If there was a suitable number of humans on the planet that didn't rape resources right and left, the planet would probably sustain human life until an asteroid of sufficient size or the red giant expansion of the sun. Getting a good portion of us off the rock might help in that.
Well, for one, not all Catholics are perfect Catholics who do everything according to the papal dogma. He's celibate, so not having kids is easy for him. For another, I think you'll find many religions, including the Catholic church in the past, have done more than their share to lower certain populations rather quickly and drastically, although spending a great amount of wealth and natural resources to do so. Ever heard of the Crusades? Ever heard of the Papal wars among European monarchies, in which the Church would offer land and money for one king to attack another king at the expense mostly of the people? The inquisitions you've heard of perhaps, at least of the Spanish Inquisition?
Also, with no Spanish Inquisition Columbus wouldn't have had his Spanish funds to discover the West Indies. They seldom mention in public schools that the public hero Christopher Columbus was backed by the very same king and queen of Spain who initiated a general inquisition to rid their country of Moorish and Islamic influence, and whose coffers became rich with the money of devout Jews and Muslims who would not convert. It's a good thing for those who love the Western hemisphere that all those people were killed and their money taken, huh? They lowered the population and funded the exploitation and murder of hundreds of thousands more in the "New World", although I'm pretty sure the Incas, Aztecs, and Zunis just called the "New World" something more like "home".
There are only five ways to force population control on people. One is to enslave or incarcerate them so you have control of their bodies. Another is to starve them. A third is to murder them outright. A fourth one is to assault them and forcibly sterilize them. A fifth is to initiate wars so that the populations are killed off generally. Hail Malthus, huh?
Why don't we instead try what generally works best for motivating people: self-interest? It's pretty clear that outside of agrarian societies fewer children born later in life tend to make you wealthier. Most of the developed countries have very slow population growth, even those with majority Catholic populations. Countries that have massive population growth tend to be ones in which the people are poor, not well educated, and largely agrarian. The answer to population growth isn't force. It's ending poverty. I know it is counterintuitive, but poor people the last several decades multiply faster than middle class or wealthy people. So share some of that global wealth in the form of moderate development and immoderate education, and maybe we'll find a way out of the mess without forcing it on people.
But then, you weren't really serious about talking about mankind as a whole, were you? You said "before we start getting hungry", which I can assure you much of the world is already.
If they can make muscles contract and relax according to the supply or lack thereof of a bright light, then a great next step is to get a bundle of nerve tissue to generate a bright light when they are excited. Then fiber optics could potentially be used as artificial nerve tissue to route around damage. The limits of bioluminescence might not allow for interaction with this, though.
Even if there's an inorganic portion needed to generate a bright enough light, having that controlled directly by nerve tissue (and maybe fueled by ATP or body heat if possible) could be a huge advance over the conductive wires and external battery bags being tested now.
My thought was that they could filter the results for Instant rather than just returning nothing, or have some options for that behavior. Returning absolutely nothing just points out to people that there's something offensive connected to the typed characters anyway.
I don't know which is more sad: Google thinking "lolita" needs to be protected or that the people keeping the list only know it as a reference to a film and never mention the book.
I also don't know which is more sad: that the Spanish word for "black" is considered so offensive just because it might be used by a racist or that a word for rooster is considered offensive just because it could be used sexually.
I just noticed, BTW, that even turning SafeSearch off doesn't prevent this particular Instant filter.
Also, I noticed that not only is "latino" okay and "latina" not, but although "negro" is filtered "negra" is not. Probably because the first page of results are a commercial product (Negra Modelo beer)?
I think their new addition is failing, too, because "marij" through "marijuan" give great results for "marijuana" even though "marijauna" itself is blocked.
I can drop or mangle the packets if they cross my router, or redirect them to random destinations. Even though VPNs make it hard to read the data inside the packets, they don't secure the arrival of those packets.
It's not so much about us conforming to them. It's about them not having to conform to us. They want a separate physical network with its own fiber, its own routers and switches, and probably its own protocols. They want it to not be part of the public Internet at all, so if they get that it won't bother us a bit.
You can tell when most people are monitoring a fiber connection. It takes some real sophistication to do it without altering the propagation times, received intensity, etc. Most techniques involve actually splicing in, and they can tell approximately where the break in the cable was made remotely.
