And if you want further proof, here is another page posting unabridged benchmark results, which indicate clearly that the Double benchmark uses iSSE2 (you can see the same ~5.5x drop in performance as with the other benchmark I linked). PcPer was just incredibly lazy marking their benchmark graphs.
Atom sucks at double precision. You can deny it all you want, but the benchmarks ring true. I can't say I'm surprised - double performance is one of the first things to go when you're targeting low-power.
Interesting. That's quite a big issue albeit probably not for the applications Atom is targeted for. Still, though, ARM will make it look terrible.
For most workloads, if the server is not busy, you'll get much better power usage from the Xeon as well.
Not really. Look at the idle power numbers on a Xeon compared to an Atom. One of the things about low power (and low-clocking) chips is that leakage is orders of magnitude less. Atom not only runs on a lower-leakage process, it also runs at a lower voltage than Xeons. So while dynamic power-per-unit-work may not be competitive in aggregate, idle power is not even close.
While "shifting features" is indeed not accurate, the point stands. Compared to ARM (or even Athlon and Nehalem/Lynnfield processors), Atom (prior to Pinetrail and Moorestown) was less integrated. So to compare power numbers, you can't just look at the CPU itself but need to include the memory controller (when compared to Athlon and Nehalem/Lynnfield) and PCI-E controller (when compared to Nehalem/Lynnfield). When compared to most ARM SoC's, you'll need to include graphics, miscellaneous I/O, GPS, video processor, image processor, memory controller, USB, CPU bus, peripheral bus, chip-to-chip interconnect, and sometimes a 3G modem.
Even Moorestown is very misleading when they did their comparison with ARM. They conveniently showed off numbers for tasks that use the dedicated components (video decoding, audio playback, talk time on GSM/CDMA, etc.) -- much of which was done by the "helper" chip and not the chip Intel fabbed themselves -- but left out any tasks that would tax the main CPU. Comparing those numbers to ARM SoC's made it look competitive. Then they showed performance numbers of the CPU compared to ARM CPU's -- which, btw, seemed to increase linearly with number of threads, notice how a dual-core A9 was twice the performance of a single? -- but left out the power numbers for when the CPU was taxed heavily.
I mean, it's marketing. But that hardly means it's not underhanded.
Atom has a TDP of ~2W and I would speculate a maximum power consumption of ~4W. A C2050 has a maximum power consumption of 247W. For the same power bugdet as a C2050, you can use 60 Atoms each with its own pocket of memory and condition code handlers.
But it's a poor choice for high-precision scientific calculations. The vector unit on the Atom is optimized for 32-bit Floats, and when you switch-up to Doubles, the performance drops off the map.
See this performance comparison (Processor Multimedia). The dual-core Atom 1.6 is actually faster than the dual-core Athlon 64 1.6 when running SSE 32-bit Floats (this is due to the dual 64-bit SSE units on the Athlon 64, and poor scheduling on AMD's part). But when you transition to Double precision, the performance on the Athlon 64 only drops by roughly 2x, whereas the Atom sees a 5-6x performance drop!
This means you have very little flexibility in workloads you can use - if you're not using 32-bit precision, you've wasted your money.
That was for x87 FP though. While it's important in that a lot of code out there might still use it, for performance on any Intel architecture, you're supposed to use SSE (scalar SSE if you're not doing SIMD). I would like to see SSE performance broken down to SP vs DP. I suspect, from what little I know of the vector pipeline, that it'd be pretty much exactly half the performance.
It's not so bad if you're running a highly-multithreaded program on the system. The Atom uses the SMT to great advantage, in most cases seeing a %30 performance increase. But yeah, for single-threaded code, the Atom is a joke.
Even for highly threaded programs, the performance/watt will not be competitive with (lightly) OoOE architectures. In-order is simply inefficient from a power standpoint. Unless they manage to cram 4 threads simultaneously on a single Atom core, even 2 heavy threads won't be able to saturate every execution slot and if you throw more at it, you'll get the overhead of context switching and cache thrashing.
If that's the case, Atom is a horrible choice for the CPU. Atom's strongest feature is its 128-bit vector (SSE) unit for SIMD FLOPS.
The in-order (albeit SMT-enabled), 2-issue integer/LDST/Conditional pipeline is lackluster at best from either a performance or a performance/watt standpoint.
