jec252

Member
  • Content Count

    166
  • Joined

  • Last visited

Everything posted by jec252

  1. Has anybody tried these? In a standard system I mean. I have one in my VGN-SZ650N and it works wonders for battery life, but has anybody heard of how they work in systems that didn't ship with them originally? Also, does anybody have any non-OEM experience with them at all?
  2. jec252

    7500-8 v.s. 7506-8

    Well, how much it can help depends on a number of factors. The most obvious is the array itself: you need ot have a pretty decent set of drives to eat up the 266 MB/s that the 7500 offers, but it can be done, especially with RAID 0. Beyond that, however, if you have a PCI-66 64 bit bus then using the 7500 will slow it, and every other device on the bus, to 33 MHz. On an AMD MPX system, since the 66 MHz 64 bit bus also hosts the southbridge and hence the PCI-33 32 bit bus, this may lead to a reduction in the performance of perhipherals on that PCI bus if the PCI-66 64 bit bus is otherwise saturated with I/O from another PCI-66 device and/or something on the southbridge (for instance the ATA channels).
  3. jec252

    Analog modems

    For that application, virtually no difference. It's a slow system and will therefore eat up more timeslices to use a winmodem over a hardware modem compared to newer systems, but the load is still going to be dominated by applications, not the modem. Also, it may have a slightly higher propensity to drop connections and lower quality connections with slowly lower download speeds. I use a winmodem all the time in my laptop though. If you can find an ISA modem and he has an ISA slot, it would be oustanding as it offers the hardware features at a lower price point and at 56 kbps the bandwidth is a non-issue, but they're difficult to find at all anymore (/me clutches his 56k ISA modems)
  4. The same applies to any RAID level with the exception of RAID 2. It protects against a single disk failure and can also provide protection from a single controller failure in SCSI and FC scenarios. It is not a backup - but it can and does enhance performance and protects against a type mechancial system failure, by far the most common type. And frankly, if you're booting off a RAID 5 array, you need to hae your head examined. There is no good reason to boot off of a RAID 5 array - the performance to run the OS off of it is pretty horrible, even if all it is is a file server. Booting off of a RAID 1 array is much preferable, and then using additional arrays for additional storage if needed. Fine, but he has that already. Remember that massive amount of network stoarge? And the SDLT? Sounds an aweful lot like my array + DDS4. And for some more background information, the ICH5R is not a PCI-E device but uses a single proprietaty IHA 1.5 connection at 266 MB/s to the northbridge, shared with the PCI bus, all of the ATA channels, onboard sound and the onboard 10/100 NIC if it's there. Similarily, a PRO/1000CT network connection also has its own dedicated connection to the northbridge at 266 MB/s. My suggestion would be to use two Raptors in RAID 1 for your OS, programs, swap and et cetera, and one large 7200 RPM PATA drives either independently or in RAID 1 for addational space. Audio files aren't particularily bandwidth intensieve, but the seek time on yoru OS and applications drive as well as having a seperate spindle for your data will help performance. FYI: CompUSA has 400 GB 7200.8 PATA's on sale for a few days for $200 after MIR. Also, there's nothing wrong with FC on the desktop, but the drives you're talking about are fairly old to being with and will not outperform a Raptor, plus you'll be dealing with a single FC link at 2 Gbps full duples for all of the drives - somewhat less than the 100-150 MB/s per drive offered by ATA.
  5. jec252

    DVI-I vs DVI-D cable

    You should be able to adjust that on the card end - at least my FGLs can do that. In the worst case scenario, you could always use an analog connection. But the answer is that yes, right now you are running a digital connection.
  6. jec252

    74gb Rapto "Spin Down" Noise

    I have this problem on some Seagate disks, caused by spindle wobble according to Seagate. It's like what happens when your washing machine gets unbalanced, except in this case a disk goes slightly unbalanced and then comes back on othogonal to the rotation. It takes less than the distance from the head to the disk to cause this noise. Or at least that's what Seagate told me. It's annoying, but I have three disks that do it and none of them have had any problems, and one's been in use for ten years since it developed the noise.
  7. jec252