IF you control router along the path, a VPN makes it really hard for you to read the information. It makes it really easy, though, for you to just drop packets that are part of a VPN, mangle them so they are corrupted upon receipt, or record them offline for later brute-force attacks.
There's no reason to hook up just one network to your facility if you need real security. You can have public Internet access on one set of systems and private internet (small 'i') access on others.
If you need to search Google (which would probably have a separate government-funded server farm and Milnet (oops I mean "private secure internet") connection anyway) you go to an Internet-connected system. If you need to access a secure remote site, you go to the private network systems.
If you need to get data from the public Internet to your private network, you use removable media only, move only non-executable data, and scan it very carefully with a number of tools for cleverly crafted data formats designed to overflow buffers or smash stacks.
If you need to move info from your secure systems to the public Internet... then you don't really need your secure systems that damn much first of all. Yet if for some reason you do need to do this, you can use removable media for that. I've even seen people read from one terminal and type into another to bridge information across networks that were kept separate for security reasons.
I used to support 14,000 customers on two dual-processor single-core MX boxes with spam filtering, two tow-proc single-core SMTP outbound box with spam filtering (to catch our customers and keep them from spamming) and four single-proc single-core POP/IMAP boxes that accepted mail only from the MX or SMTP outbound boxes and did antivirus. Webmail was on a separate web server which was also the company's web site server and hosted a couple of specific clients.
I used to support over 100k users at a different employer with four MX boxes with rudimentary spam filtering, four SMTP outbound boxes, four dedicated AV/anti-spam boxes, and 12 POP/IMAP/Perdition/webmail boxes. All of these boxes were dual processor single-core 3.06 GHz Xeons.
Yeah, I get the concepts of percentage used, load average, RAM constraints, and per-user load.
Guess what? I still like to have more cores when I'm doing something massively parellel like running 64 make jobs at once. That example is something I already mentioned, and can be done with GNU make's -j switch and a number of maximum jobs. It can also be done with a bare -j switch by itself, but that tells make it has no maximum, and on a big build it will make your machine pretty much unusable until the build is done, even on a quad-core with the preemptible kernel Linux configuration.
I said it before Intel did it, jackass. I'm not a fanboy, BTW. I'll buy Intel when it makes sense. I just prefer AMD after about two decades of experience buying products of both companies.
An MMU has nothing to do with controlling the actual SIPPs, SIMMs, or DIMMs. It's a multiprocessing ("multiprocessing" doesn't mean "multi-core") feature that allows the processor to enforce memory address range protection so that program A doesn't stomp on program B's memory range. That's a separate concern from getting data into and out of the processor from main memory.
Why don't you go get a copy of something like Upgrading and Repairing PCs and inform yourself? Here's the ISBN.nu link for the 18th edition in case you're a bargain shopper: 18th edition. I have the fifth edition myself. I might get an updated version for the handy reference tables in the back featuring things like POST codes and error codes for SCSI controllers.
BTW, have you ever actually built a PC older than, say, a Pentium 4? Or owned one?
Well, I'll admit I'm not a typical desktop user for starters. I'm also not running the typical desktop OS on most of my machines.
Most of the time that I'm filling the four cores on this box to capacity, it's running something that forks into multiple processes rather than threading inside one. Usually something like running make against a C or C++ program. My browser also uses separate processes, one per tab. Sometimes I'll be testing server software or debugging it and the client, server, and all the middleware together will peg all four cores. Most of the software I personally write uses forks rather than threads when I need multiple streams of execution. Linux balances those out among cores very well, and a fork on Linux is cheaper than the equivalent process launch on Windows.
Other people I know do things like run a CPU and GPU intensive game while running antivirus, a host firewall, some communications program (IM or voice chat) and streaming music. Until you get to a really monster number of cores, more cores will give you a speed boost even if most of those tasks only use one or two threads or processes.
Given that more and more people are using SSDs and more and more people are using multi-core chips, there's no reason a full anti-virus scan has to check single files serially. Especially with multiple SSDs in RAID 0, 5, or 6 there's no reason you couldn't fill a couple of cores with just a full filesystem scan.
