No, they are not. We have 30-40 of these we use for developer testing, servers, etc.
They are 2-4 times faster on IDE access then any 32bit P3-4 system including Xeons. Rest of the IO is also quite good (around 2 times better then comparable P3). As a result they make very good small servers.
CPU performance is nothing to shout about, but hardly slow. It is similar to PIII at the same speed. Possibly 10-20% slower, but not more. Actually it depends on what are you doing because they have smaller cache then PIII (only 64k).
Thermals are phenomenal. A C3 eats 1-5W where P3 eats 70+.
They are rumoured to be extremely sluggish for a completely unrelated reason. The early EPIA (as well as some current non-Via system) motherboards shipped with a Cyberblade on board. It has shared memory. So when a geek takes it his first reaction is to pump up the video frequency and resolution as high as the system can bare. As a result the video is accessing the memory at 150MHz pixel clock. That into a considerable portion of the memory bandwidth. In fact the slowing down between 60 and 90Hz vertical sync is clearly visible. This is no longer the case with newer motherboards which have a fairly decent 2D video with its own memory.
Overall it depends what you use it for. If you want a silent low maintenance server or test box. It is perfect. If you want a silent typewriter/mail desktop it will do the job. If you want to play doom3 you are got to be kidding.
Here went my moderations and I just started having fun.
Anyway, tabbed menus are patented by Adobe. Tough luck in trying to use them in an image editing program as quite a few people have understood (unless you have a good portfolio for crosslicensing)
In other words: the only CPUs that are still competing against the PPC have become what they are after using elements and whole sections from the alpha design. If we add to that that alpha was supposed to be the first CPU with SMT the picture becomes perfectly clear.
Not entirely unexpected after IBM wiped the floor clean with a 3 times increase in the TPC benchmark. This is something HP cannot even dream to match for a year or two with the current Itanic designs.
So much for the idea of killing alpha and HP's own risc processors and betting the ship on Itanic. If that sore cost cutting looser did not kill alpha 3 years ago it may have been able to compete with IBM now while Itanic never had the chance.
All I can say - it is nice that reason finally triumphed over marketing and believing own's PR, but it is sad that so much talent and people's time has been wasted for nothing.
It is fairly obvious that you have not managed a deployment of this size (or have done it without any extra project load) as anyone who have tried both approaches can tell you.
1. With your approach mounts are permanent, not ephemeral. As a result in order to propagate a trivial change like moving an export (share) from one server to another you need to hammer out a script and hit every single server with it. In a 200+ workstation deployment you have to do this at least several times a week.
2. With your approach moving people's home and personal shares around will once again require scripting onto every single system in the office.
3. With your approach the systems are not interchangeable while with a LDAP or NIS backend they can be made 100% so. As a result if a desktop goes down or is due for replacement (this happens every week or so in a deployment of this size) you can simply throw the first machine you can get your hands onto at the user desk and it just works because all of his personal files are on the network and the config is in the LDAP/NIS backend. With your approach you end up having to hand craft the machine in question.
4. With your approach you end up having to backup considerably more then necessary because you have to back up a lot of redundant information instead of one LDAP or NIS store.
5. With your approach you end up shit creek without a paddle the moment you have to do specials and customizations. LDAP on well supported systems (OSX, Debian, Suse) and NIS where supported have well defined means of merging local and network config (the + notation)
6. Basically, it depends how do you value your time. From your post I think you do not and you are happy to spend most of the day scripting stuff that can be pushed out via a single configuration file change. I simply do not have that time so I will stick with NIS and LDAP and systems which support them well. BSD does not. Just ask anyone who have tried to combine passwd.byname masterpasswd.byname and shadow.byname on one network. Or try it yourself.
Or migrating very big speadsheets. In my experience no matter how much I hate excell it can open file with millions of entries. Same for gnumeric. OO will also open it, but it will be unusable starting from a few tens of thousands.
I know that data of this size should not be kept as a spreadsheet. I do not keep it myself. A lot of people do though. In fact this is likely to be a bigger problem in SMBs then in large corporates which have databases for such things.
You are misunderstanding the OSX momentum as far as the enterprise is concerned.