    Giving up on Antivirus software

    Since I'm back from the weekend now, I can say difinitively that running the AV program does not increase file latency by that much at all. For a quick and dirty benchmark, I opened and saved and closed to another disk an AutoCAD file slightly under 4 GB. With the AV on it took so little additional time that when I used my stopwatch it was within human reaction time for error, and in one test actually with with AV was actually a second faster. Times were with: 34/67 and without: 35/67 (time in seconds). 4 GB is clearly over the size of my RAM (and in any event is very much clearly over the size of virtual RAM available to AutoCAD) and given those times nearly all of the file was loaded. The file was stored sequentially on the disk as well, and the "save to" disk was actually a slightly faster disk but it was transfering from the outer zone on the read disk to the inner zone on the write disk so that affected performance a bit. All of this was done with writing to the same area on the disk, roughly. This is on a Windows 2000 system with dual Pentium III-S 1.4, 1 GB RAM, three disks and a FireGL. This latency issue isn't squat, comparatively. I turned off my UPS software and it actually shaved three seconds off of the disk read times (no effect on writes, I suspect it was due to PCI bus contention or disk limitations). A good antivirus software scans the file while it's in blocks in RAM and has good hueristics so it doesn't need any additional scanning almost at all.
  8. jec252

    Giving up on Antivirus software

    Latency's the issue, huh? Like for some reason when I save a file it doesn't get written to RAM first where it ends up being scanned, and obviously it's the AV doing file IO rather than it scanning once it's into RAM. And, then again, if latency is such an issue perhaps I should stop my UPS monitoring software that uses more CPU time than my AV. Want to stop your UPS monitoring software? I didn't think so. So why should I disable my AV?
  9. jec252

    First Car.

    I have full liability and collision so I would get up to the actual cash value of my car regardless of what happened to it in an accident. The thing was, technically it was one accident in the police report and one insurance claim but the insurance company treated it like two seperate accidents: one when I hit her, the second when her son drove the car off of mine (which is where the Explorer was totaled). Since I was only responsible for the first and the second was considered as a no-fault accident, I wasn't liable for all of that damage. What I was liable for, and my insurance covered, was about $2000 for a new bumper. She didn't even get a rental car out of the deal (usually what happens when you get hit in an accident), just the $2000 for the damage I did and the scrap value, which I understand on a brand new car is about $2000 itself.
  10. jec252

    First Car.

    Supercaff: I live in the NY suburbs in CT and pay that amount every six months! And it's on a 1998 Toyota Avalon with no accidents and only one ticket on my record. Which is another interesting story ... I was in an accident, totaled my car (and ended up totaling a two month old brand new off the lot Ford Explorer with all the bells and whistles, but I'll get to that later) and because I was 17 at the time the car was in my parent's name. Well, I didn't need a replacement car so we didn't get one for over six months, and our insurance company ties the vehicle to the accident since you're insuring the vehicle, not the driver, and after six months with no replacement vehicle we didn't have any vehicle pick up the accident and saved a boatload of money there. Oh, and me totaling the Explorer wasn't technically my fault. See, I rear-ended the Explorer driving a '92 Sable and so there was not a scratch on my bumper ... I went right under up to the engine block. After the car stopped it snapped up and so the engine block was resting directly on the frame of their car, and at this point the accident was considered "over". About 30ish seconds later when the Explorer pulled over to the side of the road, the point of which I didn't exactly see since obviously my car wasn't going anywhere and I was fully blocking the lane, and more to the point the insurance company didn't see the point either, the car pulling off of my cast iron block tore part of the steel frame in two, right in front of the lady's trailer hitch (I say "the lady's" although her son who had had his permit for two months was driving). So, both cars are totaled (mine was about $8000 to a $6000 car and theirs was around $100,000 to a $40,000 car) and the lady I hit didn't have collision insurance, so she got about $2000 from my insurance company and probably a couple thousand more for the scrap value and that was it.
  11. I'm not sure if that's the ideal, but you are correct.
  12. There's no modding required: the E7505 will work with Pentium 4's already, just as the Serverworks GC series will work with the Pentium 4 as well, and the i875 chipset will run a single Xeon. Except for the sockets and onboard processor features (such as SMBus) the chipsets are interchangable technically, although there may be some liscensing issues with it. Also, the three companies which make PCI-X chipsets right now are Intel, AMD and Serverworks. The reason is that they are actually technically reasonably hard to make and require advanced logic and alot of bandwith to the northbridge.
  13. There are a few boards floating around that are S478 with 64 bit PCI, and there have been a few uniprocessor boards using the i440BX/GX chipset with 64 bit PCI in leiu of the AGP slot for a while. The point of these boards so far has been uniprocessor servers, and if you want higher end performance you need to go to dual processor boards in most cases (somewhat understandably). Since you can run multiprocessor boards as uniprocessor systems, and in the case of servers requiring higher performance most purchasers will indeed go to a higher end board, making only DP boards with high performance PCI was not a big deal for quite a period of time. Take for example the C440GX board. Also, this has a Serverworks chipset and PCI-X, not just regular 64 bit PCI. To make a uniprocessor board which includes 64 bit PCI, currently you must use a multiprocessor capable chipset such as the Serverworks GC-SL/-LE/-HE/-WS, i860, E7500/7501/7505 or similar. To use AGP in a system based on these chipsets, you limit your choices quite a bit (to the GC-WS, i860 and E7505).
  14. jec252

