Site hosted by Angelfire.com: Build your free website today!

^/\^ PeaK /\^/\



What's new... is old again

August News


  • August 14, 1998
    Just wanted to tell you that a self imposed absence from Web browsing/articling for the last 3 weeks has proved interesting. Based upon this benchmark, I have experienced better concentration and am much less moody. In the past I found that logging in to the Internet would provide a short term "fix/elation" equivalent to browsing a magazine section at my book store. Sitting perched on a chair and staring into my 15" NEC 4FG with wrists that have seen 10+ years of "carpal tunneling" is probably not the most daily healthy exercise. In the case of web surfing, I would say if you surf hard then play hard (and that means away from your terminal). The frequency of heachaches and moodiness has gone down markedly. I'm thinking, is it my monitor and me watching the equivalent of TV from 24". I think I am going to borrow a LCD screen or portable for an extended amount of time to see if the good old cathode ray tube is the culprit.

  • The patent suit by Cirrus is now focusing on the "said to be" original claims in the patent. ATI has responded by claiming that prior disclosure was made by them in generally available public literature well before the filing date by Cirrus.

  • I'll be away cycling/driving in the Maritime provinces of New Brunswick, Nova Scotia, PEI, and Maine over the next two weeks. I just got my Katadyn water filter at about the cost of VooDoo2 card. In the end, I know I am going to enjoy it a whole bunch more. Nothing quite like a drink of water after a long portage in the middle of a remote park system, three days in.

    July News


  • July 21, 1998

    • What do you think of when you see a patent ? I looked at my tennis raquet and lo and behold, there is a patent number 3999756 on it. Could I possibly understand the thinking behind the patent and why the it was patentable in the first place ? Personally, I think the original spirit and intent of patents no longer works. In those days, in return for disclosing the details of how to make and do something better, the patentee was rewarded for his disclosure on how to make the world a better place by being granted a 20 year span of time to which he was granted payment (royalties) from users of his idea. You cannot patent an idea like "make a tennis racquet more powerful", but you can patent an implementation like, in the case of HEAD/Prince, make it longer to acheive more power. It is somewhat like writing a song ( a unique combination of notes) to which you can collect licensing fees for people who in turn record it and sell it. The problem with patents today (my opinion) is that it is a bit like patenting music notes, after a while. To follow this, back when patents where introduced, we lived in a mechanical era. Figuring out how to make something like a water tap open and close by turning a handle is pretty useful and complicated stuff. An idea like a "car tire", that had cushioned bumps by having air in it was quite neat! These things took time to develop (years) and patents were a really meant as a means to "share" ideas. How does an idea to share become outdated ? In areas of fast change, such as software, the need to "share" ideas to make the world a better place is not really needed. Ideas can be implemented in "minutes" as compared to years in the era of early patents that were mostly mechanical thingabobs. Indeed, many different ways of implementing the same end goal in software differ only in efficiency. Why allow a patent for the most efficient way of doing something. This applies somewhat to chip engineering. Some of the patents to me are just straightforward engineering. This means if I locked up 10 people in separate rooms and gave them a common design problem for a week, the 9 out of 10 would come up with the same solution. This to me is straightforward engineering. There are of course some patents that are neat. Like downloading information to a watch by holding it up to your computer monitor...no cables...and useful....and not immediately obvious.

    In an area of fast change, the license period should be shortened and patent education should taught to engineers after the area of patents has gone over a joint overhaul by respected inventors/patentors with respect to different fields. One of my colleagues says that the granting of patents has gotten so out of hand in engineering that the equivalent of "a music note A flat" might be granted becuase one person at the patent office had a bad day. All songs/inventions that used this note would alter this note to be "note A" to bypass this patent and the world would be one that was more out of tune with itself.

    Accessibility of patents to households will in itself change the system that was traditionally closed to all but a select few. I think the best way to start with patents is by looking at patents granted to common objects such as frisbees/tennis racquets that are presently my different firms.