I run IDS software, backups, logging, update checker, crypto services, mail server, name server, ssh server, time synchronization, database server, intranet web server, X server, window manager, and miscellany on this box all the time, and it spreads those things out over all four cores evenly. When I, say, build a new Linux kernel with make running 32 or 64 jobs at once, having four cores really speeds things up over having two. Having six would speed it up more, I'm sure. Having SSDs or building in a RAM disk wouldn't hurt, but I've yet to pay for enough RAM to do that and I'm trying to pick which SSD I want as I type this.
How about when AMD was the first of the two to do on-die memory controllers? Or maybe when they were the first of the two to have a quality high-end graphics division? Even though that was through purchase, I think it was a smart purchase. AMD also did 386s up to 40 MHz, a DX4 up to 120 MHz, and forced Intel into the model number marketing when AMD processors were running faster clock-for-clock than Intel chips back in the Athlon XP+ days.
I think the L3 cache on the chip, multicore chips, and the on-die memory controller are good candidates for significant past upgrades to the architecture. Also, don't forget SSE and SSE2. One way to get a huge speedup from many apps is switching from 387/487 coprocessor logic to SSE2 math.
Try an AMD 790 or newer chipset board. DFI, ECS, Asus, Gigabyte, and Foxconn all make them. You should be able to find a decent board from one of those companies. MSI, AsRock, MSI, Biostar, Jetway, and even more companies also make them.
AMD's chipset takes take of the northbridge and soutbridge, the USB, SATA, PCI and PCI-E, etc. Any other chips on the board will be sound or Ethernet unless it's a board with something somewhat unusual like more than one onboard Ethernet or more than half a dozen or so SATA ports.
So if you get any board with an AMD chipset, there are really only a few other things to consider:
build quality, of course
which southbridge chip is paired with the northbridge, because sometimes they cheap out and put an older SB on a board than the main chipset normally pairs with
claimed memory speeds (Some manufacturers using the same chipset rate their boards as less capable than others and even less than the specs for the chipset. I always doubt the quality of the board if they can't say it'll do what the chipset can normally do.)
customer support in case you do get a bad board (Newegg's reviews usually work well for guessing this, even on other products from the company.)
reported problems with the board (Newegg reviews are usually handy for this for a particular product if it's been on the market for a while.)
price (of course)
return policy at your merchant of choice for that board (again, just in case)
You should of course check CPU and memory compatibility carefully if it's been a long time since you've done an AMD build. That's no fault of the chipsets or boards, but just a fact when there are so many chipsets and CPUs on the market at the same time.
My main desktop rig right now is an ECS A790GXM-AD3 Black Edition with a Phenom II x4 955 Black Edition processor and 2x2 GiB Crucial Ballistix DDR3 1333 rated at CL6 (6-6-6-20). I've had it for quite a while now, and I'm pretty happy with the reliability and with the speed for the price. It'll overclock pretty darn well, too. I don't like the extra fan noise and it's plenty fast enough for what I do with it without overclocking.
I don't play a lot of graphics-heavy games right now, so I'm running the integrated graphics. I average near 100 FPS in AssaultCube 1.1.0.1 with that, and it spikes to over 130 and never falls below 80. I should mention that's on Mandriva 2010.1 for AMD64 with the Catalyst blob. The board does have two x16 PCI-E 2.0 slots that will run full speed together. I just haven't put anything in either of them for right now because my fastest discrete PCI Express video card is a 1600 Pro.
AMD and NVidia are really the only players in AMD board chipsets these days, and AMD chipsets are on all the really well-performing boards anyway. The tight memory timings you can get with the recent AMD chipsets are great. Now, if only they'd leapfrog Intel's triple-channel and go quad-channel for memory modules, you'd see some really nice performance.
I know some Intel chips outrun AMD chips, but after the whole Randal Schwartz fiasco, Intel refusing to do 64-bit extensions to ia32 because they were counting on Itanium (then bringing out EMT64 as a knockoff of AMD) and messing around with the FSB for so long until they took AMD's lead again for the on-die memory controller, I'll keep rewarding AMD for the way they do business and push the market forward.
I do buy Intel chips sometimes, usually used. Most of my systems are AMD, though, and through the years although I've been mnostly happy with both brands I've been more happy with AMD.
Slashdot Mirror
That's the day before a House of Commons election.
It could just as easily be the last, with the first 16 being the mirrored portion.
It "isn't rocket science", but it "is like brain surgery". In fact, it is brain surgery. The two idioms, BTW, are nearly synonymous.
