Page 1 of 2 1 2 LastLast
Results 1 to 20 of 28

Thread: transfer rates

  1. #1
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    Ok macgurus....I have a problem with transfer rates; that is, how fast information is transferred from one drive to another. And to be real specific, I am referring to copying files back and forth. And in a way, this problem was generated by YOU....yes, you guys, who insist on doing scientific tests so that apples can be compared to apples....

    kaye and others published read/write data for good single drives in 40 - 60 MB/sec range. Pristine and rigorous testing methods were used, precisely tuned, so that one macguru's numbers could be compared with another macguru's results. I too have a few benchmarking tools, and I too obtained such numbers on my new MDD. But....so what???

    Once upon a time ricks took me to task for measuring performance using selected tasks. His point was that if I don't adhere to commonly accepted standards, my data can't be compared to anyone else's. So true, so logical, so down to earth that I had to acquiesce. But after experiencing the "slow" copying rates in the new expensive machine, I am beginning to feel rebellious. And thus the questions.

    I tried to copy huge files (over 1GB) from one drive on ATA100 to another drive on ATA66 (yes, good performance drives, good performance buses), and the best I can get is about 1GB/min. That is less than 20 MB/sec. Since the time is long enough to time transfer with a stop watch, this is real world data. If I try smaller files, they zip right along and I can't time them.

    So, question number one: what good is the 40+ number if the best I get is less than 20 MB/sec?? (I am restricting my comments/questions to ATA single drives -- not RAIDed, not SCSI).

    Then I read the specifications for ethernet hub, by Asante, where they brag of transfer rates up to 100 - 200 Mbps. Allowing for overhead, that is only 10 - 20 MB/sec. So, question two: what's there to brag about? These numbers are nothing to write home mother about....are they???

    Question number three: is my complaint caused by poorly maintained drives in my MDD? That is, not optimized for best performance? Can I take my single 180GXP attached to ATA100, tune it up, and realize the full 40MB/sec copy rate of a large file?

    Finally question number four: how fast do files copy in YOUR drives? Say, big ones, the kind you can time with stopwatch?
    marrand

  2. #2
    Join Date
    Feb 2001
    Location
    on the landline, Mr. Smith
    Posts
    7,784

    Default

    Good questions.

    quote:

    "Question number three: is my complaint caused by poorly maintained drives in my MDD? That is, not optimized for best performance? Can I take my single 180GXP attached to ATA100, tune it up, and realize the full 40MB/sec copy rate of a large file? "

    Unless you know that the file in question *and* the volume you are transfering to are contiguous, I would think that performance would be affected... the Gurus will say yea or nay. Of coarse, one may also say that your test is a "real world test", even though there is no way to do any comparison - every drive will be in a different level of fragmentation. Right?

    As far as the ethernet transfer speed issue, that is a seperate complicating factor: How fast can a system pump out the data, *and* how fast can ALL the pieces/parts of a network pass that data... hubs, swithces, routers, cabling, termination, NIC cards, firewalls, etc.

    I take those types of vender claims to mean: " We won't be the bottleneck in your network". Up to us as consumers or admin types to figure if it is true or even relevant.

    What if you buy a 1st quality switch with maxium thoughput for the given protocol (100baseT, Gigabit, etc), but there is an old 10baseT hub upstream somewhere? Or bad network design? Too many hops, bad cable termination, or a cable run that is too long?

    Can get ugly really quick...

    [This message has been edited by newbie (edited 14 March 2003).]

  3. #3
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    newbie,
    Good point - "We won't be the bottleneck in your network". I will buy that; something else in my "world" can impede transfer even more than the hub in question. But my point was: their claim is still a slow transfer rate, no?? Even if all else was perfect?

    Contiguous? Not fragmented? The real world is fragmented -- as soon as you used the computer, perfect ordered is destroyed. But I ask for something humble and simple -- how fast can you Mr. newbie, or you Mr. macguru, transfer a 1+GB file from one drive on one bus to another drive on another bus??
    marrand

  4. #4
    Join Date
    Jan 2001
    Location
    Mobius Strip
    Posts
    13,045

    Default

    Some benchmark tools will graph from small files of 4k up to 64MB file. It use to be that 8MB was larger than the disk's own cache, and up to 8MB all that is fully tested is how well the cache performs, more than the drive. Our benchmarking has much to be desired and mostly confirms if the drive setup is performing within norm is all, ie, cables and terminator are setup correctly, the drive works.

