There's no way in hell SMT gives you an 80% speedup; on the P4's it gives 10-15% at best. Thankfully, it comes at little-to-no-cost in terms of silicon, IIRC. SMT is nowhere near actually having two physical processors, it just utilises the power of the P4 more efficiently by just filling in the gaps in the execution stages. Some apps actually show a performance decrease under SMT.
Dual cores won't give you 200% either, even with the Opteron arch it'll still be 190% at the highest.
Hehe, I do lots of video encoding too...! Those temperatures I quoted were taken at high load.
Granted, I've got a heatsink the size of Delaware (Alpha 8045) sitting on top of my chip along with a not-underpowered 80mm fan, but even so your temps are way outta whack. If the CPU really was at 95*C you'd burn your finger if you touched the heatsink (which would be at about 60*C). The Barton cores aren't hot chips (the Athlon T-Bird and Palomino cores *were* hot chips - the two 2000MP's in one office server run about 15*C higher than the 2800MP Bartons); something has gone wrong somewhere!
D'oh, preview! I mistyped. I should have said 17 stage FPU.
Just to stop this post being totally worthless, the P4 Northwood had a 20 stage pipeline, and Prescott has a 30 stage pipeline, although google doesn't give away if that's int or float, or both, and I'm too tired to figure out the rest of it:)
I've just googled the info, apparently the Opteron has a 12 stage integer pipeline and a 15 stage FPU pipeline, half the length of the P4's. I'm assuming the 1P AMD64 chips are pretty similar.
And don't worry, I too think AMD will benefit from introducing SMT, it's just that the processing loss without it isn't as catastrophic as it is on the P4. I'd rather they go all guns on the dual core thing too before they tackle something as "soooooooo 2002" as HyperThreading;)
Firstly, parent is a troll... but then you knew that;)
Secondly, I agree that some of VIA's chipsets for the Athlon range were rubbish. The KT800 series for the AMD64's however seem to be prtty solid performers and are giving nVidia's nForce3 a good run for it's money.
It's nice to see good chipset support for AMD platforms, as this has been one of their major failings (although not really their fault) - I just hope this extends into the server/workstationb arena, where the Opteron platform *really* shines. They've always launched kick-ass chipsets at the time of a chips launch (AMD7xxx with the Athlon MP, AMD81xx with the Opteron) with very open specs - then everyone forgets the workstation platform, and the AMD server chips get left behind because of their poorly performing old chipsets. I remember how everyone raved about the Athlon MP kicking Xeon's ass (well, dollar for dollar anyway) when it came out, but in languished with the 7xx chipset for the rest of it's life.
Although now that the CPU's themselves take care of the memory, chipset design has become alot simpler I imagine. I just hope nVidia and others get offof their asses and cook me up some sweet 2P and 4P goodness!
Definitely a problem with your cooling setup there! Either that or your motherboard temp sensors are out of whack. My 2800 never goes above 50*C (temperature in my room here in the UK has been about 35 thanks to all my rackmounts). When all my fans are maxxed out, it does the 15*C above room temp on the CPU (on low it's more like 20-25*C above).
There's a decent little chart here http://www.cybercpu.net/howto/other/amdpr.asp that shows the power output of the Athlon chips; the 2500 is quite low down at ~55W, and it's only rated to live up until 85*C.
But like you say... wow! I'm of the opinion that AMD have totally pulled the rug out from Intel with the AMD64 line (HyperTransport, embedded memory controller), and Intel has yet to retaliate.
HT (Simultaneous Multi-Threading) won't benefit AMD's chips as much as it does Intel's because of the way they are constructed.
Due to the P4's incredibly long pipeline (30-odd stages?) and very high clockspeed, if the branch prediction goes wrong, the chip will stall
HyperThreading is a clever hack that runs two simultaneous threads on the same die. In this way, if one thread stalls, the other can execute in it's place while the other thread waits for the pipeline to redo itself, hence being a very clever way of making up for the design "faults". AMD's typically run at a lower clockspeed, and have a much shorter pipeline, so even when their piplines stall, the chip does not waste as many cycles - in short, they;re not really designed to take advantage of SMT. Hence AMD not having SMT support is a bit of a non-issue.
