Because they compete on price on the main items. Therefore they make little profit on those directly.
However once they have you in the store it gives them the opertunity to sell you overpriced extras such as extended warranties or hookup cables. They bet on the fact that once they have you in the store you'll buy the bits and peices there too.
Plus there are idiots who think high price means high quality for things like cables. May as well part those fools from thier money by selling them even more overpriced acessories.
The CD player reads the bits off the CD much like reading a CD-ROM, but there's a ton of CRC/ECC data My understanding is that there is a lot less error correction information on audio CDs than there is on data CDs.
Agreed it would be good to know where the breakdown in communication happened. Did it get ignored because the submitter didn't realise it was a security issue and report it as such? Did someone just miss an email somewhere? (and if so why wasn't there a system in place to keep track of current security bugs and make it bloody obvious which ones still needed fixing along with someone responsible for looking at that list and fixing them). Was the breakdown on the SUSE side or the upstream side?
Does anyone actually make decent expresscard SSDs?
Just to clarify by decent I mean something that meets the following critera
1: Uses the PCIe part of the expresscard port, not the USB part. 2: Has sequential read and write performance comparable to or better than a hard drive and random read/write performance much better than a hard drive? 3: Preferablly is bootable though i'm not sure how laptop bioses would handle booting from an expresscard.
an unspayed cat can produce 42,000 cats in 5 years I think I can see where a number like that came from. Under ideal conditions I suspect you could breed a heck of a lot of cats from a single female ancestor in 5 years.
Lets assume that a cat can get pregnant after it's 0.75 years old (what i've read online says 6-9 months), that pregancy takes a quater of a year (seems to be an overestimate) and that each litter produces four kittens (the soruce i've read says 3-6 is normal)two of which are female (guess). Further lets consider only the femal side under the assumeption that there are enough males arround to fertilise any cat that needs fertilising
From those assumptions I calculate one would have a total of 10569 cats/kittens after 5 years and a total of 340533 cats/kittens after seven years with a female lineage leading to your original cat.
I'm fairly sure using this method with slightly different timespan assumptions I could get to the articles figures. Especially if one started calculating cats with a male lineage to your cat as well.
Of course this wouldn't really happen in the wild because there are constraints on food and space.
my calculations from excel (/. will munge these but i'm sure those who care enough can figure out the meaning from my paste) are given below:
I want a fast disk for my workstation/server. In a desktop you can put in a SSD as your OS drive (which is where most random access takes place afaict) and keep a spinner for your data. Doing this is already reasonablly affordable.
However if you want a laptop with a SSD at the moment you have to either choose a SSD that can store everything you want on the laptop (which if you store a lot on your laptop means $$$), go for a monster size machine or sacrifice the optical drive (and pick your laptop from the very limited choice of machines that support replacing the optical drive with a hard drive).
With this a laptop vendor can put the SSD on the motherboard while having negligable impact on the rest of the machine.
I wonder if anyone will build a mini-itx board with one of these on? IDE is on it's way out and while you can get SATA disk on moudules a largish lump hanging out of a flimsy sata port doesn't seem like a very robust soloution. A board with one of these on would mean all you would need to add is ram to make a fully functional embedded PC.
despite it's weird aspect ratio 1280x1024 was a common resoloution for both CRTs and LCDs for quite a while and in places with slower replacement cycles is likely to still be common.
Afiact it's only fairly recently that 1280x1024 has been replaced as the common low end desktop option by various widescreen formats (I think 1440x900 is the most common though I have seen some desktops as low as 1280x800!).
even according to steams hardware survey ( http://store.steampowered.com/hwsurvey/ ) which is likely to be biased towards gamers (who tend to be on shorter replacement cycles) 1280x1024 is still the single most common resoloution.
and most low end laptops are on 1280x800 afaict (generally web pages are intended to be scrolled vertically so width is more important to web designers than hight).
I don't know what Bioware's up to, but I think Steam is different... since you're buying the game from them and getting it download-only IIRC half life 2 (and I think other valve games too) requires you to sign up to steam and activate and your copy through it (and IIRC the activation process involves a forced update to the latest version of the game) even if you bought your game as a boxed copy.