If there was a suitable number of humans on the planet that didn't rape resources right and left, the planet would probably sustain human life until an asteroid of sufficient size or the red giant expansion of the sun. Getting a good portion of us off the rock might help in that.
Well, for one, not all Catholics are perfect Catholics who do everything according to the papal dogma. He's celibate, so not having kids is easy for him. For another, I think you'll find many religions, including the Catholic church in the past, have done more than their share to lower certain populations rather quickly and drastically, although spending a great amount of wealth and natural resources to do so. Ever heard of the Crusades? Ever heard of the Papal wars among European monarchies, in which the Church would offer land and money for one king to attack another king at the expense mostly of the people? The inquisitions you've heard of perhaps, at least of the Spanish Inquisition?
Also, with no Spanish Inquisition Columbus wouldn't have had his Spanish funds to discover the West Indies. They seldom mention in public schools that the public hero Christopher Columbus was backed by the very same king and queen of Spain who initiated a general inquisition to rid their country of Moorish and Islamic influence, and whose coffers became rich with the money of devout Jews and Muslims who would not convert. It's a good thing for those who love the Western hemisphere that all those people were killed and their money taken, huh? They lowered the population and funded the exploitation and murder of hundreds of thousands more in the "New World", although I'm pretty sure the Incas, Aztecs, and Zunis just called the "New World" something more like "home".
There are only five ways to force population control on people. One is to enslave or incarcerate them so you have control of their bodies. Another is to starve them. A third is to murder them outright. A fourth one is to assault them and forcibly sterilize them. A fifth is to initiate wars so that the populations are killed off generally. Hail Malthus, huh?
Why don't we instead try what generally works best for motivating people: self-interest? It's pretty clear that outside of agrarian societies fewer children born later in life tend to make you wealthier. Most of the developed countries have very slow population growth, even those with majority Catholic populations. Countries that have massive population growth tend to be ones in which the people are poor, not well educated, and largely agrarian. The answer to population growth isn't force. It's ending poverty. I know it is counterintuitive, but poor people the last several decades multiply faster than middle class or wealthy people. So share some of that global wealth in the form of moderate development and immoderate education, and maybe we'll find a way out of the mess without forcing it on people.
But then, you weren't really serious about talking about mankind as a whole, were you? You said "before we start getting hungry", which I can assure you much of the world is already.
lifespan != age
(unless you mess with Bruce Schneier, yadda yadda...)
If they can make muscles contract and relax according to the supply or lack thereof of a bright light, then a great next step is to get a bundle of nerve tissue to generate a bright light when they are excited. Then fiber optics could potentially be used as artificial nerve tissue to route around damage. The limits of bioluminescence might not allow for interaction with this, though.
Even if there's an inorganic portion needed to generate a bright enough light, having that controlled directly by nerve tissue (and maybe fueled by ATP or body heat if possible) could be a huge advance over the conductive wires and external battery bags being tested now.
My thought was that they could filter the results for Instant rather than just returning nothing, or have some options for that behavior. Returning absolutely nothing just points out to people that there's something offensive connected to the typed characters anyway.
I don't know which is more sad: Google thinking "lolita" needs to be protected or that the people keeping the list only know it as a reference to a film and never mention the book.
I also don't know which is more sad: that the Spanish word for "black" is considered so offensive just because it might be used by a racist or that a word for rooster is considered offensive just because it could be used sexually.
I just noticed, BTW, that even turning SafeSearch off doesn't prevent this particular Instant filter.
Also, I noticed that not only is "latino" okay and "latina" not, but although "negro" is filtered "negra" is not. Probably because the first page of results are a commercial product (Negra Modelo beer)?
I think their new addition is failing, too, because "marij" through "marijuan" give great results for "marijuana" even though "marijauna" itself is blocked.
I can drop or mangle the packets if they cross my router, or redirect them to random destinations. Even though VPNs make it hard to read the data inside the packets, they don't secure the arrival of those packets.
It's not so much about us conforming to them. It's about them not having to conform to us. They want a separate physical network with its own fiber, its own routers and switches, and probably its own protocols. They want it to not be part of the public Internet at all, so if they get that it won't bother us a bit.
A VPN is not as secure as a separate point-to-point line.
You can tell when most people are monitoring a fiber connection. It takes some real sophistication to do it without altering the propagation times, received intensity, etc. Most techniques involve actually splicing in, and they can tell approximately where the break in the cable was made remotely.