  • July 8, 1998
  • My old ATI/Linux (via GIMP, of course) logo can still be found near the middle of this page but I think my unsteady artistic verve maybe had something to do with ATI having an official logo release page for those wishing to add them to their web pages. There are 5 GIF images with two of them being animated.... I won't give in to animated GIFs on these pages in case you are wondering, I just can't standing watching my hard drive LED blink on and off.

  • Dropped by local computer store to find the new Acer AX59 Pro board. The VIA chipset (MVP3) seems to have finally sorted out the teething problems with the prototypes and Acer has put together a mature product. The board supports all socket7 CPUs plus it adds an AGP slot and support of 100MHz frontside bus in addition to the standard 66, 75, and 83 modes that we use today (60, 100 and 112 MHz, too). Those PCI bus speeds are now derived by the addition of a 1/3 divider to generate the 33 PCI bus off the 100 MHZ setting. Collectively, these have been referred to as Super7 Boards. Find out more about these boards here.

    The AX59 Pro has 2 DIMM slots and 3 DIMM, 2 ISA/4 PCI/ 1 AGP slot in an ATX form factor. I like to have a third ISA (sound card, scanner card, ethernet card) and re-use my old AT based case that is built like a tank (weighs like one too). The board runs the new AMD K6-2 chip with 3DNow! extentions to gain performance over and above raw FPU power. Microsoft has already added support for this in new releases of DirectX and you can expect to see support in Linux in the future.

    Cyrix and AMD have chips performed better and more efficiently at the same clock frequency than the Intel counterparts on the bulk of applications which were coded without floating point but the migration of gaming houses towards floating point calculations in games (think ID Software and Quake) played into Intels FPU strength. Intel (luckily) had the best FPU unit amongst themselves, AMD and Cyrix.

    Like the AP5T board, it is designed well as evidence by its stability during evaluation of AX59 Pro at Anand's hardware site. Short of actually testing this system running Linux, I think Acer has a winner up it's sleeve. For an idea of the in-house testing performed on this board...click here.

  • Dave Steele over at the FPS forum pages offered a fairly good explanation of the dependency of framerate and refresh rate. I copied it here cuz things tend not to be archived on those pages: 

          Re: Could someone explain to me again how the refresh rate affects framerate 

      From what I've heard, the ATI cards wait for VSYNC to flip buffers. That means that unless a game
      uses triple-buffering (I believe most are double-buffered) then one buffer is displayed while the
      second is used to render a new image. When the rendering is complete then you wait for VSYNC,
      flip buffers to display the newly rendered image and start rendering a new image in what used to be
      the display buffer. Repeat forever. 

      The upshot of all this is that refresh rate can make a big difference in the average fps of a game.
      Suppose the Rage Pro was able to render 70 fps if it didn't have to wait for VSYNC. If your display
      refresh rate was 60 fps then the Rage Pro could start rendering after the buffers were flipped (at
      VSYNC) and be complete in plenty of time to flip the buffers again at the next VSYNC. If you
      increased you display refresh rate to 75 fps then rendering would not quite be complete in time for
      the next VSYNC and you'd end up flipping buffers every 2 VSYNCs. That would effectively reduce
      the game fps to 37.5 (75 /2). Quite a bit less than the 60 fps you could have achieved. 

      This is very much like the effect of interleaving on an old hard disk or on a CD-ROM (I don't want to
      explain that so I hope it stands on it's own). The effect is much less noticeable with a Rage Pro since
      it typically doesn't render at anywhere near the refresh rate of your monitor. If it could render 24 fps
      flat out (triple-buffered or not waiting for VSYNC) then you'd get 20 fps displayed with 60 Hz
      refresh, 23.33 fps with 70 Hz refresh and 18.875 fps with 75 Hz refresh. Hardly a staggering
      difference but a difference just the same. 

      I'd be interested to hear what games like FS98 are telling me when they indicate 28 fps (or
      whatever). Is FS98 triple-buffered? Does it report 1/render-time without considering the wait for
      VSYNC? Does it report the average fps over some arbitrary sample period? Or, perhaps I'm just full
      of crap when it comes to fps vs. monitor refresh rate for the Rage Pro. 