Hybrid drives are one answer, although I do not think they are the best answer. If buying a new laptop another option to consider is to sacrifice the optical drive, put a SSD in the main bay and demote the hard drive to the bay the optical drive came in. This does limit your choice of laptops though.
though if you are buying a new laptop you do have the option of getting one where you can replace the optical drive with a hard drive.
Since USB sticks have grown to the size of multiple DVDs I find myself using optical drives a lot less. Sure I need them for installing boxed software but I do that sufficiantly rarely that an external is fine.
Shock protection (unless there's some development I'm not aware of) measures acceleration and parks the drive's heads if the acceleration is too much IIRC it actually parks the hards if the apparent acceleration (which given relativity is all you can measure) is too little.
Because if the apparent acceleration is significantly less than gravity it means you are falling.......
1: they are forced to be AC coupled, for digitial work you really want a DC coupled scope (real scopes usually have a switch or menu option to select AC or DC coupling). Otherwise you can't tell the difference between a line stuck high and a line stuck low. 2: the sample rate and bandwidth are way too low for most microcontroller etc work 3: there is no triggering (though given the low samplerates you could do triggering in software) 4: unless you calibrate manually against a known signal source you have little idea of the absoloute levels.
If we assume feature size halve every two years then in a decades time our feature size will be down to 1 nanometer. Afaict this translates to about 2 atoms!
Fundamentally electronics is based on the idea that we can treat a doped semiconductor as a homogenous material. At sizes in the single atoms this will make no sense anymore and below one atom I just don't see how you could make the features any smaller while continuing to have a crystal lattice structure.
Within the next decade or so feature size decrease (which afaict is the main thing that has powered moores law) will no longer be an avenue open to us. Increasing clock speed is also likely to be hampered by speed of light issues.
Can we go back to i5 quads vs Phenom II X4? Lets look at the i5-750 and the AMD Phenom II X4 965.
Pros of the i5: You can upgrade to a quad-core i7 8xx. The only upgrade path for a Phenom II X4 is really a Phenom II X6 and IMO on the desktop faster cores are worth more than more cores From what i've read the i5 tends to have lower power consuption especially under idle According to toms hardware's performance index it is slightly faster
Cons of the i5: PCIe is rather limited, you get 16 fast lanes from the CPU and 2-6 slow ones from the southbridge. With AMD the fast PCIe is determined by the northbridge, with intel you have to go up to LGA1366 (LGA1366 processors tend to be more expensive for the same performance than the LGA1156 ones) to get more fast PCIe. You HAVE to use a seperate graphics card, if you were planning to use one anyway this won't matter much to you but if you weren't it's an extra cost to add on. it is a bit cheaper LGA1156 motherboards seem to be a bit more expensive than AM3 ones
Notice the performance of the 980X over the other two. There's no more than a 3x performance increase in media encoding. If we ignore the single threaded benchmark at the start and the PCmark at the end and focus on the three mulithreaded media benchmarks we get two ratios just over 3 and one just under 3
So it's just under 9x the price for about 3x the performance (assuming media encoding is your application) which is somewhat different from the figures you stated originally.
The kind of premium Intel charges for the "Extreme Edition" brand is ridiculous. The non-extreme 6-core isn't a whole lot cheaper and the dual-socket stuff with comparable (cores x clockspeed) is also just as expensive once you count the extra motherboard/case/psu cost. It's not the usual extreme edition case of paying a lot more for a slightly faster chip.
Really comparing price/performance of processors alone is fairly meaningless. If you are buying for a situation where you are limited to one machine it should be a case of whether the time saved makes up for the extra cost. If you are buying for a situation where multiple machines will work together it's about the TCO/performance ratio of the complete machines.
