I have a ATI TV Wonder VE, its based on a brooktree 878, though I think its using the connexant version of the chip. The problem I have is that the while the signal I'm receiving (coax cable) looks good on a regular Television (Sony Trinitron, a few years old) the captured version is full of noise. The closest example I can find is here:
The noise I refer to is quite noticeable in the full resolution pngs. I'll note that this is no ill reflection on tvtime. Then again, reading about the specs of the BT878, its supposed to have higher fidelity than a television monitor. Hence, I'm confused where I'm going wrong.
The EM is very limited including only a Program Automatic and an Aperture Priority Automatic system. The only semblance of manual control is using Aperture Priority and adjusting the exposure compensation. He's much better off with a FM, FE, or FA series camera if he's looking at the pre N2000 series cameras. He'd probably get a better deal buying an N70.
A camera itself is a light-tight box with two important components: A Lens and a Shutter.
The most important of the two is the Lens. The Lens warps and bends the light to react on the film. Camera is a box; Lens does all the work and has far more effect on the quality of your image than the box.
This is the crucial part of the comment: SPEND MORE ON YOUR LENS THAN THE CAMERA.
We've all taken physics so we can all appreciate how fantastically complex these lenses are. Remember how the angle light is bent when passing between two different densities is depedent on its wavelength? Well, imagine having to deal with that along with half a dozen other problems in lens design.
First, evaluate what you want to shoot and then figure out what focal lengths you will need to create the images you want. Remember that focal length has many different effects on the image. For example, magazine covers are typically shot with very, very long lenses - around 400 or 500 mm - because the telephoto effect compresses facial features. Try taking a portrait with a wide-angle some time.;) Point is, choose the focal length that has the distortion you want, then zoom with your feet. Also, remember you can always crop in the dark room.
Many of the large camera companies have very devout followings who are more than happy to tell you about notable lenses. For example, for Nikon, people will often gush over the old AI-S 80f2.5 (iirc, haven't been on the nikon list in over a year.) However, other 'solid' lenses are the AF-D 50f1.8, the AF-S 80-200f2.8.
So get yourself a *nice* lens then get yourself a cheap, old, used manual camera body like an FM-2 or FE-2. (with a few exceptions) Nikon lenses are backwards compatible. Ie, I can take my AF-S ED IF 80-200f2.8 lens and use it quite well on my Nikon FE.
Also, Their flashes are backwards compatible, I can use my SB-28DX on my FE as well as my F5.
A quick rundown of Nikon Lens tags: AF = AutoFocus AI and AI-S = Automatic Indexing (manual lenses that have a tab which mates with a ring on the lens mount to report the aperture index to the camera for metering.) ED = extremely low dispersion glass (used to make different generations of lenses distinct. ELD is a specially doped glass from nikon that, like its name, has very low dispersion.) IF = Internal Focussing (the lens will not extend/contract to focus, the focussing is done by moving internal elements within the lens.) -D = The lens can report the distance it is focussed at to the camera for flash metering. -S = The lens uses a piezoelectric ring inside the lens barrel to focus instead of relying on the mechanical linkage to the motor in the camera. These lenses also have -D functionality.
Personally, I would get an AI 50f2 or an AF 50f1.8 with a Nikon FE or FE-2 to get the hang of things. Once you get feature envy, then you can get yourself an F100, SB-80, etc.
Frankly, this is taken out of context and is rather misleading. Sometimes I wish for a moderation option entitled "Wrong."
First of all, the section above is listed under the header called "Intel Ipsec Cards" and more specifically refers to the Intel Encryption Coprocessor on the card.
Further, Intel has written and released a free, GPL ethernet driver for their EEPro 100, 1000, and 10000 ethernet cards. I shall transcribe for your benefit the top few lines from linux/drivers/net/e100/e100_main.c:
Copyright(c) 1999 - 2003 Intel Corporation. All rights reserved.
This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the Free
Software Foundation; either version 2 of the License, or (at your option)
any later version.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 59
Temple Place - Suite 330, Boston, MA 02111-1307, USA.
The full GNU General Public License is included in this distribution in the
file called LICENSE.
Contact Information:
Linux NICS
Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
That driver is a GPL implementation meaning that the OpenBSD developers are more than welcome to port it at their leisure.
Ok, so the IPSEC chip on the NIC isn't supported nor is there any data on that chip forthcoming. However, there are a number of papers that show that IPSEC and TCP offloading (not to be confused with TCP fragmentation/checksumming) are not efficient. Specifically, the "hardware" IPSEC is done by firmware downloaded to a small embedded processor on the NIC. This small, embedded NIC is not very fast, in fact, its rather slow.
Result: Processor utilization drops marginally (modern processors can encrypt 10 megabytes/s trivially) Latency shoots up (It takes the embedded processor longer to encrypt a packet than the host processor would.)
There are a number of papers corroborating that latency has a huge effect on maximum bandwidth.