OSX momentum in enterprise desktop to any extent will be due to zeroconf and possibility to command mounts via LDAP. If you have 200+ desktops you must have means of propagating unified network filesystem view to them otherwise your support will go mad.
To this respect the more common breeds of BSD are seriously behind Debian and Suse. I do not count RedHat as suitable, because using automounter on it used to lead to oops and halts as far as 9.0 and they did not give a flying f about this.
Well... Steve Linford blocked it once. So guess what they did - they closed down on spammers and opened up business for SPAMvertized websites. It is the Chinese approach when dealing with authority (and especially westerners). Smile, say yes and screw you the next second.
1. They give you 25 years warranty that it will produce some power. Not 25 years that it will produce at the same level as shipped. Read their statement again.
2. I said 10% on the roof. That assumes everything including the usual dust and grime deposits you have on the roof. I am fully aware of the 35% number. But I do not see Joe Average washing his panels every morning to get there. I am not washing them either.
In colder climate you put them spaced at as much as necessary from the roof to keep them free from snow. That is the idea of this type of panel - they are capable of heating the antifreeze while remaining cold on the outside. It is normal to get 50+ C from them on a sunny winter day when it is under -15 outside.
Sorry, I have been unclear.
Solar panels deteriorate.
The deterioration is as bad as 90% loss in 10 years for the cheap ones (or 10% of what you have started as in my post). You are pointing me to a graph of efficiency on manufacture date. Now take that graph and drop all numbers down to 10% of them which is where they will be after 10 years on your roof in the California sunshine. Looks feasible? Fsck No.
Semiconductor Solar panels are the Wrong Idea (TM). Expensive, limited life (10% efficiency in 10 years for most designs), pollution (materials and processes used are same as in the semiconductor industry which is not the cleanest thing on earth, producing consumes a lot of energy and water as well.
If you want to get extra efficiency into your house you are best off with evacuated tube solar collector running with antifreeze like this one http://www.apricus-solar.com/index.htmand a heat exchanger. If you hook them up to your heating you save considerable on bills even as far north as Sweden or high up the alps (dunno about Iowa, should be OK there as well).
Generation of energy commercially is not that different. I simply do not see how are you going to break even if you have to pay for the regeneration and disposal of old semiconductor panels. So once again a system with controlled mirrors and a boiler driving a conventional steam turbine is likely to give you much better cost performance if you can find a good place for it. It is not as easy as it seems because you have to have both sun, loads of cheap open space and water. There are not that many places that fit this description.
Microwaving the energy down is the most idiotic sci fi joke ever made. At the energies in question the air in the downlink path will get ionised right away. As a result you will have huge losses as well as power dissipation and pollution (Ozone and nitrous oxides).
That is an entire wind farm, not a turbine mind ya. If we had even half as many windfarms as we have cats we would have had enough energy to launch an expedition to Alpha Centauri every year or so.
1. There is a lobby at least in some places. The RSPSB in the UK has recently started pestering window manufacturers as well as passing around leaflets asking people to put pictures of hawks and such on their windows to decrease the number of blackbirds who die hitting a window.
2. The birds killed by wind turbines are usually birds of prey which are endangered and breed slowly, not the birds that hit your window.
3. Funnily enough in the last 5 years I have not seen a single bird hit any of my windows. Not even talking about dieing from it. At the same time I have seen at least 10 cases where pigeons have died after getting completely shitfaced eating fermented berries and apples followed by flying into a tree or a wall at top speed. In fact most city birds have learned to avoid windows as a matter of natural selection. The ones that still hit them are likely to be way above whatever is the safe alcohol limit for their species as a result of eating fermented berries or fruit.
If you have no relay and are processing your mail directly and bouncing within the SMTP transaction the spammer gets the bounce because they are sending it. Most spambots send collated reports about unreachable addresses (note, not permission denied or anything else) back to their owner. So if you have a:
domain
in control of the mail server for the domain
can fake a user unknown or any other well known error code instead of relaying denied
You are lucky.
If you have a relay which has passed the email and you are dealing with the aftermath, you bounce at the forged from. That is the boat I am in and I have been considering all kinds of solutions to allow me to circumvent my ISP port 25 filter. Quite annoying really. Similarly, if the spammer is using an open relay instead of a SPAMbot you always get the crap and the forged from always gets the bounces.