Lets look at the price of putting together a complete system (to be charitable i'll assume you are running linux and not paying any per-node licensing costs and that the rest of the system is being built on the cheap) based on the i3-530 and i5-980x
(prices rounded to nearest dollar) CPU: $115 for i3-530 , $1000 for i7-980x HDD: $60 (for 1TB) Motherboard $65 for LGA1156, $130 for LGA1366 case: $15 (much more if you want to rackmount the system) PSU: $15 (more if you want a decent brand that won't blow up on you) Graphics: $27 (not needed for i3) Ram: $48 (for 2x1GB)
I make that a total of $314 for the i3-530 system and $1295 for the i7-980x system. So looking at overall system costs you are paying about four times as much for a system that is about three times as fast at video encoding. Depending on your admin and space costs that could easilly be a better deal for machines whose primary purpose is video encoding.
I don't see any sense in paying 8x more for a CPU that only runs at about 2x the speed. I do,
For a desktop the question should be "is the extra performance worth the extra price". Depending on your workload and how much your time is worth a machine that is twice as fast may save you several computers worth of time.
For a machine in a cluster (whether a compute cluster, a server cluster or whatever) the question should be the TCO/performance ratio of the machines. The TCO of a machine involves a lot more than just the performance of the CPUs (cost of other components in the machine, cost of hosting/powerwing the machine, cost of admins).
Afaict at least for desktops (laptops usually have the CPU soldered down so you can't upgrade it at all) there are two main platforms.
The LGA1156 platform is the main one. It covers i3, i5 and i7 8xx. If you are upgrading from a dual-core to a quad-core and are using integrated graphics you will need to add a graphics card at the same time (you will also need a graphics card if building a quad-core system from scratch).
LGA1366 is the high end platform used for the i7 9xx (and also some server/workstation chips) with more PCIe and more ram channels (which means higher max ram).
So you should be able to upgrade your i5 to an i7-8xx but you can't upgrade to an i7-9xx (however the i7-870 beats everything in AMDs lineup in desktop workloads afaict).
BIOS compatibilty may sometimes be an issue but that can be a problem on the AMD side as well. Afaict that is mostly a case of whether you buy your motherboard from a vendor who can be bothered delivering bios updates.
I was just saying that the browser should simply not display the padlock at all. They shouldn't treat the connection as less secure than non-SSL A big problem with this is the page by page model of the web.
Consider I make a page which is only served over https, said page contains links and form submissions to other pages on the site. Those pages may or may not be on the same hostname as the current page.
Consider a man in the middle that only intercepts some proportion of web requests or in the case of multiple hostnames only intercepts some proportion of them. With your "just don't display the padlock" option by the time I'm warned that I don't have a secure connection it's too late. By the time I see a page without a padlock on whatever data was in the request I just made has been submitted to the man in the middle.
A site that is not encrypted at all can't process a request for a https url so this problem doesn't apply there.
Certificates implemented sensiblly work just as they were designed.
The trouble is the CA model only works when the CAs are trustworthy. For some reason browsers are allowing ever growing lists of CAs of various degrees of trustworthyness and the system as implemented in web browsers is only as secure as the least trustworthy CA in the list. Intermediate signing certificates are a good idea in theory but open up a huge can of worms of thier own by allowing CAs to add other CAs to the list without those other CAs having to prove themselves to the browser vendors.
You have totally sidestepped the question. No, he wasn't looking at LNA gain on an active antenna, he was looking at a chart on a website. PipsqueakOnAP133 appears to have been copying figures including the units (or lack of meaningfull units as I explain below) from product discritions. I don't see any evidence of a chart being involved.
Lets take a proper look at the three products he listed
GPS-09131 is a passive PCB mount antenna, a figure of 2.5dBi is given in the product description (dunno why PipsqueakOnAP133 said 2.6, maybe it was just a typo) and this seems sensible.
The other two are both active antennas so we have to determine what the gain figures listed in the product descriptions really mean. They seem highly unlikely to be an antenna gain because an antenna of that gain would be too directional* to be much use for something like GPS.
For GPS-09871 the datasheet says the gain is in a minimum of 17dBic and typically 18dBic but doesn't say how much of that gain is antenna gain and how much is LNA gain. However the polar plots given make it obvious (at least to someone with an RF background) that most of it must be LNA gain.
For GPS-00177 the datasheet says the antenna gain is 3±0.5 dB but leaves us to guess as to what reference atenna they used it also gives a LNA gain of 28±3 dB.