[I think the paper regarding TCP offloading not being worthwhile is by Mudge. The IPSEC offloading not being worthwhile is my hypothesis, untested, but I feel logically founded.]
My point is that IPSEC offloading is not an advantage - it probably was in 200MHz K6 days, but it certainly is not in 2.0 ghz K7 days.
Other notables, for example, the 3Com 3CR990 still doesn't have IPSEC offloading, despite promises from either the openbsd txp driver or the linux typhoon driver.
Frankly, that's a product of the driver not a result of chip design or flaws. Further, all Windows (NDIS?) drivers I've had to work with have had the option called "Locally Administered Address" where you can force a custom MAC address.
If you take a look at most ethernet drivers, you'll notice that most drivers need to read a MAC address out of a serial eeprom. This address is then configured into the ethernet chip so it can receive appropriate unicast frames and is used by the driver when constructing ethernet frame headers for transmission.
Further, regarding just how "junky" the Realtek stuff is, I refer you to this page: http://www.fefe.de/linuxeth/realtek.txt which states quite clearly and specifically why the realtek card is "junk."
For the record, I can spoof the MAC address on my 3Com 3C905's, Intel EEPro100 and Broadcom Tigon 3.
I've got five WD1200JB's in a raid 5 configuration; they have been operating solidly for almost a year. What problems have you noticed using these drives for RAID?
1, you can't control external IT services, external IT is hotile to you.
Fine, if you have support from your department, then treat the rest of your company's IT assets as 'hostile' and 'insecure.' Having your bosses support is crucial, it's his job as a project manager, or division head to facilitate his employees getting their work done. Further, its his job to make sure that important information and data is not compromised. He is delegating that responsibility to you, but it is still responsible. Hence, if 'hostile IT' has a fit, its his job to deal with them and to keep them off your back. This is similar to chain of command in the armed forces; your supierior officer is responsible for you and it his job to make sure you can do your job. Make sure your boss knows whats going on and do nothing behind his back. An open and honest relationship with your boss is crucial in office politics.
So, Firewall your LAN. You're worried about network based threats, so install an IDS to watch traffic that crosses your border 'your firewall'. Make sure that your windows boxes are patched; there are microsoft tools for facilitating this.
Even if you don't have support for a firewall from your IT people, you can set up a NAT machine to bridge from your old IP space to a private network (10/8, 192.168/16, etc.)
If you need to provide external access to internal resources, use a secure connection. There are systems for validating user identity (rsa tokens, smart cards.) Use them as your authentication system.
2, You don't have physical access controls and you cannot implement them.
Encrypt hard drives, I believe Windows 2000+ has the option for encrypted volumes. I would check around to proove to yourself, without a doubt, that its secure. Make sure swap partitions are encrypted, etc.
If someone has the resources and interest to physically break into your workplace, then encrypted hard drive volumes might not be enough.
Another approach is to make sure that no sensitive data is saved to the hard drive. This is difficult, as compiling or loading data may write data to swap partitions or leave other sensitive information in unguarded locations.
Consider a terminal server, and secure the terminal server. With a terminal server, you can place it behind large amounts of steel, locks, and other physical measures that make theft increasingly difficult. Remember, humans are generally the weakest key; if you are able to lock down a small part of your resources in a cage, or a reinforced closet, who has access to it?
Further, a terminal server would reduce downtime for engineers. "my power supply burned out; my hard drive died." big deal, their PC becomes essentially a stateless client.
3. Need to compartmentalize. Frankly, if you're in a microsoft world, the easiest system is probably to use CIFS with ACLs. Use Active Directory to model the "logical" break down of people in your organization. Its left up to the reader as an exercise in 'trust'. I would get two fileservers, and use fibrechannel to have redundant links to a redundant RAID array. (compaq, dell, etc. would love to sell you this.) If cost is a concern, a similar solution could be constructed. Place these servers in the secure cage with the terminal server. Give the terminal server and your file/authentication servers a private, local LAN between each other. Place a tape library in the cage as well, and automate backups. You might consider renting a machine at a colo somewhere far away and using that for off-site storage over a secure VPN.
4. "engineer hit by a bus scenario" Your fileservers will still allow the administrator to access all files, but, lock the administrator accounts to local log in only. You can only be the super user after unlocking fifteen physical locks that bolt down 100 lbs of stainless steel. However, in an emergency, you can retrieve information.
5. take files home What about allowing VPN access to the terminal se
Actually, no. The wireless chip driver supports all the broadcom line of wireless cards. They are a Software Defined Radio (SDR), and per FCC regulations, Broadcom is not allowed to divulge the programming information to the public, to prevent abuse. For more information, look at the work on the Atheros chipset drivers, they require a binary module to initialize the SDR.
Mind you, this chipset is the same as the Apple Airport Extreme's, and the Linksys WPC54G wireless cards.
That's the big deal. Also, the firmware image contains drivers for the Broadcom 10/100 chipset, too.
It's important to note that the kernel has been modified as well. Some people have suggested that they are immune to GPL violations since broadcom drivers are all modules instead of compiled into the kernel. The modules depend on custom hooks in the kernel.