By the way - I can confirm that the approach this guy used works. I have noted significant drops in SPAM levels after outages or my mailbox going over quota before.
And this is sufficient for all the vindictive idiots in Sun to go mad. If you mention IBM anywhere near a Sun employee which has been with Sun from the beginning you better hide. They completely lose their sense of reasoning and measure. I have observed it many times, including public appearances, Usenix appearances, etc
Also, I do not want to nitpick, but every time I hear the words Churchill and Gentleman in one sentence I remember the following as well:
1. Churchill new about Coventry in WW2. 2. Churchill new about PQ17 WW2. 3. Churchill new about the Turkish artillery positions at the Mare Marmaris straights in WW1. 4. Churchill personally authorised the bombing of Dresden.
The difference between Intel and AMD is that Intel has been successfull (or unfortunate depending on the point of view) to secure military and government contracts for its CPUs. Some of these contracts require at least 7 years worth of part availability for any component (some even more).
In the past Intel has been successful in moving the technology for its old CPUs to licensees and relieving itself from the burden of maintaining manufacturing facilities. For example the 80286 lifetime during the last years of the contracts was fulfilled by Harris which managed to convince the military that their parts are acceptable replacement despite them using a different semiconductor technology.
There are no full licensees for anything after i486 this is no longer the case and Intel has to ship all of the CPUs themselves. And methinks that with all the developments in CPUS even the circa 2K$ which people like US Gov pay for a Pentium 2 keeping the facilities makes it not worthwhile.
Do not be so cocky. Adobe holds a number of key patents in the field and open source is not going to be anywhere close to Adobe's products until they expire. Same for non-Adobe close source for that matter. So if you want to get a professional job done, get a professional tool: real negative/slide scanner, not a transparency adapter, a reasonable spec Mac and a full Adobe Photoshop. It will put you back by 5K+ but if you are doing it for the money you should be able to pay it back reasonably quick.
Alternatively - use a professional digital camera in first place. That is cheaper as you can get away with around 2K.
They exist and they suck. I have done minor coding on traffic modelling projects for around 4 years back in the early 90-es and I can tell you that they are the wrong idea.
The problem is that if every traffic light reacts only to input from sensors they traffic tends to get into a positive feedback state. This results in the total throughput on a road decreasing instead of increasing. I have been through this calculations several hundred times and no matter what method you use the result is still the same. It sucks rotten eggs compared to having all lights on a road set to fixed sync and having a floating speed limit to accommodate for congestion.
RT is a general issue tracking system, not a bugtracking system. I have used RT for a while now and I will recommend RT for nearly any support deployment I will not recommend it for bug tracking. The main reasons are:
1. It does not have any source code control integration. It relates well to issues/problems reported by users, but it does not relate well to a source tree.
2. It is extremely versatile and open-ended. You can adapt it to nearly any task, but there is a very high likelihood that you will hang yourself in the process by chosing a defficient workflow or issue life schematic. As far as bug tracking is concerned you are likely to have better mileage with Bugzilla.
3. At least some of the ticket/issue states are hardwired in the code so for a good bugtracking process flow you will end up having to hack perl/mason and searchbuilder stuff.
Same can be said for probes in more interesting places.
If the graph is to believed there are nearly zero probes in the circular current in the Southern Hemisphere roaring sixties, there also very few probes in the other major current systems - Gulfstream, along N and S America West Coast, Azora, etc. At the same time there are plenty of probes which are sitting in relatively silent regions like 30-40 latt in the middle of the Pacific (north and south).
I hope they put the remaining 1500 into the major current systems as these are the places that determine the weather around the globe. It will be more expensive to maintain as you have to salvage them quite often and relocate to the beginning of the current, but hopefully the data collected will pay back for the excessive maintenance costs.
Seconded.
In a few more years a significant proportion of the dogs will have cancers where it has been injected.
PEG is not a strong carcinogenic substance, but carcinogenic substance nonetheless.
Slashdot Mirror
No, they are not. We have 30-40 of these we use for developer testing, servers, etc.