(Disclaimer: I'm not much of a buff on chip architecture, this is just stuff I've picked up from reading/.)
Just be warned: nVidia's IDE drivers are still a little on the dogy side. Unless you really need the speed, stick to the Microsoft/kernel supplied drivers.
The DrayTek only polls dyndns when it is issued with a new IP address and when the connection goes and comes back. Typically, I last for about a month before I get issued with a new IP, and my dyndns acount shows that my IP gets polled... about once a month;)
I didn't want to bother installing a client on my server(s) because a) I don't want to run any more services than are neccesary and b) the DrayTek seems to handle it perfectly. And DrayTek are most definitely aimed at businesses, not consumers.
I like my binary early-P3 flavoured. Who wants 1's and 0's when you can have 0.999999999993 and -0.000000000000278? It's like a tequila slammer and a lemon meringue pie all rolled into one, baby.
I don't know about the Linksys, but all of the DrayTek DSL routers I've used have an inbuilt dyndns client (they also have clients for no-ip and all the other popular DNS providers). It grabs your dynamic IP address as soon as a connection is made, and pings it up to your DynDNS account. Works like a charm.
I wouldn't be surprised if this is included in pretty much every router these days.
However, you also have the option of creating it under Linux using iRipDB http://www.marevalo.net/iRipDB/, since I believe iRiver made the database spec open.
...trying to install Windows onto a RAID card (or indeed load any other drivers for exotic hardware at install time) without a floppy drive.
Cripes, even Debian Woody install media has the ability to load drivers from floppy, CDROM, USB, NFS, FTP and HTTP. The situation with windows has gotten to the stage where I need to make slipstream install media all the time as I, unlike Microsoft, am trying to obsolete the 3.5" floppy into a belated grave.
Now if only BIOS manafacturers and bootdisk.com would make it a tad easier for us to make bootable DOS drives out fo CDR's and USB keys.
I may be a little behind the times with this since it's eben a while since I've used an nForce2 under Linux, but last time I checked nVidia's Linux drivers for the SoundStorm APU didn't include those for a hardware mixer; hence mixing of multiple audio streams must be done via a software mixer like arts.
If nVidia can't be arsed to release the specs to the ALSA/OSS guys *or* produce a half decent driver themselves, then I'm gonna stick to my Audigy cards for the time being.
Disclaimer: I'm usually a huge fan of nVidia's drivers (having had nothing but good experiences with them), and can understand their binary-only-ness. But it's not like soundcards are a hotly contested area of computing at the moment, and it strikes me that nVidia is shooting themselves in the foot a bit with this rather annoying issue. Under Windows the SoundStorm is a helluva sound chip.
I agree completely, but once again I feel the companies will be hampered with the "but if we release the specs then company XYZ will start producing amazing graphics cards!" line on things. There's also the worry that, with full access to the specs, people will work around the "crippling" of cores that is supposed to mark the difference between a £100 card and a £400 card. Given the performance war that's been going on between nVidia and ATI since the year dot, I think the chances of either side relenting are slim.
So as a whole the problem is probably part IP, part marketing/management, like the AC in the post above mentioned.
*launches into "why can't we all just get along?" caterwauling;)*
There's too much proprietary licensed code in these drivers for them ever to be open sourced. ATI and nVidia don't have ownership of alot of the code. At least nVidia did the decent thing and GPL'd their "glue" code which they do have control over (maybe ATI have too, but I'm not familiar with their drivers).
To be frank, I'm just glad that these companies are supporting Linux at all, although I don't think we'll see a major change in the status quo until Linux CAD workstations become more popular, in which case very high quality drivers will be mandated.