      Dave Steele

    Thanks Dave. Two more points are that the above assumes a "Hurry up and wait" algorithim for frames in the buffer that just missed the next upcoming "Vsync": This is analogous to missing the Vsync "train", so regardless if you are Donovan Bailey who missed the train by a minute or me who came in two minutes after (wishful thinking ?), we both need to wait for the next train the comes in an hour. If the next train (refresh rate period) was shortened (ie. higher refresh rate), we could get a faster/shorter refresh interval. Microsoft requires the drivers for D3D to "wait on vsync". Some chips get around this by providing a over-ride of this unbeknownst to MS to better show off the hardware. Last point is that refresh rates that exceed the framerate can cause a phenomenom that I'll call "false image". The effect can be understood by the eye linearly interpolating an image between successive still frames (like those flip pages of a golfswing) to generate smooth motion "if" the changes in position are equal distant from frame to frame. When a frame is rendered twice in the same position, as happends in refresh > framerate, two images appear in your head: One interpolated as descibed above and the one that was double rendered. Users experience a bit of disorientation/vertigo that people have to refer to as lack of smoothness. So it is a tradeoff, better benchmarks/tearing vs no tearing/wait on vsync/lower numbers for the cases where frame rate (See FPS) is less that refresh rate. VooDoo2 users might actually like running with "wait on Vsync" and have less headaches but they are driven by framecounter envy :)

  • Pssst....want to know what the next killer app will be ? I predict "image processing/manipulation" software. This software abounds but the ones that will suceed are the ones that are extensible (read plugins) by the endusers. Before I go into why, you have to understand that there are a lot of "Leonards" in the world. These people are into the arts and community and they give into technology only grungingly. These are the sort of folks that ask you about Dicitonaries to run on their computers and go out and spend $149 on the Oxford Dictionary because just d'ont want to know but want to know best in the area of non-commercial areas called humanities. Fact is, they do not care what operating system they run on as long as they can get at an application that can match their creative instincts. GIMP may be the "killer" application that will give Linux another knod or two when it comes down to getting work done.

    • Linux is an unbelievably cool community. For GIMP, a central site with downloadable extentions to create new effects will be the "ace" up its sleeve. Upgrade paths will no longer be new releases but accomplished by modular "plugins". The "Leonards" of the world can draw a sketch of someone they have never met before in 9 strokes of a pencil and probably some Mondo Cool stuff with 9 presses of a keyboard under GIMP. Again, it is free and extensible and supports the concept of "community" in a world of "Leonards". Nothing worse to break up a commuity than sending something as simple as a spreadsheet called "file.xls" to a friend to look over only to find that the default file format Under Excel 97 you saved to is incompatible with the old version of Excel that came with Win95. GIMP (Graphical Image Manipulation Program) has just released and a "tomb" of a User's Manual, 591 pages, called GUM (whatelse Gimp User's Manual) was relased almost concurrently....it is quite awesome and has some really great sections on how do put together some really neat stills. The manual is 16 megabytes of Adobe PDF and the 80 minute tour/tutorial is definitely work the visit....How about commisioning GIMP to do an oil painting of your favourite dog from a favourite photograph. Perhaps the best way I could sum this up is that I wake up on Xmas day holed up in some Ski Lodge with me and my GIMP and a stack of photographs....only to find Leonard peeping through the window and eyeing my GUM :) :)

