I thought they stripped out the MP related logic (MP cache coherency protocols at least) in the pentium-M, so that does sorta count as dumbed down (for server apps anyway).
Maybe they'll glue on some of the new hublink like memory controller stuff and it will look completely like an AMD64 from an MP point of view.
I've had packet8 for 6 mos or so. Its good except for the massive latency. Its like those commercials for Nextel where everyone has to wait a turn and make short little comments while holding the push-to-talk. All the latency in the traceroute is between Chicago and SJC (I'm in Minneapolis), about 200ms worth, on their upstreams network (Level 3). Apparently they backhaul everything to SJC and back out again, even when I'm calling to the East Coast.
I still have and use the service and it is almost worth $20/month, I'm starting to itch to get something else (Qworst has a $20/month unlimited long distance plan) because I find myself avoiding calling people due to the lag. My sister uses Vonage and doesn't have this bandwidth problem, but she's in NY and I'm guessing there's a Vonage interface in NJ.
No one in packet 8 that I can get to is willing to give any real info on their architecture, based on their tech support asking everyone to 'check their ping' to SJC I suspect they really do shuffle everything through SJC, latency be damned.
I've been using Packet8 for about 6 mos, the latency is fairly bad but no worse than a bad cell phone.
I called my analog line from the Packet8 line and made some noises with a storage scope connected to a mic to time it: the one way latency calling my analog line is around 300ms (oddly enough it was closer to 200ms going the other way). This is long enough that I end up stepping on people I'm talking to and vice versa, but not so long its unmanageble. This is about the same using my cable modem (Time Warner) or DSL service.
Packet8 had me run some ping tests to San Jose, the bulk of the latency was in Level 3's network between Chicago and San Jose -- apparently Packet8 has a gateway in San Jose that everything goes through (even local calls). I got a phone number in San Jose and that helped some, I don't pay the latency back to Minnesota to get back on the POTS network (most of my calls are outgoing so this works fine).
All that said I've gotten used to the long delay, just pretend you're using a push-to-talk walkie talkie and its not so bad. The actual audio quality is excellent.
My sister uses Vonage and I never notice the delay when she calls my analog line with it.
American express did do this, they cancelled it as being too expensive mid last year.
It was great and I used it a lot. The only problem was the numbers they generated always had an expiration date of the current month and sometimes next month (if it was near the end of the month), many merchant banks rejected these as 'expired'.
I've long considered callerID as reliable as the From address in SMTP -- much of the time its useful but you sure can't trust it.
Back before the spammers really caught on it was pretty common at places I ran for mgmt (and some of my fellow engineers even) to believe there was some sacred guarantee on the From: address, its as sacred as the return address on a piece of postal mail and IMO so is caller ID.
I have one of these and am very happy with it. I don't use the mouse much but it works fine. Its small enough to sit reasonably unobtrusively in my living room (with the IR repeated via Xantech IR repeaters down to where the equipment is racked).
Only issue I've had is it sends lengthy IR codes when you use the mouse that were crashing an IRTrans infrared decoder/receiver -- the guy that makes them has a fix in the works though.
I saw these at Comdex a couple of years ago (from a taiwanese company), then recently set up a MythTV box and found a guy in the UK selling them on ebay. I looked a bit in the US but lost patience trying to find a distributor and ended up ordering from the UK guy for about $80 including shipping.
Having done development in a virtual environment for about the last 10 years IMO the most important thing is facilitating collaboration between engineers.
The first company I did this with was almost entirely virtual and we used primarily telephone and email. This is good and worked OK where the projects were small enough they could be designed and implemented by 1 or 2 persons (basically isolated development). The largest project (multithreading a legacy kernel) was 3 people and I probably spent 3-4 hours a day on the phone in some phases of it.
This pattern served well enough for the next 2 companies as well (one a startup and one a large corp), but in both cases a lot of travel was involved to keep everyone in their loops.
Its not as much the software used as the mindset that everyone has to be involved in what used to be 'hallway' talk. While you have to have some additional process other than hallway talk for a project, it is very valuable and cements a group together (if all you ever experience of your co-workers is spec and design email exchange its hard to develop a feel of how they think/work, and IMO empathy with your co-workers greases the skids significantly).
To finally get to the point: based on something I read on slashdot back in 99 or so when we did the next 'virtual' startup I pushed hard to use a broader range of tools. After 4 years of trying various mechanisms some have stuck and some have not, here's what is working really well for a smallish group of sr developers (5-10) and worked OK for a larger group (25ish) of mixed sr and jr people doing development of a 500kloc scale project involving kernel work (database and OS/networking):
IRC: this is our virtual office. The equivalent of walking to someone's cube and asking them a question happens here. We found that running structured meetings solely on IRC was not efficient, people who hate meetings would tend to do other work and not pay any attention at all.
We set up UnrealIRC as the server (with a hack to disable the throttling so people can paste blocks of code or debug output w/o getting limited to 1 line per second) inside a firewall. Everyone uses an SSH tunnel to get to it. For clients everyone uses Xchat or mIRC.
The most important trivial sounding thing about this setup is that everyone set up a trigger that watches for their name or traffic on a/query window and makes a sound. Some people set up filters to make sounds when their subsystem name is mentioned too. The key is you can say 'hey fred!' and at fred's end a noise happens. Most new employees don't see the point until a few weeks into using the system when they've missed out of good discussions regarding something they're responsible for.
a Wiki: I fought this as 'a toy' for a while, but finally came around and now I can't imagine how we worked w/o it. We tried using Frame+Visio+cvs for design documents, as well as Word + powerpoint (for drawings), also nroff+xfig. Nothing has come even close to the ease of doing collaborative design work on a Wiki.