Yes and no. It really depends on how the dual-core is architected. Assuming you had 2 cores with independent frequency and voltage scaling and assuming you were running two tasks that were fairly processor and memory intensive, it could indeed be beneficial to run them on 2 cores. Context switching is a killer on cache and memory and having a single CPU throttle between multiple processes is inefficient compared to two separate CPU's working out of their own caches.
Moreover, power is a function of voltage squared; so two CPU's running at 0.8V at 400MHz would require less power than one CPU running at 1.0V at 800MHz.
Windows 7 won't be on ARM but Windows Embedded 7 will. That's the OS Microsoft is making to target tablets and it will come out on both ARM and x86 at the same time.
Mobile chips, particularly cell phones, are constrained by power. Supercomputers have the luxury of sucking up as much wattage as they need.
Double the memory traces takes a lot of power. Using DDR3 vs DDR1 takes a lot of power. Using dual memory modules takes a lot of power.
Taking all of those restrictions in mind, it's actually fairly amazing how well modern mobile SoC's do in keeping the processor and subsystems busy through aggressive caching and good memory packaging (for higher speed without extra power).
Interestingly, Apple campus is one of the only places in Cupertino with good AT&T 3G coverage. My friend lives about 2 blocks away from Apple HQ on Homestead road and I can't even make calls there let alone actually use the web.
I'm curious about Linpack results when running under Dalvik. I believe Froyo 2.2 enabled Dalvik start using the FPU of the ARM ISA. So what were the results before? Emulated floating point?
On top of that, the ARM ISA allows for SIMD operations. I would assume the VM isn't capable of that.
If we're to compare processing power, the Cortex A8 at 1GHz (A4, OMAP 3640, overclocked Droid) is capable of a vector multiply (2 at a time) every 3 cycles and add in 2 cycles. So that's about 400 MFLOPS.
Those aren't advancements; they're political/social changes. Science discovers facts; it does not set policy. Anti-intellectualism is not choosing not to implement a particular finding or method; it's sticking your head in the sand and refusing to even acknowledge its there for some retarded concept that the danger will be gone "if you keep all that devil knowledge out".
Knowing how to, say, trade one person's life for another is very different than actually doing it. Whether or not something should be done is outside the realm of science and deep in philosophy and politics. You demonstrate the exact ignorance and knee-jerk fear that we're speaking of when you try to conflate the two.
Ya. I mean all those people who didn't want penicillin, modern plumbing, flight, electricity, smallpox vaccines, the internet you're typing this on, etc. are doing so well.
Do you even realize the insane irony of making the "we don't want your advancements" argument on the internet?
The headline is (again) inaccurate. It's not a 7-atom transistor, it's a transistor with 7 phosphorus atoms (dopant) inserted into the silicon crystal structure by placing the atoms using a tunneling microscope.
The President can swear on whatever he likes. It just so happens they've all been Christian and so they swear on the Bible. Again, this is the crucial separation here. The personal beliefs of leaders may very well be Christian, but they as public servants should know that that's as far as it goes, their personal belief.
It is also exactly what was said in your quote above. Translated: "since the majority of this nation is [at that time] Christian, they will vote in a way that Christians would vote, and as a democracy, we enact the policies they vote on".
Essentially stating the obvious. If this nation were to one day be majority Muslim or majority atheist, etc. the same statement would apply.
So yes, you can say it is a Christian Nation in the sense that the majority of its population is Christian. You cannot say it's a Christian Nation in the sense that its governing principles are Christian.
A more mature of that "ReadyBoost" exists in Windows 7, which has caught on; a lot. I imagine that something like it exists in Linux and OSX (or will if drives exist).
The OS will know enough not to thrash the flash cache with swapfile operations.
In the real world, a hybrid drive such as Seagate is proposing is a lot better in almost every way thinkable. It's just one drive, so it will fit in basically every computer in existence and it functions completely automatic, as the user is presented with just one storage medium. The tests in the article prove this type of drive is both faster than traditional drives and a lot cheaper than SSD's, so it really is best of both worlds.
It's not actually the best of both worlds. It's more along the lines of something good from both worlds. First, this only improves read speed as it's not used as a write-cache. Second, random read and write speeds are just as abysmal as traditional HD's. And lastly, it's not quite going to match the price/capacity of traditional drives due to the need for mult-GB of SLC cache.
Now, it's certainly better than a traditional drive at sustained reads and offers better price/capacity vs SSD drives. But I get the feeling that a pure hardware, blind algorithm isn't the best way to go about this.