    Spend some time looking at the reviews, benchmarks, and more important at this juncture, the methodology that Storage Review uses, their testbed, and the tools they've developed. they've got 5 yrs now as of this week, in honing their skills in this domain.

    I just installed a new SCSI drive and I know that my computer and hardware can't take full advantage of what it is capable of. But, it is quiet, even though 10K SCSI, fast (faster in heavy I/O and random use than numbers suggest), runs cooler, and can't be heard... no pitch or whine to give it away, you'd think an ATA drive was purring along. There is a slight [i]something [/] during copying a group of large files, not even a hum though. But it boots fast, loads huge programs faster, and cost so much less (60% less), than the two year old drives, that it replaces. If it was in a QS/MDD, etc it would deliver a full 70MB/s... but when you look at the specs, there is internal and external and often an average transfer rate. Look at the drive specs on Googlegear or the vendors web site or pdf data sheet.

    The PCI bus probably is one limiting factor as well.

  5. #5
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,109

    Default

    marrand,

    ¬?An issue that I don't know how to deal with that Gregory has brought up a few times is system caching. In OSX most current bench programs don't know how to bypass that, the system cache makes disk I/O look like it happens at RAM speed. Skews all the results.

    ¬?Look at ATTO ExprssProToolkit when it is run in Classic, we start seeing 387 or 425 MB/sec which we KNOW is nonsense. Except that the data DID move that fast.

    ¬?Your test of a very large file mostly removes that issue. I like your test, it needs to have some variables removed is all so that we all can do the same test and post the results. The type of data is important, I don't think that fragmentation is a very big factor, it just doesn't take that long to move the heads when compared to the time it takes to move 10 gigs of data.

    ¬?The next thing that is an uncertainty is how does the data travel from one disk to another and is anything there under our control? If it is under our control then we need to specify what the settings or conditions are. When I say 'what does the data do' on a disk to disk copy what I mean is that it isn't like the two drives are connected directly to each other with no other influencing hardware or software. When the data is read out of a drive does it go directly to ram? Does the memory controller shortstop the ram and point the data to the recipient drive? Does the cpu ever see the data and if so what actions does it make and in what cases do those actions get triggered?

    ¬?What would take time is if the data causes the system to have to look at it and decide whether it should just pass it on to the other drive or process it somehow. We need to determine the data file type carefully to minimize those events.

    ¬?Also I would like to know what other system variables are involved. The OS itself has priority, so if it needs to make a log access or a housekeeping it will preempt the bus, boom - shutdown. What happens if that is happening to you but not to me?

    ¬?What is the difference if you have 1.5 gig of RAM and I am running .75 gig? Does that have an effect? I don't know. What is happening if I am running a GeForce3 graphics card and a flatpanel and you are running an 8500 ATI and a 24" SGI monitor? What is the difference between 10.2.4's I/O package performance and 10.2.1's I/O package? What if I have SCSI mounted and you don't?

    ¬?We may very well get to bypass most of these. ataMan may have some of the I/O package information as well as some ideas on how the data involves other subsystems so we can eliminate those variables. I'm thinking on stuff that has been on my mind for a while but I have very limited knowledge of how OSX moves data. I may just be borrowing trouble.

    ¬?I would guess that we can use the mkfile Terminal command and fill it with ones. Then we'd have a plain data file that every user could duplicate. The ditto command to move the data. As soon as the transfer is finished the Terminal window reverts to login again, great way to stopwatch it.

    ¬?I'm happy to compare my disk performance to yours to see if it helps. I don't have a MDD but I do have a couple of buses and drives.

    Rick

  6. #6
    Join Date
    Jan 2001
    Location
    Mobius Strip
    Posts
    13,045

    Default

    SR has a lot of PC users that swear by the ATTO benchmark also. And it is not indigative of any real world performance. I was talking about the drive cache more than the system cache. A benchmarking program would have to be able to disable deferred writes and send 'sync' commands at the least.