We use TWiki: it keeps everything in RCS under the covers and lets you easily attach binary files to any page (for drawings and such). There are lots of fancy plugins.
Plain old email: nothing fancy; used mostly as a store and forward message system to indicate when someone updates something in the Wiki that needs review or when changes are submitted to source control.
Phone conference: we use a commercial service called ReadyConference, no scheduling required everyone just calls into the bridge whenever we internally need a meeting. For small conferences 3-way calling from the phone company (even two 3ways put together) is much cheaper and good quality (just a pain to set up). Keep the number of meetings low and to the point (always have an agenda) and the phone is a fast way to reach consensus, its a poor place to float new proposals , IRC is much better for sending up a balloon.
Source control: I know this shouldn't even require menti
Slashdot Mirror
I thought they stripped out the MP related logic (MP cache coherency protocols at least) in the pentium-M, so that does sorta count as dumbed down (for server apps anyway).
Maybe they'll glue on some of the new hublink like memory controller stuff and it will look completely like an AMD64 from an MP point of view.
I've had packet8 for 6 mos or so. Its good except for the massive latency. Its like those commercials for Nextel where everyone has to wait a turn and make short little comments while holding the push-to-talk. All the latency in the traceroute is between Chicago and SJC (I'm in Minneapolis), about 200ms worth, on their upstreams network (Level 3). Apparently they backhaul everything to SJC and back out again, even when I'm calling to the East Coast.
I still have and use the service and it is almost worth $20/month, I'm starting to itch to get something else (Qworst has a $20/month unlimited long distance plan) because I find myself avoiding calling people due to the lag. My sister uses Vonage and doesn't have this bandwidth problem, but she's in NY and I'm guessing there's a Vonage interface in NJ.
No one in packet 8 that I can get to is willing to give any real info on their architecture, based on their tech support asking everyone to 'check their ping' to SJC I suspect they really do shuffle everything through SJC, latency be damned.
I've been using Packet8 for about 6 mos, the latency is fairly bad but no worse than a bad cell phone.
I called my analog line from the Packet8 line and made some noises with a storage scope connected to a mic to time it: the one way latency calling my analog line is around 300ms (oddly enough it was closer to 200ms going the other way). This is long enough that I end up stepping on people I'm talking to and vice versa, but not so long its unmanageble. This is about the same using my cable modem (Time Warner) or DSL service.
Packet8 had me run some ping tests to San Jose, the bulk of the latency was in Level 3's network between Chicago and San Jose -- apparently Packet8 has a gateway in San Jose that everything goes through (even local calls). I got a phone number in San Jose and that helped some, I don't pay the latency back to Minnesota to get back on the POTS network (most of my calls are outgoing so this works fine).
All that said I've gotten used to the long delay, just pretend you're using a push-to-talk walkie talkie and its not so bad. The actual audio quality is excellent.
My sister uses Vonage and I never notice the delay when she calls my analog line with it.
American express did do this, they cancelled it as being too expensive mid last year. It was great and I used it a lot. The only problem was the numbers they generated always had an expiration date of the current month and sometimes next month (if it was near the end of the month), many merchant banks rejected these as 'expired'.
I've long considered callerID as reliable as the From address in SMTP -- much of the time its useful but you sure can't trust it. Back before the spammers really caught on it was pretty common at places I ran for mgmt (and some of my fellow engineers even) to believe there was some sacred guarantee on the From: address, its as sacred as the return address on a piece of postal mail and IMO so is caller ID.
I have one of these and am very happy with it. I don't use the mouse much but it works fine. Its small enough to sit reasonably unobtrusively in my living room (with the IR repeated via Xantech IR repeaters down to where the equipment is racked).
Only issue I've had is it sends lengthy IR codes when you use the mouse that were crashing an IRTrans infrared decoder/receiver -- the guy that makes them has a fix in the works though.
I saw these at Comdex a couple of years ago (from a taiwanese company), then recently set up a MythTV box and found a guy in the UK selling them on ebay. I looked a bit in the US but lost patience trying to find a distributor and ended up ordering from the UK guy for about $80 including shipping.
The first company I did this with was almost entirely virtual and we used primarily telephone and email. This is good and worked OK where the projects were small enough they could be designed and implemented by 1 or 2 persons (basically isolated development). The largest project (multithreading a legacy kernel) was 3 people and I probably spent 3-4 hours a day on the phone in some phases of it.
This pattern served well enough for the next 2 companies as well (one a startup and one a large corp), but in both cases a lot of travel was involved to keep everyone in their loops.
Its not as much the software used as the mindset that everyone has to be involved in what used to be 'hallway' talk. While you have to have some additional process other than hallway talk for a project, it is very valuable and cements a group together (if all you ever experience of your co-workers is spec and design email exchange its hard to develop a feel of how they think/work, and IMO empathy with your co-workers greases the skids significantly).
To finally get to the point: based on something I read on slashdot back in 99 or so when we did the next 'virtual' startup I pushed hard to use a broader range of tools. After 4 years of trying various mechanisms some have stuck and some have not, here's what is working really well for a smallish group of sr developers (5-10) and worked OK for a larger group (25ish) of mixed sr and jr people doing development of a 500kloc scale project involving kernel work (database and OS/networking):
We set up UnrealIRC as the server (with a hack to disable the throttling so people can paste blocks of code or debug output w/o getting limited to 1 line per second) inside a firewall. Everyone uses an SSH tunnel to get to it. For clients everyone uses Xchat or mIRC.
The most important trivial sounding thing about this setup is that everyone set up a trigger that watches for their name or traffic on a
We use TWiki: it keeps everything in RCS under the covers and lets you easily attach binary files to any page (for drawings and such). There are lots of fancy plugins.
Hello! This is a FreeBSD is dying post with Itanium sedded in. Mod this funny.