Hopefully you don't spend two hours a day standing up in a sardine can at the bottom of a deep well with your face pressed into someone's less than aromatic armpit;)
Well, for one you're probably right - I doubt London is gong to get as hot as New Orleans.
This summer, midday temperatue in the south east has averaged out at about 25 degrees C (77F) or thereabouts. Since London is so utterly huge and full of buildings, London is always a few degrees hotter.
The majot problem is that the tube (especially the deep lines) have no air conditioning, and exceptionally poor ventilation. Temperatures can reach into the 30's with high humidity from other peoples sweat, and the air is entirely stagnant and unbreathable.
Add to that you're cramming 100 people into a carriage so that no-one can move, and the tube carriages quickly become torture chambers. Heatstroke and exhaustion are common. In the winter, you just have to contend with overcrowding and all the rest of it, it's just not quite as hot.
We may not have the numbers, but I bet even hardened Florida alligators wouldn't last long on the Victoria or Northern lines during rush hour:)
Sorry, kinda OT, but I just thought some of you might like to know about the eccentric Jeremy Bentham. I was an earth sciences student at UCL, and every time you enter the South Wing you see JB's mummified corpse sitting in it's chair in it's little display case.
Some trivia: The head displayed on this body is not JB's real head. This is because students from Kings College (long time rivals of UCL - Kings was an anglican uni IIRC, and UCL was full of godless heathens) kept stealing the head and playing footbal with it
JB entrusted the university to the board on the sole condition that he attended every important meeting. Hence it is the university's obligation to wheel him into every meeting the university has.
Slashdot Mirror
Everyone who works with computers knows that the secret in getting them to work is not to let the magic smoke get out! ;)
It doesn't have iPod[TM] in the product name, if you use proper grammar...
;)
It's proper name is an Extremei Pod. Simple how a little judicious spacing can save you a lawsuit
There's no way in hell SMT gives you an 80% speedup; on the P4's it gives 10-15% at best. Thankfully, it comes at little-to-no-cost in terms of silicon, IIRC. SMT is nowhere near actually having two physical processors, it just utilises the power of the P4 more efficiently by just filling in the gaps in the execution stages. Some apps actually show a performance decrease under SMT.
Dual cores won't give you 200% either, even with the Opteron arch it'll still be 190% at the highest.
Hehe, I do lots of video encoding too...! Those temperatures I quoted were taken at high load.
Granted, I've got a heatsink the size of Delaware (Alpha 8045) sitting on top of my chip along with a not-underpowered 80mm fan, but even so your temps are way outta whack. If the CPU really was at 95*C you'd burn your finger if you touched the heatsink (which would be at about 60*C). The Barton cores aren't hot chips (the Athlon T-Bird and Palomino cores *were* hot chips - the two 2000MP's in one office server run about 15*C higher than the 2800MP Bartons); something has gone wrong somewhere!
D'oh, preview! I mistyped. I should have said 17 stage FPU.
:)
Just to stop this post being totally worthless, the P4 Northwood had a 20 stage pipeline, and Prescott has a 30 stage pipeline, although google doesn't give away if that's int or float, or both, and I'm too tired to figure out the rest of it
I've just googled the info, apparently the Opteron has a 12 stage integer pipeline and a 15 stage FPU pipeline, half the length of the P4's. I'm assuming the 1P AMD64 chips are pretty similar.
;)
And don't worry, I too think AMD will benefit from introducing SMT, it's just that the processing loss without it isn't as catastrophic as it is on the P4. I'd rather they go all guns on the dual core thing too before they tackle something as "soooooooo 2002" as HyperThreading
Firstly, parent is a troll... but then you knew that ;)
Secondly, I agree that some of VIA's chipsets for the Athlon range were rubbish. The KT800 series for the AMD64's however seem to be prtty solid performers and are giving nVidia's nForce3 a good run for it's money.