    Anyone out there have one of the 10K Cheetah's that have 16MB disk cache? They use to be sold for highend video environments. And what if the disk cache started using DDR or faster SDRAM?

    That's why SR has a way of "burning" data to a drive (sector copy), scattering data, and then running tests for literally hours in order to derive I/O numbers. And different patterns for different uses (office desktop, server, web, ftp, gaming). We're "benchmarking" empty drives, then wonder why under a full load of data they don't copy as fast or run as well? I can load up my Atlas or Cheetah and they still run well. I could ride my bike at 50 mph... downhill, too.

    Try putting a 4GB folder on the Dock that has hundreds of sub-folders. I notice on ATA that there is a real delay when opening folders that I just don't see or feel when using SCSI 10K's. response time seems faster. The idea of seeing how long it takes to open 100 folders is a nice test. No delay.

  7. #7
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    marrand,

    I think I can shed some light on your slow Finder drag copying of a large file size. For a long time, after someone posted to the forums about slow Finder drag and drop compared to testing drives with ATTO EPT, I wondered if the Finder verifies. When I do drag, things happen so fast that I can't tell if there is verify and nowhere can I find if it does, tho I'm almost positive now that it does not. But it is slower considerably than the EPT tests. So why is Finder drag/copy so slow?

    From what I've been able to glean, there are at least a couple of things going on. First, it seems the Finder typically breaks files down into small chunks, I just don't know how small. But apparently down to small enough chunk sizes where most drives are at their least efficient to read/write. Second, I had a long discussion way back with an Intech Tech guy who explained that drivers from the various manufacturers are tuned for best performance at a particular size. The drive itself will introduce other results such as its Mode Page Parameter settings (such as number of cache segments), but he said the intech driver (and he thought the SoftRAID driver) were tuned for 8MB size max thruput. My take is that it was probably done to look good in EPT, but again results also depend on the drive involved. However, when you software stripe RAID, in addition to the variables above, there are other variables which can be controlled, such as the size of stripe segments on each drive, and, in the case of SoftRAID, using more main memory for a variety of uses in striped RAID. k

  8. #8
    Join Date
    Feb 2001
    Location
    on the landline, Mr. Smith
    Posts
    7,784

    Default

    marrand:

    You can see why I picked the name newbie...I am not new to Macs, but I am new to this level of detail and analysis. Much of what I know comes from these gentlemen in particular. Thanks guys!


    Thanks Ricks; I always wondered how much fragmentation actually effects I/O. Sounds like I can say little to none... unless perhaps it was very severe? And as drive tech continues to improve it will be less and less (if all else remains the same, like bus speed)



    The world I live in (databases, email servers, fileservers, and "does my printer have to be plugged in to work?" kind of questions) does not allow time for (or require) this level of knowledge for me. Like many, I have no doubt I will never make it to their level. That's OK, it does not pay the bills for me. When one of 100 users moves a file on/off our OS X server, speed is what it is. Other than verifing that the machine is configured correctly, the only time I can really consider such lofty notions is when it is time to replace hardware every few years.

    Would never try to discount or argue with the kind of info we are lucky to have here; it is a blessing for all of us trying to understand. Learning all the time.

    But at the end of the day, all I need is an "estimated miles per gallon" type of number. As I try to absorb the details, I can know that my milage may vary depending on things like driving conditions, tire inflation, and city or highway. If the sticker says 25 city, I know I may not get 25 in the city, but, whatever I get, I *expect* to get better than the car with the sticker that says 18 city in simailar conditons....

    I took your questions to asking for that same type of speed test/information. As important as consistant, repeatable benchmarking is - apple to apples - the reason we we benchmark is to try and extrapolate real world performance, right?

    Am I oversimplifing, Gurus?



    [This message has been edited by newbie (edited 14 March 2003).]

  9. #9
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,109

    Default

    ¬?¬?¬?"Am I oversimplifing, Gurus?"


    ¬?No, couldn't a said it better. In fact, I didn't say it better

    R

  10. #10
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    Guys,
    Thanks for your input. But still.....it is so confusing. Let me approach it another way.