Repeat: is he talking about dBi or dB? Depending on which, the numbers are not going to be the same. But he mixed the two. dB on it's own is meaningless for measuing antenna gain, antenna gain has to be measured relative to a reference antenna this can be among other things a theoretical isotropic antenna (dBi), a circularly polarised theoritical isotropic antenna (dBic) a half-wave dipole (dBd).
If someone says dB when talking about antenna gain and the couple of dB of difference between different measurements matters you need to ask them to clarify.
* Antenna gain is NOT a free lunch, gain in one direction will always come at the cost of loss in another.
Of you could use a device that charges the phone from AAs, you can even use that on your iPhone. How well do these work? that is it a case of buying a set of AAs and restoring the phone to full charge? is it a case of buying a set of AAs and only having enough for a quick emergency call or is it somewhere inbetween?
4) physically making permanent marks on your now-one-time-use map One trick for getting arround this is to buy laminated maps. You can write on these with a marker pen and remove the markings with methalated spirit.
IIRC the headline figures for ICs are "feature size", that is the size of the smallest feature. The size of a complete logic gate will be significiantly bigger than that and afaict neurons do a fair bit more than your typical logic gate.
More importantly though the brain is packed in 3D. ICs still generally only have a single layer of transistors.
Slashdot Mirror
When do software developers talk about circuits?
Presumablly when they are looking for something else to blame thier screwups on....
Because they compete on price on the main items. Therefore they make little profit on those directly.
However once they have you in the store it gives them the opertunity to sell you overpriced extras such as extended warranties or hookup cables. They bet on the fact that once they have you in the store you'll buy the bits and peices there too.
Plus there are idiots who think high price means high quality for things like cables. May as well part those fools from thier money by selling them even more overpriced acessories.
The CD player reads the bits off the CD much like reading a CD-ROM, but there's a ton of CRC/ECC data
My understanding is that there is a lot less error correction information on audio CDs than there is on data CDs.
Agreed it would be good to know where the breakdown in communication happened. Did it get ignored because the submitter didn't realise it was a security issue and report it as such? Did someone just miss an email somewhere? (and if so why wasn't there a system in place to keep track of current security bugs and make it bloody obvious which ones still needed fixing along with someone responsible for looking at that list and fixing them). Was the breakdown on the SUSE side or the upstream side?
Does anyone actually make decent expresscard SSDs?
Just to clarify by decent I mean something that meets the following critera
1: Uses the PCIe part of the expresscard port, not the USB part.
2: Has sequential read and write performance comparable to or better than a hard drive and random read/write performance much better than a hard drive?
3: Preferablly is bootable though i'm not sure how laptop bioses would handle booting from an expresscard.
an unspayed cat can produce 42,000 cats in 5 years
I think I can see where a number like that came from. Under ideal conditions I suspect you could breed a heck of a lot of cats from a single female ancestor in 5 years.
Lets assume that a cat can get pregnant after it's 0.75 years old (what i've read online says 6-9 months), that pregancy takes a quater of a year (seems to be an overestimate) and that each litter produces four kittens (the soruce i've read says 3-6 is normal)two of which are female (guess). Further lets consider only the femal side under the assumeption that there are enough males arround to fertilise any cat that needs fertilising
From those assumptions I calculate one would have a total of 10569 cats/kittens after 5 years and a total of 340533 cats/kittens after seven years with a female lineage leading to your original cat.
I'm fairly sure using this method with slightly different timespan assumptions I could get to the articles figures. Especially if one started calculating cats with a male lineage to your cat as well.