It's nice to see good chipset support for AMD platforms, as this has been one of their major failings (although not really their fault) - I just hope this extends into the server/workstationb arena, where the Opteron platform *really* shines. They've always launched kick-ass chipsets at the time of a chips launch (AMD7xxx with the Athlon MP, AMD81xx with the Opteron) with very open specs - then everyone forgets the workstation platform, and the AMD server chips get left behind because of their poorly performing old chipsets. I remember how everyone raved about the Athlon MP kicking Xeon's ass (well, dollar for dollar anyway) when it came out, but in languished with the 7xx chipset for the rest of it's life.
Although now that the CPU's themselves take care of the memory, chipset design has become alot simpler I imagine. I just hope nVidia and others get offof their asses and cook me up some sweet 2P and 4P goodness!
Definitely a problem with your cooling setup there! Either that or your motherboard temp sensors are out of whack. My 2800 never goes above 50*C (temperature in my room here in the UK has been about 35 thanks to all my rackmounts). When all my fans are maxxed out, it does the 15*C above room temp on the CPU (on low it's more like 20-25*C above).
There's a decent little chart here http://www.cybercpu.net/howto/other/amdpr.asp that shows the power output of the Athlon chips; the 2500 is quite low down at ~55W, and it's only rated to live up until 85*C.
The FX-53 is actually clocked at 2.4GHz
But like you say... wow! I'm of the opinion that AMD have totally pulled the rug out from Intel with the AMD64 line (HyperTransport, embedded memory controller), and Intel has yet to retaliate.
HT (Simultaneous Multi-Threading) won't benefit AMD's chips as much as it does Intel's because of the way they are constructed.
/.)
Due to the P4's incredibly long pipeline (30-odd stages?) and very high clockspeed, if the branch prediction goes wrong, the chip will stall
HyperThreading is a clever hack that runs two simultaneous threads on the same die. In this way, if one thread stalls, the other can execute in it's place while the other thread waits for the pipeline to redo itself, hence being a very clever way of making up for the design "faults". AMD's typically run at a lower clockspeed, and have a much shorter pipeline, so even when their piplines stall, the chip does not waste as many cycles - in short, they;re not really designed to take advantage of SMT. Hence AMD not having SMT support is a bit of a non-issue.
(Disclaimer: I'm not much of a buff on chip architecture, this is just stuff I've picked up from reading
Agreed, nForce2's are the dogs danglies chipsets.
Just be warned: nVidia's IDE drivers are still a little on the dogy side. Unless you really need the speed, stick to the Microsoft/kernel supplied drivers.
The DrayTek only polls dyndns when it is issued with a new IP address and when the connection goes and comes back. Typically, I last for about a month before I get issued with a new IP, and my dyndns acount shows that my IP gets polled... about once a month ;)
I didn't want to bother installing a client on my server(s) because a) I don't want to run any more services than are neccesary and b) the DrayTek seems to handle it perfectly. And DrayTek are most definitely aimed at businesses, not consumers.
I like my binary early-P3 flavoured. Who wants 1's and 0's when you can have 0.999999999993 and -0.000000000000278? It's like a tequila slammer and a lemon meringue pie all rolled into one, baby.
(Yes yes, bad/old joke. I'll get me coat)
I don't know about the Linksys, but all of the DrayTek DSL routers I've used have an inbuilt dyndns client (they also have clients for no-ip and all the other popular DNS providers). It grabs your dynamic IP address as soon as a connection is made, and pings it up to your DynDNS account. Works like a charm.
I wouldn't be surprised if this is included in pretty much every router these days.
Indeed, the database for the iRiver is optional.
However, you also have the option of creating it under Linux using iRipDB http://www.marevalo.net/iRipDB/, since I believe iRiver made the database spec open.
This is actually a very good security measure. No-one is going to attempt to sniff the network after *that*
...trying to install Windows onto a RAID card (or indeed load any other drivers for exotic hardware at install time) without a floppy drive.
Cripes, even Debian Woody install media has the ability to load drivers from floppy, CDROM, USB, NFS, FTP and HTTP. The situation with windows has gotten to the stage where I need to make slipstream install media all the time as I, unlike Microsoft, am trying to obsolete the 3.5" floppy into a belated grave.