IF you control router along the path, a VPN makes it really hard for you to read the information. It makes it really easy, though, for you to just drop packets that are part of a VPN, mangle them so they are corrupted upon receipt, or record them offline for later brute-force attacks.
Forcing outbound or inbound traffic through a particular gateway? It can be done in software on Linux or any of the BSDs. Look up "iptables".
Any halfway competent script kiddie can DDoS a VPN.
Depending on what data is at stake, you could get fired on really quickly if you refused to stand down from the terminal.
There's no reason to hook up just one network to your facility if you need real security. You can have public Internet access on one set of systems and private internet (small 'i') access on others.
If you need to search Google (which would probably have a separate government-funded server farm and Milnet (oops I mean "private secure internet") connection anyway) you go to an Internet-connected system. If you need to access a secure remote site, you go to the private network systems.
If you need to get data from the public Internet to your private network, you use removable media only, move only non-executable data, and scan it very carefully with a number of tools for cleverly crafted data formats designed to overflow buffers or smash stacks.
If you need to move info from your secure systems to the public Internet... then you don't really need your secure systems that damn much first of all. Yet if for some reason you do need to do this, you can use removable media for that. I've even seen people read from one terminal and type into another to bridge information across networks that were kept separate for security reasons.
I used to support 14,000 customers on two dual-processor single-core MX boxes with spam filtering, two tow-proc single-core SMTP outbound box with spam filtering (to catch our customers and keep them from spamming) and four single-proc single-core POP/IMAP boxes that accepted mail only from the MX or SMTP outbound boxes and did antivirus. Webmail was on a separate web server which was also the company's web site server and hosted a couple of specific clients.
I used to support over 100k users at a different employer with four MX boxes with rudimentary spam filtering, four SMTP outbound boxes, four dedicated AV/anti-spam boxes, and 12 POP/IMAP/Perdition/webmail boxes. All of these boxes were dual processor single-core 3.06 GHz Xeons.
Yeah, I get the concepts of percentage used, load average, RAM constraints, and per-user load.
Guess what? I still like to have more cores when I'm doing something massively parellel like running 64 make jobs at once. That example is something I already mentioned, and can be done with GNU make's -j switch and a number of maximum jobs. It can also be done with a bare -j switch by itself, but that tells make it has no maximum, and on a big build it will make your machine pretty much unusable until the build is done, even on a quad-core with the preemptible kernel Linux configuration.
I said it before Intel did it, jackass. I'm not a fanboy, BTW. I'll buy Intel when it makes sense. I just prefer AMD after about two decades of experience buying products of both companies.
The 486 had a DRAM controller on its die? I'm going to have to ask for a citation. I think you're thinking of either the on-die L1 cache or the MMU (memory management unit), neither of which is a system main memory controller. Here's a citation to the counter: List of Intel Chipsets at Wikipedia. See how the chipset determines the memory specs up until the Core i Series, including the 80486? Here's another: List of Intel Chipsets at World IQ. Here's another: Intel CPU and Chipset History at Overclock 3D courtesy of a forum post there by "PV5150".
An MMU has nothing to do with controlling the actual SIPPs, SIMMs, or DIMMs. It's a multiprocessing ("multiprocessing" doesn't mean "multi-core") feature that allows the processor to enforce memory address range protection so that program A doesn't stomp on program B's memory range. That's a separate concern from getting data into and out of the processor from main memory.
Why don't you go get a copy of something like Upgrading and Repairing PCs and inform yourself? Here's the ISBN.nu link for the 18th edition in case you're a bargain shopper: 18th edition. I have the fifth edition myself. I might get an updated version for the handy reference tables in the back featuring things like POST codes and error codes for SCSI controllers.
BTW, have you ever actually built a PC older than, say, a Pentium 4? Or owned one?
Well, I'll admit I'm not a typical desktop user for starters. I'm also not running the typical desktop OS on most of my machines.
Most of the time that I'm filling the four cores on this box to capacity, it's running something that forks into multiple processes rather than threading inside one. Usually something like running make against a C or C++ program. My browser also uses separate processes, one per tab. Sometimes I'll be testing server software or debugging it and the client, server, and all the middleware together will peg all four cores. Most of the software I personally write uses forks rather than threads when I need multiple streams of execution. Linux balances those out among cores very well, and a fork on Linux is cheaper than the equivalent process launch on Windows.