M Audio 2448 is a C-Media 8738, it has a high quality SDPIF output. Unlike many consumer audio cards, SBLive, FM801, CrystalSound cards it does not resample all digital output to 48khz. If you play 44.1khz audio, you get a 44.1khz digital stream. Digital resampling introduces artifacts and deteriorates the sound quality.
No, it does not suit you to control the volume through the SPDIF output, that should be done at the analog amplification level. You shouldn't do this at the SPDIF level because you would be losing resolution.
It's possible, by making the samples smaller - you'll have softer music, but, as you make the sample values smaller, you're essentially throwing out the least significant bits of precision. You go from 16 bit audio, to 15 bit, to 14 bit, etc. It is not an advantage.
The M-Audio 2448 is your best bet, if you can get one on ebay. It has excellent Analog out - the digital output is run into some nice DACs on the card. You have good linux support.
Another option is the Zoltrix Nightingale Pro-6, it's a $50 card with SPDIF Coax/Optical in and out. Analog will not be as high quality as the M-Audio card, however, if you are using external DACs (this is a good idea) it makes more sense.
I would suggest a Zoltrix card with an external DAC that connects to a power amplifier. The external DAC is better electrically isolated, and probably will have nicer DACs than a consumer soundcard. Further, you can attach the SPDIF to a regular home theatre receiver, and send raw AC3 from DVDs/DIVX (I have encountered DIVX with AC3 channels, very nice.)
Well, why constrain yourself at the PCI level? If you recall the prilchard is exactly what you suggest: It's a Virtex FPGA attached to the PC133 bus. You interface with the board using regular push/mov/movsw instructions. The author was able to demonstrate almost a gigabyte/second DES, much faster than any other PCI accelerator. Furthermore, you could do fancy off-loading using DMA. E.G. DMA from the ethernet card into a ring buffer maintained by the FPGA, and have the FPGA preprocess.
For a real general approach, you would need another card in the system, probably PCI to be able to program the fpga, and it would allow the FPGA to raise an interrupt if necessary. Personally, I would be extremely interested in one of these because of the phenomenal possibilities it allows.
For example, it would make the gnuradio project much easier by building an FPGA with all the requisite, processor intensive FIR/IIR filters, FFTs, and Viterbi decoders.
Furthermore, you could buckle the FPGA to a high-speed A/D converter, and use the FPGA to do do the initial signal processing. That would really make a difference for very low-latency sound processing for example.
You could implement custom cache functions in hardware for databases, you could accelerate SSL like mad. With open cores, you could patch libmad, etc. to use the FPGA when its available, accelerating MPEG encoding, for example.
I have an A21p with the same screen. The interesting thing about it is that it is an 18 bit screen. 262, 144 colors per pixel. Hence, I run X at 16 bit color. I lose a bit of blue and a bit of red, but I gain much faster X performance.
I don't know if its possible to replace it with a full 24 bit LCD, or if one exists - but, from what I understand, the rage 128 chip has a bunch of LVDS outputs - perhaps the full 24 bit compliment, and as long as you get it to drive the display correctly, you're gold.
umm, no, the mdlp is atrac3, which is an atrac frame that encodes silence (zero's) followed by a hidden atrac3 frame which has the newer encoding algorithm. MD's algorithm is similar to MP3's (variable allocation of bits to encode the output of a few simultaneous DFT's into a constant bitrate stream - the most important features of the wave are encoded). This way, ATRAC3 discs will play as silence on non ATRAC3 compatible devices. The new NetMD recorders which attach via USB have the computer do the mp3->atrac3 transcoding process, then download the atrac3 stream. See the OpenMD project which is reverse engineering the NetMD usb protocol.
However, I'd like to clarify that the md and mdlp units use the same media, with the same mechanical recording system - the change is in the bitstream and the playing software.
I would suspect that sony is probably placing ATRAC3 on a CD and playing that. Simply an issue of software which is generally cheaper to develop than hardware.
I agree with you, the students should be taught how to read instead of being given a crutch to appreciate computer-learning; however, more alarming is the idea that people might propose to have computer educate our children how to read.
In my experience, I've found many people get excited by technology. Better living through chemistry has been replaced with better living through computers. However, teachers are expensive - more expensive than computers and there is a monetary as well as opportunity cost to have a human instructor teach a child to read. That instructor is then unable to teach other students.
Frankly, I think multimedia initiatives and wiring up elementary schools with ethernet and satelite links is ridiculous. I believe that the ultimate goal of school is to teach someone how to learn. This means learning reading, writing, math, scientific reasoning, etc. Most of the basic things taught in elementary school.
For example, the point of writing most papers in primary and secondary school is not to learn the subject matter, but to learn how to research a topic (any topic) and write a coherent, legible paper about it.
Any school library will have more than enough resources more papers than teachers can assign their students. Given the goals for that period in a student's education, a student's ability to learn is not limited by the scope of your average school library and is not further enabled by access to more information through the internet.