Now if only BIOS manafacturers and bootdisk.com would make it a tad easier for us to make bootable DOS drives out fo CDR's and USB keys.
I may be a little behind the times with this since it's eben a while since I've used an nForce2 under Linux, but last time I checked nVidia's Linux drivers for the SoundStorm APU didn't include those for a hardware mixer; hence mixing of multiple audio streams must be done via a software mixer like arts.
If nVidia can't be arsed to release the specs to the ALSA/OSS guys *or* produce a half decent driver themselves, then I'm gonna stick to my Audigy cards for the time being.
Disclaimer: I'm usually a huge fan of nVidia's drivers (having had nothing but good experiences with them), and can understand their binary-only-ness. But it's not like soundcards are a hotly contested area of computing at the moment, and it strikes me that nVidia is shooting themselves in the foot a bit with this rather annoying issue. Under Windows the SoundStorm is a helluva sound chip.
MPC handles playlists just fine.
View > Playlist
or
Ctrl + 7
I've got a pretty much identical setup under windows myself. Winamp 2.91 does all the audio and music, MPC does all the video, ripping by CDex.
I agree completely, but once again I feel the companies will be hampered with the "but if we release the specs then company XYZ will start producing amazing graphics cards!" line on things. There's also the worry that, with full access to the specs, people will work around the "crippling" of cores that is supposed to mark the difference between a £100 card and a £400 card. Given the performance war that's been going on between nVidia and ATI since the year dot, I think the chances of either side relenting are slim.
;)*
So as a whole the problem is probably part IP, part marketing/management, like the AC in the post above mentioned.
*launches into "why can't we all just get along?" caterwauling
There's too much proprietary licensed code in these drivers for them ever to be open sourced. ATI and nVidia don't have ownership of alot of the code. At least nVidia did the decent thing and GPL'd their "glue" code which they do have control over (maybe ATI have too, but I'm not familiar with their drivers).
To be frank, I'm just glad that these companies are supporting Linux at all, although I don't think we'll see a major change in the status quo until Linux CAD workstations become more popular, in which case very high quality drivers will be mandated.
Yowza, that's too hot for my liking.
;)
Hopefully you don't spend two hours a day standing up in a sardine can at the bottom of a deep well with your face pressed into someone's less than aromatic armpit
Well, for one you're probably right - I doubt London is gong to get as hot as New Orleans.
:)
This summer, midday temperatue in the south east has averaged out at about 25 degrees C (77F) or thereabouts. Since London is so utterly huge and full of buildings, London is always a few degrees hotter.
The majot problem is that the tube (especially the deep lines) have no air conditioning, and exceptionally poor ventilation. Temperatures can reach into the 30's with high humidity from other peoples sweat, and the air is entirely stagnant and unbreathable.
Add to that you're cramming 100 people into a carriage so that no-one can move, and the tube carriages quickly become torture chambers. Heatstroke and exhaustion are common. In the winter, you just have to contend with overcrowding and all the rest of it, it's just not quite as hot.
We may not have the numbers, but I bet even hardened Florida alligators wouldn't last long on the Victoria or Northern lines during rush hour
I believe Winston refers to the island previously known as Britain as "Airstrip One".
Sorry, kinda OT, but I just thought some of you might like to know about the eccentric Jeremy Bentham. I was an earth sciences student at UCL, and every time you enter the South Wing you see JB's mummified corpse sitting in it's chair in it's little display case.
Some trivia:
The head displayed on this body is not JB's real head. This is because students from Kings College (long time rivals of UCL - Kings was an anglican uni IIRC, and UCL was full of godless heathens) kept stealing the head and playing footbal with it
JB entrusted the university to the board on the sole condition that he attended every important meeting. Hence it is the university's obligation to wheel him into every meeting the university has.
Lickle mini-bio here http://www.ucl.ac.uk/Bentham-Project/info/jb.htm