I would've preferred it if the drive presented the flash cache as a ReadyBoost drive to Windows and have the OS manage what needs to be cached. Certainly Windows knows more about which of its own disgusting innards needs to be readily accessible better than a hardware algorithm.
Are you saying that some African Americans during the time of slavery would have been unable of creating specializations, investments, and technological developments if they were freed?
I'm saying that at a time when labor (males living past 12) were in such high demand, they contributed more through manual labor than any invention, specialization or technological development from the one in a million slaves (and for truly revolutionary tech, that number is high) would've contributed.
Think about it this way, you're talking about effectively ending the entire American economy as it was in the 1600's.
Surprisingly many. In fact, the wonderful thing about science is that it really was true to observation. There are just certain conditions that it doesn't work for. The idea of atoms, geometry, planetary bodies, etc. all come from the Greeks and for the most part, are true today. Euclidean geometry is simply true for certain scales of size and time and energy levels. It doesn't describe things at high energy or small size very well.
The Earth was known to be round-ish (still true today, last I checked), the Sun was the center of the solar system (still true) and the planetary bodies orbited in an elliptical orbit (still true).
But that's not even the point here. It isn't just "commonly accepted" data that is being rejected by Texas. It's knowledge reviewed, scrutinized and accepted by historians who've devoted their lives to studying this field. Those "experts" that Texas has to "stand up to".
He didn't say the founders weren't Christians. He said the founding principles aren't Christian. The founders were smart enough to see how politics corrupts religion and vice versa. They built the government without inserting much if any Biblical principles into it. See anything in the Constitution about coveting wives, worshiping on the 7th day or giving up worldly wealth?
The claims this country is a "Christian" country is very much false. The founders were smart enough to separate their religious beliefs from what they learned through history and philosophy as functional, fair and resilient government.
Slashdot Mirror
And if you want further proof, here is another page posting unabridged benchmark results, which indicate clearly that the Double benchmark uses iSSE2 (you can see the same ~5.5x drop in performance as with the other benchmark I linked). PcPer was just incredibly lazy marking their benchmark graphs.
Atom sucks at double precision. You can deny it all you want, but the benchmarks ring true. I can't say I'm surprised - double performance is one of the first things to go when you're targeting low-power.
Interesting. That's quite a big issue albeit probably not for the applications Atom is targeted for. Still, though, ARM will make it look terrible.
For most workloads, if the server is not busy, you'll get much better power usage from the Xeon as well.
Not really. Look at the idle power numbers on a Xeon compared to an Atom. One of the things about low power (and low-clocking) chips is that leakage is orders of magnitude less. Atom not only runs on a lower-leakage process, it also runs at a lower voltage than Xeons. So while dynamic power-per-unit-work may not be competitive in aggregate, idle power is not even close.
While "shifting features" is indeed not accurate, the point stands. Compared to ARM (or even Athlon and Nehalem/Lynnfield processors), Atom (prior to Pinetrail and Moorestown) was less integrated. So to compare power numbers, you can't just look at the CPU itself but need to include the memory controller (when compared to Athlon and Nehalem/Lynnfield) and PCI-E controller (when compared to Nehalem/Lynnfield). When compared to most ARM SoC's, you'll need to include graphics, miscellaneous I/O, GPS, video processor, image processor, memory controller, USB, CPU bus, peripheral bus, chip-to-chip interconnect, and sometimes a 3G modem.
Even Moorestown is very misleading when they did their comparison with ARM. They conveniently showed off numbers for tasks that use the dedicated components (video decoding, audio playback, talk time on GSM/CDMA, etc.) -- much of which was done by the "helper" chip and not the chip Intel fabbed themselves -- but left out any tasks that would tax the main CPU. Comparing those numbers to ARM SoC's made it look competitive. Then they showed performance numbers of the CPU compared to ARM CPU's -- which, btw, seemed to increase linearly with number of threads, notice how a dual-core A9 was twice the performance of a single? -- but left out the power numbers for when the CPU was taxed heavily.
I mean, it's marketing. But that hardly means it's not underhanded.
Atom has a TDP of ~2W and I would speculate a maximum power consumption of ~4W. A C2050 has a maximum power consumption of 247W. For the same power bugdet as a C2050, you can use 60 Atoms each with its own pocket of memory and condition code handlers.
But it's a poor choice for high-precision scientific calculations. The vector unit on the Atom is optimized for 32-bit Floats, and when you switch-up to Doubles, the performance drops off the map.
