To be fair, Sandy Bridge does include a video encoding accelerator/transcoder as well. However Atom doesn't at the moment, although Medfield will presumably include one.
The thing about OMAP 44xx is that the accelerator is a programmable DSP, not fixed function hardware. The specs say it can do 1080p30 (at 30fps). That means it could do VP8, or presumably even your own custom DSP code.
We should also be considering the quality of the output though.
The latest ARM processors performance is comparable to a 1996 Pentium class processor.
Troll much?
These SoCs can devote a dozen or so mm^2 for the CPU cores, yet they still achieve performance comparable to a 1GHz Atom but at far far lower power consumption levels. In addition the rest of the SoC contains application specific accelerators (graphics, video, security are the most common) for the difficult tasks. The Phoronix benchmarks sadly didn't test the SoC as a whole, just the CPU cores. It's also up in the air whether ARM Neon was actually utilised in all the tests.
It had 22x23 character blocks - hence generating graphics required some effort!
Namely - set VIC20 into high-characters mode and set the display size to 22x11 with each character cell being assigned a unique character (do this vertically to get a display that is 22 characters of 176 pixels in height), then write a graphics library that will treat characters 0..241 as a funkily arranged bitmap.
Of course coding on a 22x23 display is extremely unpleasant in the first place. It is a shame that the VIC couldn't create a 44x23 display mode in higher resolution.
Most home computers since then had pixel-addressable graphics modes and higher resolution editing modes thankfully, making BASIC far more visual on those systems.
Yeah, because a beginner wants to deal with File I/O...
1000 DATA "A small kitten, miaowing." 1010 DATA "A glowing kitten, chasing a mouse." 1020 DATA "..."
VB.NET is not the BASIC this article is talking about. It's about removing complexity. If you have file I/O then you want it on the level of the following:
I remember learning Pascal during A-Level (UK - 16-18 post-secondary; pre-university level), after coming from BASIC on the Amstrad CPC.
What BASIC did was allow you to create visual things very easily - this is a great way to learn to program because you get a lot of feedback. You didn't have to worry about files or streams or so on, you just INPUTed and PRINTed, you DRAWed and so forth. If you needed data, you'd put it in DATA statements, rather than read it from storage (although you could do that too), thus making the language even more accessible (and allowing for program listings in magazines!).
Pascal, however, taught programming concepts far better. Variable declaration, structure, etc. But the I/O was purely text, it was very dry and un-engaging. This is also the problem with many other modern languages - although Javascript now has a canvas. There's probably a BASIC interpreter online implemented in Javascript that draws on a canvas...
The hidden opportunity was LOGO. A language taught in secondary school merely as a "move the turtle around to draw" mechanism. Underneath was a powerful functional language that was never touched.:-(
Anyway, BASIC was a 'hook' back then, it got many people into computers, and acted as a gateway to assembler and other languages. Some BASICs even included assembler - e.g., BBC BASIC - which was probably the best home computer BASIC of the 80s as it also included PROCedures.
The reason that Python is the introductory language at MIT is probably the same as why ML was (is?) the introductory language at Cambridge. The students are not likely to have used it before. It puts the students on a more equal footing at the beginning of the university course as they all have to learn something new.
Does Python make a good introductory language in the modern age? Why not? I don't know enough about it to comment one way or the other, but compared to interpreted BASIC it's probably a reasonable match. Indentation thing is annoying, then again FOR... NEXT or WHILE... WEND aren't ideal either.
What is really missing however is magazines with coding tutorials and type-ins, that make people learn to program by rote repetition of entering programs, then wanting to edit the programs to add features, change the maps, graphics, etc, and so on.
For some reason they are using a physical PCMCIA interface, but electrically it's completely different, with USB2, video, ethernet, misc I/O, I2C, etc being carried over the pins.
I guess it's prettier than a 60-pin header.
But at some point it needs something to plug into that makes it into a system that can be used.
I'm amazed at the cost that some of these ARM SoCs are being priced at. $7 is really rather cheap, especially for a 1.5GHz ARM Cortex A8 (sure, it's an older design, but next year's $7 SoC might have a dual-core ARM A7 instead) with ARM Mali 400 graphics, "2160p" video decode, and so on.
However the Raspberry Pi will probably be out a lot sooner than this hardware, and then I guess they'll start work on the Raspberry Flan or something.
The article is a bit over the top really...
on
The Condescending UI
·
· Score: 2
I fail to see how a UI THEME somehow makes a UI less 'condescending'. The Ribbon is still there in Windows 7 'classic', you just lose the GPU acceleration, and instead get the plain grey (with wider window borders) ugliness that Windows has had since Win95 (when it had to look okay in 16 colours). Of course, the Ribbon looks really out of place in the classic theme.