Of course this wouldn't really happen in the wild because there are constraints on food and space.
my calculations from excel (/. will munge these but i'm sure those who care enough can figure out the meaning from my paste) are given below:
time elapsed in years female adult cats 0.5 year old female kittens 0.25 year old female kittens neborn female kittens total
0 1 0 0 0 1
0.25 1 0 0 2 3
0.5 1 0 2 2 5
0.75 1 2 2 2 7
1 3 2 2 2 9
1.25 5 2 2 6 15
1.5 7 2 6 10 25
1.75 9 6 10 14 39
2 15 10 14 18 57
2.25 25 14 18 30 87
2.5 39 18 30 50 137
2.75 57 30 50 78 215
3 87 50 78 114 329
3.25 137 78 114 174 503
3.5 215 114 174 274 777
3.75 329 174 274 430 1207
4 503 274 430 658 1865
4.25 777 430 658 1006 2871
4.5 1207 658 1006 1554 4425
4.75 1865 1006 1554 2414 6839
5 2871 1554 2414 3730 10569
5.25 4425 2414 3730 5742 16311
5.5 6839 3730 5742 8850 25161
5.75 10569 5742 8850 13678 38839
6 16311 8850 13678 21138 59977
6.25 25161 13678 21138 32622 92599
6.5 38839 21138 32622 50322 142921
6.75 59977 32622 50322 77678 220599
7 92599 50322 77678 119954 340553
I want a fast disk for my workstation/server.
In a desktop you can put in a SSD as your OS drive (which is where most random access takes place afaict) and keep a spinner for your data. Doing this is already reasonablly affordable.
However if you want a laptop with a SSD at the moment you have to either choose a SSD that can store everything you want on the laptop (which if you store a lot on your laptop means $$$), go for a monster size machine or sacrifice the optical drive (and pick your laptop from the very limited choice of machines that support replacing the optical drive with a hard drive).
With this a laptop vendor can put the SSD on the motherboard while having negligable impact on the rest of the machine.
I wonder if anyone will build a mini-itx board with one of these on? IDE is on it's way out and while you can get SATA disk on moudules a largish lump hanging out of a flimsy sata port doesn't seem like a very robust soloution. A board with one of these on would mean all you would need to add is ram to make a fully functional embedded PC.
despite it's weird aspect ratio 1280x1024 was a common resoloution for both CRTs and LCDs for quite a while and in places with slower replacement cycles is likely to still be common.
Afiact it's only fairly recently that 1280x1024 has been replaced as the common low end desktop option by various widescreen formats (I think 1440x900 is the most common though I have seen some desktops as low as 1280x800!).
even according to steams hardware survey ( http://store.steampowered.com/hwsurvey/ ) which is likely to be biased towards gamers (who tend to be on shorter replacement cycles) 1280x1024 is still the single most common resoloution.
and most low end laptops are on 1280x800 afaict (generally web pages are intended to be scrolled vertically so width is more important to web designers than hight).
I don't know what Bioware's up to, but I think Steam is different... since you're buying the game from them and getting it download-only
IIRC half life 2 (and I think other valve games too) requires you to sign up to steam and activate and your copy through it (and IIRC the activation process involves a forced update to the latest version of the game) even if you bought your game as a boxed copy.
Hybrid drives are one answer, although I do not think they are the best answer.
If buying a new laptop another option to consider is to sacrifice the optical drive, put a SSD in the main bay and demote the hard drive to the bay the optical drive came in. This does limit your choice of laptops though.
though if you are buying a new laptop you do have the option of getting one where you can replace the optical drive with a hard drive.
Since USB sticks have grown to the size of multiple DVDs I find myself using optical drives a lot less. Sure I need them for installing boxed software but I do that sufficiantly rarely that an external is fine.
Shock protection (unless there's some development I'm not aware of) measures acceleration and parks the drive's heads if the acceleration is too much
IIRC it actually parks the hards if the apparent acceleration (which given relativity is all you can measure) is too little.
Because if the apparent acceleration is significantly less than gravity it means you are falling.......
There are a few issues I see withh sound cards.
1: they are forced to be AC coupled, for digitial work you really want a DC coupled scope (real scopes usually have a switch or menu option to select AC or DC coupling). Otherwise you can't tell the difference between a line stuck high and a line stuck low.
2: the sample rate and bandwidth are way too low for most microcontroller etc work
3: there is no triggering (though given the low samplerates you could do triggering in software)
4: unless you calibrate manually against a known signal source you have little idea of the absoloute levels.
If we assume feature size halve every two years then in a decades time our feature size will be down to 1 nanometer. Afaict this translates to about 2 atoms!