Frankly, I think we need to reevaluate priorities. Teachers, after all, are the most versatile educational instrument available. I think we should focus less on computers for students, and more teachers for children. (E.g, Maine providing ibooks for students; Wiring up impoverished third-world villages.)
A single teacher with no other resources can achieve more than ten times his/her salary in educational tools.
P.S. I realize this is semi-offtopic. It's an observation on the use of computers in school and how they are utilized.
Not only is it possible, but, I have a U3 in my backpack running linux. There are a bunch of patches to get a lot of the custom stuff working, and not all of it is entirely functional, but, it runs like a champ. Supposedly it works betterw ith devel, 2.5.x kernels. Only qualm I have with it is the size of the keyboard, keys are very small - I'd try one before you buy it.
Speaking of which, I run SPDIF between my pc and my receiver over a 50 foot length of speaker wire that I terminated with el-cheapo radioshack rca plugs. You can tell when it loses sync, 'cause the receiver will cut out for a moment and display "PCM 48Khz" briefly. Otherwise, it works well.
Woah, cut the guy some credit! This isn't just a case hack; he didn't just spend 4 hours with a dremel.
He went to the trouble of using the orignal keyboard. That means he designed and programmed a microprocessor to convert the matrix from the keyboard into the serial stream for the ps/2 port.
According to his tech page, he's going to build a switching power supply, which isn't easy to do right, to replace the 1U power supply he's currently using.
Oh, and there's the sound interface, too, more custom hardware.
Sure, he's using an emulator, but, he's also using a lot of the original hardware.
Actually, an interesting thing I noticed the other day, is that a bunch of divx's are being released with downsampled or original AC3 tracks, so if you have alsa and a spdif capable soundcard (cmi, fm801), you can get dolby goodness. A websearch indicates they were probably encoded with a patched version of nandub. mplayer handles this great to: mplayer -ao alsa9 -ac hwac3.avi
Of course, if you don't have all that, mplayer will happily do the AC3 decoding and output regular audio.
Frankly, If you have tasks that require a 64 bit processor, there are a number of better options than the itanium. In all likelihood, if you really NEED 64 bit, you probably have large memory demands, at least, more than 3 gb/process. There are a number of architectures that would be more cost effective. For example, you may be able to make do with a Xeon solution. Of course, it may require you to adapt your application to work with the paging model (you're still limited to 32 bit pointers.)
If you need real 64 bit pointers, consider Sun or IBM, SGI, etc. Some information about what you want to do or how much money you have to spend would be more useful.
If you absolutely need 64 bit data types, and less than 3 gb of ram, use PPC's. They have a 64bit/32bit split. 32bit address space, 64 bit data types.
Remember, 64 bit processors are not inherantly faster than 32 bit ones. In fact, if you have no real dependency on 64 bit processors, then get a bunch of 32 bit processors. Get extra for redundancy, parralel processing, or as a spare space warmer.
Remember, the more established options are generally a better idea. They're mass-produced, cheaper, more supported, more spares, etc.
I doubt there's any application that specifically requires an Itanium. Evaluate what you need, and purchase appropriately.
Add to this one of those nice Altera FPGA's buckled to a DDR-333 DIMM. Dynamic, configurable, arbitrary logic at FSB speeds. Two products I would buy at cost, or a little over cost. Unfortunately there is only a small market.
Am I the only person disturbed by the fact that Gibson was born in 1948, SFFWorld Another Another etc. ?
Also, there is no mention that William Gibson died.
Either, this is another canadian sci-fi enthusiast, with the same name as the William Gibson that wrote Neuromancer, or, Someone's trying to pull a fast one.
A, a Pentium 90 can saturate 100Mb/s ethernet. It's trivial, after all, it's only around 10-12 megs of data, a second.
B, 100Mb/s ethernet operates at 31.25Mhz, not 100Mhz.
C, At PIO 4, It's not going to get near 16MB/s. They processor would waste an insane amount of time copying the bytes from the interface.
D, I suspect you have something horibly misconfigured with your router/file server/print machine; perhaps you are using a PIO mode, as I mentioned above. I used a pent 200 w/ a 20 gig drive on a UDMA-33 card. It pushed an easy 8 to 9 megs a second.
Slashdot Mirror
I have a ATI TV Wonder VE, its based on a brooktree 878, though I think its using the connexant version of the chip. The problem I have is that the while the signal I'm receiving (coax cable) looks good on a regular Television (Sony Trinitron, a few years old) the captured version is full of noise. The closest example I can find is here:
tvtime screenshots
The noise I refer to is quite noticeable in the full resolution pngs. I'll note that this is no ill reflection on tvtime. Then again, reading about the specs of the BT878, its supposed to have higher fidelity than a television monitor. Hence, I'm confused where I'm going wrong.
Any suggestions?
The EM is very limited including only a Program Automatic and an Aperture Priority Automatic system. The only semblance of manual control is using Aperture Priority and adjusting the exposure compensation. He's much better off with a FM, FE, or FA series camera if he's looking at the pre N2000 series cameras. He'd probably get a better deal buying an N70.
A camera itself is a light-tight box with two important components: A Lens and a Shutter.
;) Point is, choose the focal length that has the distortion you want, then zoom with your feet. Also, remember you can always crop in the dark room.
The most important of the two is the Lens. The Lens warps and bends the light to react on the film. Camera is a box; Lens does all the work and has far more effect on the quality of your image than the box.
This is the crucial part of the comment: SPEND MORE ON YOUR LENS THAN THE CAMERA.
We've all taken physics so we can all appreciate how fantastically complex these lenses are. Remember how the angle light is bent when passing between two different densities is depedent on its wavelength? Well, imagine having to deal with that along with half a dozen other problems in lens design.
First, evaluate what you want to shoot and then figure out what focal lengths you will need to create the images you want. Remember that focal length has many different effects on the image. For example, magazine covers are typically shot with very, very long lenses - around 400 or 500 mm - because the telephoto effect compresses facial features. Try taking a portrait with a wide-angle some time.
Many of the large camera companies have very devout followings who are more than happy to tell you about notable lenses. For example, for Nikon, people will often gush over the old AI-S 80f2.5 (iirc, haven't been on the nikon list in over a year.) However, other 'solid' lenses are the AF-D 50f1.8, the AF-S 80-200f2.8.
So get yourself a *nice* lens then get yourself a cheap, old, used manual camera body like an FM-2 or FE-2. (with a few exceptions) Nikon lenses are backwards compatible. Ie, I can take my AF-S ED IF 80-200f2.8 lens and use it quite well on my Nikon FE.
Also, Their flashes are backwards compatible, I can use my SB-28DX on my FE as well as my F5.
A quick rundown of Nikon Lens tags:
AF = AutoFocus
AI and AI-S = Automatic Indexing (manual lenses that have a tab which mates with a ring on the lens mount to report the aperture index to the camera for metering.)
ED = extremely low dispersion glass (used to make different generations of lenses distinct. ELD is a specially doped glass from nikon that, like its name, has very low dispersion.)
IF = Internal Focussing (the lens will not extend/contract to focus, the focussing is done by moving internal elements within the lens.)
-D = The lens can report the distance it is focussed at to the camera for flash metering.
-S = The lens uses a piezoelectric ring inside the lens barrel to focus instead of relying on the mechanical linkage to the motor in the camera. These lenses also have -D functionality.
Personally, I would get an AI 50f2 or an AF 50f1.8 with a Nikon FE or FE-2 to get the hang of things. Once you get feature envy, then you can get yourself an F100, SB-80, etc.
First of all, the section above is listed under the header called "Intel Ipsec Cards" and more specifically refers to the Intel Encryption Coprocessor on the card.
Further, Intel has written and released a free, GPL ethernet driver for their EEPro 100, 1000, and 10000 ethernet cards. I shall transcribe for your benefit the top few lines from linux/drivers/net/e100/e100_main.c:
That driver is a GPL implementation meaning that the OpenBSD developers are more than welcome to port it at their leisure.
Oh, you want real documentation too. Take a look at developer.intel.com:
Intel 82551ER Fast Ethernet Controller Networking Silicon Datasheet
Intel 82559ER Fast Ethernet Controller Networking Silicon Datasheet
Intel 82559ER EEPROM Map and Programming information
True, Those are for their FastEthernet chipsets, not the Gigabit chipsets that, "Intel steadfastly refuses to provide us with documentation"
Well, what about these?
Intel(R) 82541ER Gigabit Ethernet Controller Networking Silicon Datasheet
Intel(R) 82547GI(EI)/82541ER EEPROM Map and Programming Information Guide
Ok, so the IPSEC chip on the NIC isn't supported nor is there any data on that chip forthcoming. However, there are a number of papers that show that IPSEC and TCP offloading (not to be confused with TCP fragmentation/checksumming) are not efficient. Specifically, the "hardware" IPSEC is done by firmware downloaded to a small embedded processor on the NIC. This small, embedded NIC is not very fast, in fact, its rather slow.
Result:
Processor utilization drops marginally (modern processors can encrypt 10 megabytes/s trivially)
Latency shoots up (It takes the embedded processor longer to encrypt a packet than the host processor would.)
There are a number of papers corroborating that latency has a huge effect on maximum bandwidth.
[I think the paper regarding TCP offloading not being worthwhile is by Mudge. The IPSEC offloading not being worthwhile is my hypothesis, untested, but I feel logically founded.]
My point is that IPSEC offloading is not an advantage - it probably was in 200MHz K6 days, but it certainly is not in 2.0 ghz K7 days.
Other notables, for example, the 3Com 3CR990 still doesn't have IPSEC offloading, despite promises from either the openbsd txp driver or the linux typhoon driver.
Frankly, as far as Gigabit
Frankly, that's a product of the driver not a result of chip design or flaws. Further, all Windows (NDIS?) drivers I've had to work with have had the option called "Locally Administered Address" where you can force a custom MAC address.
If you take a look at most ethernet drivers, you'll notice that most drivers need to read a MAC address out of a serial eeprom. This address is then configured into the ethernet chip so it can receive appropriate unicast frames and is used by the driver when constructing ethernet frame headers for transmission.
Further, regarding just how "junky" the Realtek stuff is, I refer you to this page: http://www.fefe.de/linuxeth/realtek.txt which states quite clearly and specifically why the realtek card is "junk."
For the record, I can spoof the MAC address on my 3Com 3C905's, Intel EEPro100 and Broadcom Tigon 3.
He does work while sitting in the coffee shop; he does not, however, work for the coffee shop.
I've got five WD1200JB's in a raid 5 configuration; they have been operating solidly for almost a year. What problems have you noticed using these drives for RAID?
Replying to each in turn:
1, you can't control external IT services, external IT is hotile to you.
Fine, if you have support from your department, then treat the rest of your company's IT assets as 'hostile' and 'insecure.' Having your bosses support is crucial, it's his job as a project manager, or division head to facilitate his employees getting their work done. Further, its his job to make sure that important information and data is not compromised. He is delegating that responsibility to you, but it is still responsible. Hence, if 'hostile IT' has a fit, its his job to deal with them and to keep them off your back. This is similar to chain of command in the armed forces; your supierior officer is responsible for you and it his job to make sure you can do your job. Make sure your boss knows whats going on and do nothing behind his back. An open and honest relationship with your boss is crucial in office politics.
So, Firewall your LAN. You're worried about network based threats, so install an IDS to watch traffic that crosses your border 'your firewall'. Make sure that your windows boxes are patched; there are microsoft tools for facilitating this.
Even if you don't have support for a firewall from your IT people, you can set up a NAT machine to bridge from your old IP space to a private network (10/8, 192.168/16, etc.)
If you need to provide external access to internal resources, use a secure connection. There are systems for validating user identity (rsa tokens, smart cards.) Use them as your authentication system.
2, You don't have physical access controls and you cannot implement them.
Encrypt hard drives, I believe Windows 2000+ has the option for encrypted volumes. I would check around to proove to yourself, without a doubt, that its secure. Make sure swap partitions are encrypted, etc.
If someone has the resources and interest to physically break into your workplace, then encrypted hard drive volumes might not be enough.
Another approach is to make sure that no sensitive data is saved to the hard drive. This is difficult, as compiling or loading data may write data to swap partitions or leave other sensitive information in unguarded locations.
Consider a terminal server, and secure the terminal server. With a terminal server, you can place it behind large amounts of steel, locks, and other physical measures that make theft increasingly difficult. Remember, humans are generally the weakest key; if you are able to lock down a small part of your resources in a cage, or a reinforced closet, who has access to it?
Further, a terminal server would reduce downtime for engineers. "my power supply burned out; my hard drive died." big deal, their PC becomes essentially a stateless client.
3. Need to compartmentalize.
Frankly, if you're in a microsoft world, the easiest system is probably to use CIFS with ACLs. Use Active Directory to model the "logical" break down of people in your organization. Its left up to the reader as an exercise in 'trust'. I would get two fileservers, and use fibrechannel to have redundant links to a redundant RAID array. (compaq, dell, etc. would love to sell you this.) If cost is a concern, a similar solution could be constructed. Place these servers in the secure cage with the terminal server. Give the terminal server and your file/authentication servers a private, local LAN between each other. Place a tape library in the cage as well, and automate backups. You might consider renting a machine at a colo somewhere far away and using that for off-site storage over a secure VPN.
4. "engineer hit by a bus scenario"
Your fileservers will still allow the administrator to access all files, but, lock the administrator accounts to local log in only. You can only be the super user after unlocking fifteen physical locks that bolt down 100 lbs of stainless steel. However, in an emergency, you can retrieve information.
5. take files home
What about allowing VPN access to the terminal se
PCI:
33mhz*32bit==105 MB/s
66mhz*32bit==211 MB/s
66mhz*64bit==422 MB/s
133mhz*64bit==851 MB/s (PCI-X)
NET:
10BaseT == 1MB/s
100BaseT == 10MB/s
1GbE == 100MB/s
DS1 == 150kB/s
DS3 == 6.75MB/s
Multiple DS3's on the same bus would pose about as much problem as multiple 100BaseT cards. Ie, no significant challenge.
Actually, no. The wireless chip driver supports all the broadcom line of wireless cards. They are a Software Defined Radio (SDR), and per FCC regulations, Broadcom is not allowed to divulge the programming information to the public, to prevent abuse. For more information, look at the work on the Atheros chipset drivers, they require a binary module to initialize the SDR.
Mind you, this chipset is the same as the Apple Airport Extreme's, and the Linksys WPC54G wireless cards.
That's the big deal. Also, the firmware image contains drivers for the Broadcom 10/100 chipset, too.