They are 2-4 times faster on IDE access then any 32bit P3-4 system including Xeons. Rest of the IO is also quite good (around 2 times better then comparable P3). As a result they make very good small servers.
CPU performance is nothing to shout about, but hardly slow. It is similar to PIII at the same speed. Possibly 10-20% slower, but not more. Actually it depends on what are you doing because they have smaller cache then PIII (only 64k).
Thermals are phenomenal. A C3 eats 1-5W where P3 eats 70+.
They are rumoured to be extremely sluggish for a completely unrelated reason. The early EPIA (as well as some current non-Via system) motherboards shipped with a Cyberblade on board. It has shared memory. So when a geek takes it his first reaction is to pump up the video frequency and resolution as high as the system can bare. As a result the video is accessing the memory at 150MHz pixel clock. That into a considerable portion of the memory bandwidth. In fact the slowing down between 60 and 90Hz vertical sync is clearly visible. This is no longer the case with newer motherboards which have a fairly decent 2D video with its own memory.
Overall it depends what you use it for. If you want a silent low maintenance server or test box. It is perfect. If you want a silent typewriter/mail desktop it will do the job. If you want to play doom3 you are got to be kidding.
Here went my moderations and I just started having fun.
Anyway, tabbed menus are patented by Adobe. Tough luck in trying to use them in an image editing program as quite a few people have understood (unless you have a good portfolio for crosslicensing)
In other words: the only CPUs that are still competing against the PPC have become what they are after using elements and whole sections from the alpha design. If we add to that that alpha was supposed to be the first CPU with SMT the picture becomes perfectly clear.
Not entirely unexpected after IBM wiped the floor clean with a 3 times increase in the TPC benchmark. This is something HP cannot even dream to match for a year or two with the current Itanic designs.
So much for the idea of killing alpha and HP's own risc processors and betting the ship on Itanic. If that sore cost cutting looser did not kill alpha 3 years ago it may have been able to compete with IBM now while Itanic never had the chance.
All I can say - it is nice that reason finally triumphed over marketing and believing own's PR, but it is sad that so much talent and people's time has been wasted for nothing.
It is fairly obvious that you have not managed a deployment of this size (or have done it without any extra project load) as anyone who have tried both approaches can tell you.
1. With your approach mounts are permanent, not ephemeral. As a result in order to propagate a trivial change like moving an export (share) from one server to another you need to hammer out a script and hit every single server with it. In a 200+ workstation deployment you have to do this at least several times a week.
2. With your approach moving people's home and personal shares around will once again require scripting onto every single system in the office.
3. With your approach the systems are not interchangeable while with a LDAP or NIS backend they can be made 100% so. As a result if a desktop goes down or is due for replacement (this happens every week or so in a deployment of this size) you can simply throw the first machine you can get your hands onto at the user desk and it just works because all of his personal files are on the network and the config is in the LDAP/NIS backend. With your approach you end up having to hand craft the machine in question.
4. With your approach you end up having to backup considerably more then necessary because you have to back up a lot of redundant information instead of one LDAP or NIS store.
5. With your approach you end up shit creek without a paddle the moment you have to do specials and customizations. LDAP on well supported systems (OSX, Debian, Suse) and NIS where supported have well defined means of merging local and network config (the + notation)
6. Basically, it depends how do you value your time. From your post I think you do not and you are happy to spend most of the day scripting stuff that can be pushed out via a single configuration file change. I simply do not have that time so I will stick with NIS and LDAP and systems which support them well. BSD does not. Just ask anyone who have tried to combine passwd.byname masterpasswd.byname and shadow.byname on one network. Or try it yourself.
Or migrating very big speadsheets. In my experience no matter how much I hate excell it can open file with millions of entries. Same for gnumeric. OO will also open it, but it will be unusable starting from a few tens of thousands.
I know that data of this size should not be kept as a spreadsheet. I do not keep it myself. A lot of people do though. In fact this is likely to be a bigger problem in SMBs then in large corporates which have databases for such things.
You are misunderstanding the OSX momentum as far as the enterprise is concerned.