Fundamentally electronics is based on the idea that we can treat a doped semiconductor as a homogenous material. At sizes in the single atoms this will make no sense anymore and below one atom I just don't see how you could make the features any smaller while continuing to have a crystal lattice structure.
Within the next decade or so feature size decrease (which afaict is the main thing that has powered moores law) will no longer be an avenue open to us. Increasing clock speed is also likely to be hampered by speed of light issues.
Can we go back to i5 quads vs Phenom II X4?
Lets look at the i5-750 and the AMD Phenom II X4 965.
Pros of the i5:
You can upgrade to a quad-core i7 8xx. The only upgrade path for a Phenom II X4 is really a Phenom II X6 and IMO on the desktop faster cores are worth more than more cores
From what i've read the i5 tends to have lower power consuption especially under idle
According to toms hardware's performance index it is slightly faster
Cons of the i5:
PCIe is rather limited, you get 16 fast lanes from the CPU and 2-6 slow ones from the southbridge. With AMD the fast PCIe is determined by the northbridge, with intel you have to go up to LGA1366 (LGA1366 processors tend to be more expensive for the same performance than the LGA1156 ones) to get more fast PCIe.
You HAVE to use a seperate graphics card, if you were planning to use one anyway this won't matter much to you but if you weren't it's an extra cost to add on.
it is a bit cheaper
LGA1156 motherboards seem to be a bit more expensive than AM3 ones
Notice the performance of the 980X over the other two. There's no more than a 3x performance increase in media encoding.
If we ignore the single threaded benchmark at the start and the PCmark at the end and focus on the three mulithreaded media benchmarks we get two ratios just over 3 and one just under 3
So it's just under 9x the price for about 3x the performance (assuming media encoding is your application) which is somewhat different from the figures you stated originally.
The kind of premium Intel charges for the "Extreme Edition" brand is ridiculous.
The non-extreme 6-core isn't a whole lot cheaper and the dual-socket stuff with comparable (cores x clockspeed) is also just as expensive once you count the extra motherboard/case/psu cost. It's not the usual extreme edition case of paying a lot more for a slightly faster chip.
Really comparing price/performance of processors alone is fairly meaningless. If you are buying for a situation where you are limited to one machine it should be a case of whether the time saved makes up for the extra cost. If you are buying for a situation where multiple machines will work together it's about the TCO/performance ratio of the complete machines.
Lets look at the price of putting together a complete system (to be charitable i'll assume you are running linux and not paying any per-node licensing costs and that the rest of the system is being built on the cheap) based on the i3-530 and i5-980x
(prices rounded to nearest dollar)
CPU: $115 for i3-530 , $1000 for i7-980x
HDD: $60 (for 1TB)
Motherboard $65 for LGA1156, $130 for LGA1366
case: $15 (much more if you want to rackmount the system)
PSU: $15 (more if you want a decent brand that won't blow up on you)
Graphics: $27 (not needed for i3)
Ram: $48 (for 2x1GB)
I make that a total of $314 for the i3-530 system and $1295 for the i7-980x system. So looking at overall system costs you are paying about four times as much for a system that is about three times as fast at video encoding. Depending on your admin and space costs that could easilly be a better deal for machines whose primary purpose is video encoding.
I don't see any sense in paying 8x more for a CPU that only runs at about 2x the speed.
I do,
For a desktop the question should be "is the extra performance worth the extra price". Depending on your workload and how much your time is worth a machine that is twice as fast may save you several computers worth of time.
For a machine in a cluster (whether a compute cluster, a server cluster or whatever) the question should be the TCO/performance ratio of the machines. The TCO of a machine involves a lot more than just the performance of the CPUs (cost of other components in the machine, cost of hosting/powerwing the machine, cost of admins).
Afaict at least for desktops (laptops usually have the CPU soldered down so you can't upgrade it at all) there are two main platforms.
The LGA1156 platform is the main one. It covers i3, i5 and i7 8xx. If you are upgrading from a dual-core to a quad-core and are using integrated graphics you will need to add a graphics card at the same time (you will also need a graphics card if building a quad-core system from scratch).