  • July 4, 1998
    Hmmm.... stamp collecting or driver upgrade ? I loaded the 5.20(2411) driver and saw slight improvements in overall "image rate" in the Final Reality Robot scene and City scene. FR has is a true gaming benchmark written by a consortium of people in the industry including S3, Matrox, Intel, Cirrus-Logic, ATI, AMD etc. Wait for the credits at the end of the sound demo to see Gordon get his 5 seconds of fame. My system is a Acer AP5T board running a Cyrix M2 (2x75) with a Rage Pro based card. I have quite a few friends of mine that are snapping up this affordable board and PR233 chips to run Linux....they are very impressed with the balance of stability and snappy performance of this combination of pratical/powerful hardware technology and operating system technology. OK, I'll get off my Linux high horse and come back down to Win95 land. I have included a table below that summarizes the performance differences using the last four driver releases.
    • Driver
    Release    		 FR
    25 pixel
    (kbp/s)    		 FR
    Robot Scene
    (fps)    		 FR
    Fill Rate
    (mp/s)    		 FR
    City Scene
    (fps)    		 FR
    3D Transfer
    (Mb/s)    		 FR
    Overall
    (Reality
    Marks)
    2237
    (431c14k2)    		 42.61 10.01 52.42 11.31 5.06 2.19
    2278
    		 41.92 11.72 54.35 13.05 3.92 2.31
    2312
    (turbo)    		 43.43 11.87 54.34 13.33 3.92 2.32
    2411
    (5.20)    		 35.56
    (-18.1)    		 12.44
    (+6.9% *)    		 31.83
    (-41%)    		 13.73
    (+4.9% *)    		 3.80
    		 2.16
    Driver
    Release    		 FR
    25 pixel
    (kbp/s)    		 FR
    Robot Scene
    (fps)    		 FR
    Fill Rate
    (mp/s)    		 FR
    City Scene
    (fps)    		 FR
    3D Transfer
    (Mb/s    		 FR
    Overall
    (Reality
    Marks)
    2278 292.7 42.50 40.01 56.88 18.02 4.23
    2411
    (5.20)
    		 196.9
    (-32%)    		 45.6
    (+7.2%)    		 53.50
    (+33%)    		 60.99
    (+7.2%)    		 10.09
    		 4.07

    Note that the image rates are up about 7% (compared to 2278) for both processors but this is accompanied in a drop in the polygon rendering rate. The pixel fillrate shows a increase for the P2 CPU but a drop for the M2.

    The results above show some of the best FR numbers, ever, but do not approach the +30% figure (fast 2241 vs 2278) seen in earlier in his Final Reality results using P2 systems under some souped up beta driver (fast 2241). Comments around the Rage Pro sites are that the new 2411 drivers are very stable and frames do not periodically stutter anymore. This seems to be true of games such as Quake which play smoother (less jerky) and in my opinion look visually better (less blocky when you get up close and better shadow detail...more of the you are there feel...too bad I cannot benchmark it). The good news is that this indicates (to me) that further improvements is possible in the drivers.The confusing thing is that the rare EndUser (fast 2241) drivers were first seen early in the year but deemed not stable enough for general release. I like to try them out even in their unstable form to get a glimpse of usefulness across different applications. Thus, I do not expect latest 2411 to be the last driver for Rage Pro chipsets and still fully anticipate a driver with significant performance improvement plus better 3D bus transfer. I think the recent release of Win98 made ATI shy in adding in new features for 3D that may have jeopardize the running of 2D business apps running in Win98. With the way the the French cars have been running on the Formula One, I'll even aspire to having it codenamed after a french bicyle manufacturer... Mercier ? Gitane ? Peugeot anyone ?

    The new version (1.01) of the Final Reality Benchmark is available. The user sumbmitted scores show a second place finish for the Rage Pro. If the 3D transfer were better it would be in first in a grouping including i740, VoodDoo2 and Nvidia. The interesting thing to look at are the relative rankings of chips and areas were chips with better gaming framerate performance exceed those of the Rage Pro. For instance, look at the Nvidia chip in a Diamond Viper coming in at sixth place. It has better fillrate, polygon rate and 3D transfer rate but loses out in image rate. I am not sure what image rate is but I think it takes away the advantage in framerate for chips that do not wait on V-sync. The margin of this difference in these areas is precisely the amount of increase hinted at by EndUser's earlier results. So hang on to your bicycle seat.