    Gregory,
    I have seen the graphs from 4K to 64MB on my drives. I tried to understand SR's IPEAK test results, and the only one which registers on my brain is the MB/sec vs. radius plot -- fastest read/write on the outside, and slowest on the inside. I stare at the Read Service Time distribution and scratch my head.....duh. I hear about disabling cache, but then SR proceeds to tell me that external controller and its driver make it difficult.....etc, etc, and I ask myself.....where am I???

    I understand your analogy better....you can ride the bike at 50mph....down hill. Oh yeah!! That I dig. So, when I ask how fast you can ride your bike, and you answer 50mph, I ask you to repeat the test. You may come up with 50mph again, but soon you will run out of hills and your numbers will change drastically. When I ask you how fast you can copy a huge file, that task you CAN repeat...on and on and on.....and if you don't alter your equipment much, your answers won't change that much.

    kaye,
    Aha. So you too observed something funny going on with plain copy, drag/copy as you call it. Is it as funny when you back up files to a different drive? With Restrospect Express? With FWB Backup Toolkit? It is still transfer of data from one drive to another. Then you mention in passing that ATTO may have tuned their utility to 8MB max file size to make their results look good. Hmmm.....thus results artificial??

    newbie,
    You make me feel good!. You are much more experienced and skilled with computers than I am, yet you admit to being awed by the experts in our midst. I agree with you....at the end of the day, all I need is the "estimated miles per gallon". So, if you copy a huge file from one of your drives to another, and we exchange a bit of info about our systems, won't that be worthwhile and sufficient? I like your analogy to miles/gallon; if you tell me nothing about your vehicle, nothing about city vs. hiway traffic, then yes, your number won't mean much to me. But if we exchange a bit of info about our cars and roads and traffic patterns, won't the comparisons mean something??? By the way, I like your simplification.....I can understand that.

    ricks,
    You agonize. And I am glad, because you are getting closer to understanding my pain. LOL. How does the data travel, you ask. Well, I have asked the same. Simplistically, 32 bits (32 little voltage signals) are gathered on the uphill slope of the pulse, another 32 on the downhill, and the pulses travel at near the speed of light to the sensor in the head, then through the circiutry to disk's cache, then further and further down the path until they find themselves in the sensor of the head of the receiving drive, and now writing begins. Read one disk, pump info through wires, and write on the other disk. Too simplistic, however; does the data go to RAM, you ask? By your questions you have disassembled the process into a myriad of jigsaw puzzle pieces...and now what? Who will put them together? To put the question in another way, ricks: are we analyzing too much? Reaching paralysis by analysis?

    I just don't know. At times it seems to me that copying huge files is a good idea. When I read the macgurus answers, however, the idea loses its appeal.
    marrand

  11. #11
    Join Date
    Aug 2000
    Location
    Concord, CA
    Posts
    7,056

    Default

    marrand,

    I have been copying in the Finder from one drive to another. I use Retrospect (not the Express version) and it takes longer because I have it verify every file it copies. With BackUp Toolkit I was surprised at how fast it copied to another drive but I'll have to see how fast it actually was around the first of April when I go thru my maintenance routine. But neither Retrospect nor BUT are using Finder copy.

    The last comment about who is tuning to 8MB you have backwards. The driver manufacturers such as Intech are/were tuning their driver for max I/O at 8MB to better their results with EPT on the Mac. In other words, EPT sort of became the de facto standard for benching drives for lack of better utilities.

    The point I guess is that what do you do most on your computer, copying files or running applications that work better with faster I/O, like Photoshop or iMovie/FCP? If you are working with applications that work better with faster I/O, and particularly making a living at it and spending a lot of time watching progress bars, then I think EPT and some of the MB5 tests can give you a ballpark idea of where you need to spend your upgrade money.

    Generally I only do the maintenance once a month that entails copying files or Retrospect or BUT. The rest of the month, I'm trying to take advantage of my upgrades. Unfortunately, keeping up is hard. I now have 10 one hour SVHS tapes to get into iMovie then iDVD for burning. Running out of tape and have to get to that. But, thanks to Gregory, I also have a nearly new 50-pin FAST-20 Cuda to replace the original drive in my PTP and one off those Sonnet HardRAID cards to test in the Buick, plus I found out last night that Princess has company coming today. Guess what comes first. k

  12. #12
    Join Date
    Jan 2001
    Location
    Mobius Strip
    Posts
    13,045

    Default

    Makes me think of those black & white photos that have "mirror images" of faces if you stare but don't focus, something will pop out!