As for the faux-real-world UIs in the Apple apps, the real issue they have is that they present an expectation of more functionality that should be present because they are so similar to the real world example. They're only offensive if you think of the wasted RAM and CPU cycles that go into rendering these interfaces. Then again if you sync with Google you can just use Google's rather stark online applications instead and not worry about the bit of faux leather or torn paper UI silliness.
At least Office on the Mac still gives you the menu alongside the Ribbon. Best of both worlds eh?
The tile-based rendering approach that mobile GPUs use (including the ARM Mali and the PowerVR SGX) means that 4x AA is effectively free anyway. According to their websites, anyway.
The ARM Mali 6xx series also supports OpenCL (and DirectCompute), and in terms of GPU performance it looks like it ranges from AMD Ontario up to half-a-Llano.
Coupled with the high performance ARM CPUs (high performance meaning relative to older ARM cores), these will form a compelling offering for lots of products. And with SoCs costing under $40, it is an obvious way for future OEMs to reduce the cost of Ultrabooks, where the cost of the low-power Intel CPU is crippling the price of the devices.
And as for your other respondent, an ARM Cortex A9 is roughly equivalent or better than an equivalent clocked Intel Atom (but obviously available in quad configurations), and Cortex A15 is even faster per clock, and clocks even higher (2.5GHz+).
Tablets are approaching 1280x800 today. Most PS3 games render at sub 720p resolution (1024x600 for example, then scaled up).
In terms of GFLOPS this ARM Mali T658 will be pushing ~270 GFLOPS. That's comparable to the PS3 RSX's throughput, although not directly comparable due to architectural differences (unified shaders in Mali vs. distinct vertex and pixel shaders in the RSX).
Double the clockspeed for a netbook/laptop implementation, and you're over half a TFLOP. This can be achieved by merely fabbing it on a HP (General) process rather than an LP process - presumably 28nm or below. Sure, this won't apply to a smartphone or tablet, but why not a cheap future console / media player?
Nah, it's a lot less than that, mobile phones don't have 24WHr batteries (all-day computing).
Hell, an iPad lasts 10 hours when being used from a 25Whr battery - that's 2.5W load, including the display.
Mobile phones have 8Whr batteries. Samsung Galaxy S2 is 6.6Whr for example. It can standby for a couple of days (with no apps running in the background) so standby power consumption for the entire phone has to be 0.125W. In reality power consumption in use is more like an average of 0.25W but that does include light usage throughout the day.
A T604 core achieve around 68 GFLOPS according to ARM's documentation.
ARM say "On the compute side, Mali-T658 provides four times the processing power of Mali-T604" (http://blogs.arm.com/multimedia/625-arm-mali-t658-gpu-arrives-at-the-japan-technical-symposium/?sf2518165=1) so that's 272 GFLOPS. The PS3 RSX does 400 GFLOPS according to Wikipedia, but 255 GFLOPS according to other sources. Regardless, it's in the same ballpark. Most PS3 games are rendered at something like 1024x600 to 1280x720, which is roughly equivalent to high-end smartphone resolutions too.
This is ten times faster than ARM Mali 400 graphics, as used in the Samsung Galaxy SII (in a Quad-core configuration). That's a 45nm chip as far as I am aware. It's in a phone, so the graphics can't be using a significant amount of power even when running flat out.
This GPU is coming out in two years time, and will probably be built on a mature 28nm process or even a 22nm/20nm process. That is 1.5 - 2.5 full node shrinks - you can fit 3 - 6 Mali 400s into the same area of die - that's 3 - 6x the performance (GPU scales very well, and we're also ignoring that not all of the GPU would need to be duplicated) without increasing the clock or other enhancements that are surely in this new GPU.
10x the performance at 1W would seem to be very achievable to me, if the current Mali 400 uses under 1W currently.
PS3 level graphics on a mobile device is nothing to be sniffed at.
It's also natural progression, by 2013 it will be seven or eight years since the PS3 was released.
Also maybe the graphics chip can be run at higher speeds to achieve even better performance (at higher power consumption), even if that wouldn't be an option for mobile devices apart from netbooks and above.
Indeed, the obvious defence in court is that you shook hands with someone prior to the test.
The only way for the system to work is for the person to wash their hands, wait for some sweating to occur, then take the test. And that's removes the rapid aspect of the system.
The thought processes of politicians in most countries have nothing to do with sanity, being reasonable or taking facts into account.
Create a furore, or wait for the media to create one, and then overreact to be seen to be doing something. Once the law is enacted, never review it for efficacy.
So did someone put all those names and email addresses into the To or Cc field of an email? That would be a rather large email to receive!!!
In addition, this is why proper mailer software that they should have used handles the email composition and sending internally - so that the addresses will be in the BCC field or each address will get its own email sent just for itself.
If they just sent out an excel file with the details in, that's even worse. There is no excuse for a workflow that involves someone manually getting a particular list of recipients from a CRM and then emailing that file in any way. Unless you have five employees.
So... how does this happen still? Stupidity and poor IT workflows.
So you would prefer that they had 16 of the T3 cores at 0.25x per core? Running at maybe 2GHz each because of the design?
Nah, 8 T4 cores at 0.5x per core, and far greater per-core performance, seems to be a better deal to me. You might lose half of your threads per CPU, but as the article says, that level of threading is getting more and more niche.
But not as good a deal as not using Oracle in the first place.
This chip was probably designed before Oracle finished the takeover of Sun, so saying that Oracle is making this change is quite a leap.
Also the linked article suggests why the design decisions were made in this chip - namely extreme threading compared to the mass market CPUs is getting more and more niche as the mass market CPUs get to 16 threads per socket themselves. Hence this chip going down a higher performance per thread model, with a mode for running a single thread very fast on a core should it need it.
And ultimately you still have the OS scheduler if you want to run more threads than the hardware can handle, and given that the per-thread hardware performance is far higher on this CPU you're not giving up much, if any, threading capability compared to the previous generation.
Sure, you can make the figures look good in the short term by slashing R&D and firing everybody, but in the longer run you have no new products coming down the line.
Bye bye HP. In ten years time you'll be a niche printer ink seller.
Having looked at the video and some photos, it looks like the plane crashed just in front of the boxed areas rather than inside.
But the shrapnel coming off the crash would surely have sliced anyone within 30ft to ribbons, and the boxed area was very busy. And there are some very organic looking lumps in the air in the photos too.
Slashdot Mirror
To be fair, Sandy Bridge does include a video encoding accelerator/transcoder as well. However Atom doesn't at the moment, although Medfield will presumably include one.
The thing about OMAP 44xx is that the accelerator is a programmable DSP, not fixed function hardware. The specs say it can do 1080p30 (at 30fps). That means it could do VP8, or presumably even your own custom DSP code.
We should also be considering the quality of the output though.
The TI OMAP 4460 actually does use a full programmable DSP, and can decode and encode 1080p30 apparently.
http://www.ti.com/general/docs/wtbu/wtbuproductcontent.tsp?templateId=6123&navigationId=12843&contentId=53243
The 1.8GHz OMAP 4470 should be out soon as well - that will surely perform even more comparably to the Intel Medfield offering coming in the same timeframe.
http://www.ti.com/general/docs/wtbu/wtbuproductcontent.tsp?templateId=6123&navigationId=12869&contentId=123362
Troll much?
These SoCs can devote a dozen or so mm^2 for the CPU cores, yet they still achieve performance comparable to a 1GHz Atom but at far far lower power consumption levels. In addition the rest of the SoC contains application specific accelerators (graphics, video, security are the most common) for the difficult tasks. The Phoronix benchmarks sadly didn't test the SoC as a whole, just the CPU cores. It's also up in the air whether ARM Neon was actually utilised in all the tests.
BASIC on any mobile phone today would be sandboxed so it couldn't get access to the telephony capabilities.
Writing a robo-dialler in the native APIs probably isn't that hard either.
It had 22x23 character blocks - hence generating graphics required some effort!
Namely - set VIC20 into high-characters mode and set the display size to 22x11 with each character cell being assigned a unique character (do this vertically to get a display that is 22 characters of 176 pixels in height), then write a graphics library that will treat characters 0..241 as a funkily arranged bitmap.
Of course coding on a 22x23 display is extremely unpleasant in the first place. It is a shame that the VIC couldn't create a 44x23 display mode in higher resolution.
Most home computers since then had pixel-addressable graphics modes and higher resolution editing modes thankfully, making BASIC far more visual on those systems.
Yeah, because a beginner wants to deal with File I/O...
1000 DATA "A small kitten, miaowing."
1010 DATA "A glowing kitten, chasing a mouse."
1020 DATA "..."
VB.NET is not the BASIC this article is talking about. It's about removing complexity. If you have file I/O then you want it on the level of the following:
kittah$[] = FILE "kitten.txt"
kitten = IMAGE "kitten.png"
Maybe people who have been programming for years and years aren't the best people to talk to about introductory programming languages!
I remember learning Pascal during A-Level (UK - 16-18 post-secondary; pre-university level), after coming from BASIC on the Amstrad CPC.