  • July 1/Canada Day
    Happy Canada Day! Co-incidentally, ATI released the Version 5.2 in the form of Win95/98 combined driver to run the RagePro/RageIIC chips. Not much has been said about the RageIIC but it is a third generation version of the RageII chip borrowing the "3D setup engine" from the RagePro. Until we see more actual AGP applications, it will be the right chip for most people and can be configured to fit and run in an AGP slot at twice the PCI bus rate. The drop in memory prices allows for larger memory configurations on the video board to enable local texture storage. OEMs and consumers have to look at their needs and get away from their "checkmark" mentality when specifying system requirements. In this regard, Celeron's lack of an L2 cache is not an issue in games where data is seldom reused and floating point performance is more important. But most of us do care about fast performance in Word, Excel and dBase applications which are written with small tight code loops that benefit from the L2 cache.Back to the recent driver release: The FAQ mentions unresolved bugs with the Beta/Preview Win98 Operating systems available for preview that do not exist with the official release. The Readme suggests the 2278 be used for those who wish to continue running older 16 bit applications (Win3.1) and that one more "legacy" driver will be released in the future to handle this situation. Seems like a workaround in the driver could not be worked out to acoomodate the preview/beta versions of Win98 and probably explains the small delay in releasing this driver. A classic case of bait and switch ? Looks like those ATI video card owners who thought they could upgrade to Win98 by uploading the "preview" will be forced to upgrade officially and contribute to the Bill's building fund and college education fund.

    • June News

  • June 25, 1998
    ATI is located in a city with four seasons. You only really need your air conditioner for two months. ATI has given birth to two ball hockey teams: The ATI Wildcats and the ATI Raging Bulls. Last year, I played on the Wildcats and this year the Raging Bulls. These two teams play in the same division against each other a couple of times during the season. In this regard, the rumours that we have "parallel" teams is true. The Wildcats are in first place, while the Raging Bulls are experiencing the growing pains of playing in their first year: Here is a list of game dates for the Raging Bulls and scores to date. All games are played at Stephen Leacock Arena:

  • After having seen some of the "much" lower "fps" numbers posted, try this experiment: Re-run your demos using higher vertical refresh rates. I';ll explain later but some of you will experience quite a jump and some of you will not.
  • Balancing peformance with stability for OEMs, passing WHQL, providing hooks for everchanging OpenGL "top ten list" is probably addressed by multiple possible releases of a driver. It becomes a bit like filming multiple endings for the season finale of some popular TV series. One thing for sure is that drivers will be optimized "also" for Direct3D games such as Turok, Forsaken, and Incoming. Before installing ATI's next set of Win95 drivers (>5.0), re-run framerate benchmarks in these games to get a feel/record for the improvement. Ziff-Davis has stated that Winbench has made certain assumptions about how pages were to be presented by game developers and enforced this usage during their benchmarks. The mess occurs when drivers are written to optimize page handling, the ZD way, while greater numbers are achievable at the expense of slight imperceptable page corruption. Final Reality and Winbench do provide Apples-to-apples comparisons and are good benchmarks in this regard. The question about how things compare when they are oranges is an open question. ATI's centricity reflects those of OEM's: They are becoming fans of built-in frame counters found in almost recent popular games and as a result...so is ATI. Back to OpenGL and multiple endings. The fate of multi-texture support in the next Win95 driver has already been decided and is awaiting WHQL (pronounced "Wickel") approval. If it is in, does the recent "1036" OpenGL driver take advantage of it or do we need to wait for the next mini-GL release. My prediction: expect another mini-GL to be quietly released if the next Win95 driver has the multi-texture hooks in place.
  • They are quaking hereand called 1036 . They run the existing release 5.0 (2312) drivers so I d'ont think these things are going to be "intercooled" in a textured manner. This is the 1036 readme. Check in at FPS for the numbers. You can find out more from the ATI OpenGL download page. After reading the "1036 readme", I can't help feeling that it might have been much more but thats human nature :( The use of the 2312 driver indicates that this may be an interim release worthy enough to be non-beta. Look at the bright side...they are early and they are free ...just like stamps!
  • This year's annual OpenGL meeting was held in Japan. Formal announcement by SGI and Microsoft of OpenGL 1.2 shipping this month along with an ICD kit with "Designed for Windows" logo. Could it be that Microsoft and SGI are making a real go of their joint efforts. Its certainly seems so. This could significantly impact application vendors and hardware vendor developments for the better. Bravo...but what does it mean to the ATI ICD OpenGL development ? The OpenGL chess games gets another piece.
  • DFP or Digital Flat Panel is a new paradigm for display technology welcomed by both display manufacturers and video graphic card manufacturers. It promises to be as significant as the announcement of VGA at the end of the 80's. Most of the computers were hooked up to monochrome TTL monitors that transferred information digitally. The advent of colour and 8 bit RGB (24 bits) and higher resolution monitors made the VGA standard and derivatives (SVGA@800x600 and XGA@1024*768) the only way to get information across without violating FCC emissions. The VGA graphics card converted the 24 bit information to three outputs with similar analog characteristics to a TV signal to drive each Red, Green and Blue input of the analog monitor. The question is why should this interface become digital again and wouldn't the same problems that plague TTL display interface surface again ? The basic reasons are due a convergence of low cost flat panels, LVDS (low voltage differential swing) protocol to reduce emissions using twisted pairs and reduced voltage swings, integrated PLL technology that did not exist in the late 80's, and the fact that much of this technology is being adapted to meet EMI emissions requirements with higher resolution laptop designs...what was missing was the standardization of a connector between external flat panel (digital monitor) and a graphics card generating an LVDS digital video signal to provide for a desktop solution. ATI has had the LVDS technology integrated into their chips for close to two years to allow low emission connections to flat panel displays found on laptops. Compaq and ATI have already made a joint announcement on the first DFP card and monitor combination that will ship as a desktop solution to gain everyone about 3 square feet on your desk. Do not be confused with some of the flat panel VGA monitors that are appearing now and that will plug into your existing graphics card. The flat panel VGA monitors need to convert the analog VGA signal back to digital using a A to D converter...results will vary depending on the DAC of the graphics controller and the ADC on the monitor. DFP will consist of a new 20 pin connector that will support up to two parallel digital streams (one is currently used but the other is to be used when higher resolution displays become available).