    The drive specs usually show an average transfer rate, as well as megabit (not byte) transfer rate. The Atlas 10K-4 is 820Mbps, 55MB/s average sustained, 44-72MB/s range, and 92MB/s internal transfer rate. That 820 is what has been inching up a lot in the last 2-3 yrs. At one time, 400Mbps was what you might expect, broke 600 level, next up for the 10K-5 Atlas is 860Mbps and 100MB/s internal transfer rate.
    quote:

    The Atlas™ 15K drive is the world’s fastest SCSI drive, providing seek times as fast as 3.2ms and maximum sustained data transfer rate up to 75MB/sec. The Atlas 15K drive provides up to 45% more IO/sec than average 10K RPM drives. This allows system integrators to offer end users the highest performance solution in the industry.

    The Atlas 15K is designed for the most demanding high-performance and I/O-intensive applications, such as OLTP, enterprise servers, 3-D animation, high-performance workstations, data mining, and NAS and SAN environments.

    The Maxtor-developed Ultra320 interface features MaxAdapt¬ô (adaptive active filtration), a closed-loop method of improving signal quality. MaxAdapt allows the drive to adapt to changing system conditions and components, which translates into lower error rates, easier integration, and increased bus efficiency for optimal system performance.


    This pdf article shows how scsi fits into video creation helps bridge the gap from the numbers to what they mean for editing digital video for the user. (And, rendering the pdf document seems to be a good test of your system's horsepower, it took a while on my lowly G4/500!)

    -G.

    Atlas 10K U320 drives have done their homework to make video editing as smooth as possible, in fact, the Ultra320 spec should be a big boost for those doing video (iMovie or FCP, or developing for HDTV).

    [This message has been edited by Gregory (edited 15 March 2003).]

  13. #13
    Join Date
    Feb 2003
    Location
    ex-Cupertino, CA, USA. Now HU+SRB (Europe).
    Posts
    163

    Default

    A silly idea, but let's do an experiment:

    1) Partition an ATA drive (180XP is good) into 3 partitions:

    - one partition is only 1 GB small
    - the second partition is almost the entire drive
    - the third partition is 1 GB small.

    So we have 3 partitions, one small is at begin of the drive, one large is almost the entire drive and one is at the end of the drive.

    2) Let's take a nice machine (2002 MDD or any AGP machine with 4 PCI slots; we do need a clean PCI bus). To simplify our life, take MacOS 9.2.2

    3) Take any UDMA-133 controller. All of them (Acard, Tempo, FirmTek, whoever) are faster than the internal bus. However, if there is no controller, even the internal bus is fine.

    4) Start the ATTO benchmark on the first and the third partition and IGNORE the middle partition.

    Now I am curious - what is the difference on the 180 XP between the inner tracks and the outer tracks? I have a feeling, there is a very significant difference and it can account for the most of the problem. The rest could be due the fragmentation and the file system itself.

  14. #14
    Join Date
    Jan 2001
    Location
    Mobius Strip
    Posts
    13,045

    Default

    ataMan ...did someone hit the back key and trigger two messages? No harm, I've seen it happen a couple times recently.

    The trouble with ATTO with some drives that have 8MB cache is that it may not be testing the disk drive as much as the cache. If there was a 64MB file size (like there is in ATTO ExpressStripe) that would reduce the effect of cache.