LGA1366 is the high end platform used for the i7 9xx (and also some server/workstation chips) with more PCIe and more ram channels (which means higher max ram).
So you should be able to upgrade your i5 to an i7-8xx but you can't upgrade to an i7-9xx (however the i7-870 beats everything in AMDs lineup in desktop workloads afaict).
BIOS compatibilty may sometimes be an issue but that can be a problem on the AMD side as well. Afaict that is mostly a case of whether you buy your motherboard from a vendor who can be bothered delivering bios updates.
I was just saying that the browser should simply not display the padlock at all. They shouldn't treat the connection as less secure than non-SSL
A big problem with this is the page by page model of the web.
Consider I make a page which is only served over https, said page contains links and form submissions to other pages on the site. Those pages may or may not be on the same hostname as the current page.
Consider a man in the middle that only intercepts some proportion of web requests or in the case of multiple hostnames only intercepts some proportion of them. With your "just don't display the padlock" option by the time I'm warned that I don't have a secure connection it's too late. By the time I see a page without a padlock on whatever data was in the request I just made has been submitted to the man in the middle.
A site that is not encrypted at all can't process a request for a https url so this problem doesn't apply there.
Certificates implemented sensiblly work just as they were designed.
The trouble is the CA model only works when the CAs are trustworthy. For some reason browsers are allowing ever growing lists of CAs of various degrees of trustworthyness and the system as implemented in web browsers is only as secure as the least trustworthy CA in the list. Intermediate signing certificates are a good idea in theory but open up a huge can of worms of thier own by allowing CAs to add other CAs to the list without those other CAs having to prove themselves to the browser vendors.
You have totally sidestepped the question. No, he wasn't looking at LNA gain on an active antenna, he was looking at a chart on a website.
PipsqueakOnAP133 appears to have been copying figures including the units (or lack of meaningfull units as I explain below) from product discritions. I don't see any evidence of a chart being involved.
Lets take a proper look at the three products he listed
GPS-09131 is a passive PCB mount antenna, a figure of 2.5dBi is given in the product description (dunno why PipsqueakOnAP133 said 2.6, maybe it was just a typo) and this seems sensible.
The other two are both active antennas so we have to determine what the gain figures listed in the product descriptions really mean. They seem highly unlikely to be an antenna gain because an antenna of that gain would be too directional* to be much use for something like GPS.
For GPS-09871 the datasheet says the gain is in a minimum of 17dBic and typically 18dBic but doesn't say how much of that gain is antenna gain and how much is LNA gain. However the polar plots given make it obvious (at least to someone with an RF background) that most of it must be LNA gain.
For GPS-00177 the datasheet says the antenna gain is 3±0.5 dB but leaves us to guess as to what reference atenna they used it also gives a LNA gain of 28±3 dB.
Repeat: is he talking about dBi or dB? Depending on which, the numbers are not going to be the same. But he mixed the two.
dB on it's own is meaningless for measuing antenna gain, antenna gain has to be measured relative to a reference antenna this can be among other things a theoretical isotropic antenna (dBi), a circularly polarised theoritical isotropic antenna (dBic) a half-wave dipole (dBd).
If someone says dB when talking about antenna gain and the couple of dB of difference between different measurements matters you need to ask them to clarify.
* Antenna gain is NOT a free lunch, gain in one direction will always come at the cost of loss in another.
Of you could use a device that charges the phone from AAs, you can even use that on your iPhone.
How well do these work? that is it a case of buying a set of AAs and restoring the phone to full charge? is it a case of buying a set of AAs and only having enough for a quick emergency call or is it somewhere inbetween?
4) physically making permanent marks on your now-one-time-use map
One trick for getting arround this is to buy laminated maps. You can write on these with a marker pen and remove the markings with methalated spirit.
IIRC the headline figures for ICs are "feature size", that is the size of the smallest feature. The size of a complete logic gate will be significiantly bigger than that and afaict neurons do a fair bit more than your typical logic gate.
More importantly though the brain is packed in 3D. ICs still generally only have a single layer of transistors.