Other people I know do things like run a CPU and GPU intensive game while running antivirus, a host firewall, some communications program (IM or voice chat) and streaming music. Until you get to a really monster number of cores, more cores will give you a speed boost even if most of those tasks only use one or two threads or processes.
Given that more and more people are using SSDs and more and more people are using multi-core chips, there's no reason a full anti-virus scan has to check single files serially. Especially with multiple SSDs in RAID 0, 5, or 6 there's no reason you couldn't fill a couple of cores with just a full filesystem scan.
I run IDS software, backups, logging, update checker, crypto services, mail server, name server, ssh server, time synchronization, database server, intranet web server, X server, window manager, and miscellany on this box all the time, and it spreads those things out over all four cores evenly. When I, say, build a new Linux kernel with make running 32 or 64 jobs at once, having four cores really speeds things up over having two. Having six would speed it up more, I'm sure. Having SSDs or building in a RAM disk wouldn't hurt, but I've yet to pay for enough RAM to do that and I'm trying to pick which SSD I want as I type this.
How about when AMD was the first of the two to do on-die memory controllers? Or maybe when they were the first of the two to have a quality high-end graphics division? Even though that was through purchase, I think it was a smart purchase. AMD also did 386s up to 40 MHz, a DX4 up to 120 MHz, and forced Intel into the model number marketing when AMD processors were running faster clock-for-clock than Intel chips back in the Athlon XP+ days.
I think the L3 cache on the chip, multicore chips, and the on-die memory controller are good candidates for significant past upgrades to the architecture. Also, don't forget SSE and SSE2. One way to get a huge speedup from many apps is switching from 387/487 coprocessor logic to SSE2 math.
Try an AMD 790 or newer chipset board. DFI, ECS, Asus, Gigabyte, and Foxconn all make them. You should be able to find a decent board from one of those companies. MSI, AsRock, MSI, Biostar, Jetway, and even more companies also make them.
AMD's chipset takes take of the northbridge and soutbridge, the USB, SATA, PCI and PCI-E, etc. Any other chips on the board will be sound or Ethernet unless it's a board with something somewhat unusual like more than one onboard Ethernet or more than half a dozen or so SATA ports.
So if you get any board with an AMD chipset, there are really only a few other things to consider:
You should of course check CPU and memory compatibility carefully if it's been a long time since you've done an AMD build. That's no fault of the chipsets or boards, but just a fact when there are so many chipsets and CPUs on the market at the same time.
My main desktop rig right now is an ECS A790GXM-AD3 Black Edition with a Phenom II x4 955 Black Edition processor and 2x2 GiB Crucial Ballistix DDR3 1333 rated at CL6 (6-6-6-20). I've had it for quite a while now, and I'm pretty happy with the reliability and with the speed for the price. It'll overclock pretty darn well, too. I don't like the extra fan noise and it's plenty fast enough for what I do with it without overclocking.
I don't play a lot of graphics-heavy games right now, so I'm running the integrated graphics. I average near 100 FPS in AssaultCube 1.1.0.1 with that, and it spikes to over 130 and never falls below 80. I should mention that's on Mandriva 2010.1 for AMD64 with the Catalyst blob. The board does have two x16 PCI-E 2.0 slots that will run full speed together. I just haven't put anything in either of them for right now because my fastest discrete PCI Express video card is a 1600 Pro.
AMD and NVidia are really the only players in AMD board chipsets these days, and AMD chipsets are on all the really well-performing boards anyway. The tight memory timings you can get with the recent AMD chipsets are great. Now, if only they'd leapfrog Intel's triple-channel and go quad-channel for memory modules, you'd see some really nice performance.
I know some Intel chips outrun AMD chips, but after the whole Randal Schwartz fiasco, Intel refusing to do 64-bit extensions to ia32 because they were counting on Itanium (then bringing out EMT64 as a knockoff of AMD) and messing around with the FSB for so long until they took AMD's lead again for the on-die memory controller, I'll keep rewarding AMD for the way they do business and push the market forward.
I do buy Intel chips sometimes, usually used. Most of my systems are AMD, though, and through the years although I've been mnostly happy with both brands I've been more happy with AMD.