OSX momentum in enterprise desktop to any extent will be due to zeroconf and possibility to command mounts via LDAP. If you have 200+ desktops you must have means of propagating unified network filesystem view to them otherwise your support will go mad.
To this respect the more common breeds of BSD are seriously behind Debian and Suse. I do not count RedHat as suitable, because using automounter on it used to lead to oops and halts as far as 9.0 and they did not give a flying f about this.
Can you point us to a ref please?
Well... Steve Linford blocked it once. So guess what they did - they closed down on spammers and opened up business for SPAMvertized websites. It is the Chinese approach when dealing with authority (and especially westerners). Smile, say yes and screw you the next second.
1. They give you 25 years warranty that it will produce some power. Not 25 years that it will produce at the same level as shipped. Read their statement again.
2. I said 10% on the roof. That assumes everything including the usual dust and grime deposits you have on the roof. I am fully aware of the 35% number. But I do not see Joe Average washing his panels every morning to get there. I am not washing them either.
In colder climate you put them spaced at as much as necessary from the roof to keep them free from snow. That is the idea of this type of panel - they are capable of heating the antifreeze while remaining cold on the outside. It is normal to get 50+ C from them on a sunny winter day when it is under -15 outside.
Sorry, I have been unclear. Solar panels deteriorate. The deterioration is as bad as 90% loss in 10 years for the cheap ones (or 10% of what you have started as in my post). You are pointing me to a graph of efficiency on manufacture date. Now take that graph and drop all numbers down to 10% of them which is where they will be after 10 years on your roof in the California sunshine. Looks feasible? Fsck No.
Semiconductor Solar panels are the Wrong Idea (TM). Expensive, limited life (10% efficiency in 10 years for most designs), pollution (materials and processes used are same as in the semiconductor industry which is not the cleanest thing on earth, producing consumes a lot of energy and water as well.
If you want to get extra efficiency into your house you are best off with evacuated tube solar collector running with antifreeze like this one http://www.apricus-solar.com/index.htmand a heat exchanger. If you hook them up to your heating you save considerable on bills even as far north as Sweden or high up the alps (dunno about Iowa, should be OK there as well).
Generation of energy commercially is not that different. I simply do not see how are you going to break even if you have to pay for the regeneration and disposal of old semiconductor panels. So once again a system with controlled mirrors and a boiler driving a conventional steam turbine is likely to give you much better cost performance if you can find a good place for it. It is not as easy as it seems because you have to have both sun, loads of cheap open space and water. There are not that many places that fit this description.
Microwaving the energy down is the most idiotic sci fi joke ever made. At the energies in question the air in the downlink path will get ionised right away. As a result you will have huge losses as well as power dissipation and pollution (Ozone and nitrous oxides).
That is an entire wind farm, not a turbine mind ya. If we had even half as many windfarms as we have cats we would have had enough energy to launch an expedition to Alpha Centauri every year or so.
A few minor points:
1. There is a lobby at least in some places. The RSPSB in the UK has recently started pestering window manufacturers as well as passing around leaflets asking people to put pictures of hawks and such on their windows to decrease the number of blackbirds who die hitting a window.
2. The birds killed by wind turbines are usually birds of prey which are endangered and breed slowly, not the birds that hit your window.
3. Funnily enough in the last 5 years I have not seen a single bird hit any of my windows. Not even talking about dieing from it. At the same time I have seen at least 10 cases where pigeons have died after getting completely shitfaced eating fermented berries and apples followed by flying into a tree or a wall at top speed. In fact most city birds have learned to avoid windows as a matter of natural selection. The ones that still hit them are likely to be way above whatever is the safe alcohol limit for their species as a result of eating fermented berries or fruit.
If you have no relay and are processing your mail directly and bouncing within the SMTP transaction the spammer gets the bounce because they are sending it. Most spambots send collated reports about unreachable addresses (note, not permission denied or anything else) back to their owner. So if you have a:
- domain
- in control of the mail server for the domain
- can fake a user unknown or any other well known error code instead of relaying denied
You are lucky.If you have a relay which has passed the email and you are dealing with the aftermath, you bounce at the forged from. That is the boat I am in and I have been considering all kinds of solutions to allow me to circumvent my ISP port 25 filter. Quite annoying really. Similarly, if the spammer is using an open relay instead of a SPAMbot you always get the crap and the forged from always gets the bounces.
