ones but highly cherry picked ones as they boot and overclock very high.
- tambourine_man 10y agoI had no idea either. Powering all that useless (to us) real estate must have profound consequences on our energy budget
- Terribledactyl 10y agoThey most probably have the ability to turn off those sections. Similar to turning off unused cores when turbo boost is used.
- rasz_pl 10y agoReading this you might get excited thinking wow a secret FPGA hidden in a cpu, in reality they are talking about disabling HT, VT-x, AVX, multiplier change, video encoder/decoder block, AES etc Basically binning for different market segments.