    .Net is blatant Java ripoff, half as fast...

    One questions for everyone: are you really suprised that .NET is a blatant Java ripoff? I mean, Microsoft had to do something with their now illegal JRE.
  15. To be more specific on the PCI issue, IBM actually developed the MCA bus because they didn't feel PCs could compete in the server market, and PCI of course changed all of that, and it was developed by (gasp!) Intel in an effort to move their chips into the higher end markets, and they used EISA as a kind of bridge between the two camps. The original use for PCI busses was to create SMP i486 machines to run Novel Netware on, but that has obviously since changed significantly. As an aside, PCI was actually designed on Sun workstations using the bus that PCI would eventually replace - SBus. And on the 640k of memory thing, that again wasn't IBM, that was actually Intel and Microsoft. While 32 bit processors can address 4,294,967,269 bytes of memory (4 GibiB), the 80286 that DOS made the most headway on used 16 bit addressing, limiting it to 65,536 bytes of memory (64 KibiB), but using some nifty tricks it was alocated in ten byte segments, giving rise to the 640 KibiB memory limitation (as opposed to the 64 KibiB on the older 8086 parts). When MS moved DOS into the 386 with its 32 bits of virtual address space, they left the first 640 KibiB as a kind of legacy section for 16 bit processes and then used Extended and Expanded memory to fill the gap upwards (Expaned used larger pages and 16 bit addressing of those pages and Extended actually used 32 bit addressing) as well as the high memory area and created the need for the 15-16 MB memory hole. All of this was Intel/Microsoft, IBM didn't have anything to do with it, it was simply a design limitation of the processors that got built into DOS and has been with us ever since, even though all of the way back to the 386 Intel's chips could put 16 bit address space anywhere it wanted in the 32 bit address space, Microsoft left it down in the first 16 bits of conventional memory. Now you might be asking, what has IBM thought up of? Well, if you own a laptop made within the past ten years, pay homage to IBM. With their ThinkPad, they changed the face of laptop making by creating a form factor that everybody uses today. Like your keyboard? Again, IBM. Also, the keyboards that today are considered the most rugged and have the most tactile feel and best layouts are IBMs, as they have been since the days of the IBM Selectric. That little pointing stick in the middle of your keyboard on most laptops and some desktop keyboards? Thank IBM again, it actually increases productivity by up to 20% (you don't move your hands from the keyboard), and since the patent ran out some years back everybody's been using it. RS232? Partially IBM, if not completely. ATA? IBM again for the most part did alot of work in the begening, hence its name being drived from the PC/AT. VGA interface?... did you think it was nVidia? Parallel ports? The floppy drive interface? The size of your motherboard (ATX is simply AT rotated 90 degrees)? APICs for dual processing (IBM had alot of invluence in this one)? Warm-swap devices (ThinkPad 760CD) and PCMCIA (IBM ThinkPads)? The shape of your power cables? Size of the expansion slots? Size of the expansion bays? Need I say more. IBM has had its had quite mucking around in the PC industry since they invented it, and the laptop industry as we know it as well.
  16. jec252

    Ways to make partitions?