By the way - I can confirm that the approach this guy used works. I have noted significant drops in SPAM levels after outages or my mailbox going over quota before.
And this is sufficient for all the vindictive idiots in Sun to go mad. If you mention IBM anywhere near a Sun employee which has been with Sun from the beginning you better hide. They completely lose their sense of reasoning and measure. I have observed it many times, including public appearances, Usenix appearances, etc
Also, I do not want to nitpick, but every time I hear the words Churchill and Gentleman in one sentence I remember the following as well:
1. Churchill new about Coventry in WW2.
2. Churchill new about PQ17 WW2.
3. Churchill new about the Turkish artillery positions at the Mare Marmaris straights in WW1.
4. Churchill personally authorised the bombing of Dresden.
Gentleman my arse.
The difference between Intel and AMD is that Intel has been successfull (or unfortunate depending on the point of view) to secure military and government contracts for its CPUs. Some of these contracts require at least 7 years worth of part availability for any component (some even more).
In the past Intel has been successful in moving the technology for its old CPUs to licensees and relieving itself from the burden of maintaining manufacturing facilities. For example the 80286 lifetime during the last years of the contracts was fulfilled by Harris which managed to convince the military that their parts are acceptable replacement despite them using a different semiconductor technology.
There are no full licensees for anything after i486 this is no longer the case and Intel has to ship all of the CPUs themselves. And methinks that with all the developments in CPUS even the circa 2K$ which people like US Gov pay for a Pentium 2 keeping the facilities makes it not worthwhile.
Do not be so cocky. Adobe holds a number of key patents in the field and open source is not going to be anywhere close to Adobe's products until they expire. Same for non-Adobe close source for that matter. So if you want to get a professional job done, get a professional tool: real negative/slide scanner, not a transparency adapter, a reasonable spec Mac and a full Adobe Photoshop. It will put you back by 5K+ but if you are doing it for the money you should be able to pay it back reasonably quick. Alternatively - use a professional digital camera in first place. That is cheaper as you can get away with around 2K.
The problem is that if every traffic light reacts only to input from sensors they traffic tends to get into a positive feedback state. This results in the total throughput on a road decreasing instead of increasing. I have been through this calculations several hundred times and no matter what method you use the result is still the same. It sucks rotten eggs compared to having all lights on a road set to fixed sync and having a floating speed limit to accommodate for congestion.
RT is a general issue tracking system, not a bugtracking system. I have used RT for a while now and I will recommend RT for nearly any support deployment I will not recommend it for bug tracking. The main reasons are:
1. It does not have any source code control integration. It relates well to issues/problems reported by users, but it does not relate well to a source tree.
2. It is extremely versatile and open-ended. You can adapt it to nearly any task, but there is a very high likelihood that you will hang yourself in the process by chosing a defficient workflow or issue life schematic. As far as bug tracking is concerned you are likely to have better mileage with Bugzilla.
3. At least some of the ticket/issue states are hardwired in the code so for a good bugtracking process flow you will end up having to hack perl/mason and searchbuilder stuff.
Methinks that the Sybase "one-eyed monster whacking a luser" advert was a perfect fit for this article.
:-)
Dunno if it is luck or the usual Slashdot editor impartial and unbiased reporting
Hopefully a few more whacks and it will be gone for good.
If the graph is to believed there are nearly zero probes in the circular current in the Southern Hemisphere roaring sixties, there also very few probes in the other major current systems - Gulfstream, along N and S America West Coast, Azora, etc. At the same time there are plenty of probes which are sitting in relatively silent regions like 30-40 latt in the middle of the Pacific (north and south).
I hope they put the remaining 1500 into the major current systems as these are the places that determine the weather around the globe. It will be more expensive to maintain as you have to salvage them quite often and relocate to the beginning of the current, but hopefully the data collected will pay back for the excessive maintenance costs.
Seconded. In a few more years a significant proportion of the dogs will have cancers where it has been injected. PEG is not a strong carcinogenic substance, but carcinogenic substance nonetheless.