What BASIC did was allow you to create visual things very easily - this is a great way to learn to program because you get a lot of feedback. You didn't have to worry about files or streams or so on, you just INPUTed and PRINTed, you DRAWed and so forth. If you needed data, you'd put it in DATA statements, rather than read it from storage (although you could do that too), thus making the language even more accessible (and allowing for program listings in magazines!).
Pascal, however, taught programming concepts far better. Variable declaration, structure, etc. But the I/O was purely text, it was very dry and un-engaging. This is also the problem with many other modern languages - although Javascript now has a canvas. There's probably a BASIC interpreter online implemented in Javascript that draws on a canvas...
The hidden opportunity was LOGO. A language taught in secondary school merely as a "move the turtle around to draw" mechanism. Underneath was a powerful functional language that was never touched. :-(
Anyway, BASIC was a 'hook' back then, it got many people into computers, and acted as a gateway to assembler and other languages. Some BASICs even included assembler - e.g., BBC BASIC - which was probably the best home computer BASIC of the 80s as it also included PROCedures.
The reason that Python is the introductory language at MIT is probably the same as why ML was (is?) the introductory language at Cambridge. The students are not likely to have used it before. It puts the students on a more equal footing at the beginning of the university course as they all have to learn something new.
Does Python make a good introductory language in the modern age? Why not? I don't know enough about it to comment one way or the other, but compared to interpreted BASIC it's probably a reasonable match. Indentation thing is annoying, then again FOR ... NEXT or WHILE ... WEND aren't ideal either.
What is really missing however is magazines with coding tutorials and type-ins, that make people learn to program by rote repetition of entering programs, then wanting to edit the programs to add features, change the maps, graphics, etc, and so on.
For some reason they are using a physical PCMCIA interface, but electrically it's completely different, with USB2, video, ethernet, misc I/O, I2C, etc being carried over the pins.
I guess it's prettier than a 60-pin header.
But at some point it needs something to plug into that makes it into a system that can be used.
I'm amazed at the cost that some of these ARM SoCs are being priced at. $7 is really rather cheap, especially for a 1.5GHz ARM Cortex A8 (sure, it's an older design, but next year's $7 SoC might have a dual-core ARM A7 instead) with ARM Mali 400 graphics, "2160p" video decode, and so on.
However the Raspberry Pi will probably be out a lot sooner than this hardware, and then I guess they'll start work on the Raspberry Flan or something.
I fail to see how a UI THEME somehow makes a UI less 'condescending'. The Ribbon is still there in Windows 7 'classic', you just lose the GPU acceleration, and instead get the plain grey (with wider window borders) ugliness that Windows has had since Win95 (when it had to look okay in 16 colours). Of course, the Ribbon looks really out of place in the classic theme.
As for the faux-real-world UIs in the Apple apps, the real issue they have is that they present an expectation of more functionality that should be present because they are so similar to the real world example. They're only offensive if you think of the wasted RAM and CPU cycles that go into rendering these interfaces. Then again if you sync with Google you can just use Google's rather stark online applications instead and not worry about the bit of faux leather or torn paper UI silliness.
At least Office on the Mac still gives you the menu alongside the Ribbon. Best of both worlds eh?
The tile-based rendering approach that mobile GPUs use (including the ARM Mali and the PowerVR SGX) means that 4x AA is effectively free anyway. According to their websites, anyway.
The ARM Mali 6xx series also supports OpenCL (and DirectCompute), and in terms of GPU performance it looks like it ranges from AMD Ontario up to half-a-Llano.
Coupled with the high performance ARM CPUs (high performance meaning relative to older ARM cores), these will form a compelling offering for lots of products. And with SoCs costing under $40, it is an obvious way for future OEMs to reduce the cost of Ultrabooks, where the cost of the low-power Intel CPU is crippling the price of the devices.
And as for your other respondent, an ARM Cortex A9 is roughly equivalent or better than an equivalent clocked Intel Atom (but obviously available in quad configurations), and Cortex A15 is even faster per clock, and clocks even higher (2.5GHz+).
Tablets are approaching 1280x800 today. Most PS3 games render at sub 720p resolution (1024x600 for example, then scaled up).
In terms of GFLOPS this ARM Mali T658 will be pushing ~270 GFLOPS. That's comparable to the PS3 RSX's throughput, although not directly comparable due to architectural differences (unified shaders in Mali vs. distinct vertex and pixel shaders in the RSX).
Double the clockspeed for a netbook/laptop implementation, and you're over half a TFLOP. This can be achieved by merely fabbing it on a HP (General) process rather than an LP process - presumably 28nm or below. Sure, this won't apply to a smartphone or tablet, but why not a cheap future console / media player?
Nah, it's a lot less than that, mobile phones don't have 24WHr batteries (all-day computing).
Hell, an iPad lasts 10 hours when being used from a 25Whr battery - that's 2.5W load, including the display.
Mobile phones have 8Whr batteries. Samsung Galaxy S2 is 6.6Whr for example. It can standby for a couple of days (with no apps running in the background) so standby power consumption for the entire phone has to be 0.125W. In reality power consumption in use is more like an average of 0.25W but that does include light usage throughout the day.
A T604 core achieve around 68 GFLOPS according to ARM's documentation.
ARM say "On the compute side, Mali-T658 provides four times the processing power of Mali-T604" (http://blogs.arm.com/multimedia/625-arm-mali-t658-gpu-arrives-at-the-japan-technical-symposium/?sf2518165=1) so that's 272 GFLOPS. The PS3 RSX does 400 GFLOPS according to Wikipedia, but 255 GFLOPS according to other sources. Regardless, it's in the same ballpark. Most PS3 games are rendered at something like 1024x600 to 1280x720, which is roughly equivalent to high-end smartphone resolutions too.
This is ten times faster than ARM Mali 400 graphics, as used in the Samsung Galaxy SII (in a Quad-core configuration). That's a 45nm chip as far as I am aware. It's in a phone, so the graphics can't be using a significant amount of power even when running flat out.
This GPU is coming out in two years time, and will probably be built on a mature 28nm process or even a 22nm/20nm process. That is 1.5 - 2.5 full node shrinks - you can fit 3 - 6 Mali 400s into the same area of die - that's 3 - 6x the performance (GPU scales very well, and we're also ignoring that not all of the GPU would need to be duplicated) without increasing the clock or other enhancements that are surely in this new GPU.
10x the performance at 1W would seem to be very achievable to me, if the current Mali 400 uses under 1W currently.
PS3 level graphics on a mobile device is nothing to be sniffed at.
It's also natural progression, by 2013 it will be seven or eight years since the PS3 was released.
Also maybe the graphics chip can be run at higher speeds to achieve even better performance (at higher power consumption), even if that wouldn't be an option for mobile devices apart from netbooks and above.
Indeed, the obvious defence in court is that you shook hands with someone prior to the test.
The only way for the system to work is for the person to wash their hands, wait for some sweating to occur, then take the test. And that's removes the rapid aspect of the system.
The thought processes of politicians in most countries have nothing to do with sanity, being reasonable or taking facts into account.
Create a furore, or wait for the media to create one, and then overreact to be seen to be doing something. Once the law is enacted, never review it for efficacy.
So did someone put all those names and email addresses into the To or Cc field of an email? That would be a rather large email to receive!!!
In addition, this is why proper mailer software that they should have used handles the email composition and sending internally - so that the addresses will be in the BCC field or each address will get its own email sent just for itself.
If they just sent out an excel file with the details in, that's even worse. There is no excuse for a workflow that involves someone manually getting a particular list of recipients from a CRM and then emailing that file in any way. Unless you have five employees.
So ... how does this happen still? Stupidity and poor IT workflows.
So you would prefer that they had 16 of the T3 cores at 0.25x per core? Running at maybe 2GHz each because of the design?
Nah, 8 T4 cores at 0.5x per core, and far greater per-core performance, seems to be a better deal to me. You might lose half of your threads per CPU, but as the article says, that level of threading is getting more and more niche.
But not as good a deal as not using Oracle in the first place.
This chip was probably designed before Oracle finished the takeover of Sun, so saying that Oracle is making this change is quite a leap.
Also the linked article suggests why the design decisions were made in this chip - namely extreme threading compared to the mass market CPUs is getting more and more niche as the mass market CPUs get to 16 threads per socket themselves. Hence this chip going down a higher performance per thread model, with a mode for running a single thread very fast on a core should it need it.
And ultimately you still have the OS scheduler if you want to run more threads than the hardware can handle, and given that the per-thread hardware performance is far higher on this CPU you're not giving up much, if any, threading capability compared to the previous generation.
Sure, you can make the figures look good in the short term by slashing R&D and firing everybody, but in the longer run you have no new products coming down the line.
Bye bye HP. In ten years time you'll be a niche printer ink seller.
Having looked at the video and some photos, it looks like the plane crashed just in front of the boxed areas rather than inside.
But the shrapnel coming off the crash would surely have sliced anyone within 30ft to ribbons, and the boxed area was very busy. And there are some very organic looking lumps in the air in the photos too.