See this performance comparison (Processor Multimedia). The dual-core Atom 1.6 is actually faster than the dual-core Athlon 64 1.6 when running SSE 32-bit Floats (this is due to the dual 64-bit SSE units on the Athlon 64, and poor scheduling on AMD's part). But when you transition to Double precision, the performance on the Athlon 64 only drops by roughly 2x, whereas the Atom sees a 5-6x performance drop!
This means you have very little flexibility in workloads you can use - if you're not using 32-bit precision, you've wasted your money.
That was for x87 FP though. While it's important in that a lot of code out there might still use it, for performance on any Intel architecture, you're supposed to use SSE (scalar SSE if you're not doing SIMD). I would like to see SSE performance broken down to SP vs DP. I suspect, from what little I know of the vector pipeline, that it'd be pretty much exactly half the performance.
It's not so bad if you're running a highly-multithreaded program on the system. The Atom uses the SMT to great advantage, in most cases seeing a %30 performance increase. But yeah, for single-threaded code, the Atom is a joke.
Even for highly threaded programs, the performance/watt will not be competitive with (lightly) OoOE architectures. In-order is simply inefficient from a power standpoint. Unless they manage to cram 4 threads simultaneously on a single Atom core, even 2 heavy threads won't be able to saturate every execution slot and if you throw more at it, you'll get the overhead of context switching and cache thrashing.
Stalls are bad and in-order has a lot of them.
If that's the case, Atom is a horrible choice for the CPU. Atom's strongest feature is its 128-bit vector (SSE) unit for SIMD FLOPS.
The in-order (albeit SMT-enabled), 2-issue integer/LDST/Conditional pipeline is lackluster at best from either a performance or a performance/watt standpoint.
Yes and no. It really depends on how the dual-core is architected. Assuming you had 2 cores with independent frequency and voltage scaling and assuming you were running two tasks that were fairly processor and memory intensive, it could indeed be beneficial to run them on 2 cores. Context switching is a killer on cache and memory and having a single CPU throttle between multiple processes is inefficient compared to two separate CPU's working out of their own caches.
Moreover, power is a function of voltage squared; so two CPU's running at 0.8V at 400MHz would require less power than one CPU running at 1.0V at 800MHz.
Windows 7 won't be on ARM but Windows Embedded 7 will. That's the OS Microsoft is making to target tablets and it will come out on both ARM and x86 at the same time.
My ex made a /. headline!
Mobile chips, particularly cell phones, are constrained by power. Supercomputers have the luxury of sucking up as much wattage as they need.
Double the memory traces takes a lot of power. Using DDR3 vs DDR1 takes a lot of power. Using dual memory modules takes a lot of power.
Taking all of those restrictions in mind, it's actually fairly amazing how well modern mobile SoC's do in keeping the processor and subsystems busy through aggressive caching and good memory packaging (for higher speed without extra power).
ARM v7-A support double precision in both NEON (SIMD form) and VFP (scalar form). The Cortex A8 and A9 as well as Qualcomm's Scorpion implement these.
Interestingly, Apple campus is one of the only places in Cupertino with good AT&T 3G coverage. My friend lives about 2 blocks away from Apple HQ on Homestead road and I can't even make calls there let alone actually use the web.
I'm curious about Linpack results when running under Dalvik. I believe Froyo 2.2 enabled Dalvik start using the FPU of the ARM ISA. So what were the results before? Emulated floating point?
On top of that, the ARM ISA allows for SIMD operations. I would assume the VM isn't capable of that.
If we're to compare processing power, the Cortex A8 at 1GHz (A4, OMAP 3640, overclocked Droid) is capable of a vector multiply (2 at a time) every 3 cycles and add in 2 cycles. So that's about 400 MFLOPS.
Those aren't advancements; they're political/social changes. Science discovers facts; it does not set policy. Anti-intellectualism is not choosing not to implement a particular finding or method; it's sticking your head in the sand and refusing to even acknowledge its there for some retarded concept that the danger will be gone "if you keep all that devil knowledge out".
Knowing how to, say, trade one person's life for another is very different than actually doing it. Whether or not something should be done is outside the realm of science and deep in philosophy and politics. You demonstrate the exact ignorance and knee-jerk fear that we're speaking of when you try to conflate the two.
Ya. I mean all those people who didn't want penicillin, modern plumbing, flight, electricity, smallpox vaccines, the internet you're typing this on, etc. are doing so well.