It's important to note that the kernel has been modified as well. Some people have suggested that they are immune to GPL violations since broadcom drivers are all modules instead of compiled into the kernel. The modules depend on custom hooks in the kernel.
M Audio 2448 is a C-Media 8738, it has a high quality SDPIF output. Unlike many consumer audio cards, SBLive, FM801, CrystalSound cards it does not resample all digital output to 48khz. If you play 44.1khz audio, you get a 44.1khz digital stream. Digital resampling introduces artifacts and deteriorates the sound quality.
No, it does not suit you to control the volume through the SPDIF output, that should be done at the analog amplification level. You shouldn't do this at the SPDIF level because you would be losing resolution.
It's possible, by making the samples smaller - you'll have softer music, but, as you make the sample values smaller, you're essentially throwing out the least significant bits of precision. You go from 16 bit audio, to 15 bit, to 14 bit, etc. It is not an advantage.
The M-Audio 2448 is your best bet, if you can get one on ebay. It has excellent Analog out - the digital output is run into some nice DACs on the card. You have good linux support.
Another option is the Zoltrix Nightingale Pro-6, it's a $50 card with SPDIF Coax/Optical in and out. Analog will not be as high quality as the M-Audio card, however, if you are using external DACs (this is a good idea) it makes more sense.
I would suggest a Zoltrix card with an external DAC that connects to a power amplifier. The external DAC is better electrically isolated, and probably will have nicer DACs than a consumer soundcard. Further, you can attach the SPDIF to a regular home theatre receiver, and send raw AC3 from DVDs/DIVX (I have encountered DIVX with AC3 channels, very nice.)
Hope this helps.
Well, why constrain yourself at the PCI level? If you recall the prilchard is exactly what you suggest: It's a Virtex FPGA attached to the PC133 bus. You interface with the board using regular push/mov/movsw instructions. The author was able to demonstrate almost a gigabyte/second DES, much faster than any other PCI accelerator. Furthermore, you could do fancy off-loading using DMA. E.G. DMA from the ethernet card into a ring buffer maintained by the FPGA, and have the FPGA preprocess.
For a real general approach, you would need another card in the system, probably PCI to be able to program the fpga, and it would allow the FPGA to raise an interrupt if necessary. Personally, I would be extremely interested in one of these because of the phenomenal possibilities it allows.
For example, it would make the gnuradio project much easier by building an FPGA with all the requisite, processor intensive FIR/IIR filters, FFTs, and Viterbi decoders.
Furthermore, you could buckle the FPGA to a high-speed A/D converter, and use the FPGA to do do the initial signal processing. That would really make a difference for very low-latency sound processing for example.
You could implement custom cache functions in hardware for databases, you could accelerate SSL like mad. With open cores, you could patch libmad, etc. to use the FPGA when its available, accelerating MPEG encoding, for example.
For some reason, I can't access his webpage, but, his work is at, http://www.cse.cuhk.edu.hk/~phwl/
If you're interested in this route, and want some help, shout.
I have an A21p with the same screen. The interesting thing about it is that it is an 18 bit screen. 262, 144 colors per pixel. Hence, I run X at 16 bit color. I lose a bit of blue and a bit of red, but I gain much faster X performance.
p df
The panel is an LG.PHILIPS LP150U1-A2 and its spec is at http://www.lcdspecifications.com/lgepdfs/LP150U1.
I don't know if its possible to replace it with a full 24 bit LCD, or if one exists - but, from what I understand, the rage 128 chip has a bunch of LVDS outputs - perhaps the full 24 bit compliment, and as long as you get it to drive the display correctly, you're gold.
Ahh.. The future,
"Organisms'r'US, this is Charlene, how may I help you?"
"Hi, yeah - I'd like to order a Natalie Portman with a side of grits."
"Mmmhmm, I'll need a credit card and a shipping address - would you like that overnight?"
umm, no, the mdlp is atrac3, which is an atrac frame that encodes silence (zero's) followed by a hidden atrac3 frame which has the newer encoding algorithm. MD's algorithm is similar to MP3's (variable allocation of bits to encode the output of a few simultaneous DFT's into a constant bitrate stream - the most important features of the wave are encoded). This way, ATRAC3 discs will play as silence on non ATRAC3 compatible devices. The new NetMD recorders which attach via USB have the computer do the mp3->atrac3 transcoding process, then download the atrac3 stream. See the OpenMD project which is reverse engineering the NetMD usb protocol.
However, I'd like to clarify that the md and mdlp units use the same media, with the same mechanical recording system - the change is in the bitstream and the playing software.
I would suspect that sony is probably placing ATRAC3 on a CD and playing that. Simply an issue of software which is generally cheaper to develop than hardware.
I agree with you, the students should be taught how to read instead of being given a crutch to appreciate computer-learning; however, more alarming is the idea that people might propose to have computer educate our children how to read.
In my experience, I've found many people get excited by technology. Better living through chemistry has been replaced with better living through computers. However, teachers are expensive - more expensive than computers and there is a monetary as well as opportunity cost to have a human instructor teach a child to read. That instructor is then unable to teach other students.