    I'll second Occupant's comment. Unless you're doing some fancy schmancy mounting, such as making a seperate partition for c:\documents and settings (useful if you feel you might have to reinstall) I'd just use the LVM in Windows NT/2000/XP/2003.
  17. Sandra is only valid when taken relative to itself and not to anything else, and does not provide absolute answers. However, it is a good diagnostic tool, just not exactly what I would consider a benchmarking tool.
  18. jec252

    US being smart with contracts in Iraq

    The only time companies do that is when "for the good of the world" = "save your own ass because war is on the line."
  19. Yes, but the Typhoons can be fought, and you cannot guarantee that it is destroyed before the war warning bouy (in the case of a hit by the attack sub) is deployed or 24x8 missiles (in the case of the Typhoon winning against the attack sub) are launched. The idea behind the subsurface deterrant is it's impossible to guarantee that all of them are taken out before at least some of them are launched, and if even a few of them are the devestation would be tremendous (think NYC, DC, LA, Chicago, San Diego, Oahu and Norfolk wiped off the map and a swipe NORAD).
  20. jec252

    Giving up on Antivirus software

    Lol, that part has quieted earlier, what I was doing was just making a slight correction to the earlier comment (not disagreeing, just pointing out something).
  21. The only thing that disturbs me about these nukes is that, although they have great capability as a deterrance, they also have great capability as a first-strike option. If you land a nuke directly on top of a hardened bunker containing a nuclear missile, you can take it out, and with minimal destruction (the "bunker busters" cannot go through that much concrete). This would be first strikes against North Korea, China, India and Pakistan if we got them (not against Russia, their Typhoons will still work as a deterrant) and if we went to war with them. So far though, it's only a feasability study, and the idea is to create nuclear weapons that would destroy the minimum amount of area and still take out the bunker, which you can do effectively with nuclear weapons and laser guidance systems and only take out a few city blocks (the pressure decreases with distance on a cubic curve). Also, as an aside these nukes could take out targets which the Minuteman, Trident or even the 10 MT Titan II "rubble bouncer" could not.
  22. jec252

    Giving up on Antivirus software

    Actually, I would say that it's the other way around, but it's way easier to not need support or documentation at all on Windows (aside from the included help file) than on Linux is more then point.
  23. jec252

    Networking Crazyness

    Oh, by "between them" I mean "between the 10 base T machines and the 100 base T machines on the same segment."
  24. jec252

    Networking Crazyness

    Is the GoC NIC actually running at gigabit speeds? If so, you might be getting some congestion from the step down and/or step up. I now when connecting to my 10 base T machines to a 100 base T segment performance is horrible talking in between them, but when I put the 10 base T machines on a 10 base T segment of the LAN the performance is about what I would expect (10 base T switch with a 100 base T uplink to the backbone).
  25. jec252

    Giving up on Antivirus software

    True for Linux out of the box, but not commercial UNIX, Slackware or Debian if you know what you're doing. Frankly, when my mother is finally in the market for a new desktop class machine (probably a laptop by then) I'll be pointing her to MacOS most likely, and if not then an IBM ThinkPad with Debian or Slackware on it, since she would sure appreciate a UI facelift and wouldn't miss the macro functionality at all. Problem is, she'll probably be looking into an MBA soon so she might have to keep Windows for a bit longer than she (or I) would like, but that's life. However, there are alot of reasons why Windows is necessary, and it's not due to Windows per se but due to application support for Linux or UNIX.