  15. #15
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    ataMan,
    Interesting. SR reported their results on 180GXP, plotting MB/sec vs. radius. The first and largest value was about 56 MB/sec (as I recall), and the last and slowest was about 30MB/sec. This is what IBM claims in their specs. Also, this is what all drive manufacturers say, that first and outer sectors allow for fastest data transfer, and last and inner sectors for slowest.
    So, given the backround information on this type of experiment, wouldn't one expect similar behavior with benchmarking utilities of your choosing? The absolute numbers may differ, but the relative drop-off from max to min should be about the same.
    Ok, only my $0.02 worth.
    marrand

  16. #16
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    Gregory,
    The cache thing. Interesting as well.
    I have used the 8MB file of ATTO on my 180GXP and 120GXP, 8 and 2 MB cache respectively.
    With 180GXP and its 8MB cache (on ATA100 bus), the read (or was it write?) curve reaches its peak and continues merrily along with only a slight drop-off.
    But with 120GXP (on ATA66 bus) the same curve peaks at the same value and drops suddenly to something lower after the first 2MB -- its cache! A sudden drop, not gradual, as if the 2MB cache is now filled, and transfer is slowed down.
    Perhaps this is boring and old news to everyone, but to me it was a revelation...
    (PS: Gregory, I am preparing a separate response to your special pdf file on Maxtor320 and the slow rendering).
    marrand

  17. #17
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    Gregory,
    Your "pdf article" brought my new MDD to its knees. Rendering/scrolling through all 8 pages took about 80 sec. That's the longest I have ever encountered for an acrobat file; I suspect Maxtor¬?embedded slow loading graphics to stress the point that MY drive is a slow poke, and I should buy their super SCSI.
    I tried to remove the pictures by cropping out the text only, but the document was fighting and resisting me all the way. I tried text/select all/copy and then paste into PowerPoint - no go.
    And on my lowly G3/500? Actually, when the rendering/scrolling became unbearably slow, I switched to my MDD, with its sexy video card, to speed things up. It didn't work; my MDD, 1.25 dual, just isn't suitable for special graphics (or whatever special "something" Maxtor put into that file to humiliate people).
    The file size isn't big -- less than 4MB. It has only 8 pages; only a few pictures, and primarily text. A typical instruction/specification document. But can't be handled by one of the best Macs??? It seems I can test and tune my machine till kingdom come, and still be foiled by a simple file on a computer hardware site.

    To verify, I took a 63 page acrobat file I created, primarily pictures and colorful, about 3MB. And tried scrolling through it. Each page opened instantly,and it took me about 50 seconds just to click through 63 pages. Thus not limited by computer speed. Then I took a 55 MB PowerPoint file, 15 pages, containing heavy scanned tiffs. Each opened instantly - total about15 minutes to scroll and render. Again, not limited by computer speed; I am just a slow mouse clicker.
    marand

    [This message has been edited by marrand (edited 15 March 2003).]

  18. #18
    Join Date
    Jan 2002
    Location
    NW Montana
    Posts
    8,197

    Default

    Brought your MDD "to it's knees" Look at my poor QS. I have been following this thread - nice info thanks guys - but I have so many other things going on to learn this in "detail" really wanting to learn X like I knew 9.x.

    But got curious and.... thought I would scroll through that pdf file with my QS 733, 10.2.4, 896MB RAM, GeForce 2 and a 60 Gig IBM GXP. Most pdf's scroll way to fast, but then most are 90/95% text.

    6 minutes 10 seconds. What can I say

    Randy

    [This message has been edited by rwm2 (edited 15 March 2003).]

  19. #19
    Join Date
    Jan 2001
    Location
    Mobius Strip
    Posts
    13,045

    Default

    Amazing what a poorly designed file can do. It took minutes to scroll one page ... so no complaining, okay? And using 85-92% cpu. In fact, I killed it eventually.

  20. #20
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    Gregory,
    That killer pdf file appears phony to me; perhaps designed to induce the reader to buy Maxtor stuff.
    When scrolling and rendering, the CPU in the 1.25dual goes wild for a long time, much longer than when opening up a picture ten times the size. I set the monitor for thousand colors.
    Finally I was able extract a page and save it as tiff. When opening it with Photoshop, it appeared instantly.
    The little picture at the bottom of the page, the colorful one, was also extracted as tiff and saved. Again, opened instantly, no "rendering" observed. Yet, when opening part of the original pdf file, the entire process slows down to a crawl.
    Also, the pdf equivalent of a page is larger than the tiff. Uh? I thought it was supposed to be the other way around.
    It's as if Maxtor embedded a little trojan horse, an innocent little program to make your cpu go through useless little loops and make you think that your hard drives are so, so bad. What do you think?
    marrand

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •