Hi Anand, Just wanted to say that I look forward to your podcast. I listen to them during my commute and sometime find my self talking back to them! (I know. Some of us tech guys are a little weird!) Don't make us wait soooo long!!! -Signed, Huge Huge fan of the site and it's writers. Chris.
Another great podcast guys. I really like the dynamic between Anand and Brian but it is a shame the other guys aren't included in this one as I think their different POVs add a lot to the discussion (when they can get a word in :P)
Hopefully we don't have to wait 2 months for the next one, I was starting to dread that you guys had decided to discontinue them. I'm glad that worry wasn't warranted. Thanks again guys.
Can you you guys do a detailed article on how the CCI is broken in the Exynos Octa? I would like that.
I have a question:- If the CCI is broken and the L2 Cache data has to pushed into the DRAM and back into the L2 Cache of the other cluster. Won't it improve performance consistency if you disable the A15 cluster all together?
So... This is actually podcast 20, which you posted before. Sorry to see you guys are having issues today, between this and the images not loading in the Richland mobile article this morning.
Excellent, was looking forward to this (what can I say, I like to listen to some interesting stuff when I'm driving). I want to know how soon before I can get the GPE ROM on my AT&T HTC One. It's already as smooth as butter with Sense 5, but would love to play around with almost-stock Android just to see how it runs.
Also, I'm surprised you guys don't offer a smaller MP3 size at least as a secondary option. 32 kbps (CBR preferably) is enough for speech, unless there is some background noises you want us to pick up on, ha ha. 185MB is overkill for a podcast, unless it was a music podcast.
This. I've been putting off commenting on the bitrate of your podcasts, but it is quite unnecessary to use a 204 kbps bitrate on a podcast as you currently are for the MP3 versions. I'd say 128 kbps is the absolute max needed for (non-music) podcasts, with 96 kbps being ideal in my opinion, and 80 or 64 kbps sounding fine as well.
I would prefer that you not go any lower than that as it starts to sound significantly worse when you go below 64 kbps.
This would not only decrease the strain on your servers, but allow us to start listening to the podcast sooner (for those of us who download it to listen on-the-go), as well as filling up less space in portable devices and using less bandwidth.
I actually really appreciate the high bitrate. They use quality microphones and the podcast sounds really good. 200 kbs is perhaps a little more than necessary, but it would be a shame to add that gargley sound to such a nice recording. I'm certainly not listening to the quality of their voices more than what they voice, but I appreciate the quality nonetheless. I would prefer variable bitrate (constant quality) just to save space, but I'm not to picky about that.
It already is variable bitrate, the bitrate I posted was the average.
I don't really personally care if AnandTech goes variable or constant bitrate, as it doesn't affect me, but I've been told that constant is more compatible with older devices.
My recommendation would apply to the max bitrate if it's variable..
This is something we've been debating for a while, and even though I'm all for the really high quality audio I agree for mobile use it makes sense to have a lower bitrate. I'm pretty sure that's coming :)
Even your M4A feed has an average bitrate of lower than 128 kbps and is close to 100MB vs the MP3's 185MB with an average of over 200 kbps. Obviously there is the codec difference, but I honestly cannot hear any difference between the two, and probably wouldn't even if you went even lower.
The filesize is especially noticeable to me because the bandwidth when downloading your podcasts seems to be bottlenecked on your end. When I was downloading it near release I think I was getting ~1 Mbps :(
I don't mean to disrespect you guys, I just wish for a smoother and more convenient experience when downloading my favorite podcast.
Entertaining as always. Especially looking forward to that Haswell MBA vs iPad4 efficiency article!
One other point that caught my attention - Intel's 100mV advantage on 22nm is on the threshold voltage. Which is actually a surprising amount when you look at the plot they provided with the release materials and see that they're talking about a ~250mV threshold voltage for 22nm versus 350mV on 32nm. But it's also not really applicable to actual products as near threshold computing is still in the research phase for the most part. However there is a very interesting hint regarding likely operating voltages for Silvermont from the information Intel released on their 22nm SoC process last December. Specifically, at the same leakage levels the 22nm SoC drive currents at 0.75V are roughly 2/3 those of the 32nm process at 1.0V. (I also tend to believe the fact that they provide figures at 0.75V instead of the more standard 1.0V just might be implying something.) While I somewhat doubt that Silvermont will run full turbo frequency at 0.75V, that may well be what it needs to run in its sweet spot, and that kind of voltage reduction would certainly explain the marked jump in efficiency.
The comparison between ARM's and Intel's strategies with regards to the number of their architectures is incredibly insightful. I am curious though how you view this dynamic changing in the long term. While its true that as processor tech continues to advance the middle ground will become increasingly competitive, I would venture that Intel's profit model precludes them from ever truly competing with ARM on the low end. Would you disagree?
Nope. Intel will never, EVER, be "competitive" with Core chips against ARM. The dual core chip alone *starts* at $350 for crying out loud. Their only chance is Atom, but last I checked, even that is at least 2x more expensive than the ARM chips in the same category. There's little reason why any Android/Chrome OS OEM, at least, would pick them over the ARM chips.
Plus, Intel still stands no chance in the mobile GPU race with Atom, Imagination's Series 6 and Nvidia's Kepler will obliterate whatever Intel comes out with in the next few years.
This band is my favourite. I've been listening to some others but you guys are great together, are really knowledgable about the low level aspects of products, and talk about all of the topics that are most interesting instead of sticking to a structure and forcing discussion where there is none.
Google Glass needs to have HTC One's sound recording capability [1], and OIS and good low-light performance (just like any camera should today, but it's even more important for Google Glass, where you shake your head a lot more than your hand when moving).
Anand, did you forget ARM chips will be 20nm next year, while Intel's Merrifield will be 22nm? You were talking as if Intel will have an advantage in process for smartphones next year. It won't.
Brian, why don't you get a Nokia Lumia 920? It's rugged phone by default, maybe not water resistant but I left it in snow for about 20 - 30 Minutes by mistake and it didn't even turn off. The Lumia also didn't break, and I tell you, I dropped it several times on different surfaces from about face - height.
I believe, Lumia EOS with the 41MP sensor will be also as good as Galaxy Camera/ Galaxy NX, because it will still be great phone and a decent camera and still quite rugged. Maybe, I am biased a bit - Microsoft really has to do a lot of work and polish on Windows Phone
Anand, give some thought to how successful ARM has been with n-1 foundries. Once a foundry becomes fully depreciated, it's cost to produce a SoC drops dramatically. It is a key to ARM's $10 SoC's.
Until Baytrail, Intel has never had a n-1 product that could compete in the tablet space. When Clovertrail+ becomes n-1, it's cost drops dramatically and under classic microelectronic theory, Intel is supposed to use it to trash the profitability of the tablet middle market. Clovertrail+ is your missing third Intel architecture. The Street cares more about market share than profit margin on products from Intel's obsolete fabs.
@ps4/Xbox 1. One thing that confuses me, if you saw e3 the Xbox 1 games looked real good in my book pertaining to graphics and frame rate. If you look at forza compared to drive club , both first party games, forza looked to be maintaining 60fps and running in 1080p where as drive club club did not look no where near as amazing as forza and also was struggling to maintain 30fps. Does Sony really have a huge real world performance advantage over Xbox 1? Or at the end of the day is it all smoke and mirrors? Not a fan boy BTW. I own both current generation condoles now. I am a huge fps multiplayer gamer though and am really liking the idea of Xbox 1 providing dedicated servers. Can not stand peer to peer connections.
It is odd that a website that has a large enthusiast reader base and reviews enthusiast gear believes that hardcore gamers choose to game on xbox or play station. It is odd that all Brian's friends who he calls "gamers" game on consoles. And it is odd that Anand pretty much plays Halo on xbox. Really! It is even odder that Brian would put a console on his desk and hook it up to his monitor instead of use his pc to game. Not only is the new look for advertisers rather than readers, but this is now pretty much a gadget/mobile site catering to, I guess, not enthusiasts.
Enjoy the teckie info included in your podcasts. Thank you, host (not sure of everyone's name) for explaining acronyms and higher technical references. Brian is enjoyable, but can get a bit tedious with rants. Overall very enjoyable!
Obviously Brian is smart and I do love his rants. However, he is obviously not smart enough to explain complicated concepts in simple terms so that he would not need to disparage people for not getting it the first time. Really, Brian, you are just a rude, arrogant, condescending punk. You are way too full of yourself. But I do love your honesty and your rants. Of course the vast majority of people are pretty stupid, but they don't need to be reminded of that by a punk such as yourself.
I agree. Brian Klug is sort of antisocial and arrogant. In addition to what you mentioned, he constantly talks over everyone else in the podcast, often mentioning his credentials or skills or important friends for no reason at all. I notice it too, but it usually doesn't bother me.
On the other hand I think the "technology arrogance" is perfect for a site like Anandtech, it's extremely useful to have the point of view of someone who actually cares deeply about technology.
What really bothers me though, is when that condescending attitude is combined with ignorance. A couple podcasts back, regarding removable batteries for smartphones, it was "Uhhh! You got what you got! No removable batteries ever again! MicroSD is the worst storage medium and all MicroSD cards are constantly corrupt! Nobody uses it anyway! It costs too much to put it in!"
Really? You're the mobile reviewer on an enthusiast site, and you think people are stupid for not wanting to settle for less battery life and less storage space on a $600+ device? Lol. Removing features that directly affect the usability of the phone in order to save like 7 cents in manufacturing costs makes no sense whatsoever. Meanwhile you won't stop talking about other stuff like the One 4MP camera which nobody except you would ever notice or care about..
"Sell me an Xbox license." Brilliant! For years I've been playing computer games while my Playstation sits lonely in the basement dedicated for HT use. I talk to one group of friends who share the computer interest\hardware and ignore the other group who are gaming on Xbox together. I'm tired of buying computer hardware and 2 different consoles to share the experience with all my friends. One set of hardware, virtualized console OS's and independent DLC\community environments FTW...
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
43 Comments
Back to Article
cacosta619 - Friday, June 28, 2013 - link
Hi Anand, Just wanted to say that I look forward to your podcast. I listen to them during my commute and sometime find my self talking back to them! (I know. Some of us tech guys are a little weird!) Don't make us wait soooo long!!!-Signed,
Huge Huge fan of the site and it's writers.
Chris.
B - Saturday, June 29, 2013 - link
^ +1jimbo2779 - Saturday, June 29, 2013 - link
Another great podcast guys. I really like the dynamic between Anand and Brian but it is a shame the other guys aren't included in this one as I think their different POVs add a lot to the discussion (when they can get a word in :P)Hopefully we don't have to wait 2 months for the next one, I was starting to dread that you guys had decided to discontinue them. I'm glad that worry wasn't warranted. Thanks again guys.
blanarahul - Sunday, June 30, 2013 - link
Can you you guys do a detailed article on how the CCI is broken in the Exynos Octa? I would like that.I have a question:- If the CCI is broken and the L2 Cache data has to pushed into the DRAM and back into the L2 Cache of the other cluster. Won't it improve performance consistency if you disable the A15 cluster all together?
blanarahul - Sunday, June 30, 2013 - link
Found a useful article but it still doesn't answer my question:- http://www.sammobile.com/2013/05/30/samsung-proces...Brian Klug - Monday, July 1, 2013 - link
Absolutely, that's coming in the Part 2 SGS4 review (where we look at Exynos 5410).-Brian
aravindbappanadu - Monday, July 15, 2013 - link
Hey! Could you tell everyone when the next podcast will be out? the wait is killing me!!phillyry - Sunday, June 30, 2013 - link
Yeah!Montago - Saturday, July 13, 2013 - link
+1smartypnt4 - Friday, June 28, 2013 - link
So... This is actually podcast 20, which you posted before. Sorry to see you guys are having issues today, between this and the images not loading in the Richland mobile article this morning.smartypnt4 - Friday, June 28, 2013 - link
Fixed now. Disregard.Crono - Friday, June 28, 2013 - link
Excellent, was looking forward to this (what can I say, I like to listen to some interesting stuff when I'm driving). I want to know how soon before I can get the GPE ROM on my AT&T HTC One. It's already as smooth as butter with Sense 5, but would love to play around with almost-stock Android just to see how it runs.Crono - Friday, June 28, 2013 - link
Also, I'm surprised you guys don't offer a smaller MP3 size at least as a secondary option. 32 kbps (CBR preferably) is enough for speech, unless there is some background noises you want us to pick up on, ha ha. 185MB is overkill for a podcast, unless it was a music podcast.Zink - Saturday, June 29, 2013 - link
+1 podcasts are half silence and speech doesn't need nearly the bitrate used for music. Less bits please.speconomist - Saturday, June 29, 2013 - link
I completely agree, this is waaay overkill.Half or even one third of the current bitrate would make more sense without compromising audio quality at all.
Terminator02 - Saturday, June 29, 2013 - link
This. I've been putting off commenting on the bitrate of your podcasts, but it is quite unnecessary to use a 204 kbps bitrate on a podcast as you currently are for the MP3 versions. I'd say 128 kbps is the absolute max needed for (non-music) podcasts, with 96 kbps being ideal in my opinion, and 80 or 64 kbps sounding fine as well.I would prefer that you not go any lower than that as it starts to sound significantly worse when you go below 64 kbps.
This would not only decrease the strain on your servers, but allow us to start listening to the podcast sooner (for those of us who download it to listen on-the-go), as well as filling up less space in portable devices and using less bandwidth.
Crono - Saturday, June 29, 2013 - link
64kbps is a good bitrate. I think 48kbps AAC is roughly equivalent.Eggrenade - Saturday, June 29, 2013 - link
I actually really appreciate the high bitrate. They use quality microphones and the podcast sounds really good. 200 kbs is perhaps a little more than necessary, but it would be a shame to add that gargley sound to such a nice recording. I'm certainly not listening to the quality of their voices more than what they voice, but I appreciate the quality nonetheless. I would prefer variable bitrate (constant quality) just to save space, but I'm not to picky about that.Great podcast, as usual.
Terminator02 - Sunday, June 30, 2013 - link
It already is variable bitrate, the bitrate I posted was the average.I don't really personally care if AnandTech goes variable or constant bitrate, as it doesn't affect me, but I've been told that constant is more compatible with older devices.
My recommendation would apply to the max bitrate if it's variable..
Brian Klug - Sunday, June 30, 2013 - link
This is something we've been debating for a while, and even though I'm all for the really high quality audio I agree for mobile use it makes sense to have a lower bitrate. I'm pretty sure that's coming :)-Brian
Terminator02 - Sunday, June 30, 2013 - link
Even your M4A feed has an average bitrate of lower than 128 kbps and is close to 100MB vs the MP3's 185MB with an average of over 200 kbps. Obviously there is the codec difference, but I honestly cannot hear any difference between the two, and probably wouldn't even if you went even lower.The filesize is especially noticeable to me because the bandwidth when downloading your podcasts seems to be bottlenecked on your end. When I was downloading it near release I think I was getting ~1 Mbps :(
I don't mean to disrespect you guys, I just wish for a smoother and more convenient experience when downloading my favorite podcast.
Khato - Friday, June 28, 2013 - link
Entertaining as always. Especially looking forward to that Haswell MBA vs iPad4 efficiency article!One other point that caught my attention - Intel's 100mV advantage on 22nm is on the threshold voltage. Which is actually a surprising amount when you look at the plot they provided with the release materials and see that they're talking about a ~250mV threshold voltage for 22nm versus 350mV on 32nm. But it's also not really applicable to actual products as near threshold computing is still in the research phase for the most part. However there is a very interesting hint regarding likely operating voltages for Silvermont from the information Intel released on their 22nm SoC process last December. Specifically, at the same leakage levels the 22nm SoC drive currents at 0.75V are roughly 2/3 those of the 32nm process at 1.0V. (I also tend to believe the fact that they provide figures at 0.75V instead of the more standard 1.0V just might be implying something.) While I somewhat doubt that Silvermont will run full turbo frequency at 0.75V, that may well be what it needs to run in its sweet spot, and that kind of voltage reduction would certainly explain the marked jump in efficiency.
crypticsaga - Saturday, June 29, 2013 - link
The comparison between ARM's and Intel's strategies with regards to the number of their architectures is incredibly insightful. I am curious though how you view this dynamic changing in the long term. While its true that as processor tech continues to advance the middle ground will become increasingly competitive, I would venture that Intel's profit model precludes them from ever truly competing with ARM on the low end. Would you disagree?Krysto - Sunday, June 30, 2013 - link
Nope. Intel will never, EVER, be "competitive" with Core chips against ARM. The dual core chip alone *starts* at $350 for crying out loud. Their only chance is Atom, but last I checked, even that is at least 2x more expensive than the ARM chips in the same category. There's little reason why any Android/Chrome OS OEM, at least, would pick them over the ARM chips.Plus, Intel still stands no chance in the mobile GPU race with Atom, Imagination's Series 6 and Nvidia's Kepler will obliterate whatever Intel comes out with in the next few years.
Devfarce - Saturday, June 29, 2013 - link
Glad it's back, I was worried you had stopped making them. Keep them coming!sherlockwing - Saturday, June 29, 2013 - link
I am having trouble playing the m4a feed on Pocket Casts, Does anyone have similar issues?sherlockwing - Saturday, June 29, 2013 - link
OK, I found a fix. Seems like Pocket Casts view the m4a feed as a video podcast, once I press the "audio only button it started playing fine.SniperWulf - Saturday, June 29, 2013 - link
Woohoo! The band is back together! lolZink - Saturday, June 29, 2013 - link
This band is my favourite. I've been listening to some others but you guys are great together, are really knowledgable about the low level aspects of products, and talk about all of the topics that are most interesting instead of sticking to a structure and forcing discussion where there is none.Krysto - Sunday, June 30, 2013 - link
Google Glass needs to have HTC One's sound recording capability [1], and OIS and good low-light performance (just like any camera should today, but it's even more important for Google Glass, where you shake your head a lot more than your hand when moving).[1] https://www.youtube.com/watch?v=SaL87kEqZ4Y
Also Bryan, the battery is small because of weight, but I agree, they should try to put a bigger one in there.
Krysto - Sunday, June 30, 2013 - link
Anand, did you forget ARM chips will be 20nm next year, while Intel's Merrifield will be 22nm? You were talking as if Intel will have an advantage in process for smartphones next year. It won't.blanarahul - Sunday, June 30, 2013 - link
22 nm + FinFET + HKMG v/s 20 nmPlus, I'm not even sure if the first products based on 20 nm chips will use HKMG or not.
blanarahul - Sunday, June 30, 2013 - link
Did some research. It's:-Intel:- 22 nm + 1st gen FinFET + 3rd gen HKMG
v/s
TSMC:- 20 nm + 2nd gen HKMG
Sprigjr - Sunday, June 30, 2013 - link
almost forgot yalls voices, glad to see your back. you make setting up for work a little less tedious.KaliszAd - Sunday, June 30, 2013 - link
Brian, why don't you get a Nokia Lumia 920? It's rugged phone by default, maybe not water resistant but I left it in snow for about 20 - 30 Minutes by mistake and it didn't even turn off. The Lumia also didn't break, and I tell you, I dropped it several times on different surfaces from about face - height.I believe, Lumia EOS with the 41MP sensor will be also as good as Galaxy Camera/ Galaxy NX, because it will still be great phone and a decent camera and still quite rugged.
Maybe, I am biased a bit - Microsoft really has to do a lot of work and polish on Windows Phone
dealcorn - Monday, July 1, 2013 - link
Anand, give some thought to how successful ARM has been with n-1 foundries. Once a foundry becomes fully depreciated, it's cost to produce a SoC drops dramatically. It is a key to ARM's $10 SoC's.Until Baytrail, Intel has never had a n-1 product that could compete in the tablet space. When Clovertrail+ becomes n-1, it's cost drops dramatically and under classic microelectronic theory, Intel is supposed to use it to trash the profitability of the tablet middle market. Clovertrail+ is your missing third Intel architecture. The Street cares more about market share than profit margin on products from Intel's obsolete fabs.
entrer1 - Tuesday, July 2, 2013 - link
@ps4/Xbox 1. One thing that confuses me, if you saw e3 the Xbox 1 games looked real good in my book pertaining to graphics and frame rate. If you look at forza compared to drive club , both first party games, forza looked to be maintaining 60fps and running in 1080p where as drive club club did not look no where near as amazing as forza and also was struggling to maintain 30fps. Does Sony really have a huge real world performance advantage over Xbox 1? Or at the end of the day is it all smoke and mirrors? Not a fan boy BTW. I own both current generation condoles now. I am a huge fps multiplayer gamer though and am really liking the idea of Xbox 1 providing dedicated servers. Can not stand peer to peer connections.pdjblum - Wednesday, July 3, 2013 - link
It is odd that a website that has a large enthusiast reader base and reviews enthusiast gear believes that hardcore gamers choose to game on xbox or play station. It is odd that all Brian's friends who he calls "gamers" game on consoles. And it is odd that Anand pretty much plays Halo on xbox. Really! It is even odder that Brian would put a console on his desk and hook it up to his monitor instead of use his pc to game. Not only is the new look for advertisers rather than readers, but this is now pretty much a gadget/mobile site catering to, I guess, not enthusiasts.jeffkoe - Friday, July 5, 2013 - link
Enjoy the teckie info included in your podcasts. Thank you, host (not sure of everyone's name) for explaining acronyms and higher technical references. Brian is enjoyable, but can get a bit tedious with rants. Overall very enjoyable!pdjblum - Saturday, July 6, 2013 - link
Obviously Brian is smart and I do love his rants. However, he is obviously not smart enough to explain complicated concepts in simple terms so that he would not need to disparage people for not getting it the first time. Really, Brian, you are just a rude, arrogant, condescending punk. You are way too full of yourself. But I do love your honesty and your rants. Of course the vast majority of people are pretty stupid, but they don't need to be reminded of that by a punk such as yourself.flyingpants1 - Wednesday, July 10, 2013 - link
I agree. Brian Klug is sort of antisocial and arrogant. In addition to what you mentioned, he constantly talks over everyone else in the podcast, often mentioning his credentials or skills or important friends for no reason at all. I notice it too, but it usually doesn't bother me.On the other hand I think the "technology arrogance" is perfect for a site like Anandtech, it's extremely useful to have the point of view of someone who actually cares deeply about technology.
What really bothers me though, is when that condescending attitude is combined with ignorance. A couple podcasts back, regarding removable batteries for smartphones, it was "Uhhh! You got what you got! No removable batteries ever again! MicroSD is the worst storage medium and all MicroSD cards are constantly corrupt! Nobody uses it anyway! It costs too much to put it in!"
Really? You're the mobile reviewer on an enthusiast site, and you think people are stupid for not wanting to settle for less battery life and less storage space on a $600+ device? Lol. Removing features that directly affect the usability of the phone in order to save like 7 cents in manufacturing costs makes no sense whatsoever. Meanwhile you won't stop talking about other stuff like the One 4MP camera which nobody except you would ever notice or care about..
Chemsol - Friday, July 12, 2013 - link
"Sell me an Xbox license." Brilliant! For years I've been playing computer games while my Playstation sits lonely in the basement dedicated for HT use. I talk to one group of friends who share the computer interest\hardware and ignore the other group who are gaming on Xbox together. I'm tired of buying computer hardware and 2 different consoles to share the experience with all my friends. One set of hardware, virtualized console OS's and independent DLC\community environments FTW...rpsweb - Sunday, May 11, 2014 - link
good grief ! I had no idea you did a podcast, how gratifying,I have feel this this will be a great podcast.