Do you even realize the insane irony of making the "we don't want your advancements" argument on the internet?
It's not just 7 atoms. It's 7 phosphorus atoms inserted into a silicon crystal that's 4nm. A silicon atom is roughly ~0.4nm in diameter.
The headline is (again) inaccurate. It's not a 7-atom transistor, it's a transistor with 7 phosphorus atoms (dopant) inserted into the silicon crystal structure by placing the atoms using a tunneling microscope.
The President can swear on whatever he likes. It just so happens they've all been Christian and so they swear on the Bible. Again, this is the crucial separation here. The personal beliefs of leaders may very well be Christian, but they as public servants should know that that's as far as it goes, their personal belief.
It is also exactly what was said in your quote above. Translated: "since the majority of this nation is [at that time] Christian, they will vote in a way that Christians would vote, and as a democracy, we enact the policies they vote on".
Essentially stating the obvious. If this nation were to one day be majority Muslim or majority atheist, etc. the same statement would apply.
So yes, you can say it is a Christian Nation in the sense that the majority of its population is Christian. You cannot say it's a Christian Nation in the sense that its governing principles are Christian.
A more mature of that "ReadyBoost" exists in Windows 7, which has caught on; a lot. I imagine that something like it exists in Linux and OSX (or will if drives exist).
The OS will know enough not to thrash the flash cache with swapfile operations.
That's where you're wrong, according to this, the random reads at least on the smaller files are significantly faster than a normal HDD.
http://www.overclockersclub.com/reviews/seagate_momentus_xt_500gb/4.htm
I was going by Anandtech's results:
http://www.anandtech.com/show/3734/seagates-momentus-xt-review-finally-a-good-hybrid-hdd/3
Of course, if you go in 1MB files, it's going to be faster (assuming the file isn't fragmented).
In the real world, a hybrid drive such as Seagate is proposing is a lot better in almost every way thinkable. It's just one drive, so it will fit in basically every computer in existence and it functions completely automatic, as the user is presented with just one storage medium. The tests in the article prove this type of drive is both faster than traditional drives and a lot cheaper than SSD's, so it really is best of both worlds.
It's not actually the best of both worlds. It's more along the lines of something good from both worlds. First, this only improves read speed as it's not used as a write-cache. Second, random read and write speeds are just as abysmal as traditional HD's. And lastly, it's not quite going to match the price/capacity of traditional drives due to the need for mult-GB of SLC cache.
Now, it's certainly better than a traditional drive at sustained reads and offers better price/capacity vs SSD drives. But I get the feeling that a pure hardware, blind algorithm isn't the best way to go about this.
I would've preferred it if the drive presented the flash cache as a ReadyBoost drive to Windows and have the OS manage what needs to be cached. Certainly Windows knows more about which of its own disgusting innards needs to be readily accessible better than a hardware algorithm.
Are you saying that some African Americans during the time of slavery would have been unable of creating specializations, investments, and technological developments if they were freed?
I'm saying that at a time when labor (males living past 12) were in such high demand, they contributed more through manual labor than any invention, specialization or technological development from the one in a million slaves (and for truly revolutionary tech, that number is high) would've contributed.
Think about it this way, you're talking about effectively ending the entire American economy as it was in the 1600's.
History is taught by the victors.
Surprisingly many. In fact, the wonderful thing about science is that it really was true to observation. There are just certain conditions that it doesn't work for. The idea of atoms, geometry, planetary bodies, etc. all come from the Greeks and for the most part, are true today. Euclidean geometry is simply true for certain scales of size and time and energy levels. It doesn't describe things at high energy or small size very well.
The Earth was known to be round-ish (still true today, last I checked), the Sun was the center of the solar system (still true) and the planetary bodies orbited in an elliptical orbit (still true).
But that's not even the point here. It isn't just "commonly accepted" data that is being rejected by Texas. It's knowledge reviewed, scrutinized and accepted by historians who've devoted their lives to studying this field. Those "experts" that Texas has to "stand up to".
He didn't say the founders weren't Christians. He said the founding principles aren't Christian. The founders were smart enough to see how politics corrupts religion and vice versa. They built the government without inserting much if any Biblical principles into it. See anything in the Constitution about coveting wives, worshiping on the 7th day or giving up worldly wealth?
The claims this country is a "Christian" country is very much false. The founders were smart enough to separate their religious beliefs from what they learned through history and philosophy as functional, fair and resilient government.