Frankly, I think multimedia initiatives and wiring up elementary schools with ethernet and satelite links is ridiculous. I believe that the ultimate goal of school is to teach someone how to learn. This means learning reading, writing, math, scientific reasoning, etc. Most of the basic things taught in elementary school.
For example, the point of writing most papers in primary and secondary school is not to learn the subject matter, but to learn how to research a topic (any topic) and write a coherent, legible paper about it.
Any school library will have more than enough resources more papers than teachers can assign their students. Given the goals for that period in a student's education, a student's ability to learn is not limited by the scope of your average school library and is not further enabled by access to more information through the internet.
Frankly, I think we need to reevaluate priorities. Teachers, after all, are the most versatile educational instrument available. I think we should focus less on computers for students, and more teachers for children. (E.g, Maine providing ibooks for students; Wiring up impoverished third-world villages.)
A single teacher with no other resources can achieve more than ten times his/her salary in educational tools.
P.S. I realize this is semi-offtopic. It's an observation on the use of computers in school and how they are utilized.
Not only is it possible, but, I have a U3 in my backpack running linux. There are a bunch of patches to get a lot of the custom stuff working, and not all of it is entirely functional, but, it runs like a champ. Supposedly it works betterw ith devel, 2.5.x kernels. Only qualm I have with it is the size of the keyboard, keys are very small - I'd try one before you buy it.
Speaking of which, I run SPDIF between my pc and my receiver over a 50 foot length of speaker wire that I terminated with el-cheapo radioshack rca plugs. You can tell when it loses sync, 'cause the receiver will cut out for a moment and display "PCM 48Khz" briefly. Otherwise, it works well.
Woah, cut the guy some credit! This isn't just a case hack; he didn't just spend 4 hours with a dremel.
He went to the trouble of using the orignal keyboard. That means he designed and programmed a microprocessor to convert the matrix from the keyboard into the serial stream for the ps/2 port.
According to his tech page, he's going to build a switching power supply, which isn't easy to do right, to replace the 1U power supply he's currently using.
Oh, and there's the sound interface, too, more custom hardware.
Sure, he's using an emulator, but, he's also using a lot of the original hardware.
I'd like to see you do that.
Actually, an interesting thing I noticed the other day, is that a bunch of divx's are being released with downsampled or original AC3 tracks, so if you have alsa and a spdif capable soundcard (cmi, fm801), you can get dolby goodness. A websearch indicates they were probably encoded with a patched version of nandub. mplayer handles this great to: mplayer -ao alsa9 -ac hwac3 .avi
Of course, if you don't have all that, mplayer will happily do the AC3 decoding and output regular audio.
Funny, you must go to umich.
Frankly, If you have tasks that require a 64 bit processor, there are a number of better options than the itanium. In all likelihood, if you really NEED 64 bit, you probably have large memory demands, at least, more than 3 gb/process. There are a number of architectures that would be more cost effective. For example, you may be able to make do with a Xeon solution. Of course, it may require you to adapt your application to work with the paging model (you're still limited to 32 bit pointers.)
If you need real 64 bit pointers, consider Sun or IBM, SGI, etc. Some information about what you want to do or how much money you have to spend would be more useful.
If you absolutely need 64 bit data types, and less than 3 gb of ram, use PPC's. They have a 64bit/32bit split. 32bit address space, 64 bit data types.
Remember, 64 bit processors are not inherantly faster than 32 bit ones. In fact, if you have no real dependency on 64 bit processors, then get a bunch of 32 bit processors. Get extra for redundancy, parralel processing, or as a spare space warmer.
Remember, the more established options are generally a better idea. They're mass-produced, cheaper, more supported, more spares, etc.
I doubt there's any application that specifically requires an Itanium. Evaluate what you need, and purchase appropriately.
Add to this one of those nice Altera FPGA's buckled to a DDR-333 DIMM. Dynamic, configurable, arbitrary logic at FSB speeds. Two products I would buy at cost, or a little over cost. Unfortunately there is only a small market.
Am I the only person disturbed by the fact that Gibson was born in 1948, SFFWorld Another Another etc. ?
Also, there is no mention that William Gibson died.
Either, this is another canadian sci-fi enthusiast, with the same name as the William Gibson that wrote Neuromancer, or, Someone's trying to pull a fast one.
A, a Pentium 90 can saturate 100Mb/s ethernet. It's trivial, after all, it's only around 10-12 megs of data, a second.
B, 100Mb/s ethernet operates at 31.25Mhz, not 100Mhz.
C, At PIO 4, It's not going to get near 16MB/s. They processor would waste an insane amount of time copying the bytes from the interface.
D, I suspect you have something horibly misconfigured with your router/file server/print machine; perhaps you are using a PIO mode, as I mentioned above. I used a pent 200 w/ a 20 gig drive on a UDMA-33 card. It pushed an easy 8 to 9 megs a second.