    • With the video signal staying digital and the non flicker characteristics of flat panels displays, new paradigms/standards will emerge. We can gain resoution by trading off refresh rates to rates around 30 fps that are not possible with CRT technology . TCO, MPR-II, and headaches related to CRT emissions should become a non issue with the emergence of FPD. See them soon at a hotel check-in desks, everywhere.
    Digital Flat Panel Links

  • June 22, 1998
    "Take this, its for your own good", Mom would say. In the case of drivers, the good refers to a large majority of end users who would never think of: opening up their computrs, adding memory, or upgrading drivers. An ATI Win95 driver upgrade is typically a well engineered self-extracting affair that is not equiavalent to process of dental surgery. In order to ensure the general "good", it is mutually blessed via a third party known as WHQL. ATI has had a good track record of issuing "stable" drivers which is akin to "good cough medicine". WHQL is an FDA like process that ensures that "drugs" such as Excel and Word can be taken in combination with driver upgrades. The main difference is that the FDA does not sell the goods themselves. If you have a look at the sheer number of "drivers" that exist on the MS driver site that are qualified then you begin to understand the need for scheduling the qualification process. Unfortunately, software development which share this base of code will hinge on this activity...like I said...for the general good. Not much has been said about the ICD by ATI publicly other than the June 22, 1998 first beta prototyping release date and various SGI ICD information pages This ICD affects the curious more so than the "general good" which ATI is looking after. The final Beta Quake-subset OpenGL is nearing its third release with complimentary changes to the Win95 driver/library with technical refinements flushed out through the early Betas. Judging by the releases of 2278 and 2312, the next update will not be for a while. Some of the technical goods outlined in the ATI OpenGL pages (like multitexutre support) should be there for the final release but will there functions in the full OpenGL implementation that will warrant delaying the the release of the Win95 driver that might contain new functionality in the shared Win95 driver libraries ? I think we can wait until Friday June 26 to find out. The squeaky wheel does get the oil. Send your queries about delays in the ICD Beta release to betadriver@atitech.ca to find out if the "squeaky wheel" was due to a forgotten update on a web page or a change of plans. Again, please be constructive. Its summer...get out there...your screen saver of a pictureque sand beach at high noon just doesn't cut it :) . Later....
  • June 17, 1988
    On April 5, I wrote:

    • I got the 2 CD set from the Missus about a month before my birthdate and had to hold off until then to open it. Turns out I played it through about a week after that busy day....How was the CD ? It sounded compressed and a whole lot worst than the vinyl LP version. I popped the NAD amplifier unit out, cleaned some of the RCA jacks and cable connections with Cramolin contact cleaner (it goes under other names nowadays), played with speaker and power plug polarity(the missus pulled a double reverse on me) and VOILA...it sounds GREAT. If want to read more about what you can do to tweak your STEREO (as they used to say)...See the audio upgrade page. Yes, even my better half is smiling about the end result after having played through the Neil Young CD. Moral: contacts oxidize, polarity conunts, cheater plugs make more of a difference than I ( a electrical engineer by profession) am willing to admit.

  • Have you ever had a timing belt replaced in your car ? Most modern cars use them due to the quieter noise levels achievable at the expense of having them replaced every 100,000 Km ( 60,000 miles). The idea of just running it until it breaks falls when you find out that major damage(read $$$) to engine valves can result from out of order firing of pistons. I'll term this a destructive cycle . Some people trade their cars every four years in fear of that one destructive cycle that made hurt their potential resale value. In the case of the well known timing belt we are basically following the rule: Replace components that can cauase destructive cycles before their design life. People have also termed this prventative maintenance. Without doing the routine/"well documentented" procuedures fundamental to preventative maintenance, some people incurr the wrath of destructive cycles by not changing their oil in timely fashion. I'm looking for information (books, web sites, clubs, mechanics) who have clear guidelines about long term ownership (10+ years) of cars by giving suggested replacement/"design life" info for major components to help develop a preventative maintenance schedule for those less routine peventative maintenance tasks. Maybe we should have someone write a book on this...I'm sure it will be a best seller. How about "Long term ownership of Honda Civics for Dummies". I better write to IDG.

    • Please email-me if you can help. So far the RagePro chip in my computer has shown no known destructive cycles :) Before I forget, here is one link to a Honda Civic web pages that I will add to as I get info:

    ATI has garnered the "Channel Champions" title for Graphics Video boards based upon reseller voting. Resellers are the ones who ultimately put ATI products in their stores and allow end users to upgrade their systems or buy ATI cards pre-install with new computers. Stability means working with the widest range of software without hangs or bugs. This means games and business apps. Resellers and OEMs like bullet proof installs because they do not want to see the units coming back to them. They like ATI for the same reasons that you buyers like them ....see the End User Testimonials.. June 26 is just over a week away for the final OpenGL Quake Driver and associated Win95 driver. I think all of you will consider holding onto your RagePro based cards to experience what I would call a major performance upgrade....you might even call it "Intercooled". There is also a June 22 date for a full OpenGL ICD in Beta form....that supports all the OpenGL calls and all OpenGL applications. The release schedueles are synchronized to that of WHQL's periodic release schedule for Window drivers, so the only thing possibly holding back the train might be this dependency. Discussion on veraious RagePro forums have made assertions that the Quake OpenGL driver and the ICD Full OpenGL driver are related due to various OpenGL screensavers running under the Beta mini-OpenGL released to date. We'll have to wait for the accompanying "readmes" to clarify this. The above is just a reminder of the promised dates, I have no control or inside information (shucks) at this point as to whether the drivers will be available as stated in the ATI web pages. Your end user voice on the FPS Forum is being heard. Getting back to this "Channel Champion" stuff... here is a snippet from CRN:

  • My 1988 Honda Civic Hatchback is still running strong with 240Km on it but I began test driving cars again. I dropped by Honda again and test drove their CR-V and later on at the Saturn Dealer. What I want is basically a bare bones CR-V (manual windows, non-power door locks, 2 Wheel drive) with polymer side panels starting at about $16K instead of a one size $26K sticker shock. The Saturns have been around since about 1992 and are not too shabby if you like their models. They actually use a "timing chain" instead of a "belt", have a user serviceable transmission fluid and filter that does not require dropping the pan, polymer dent/rust resistant panels, stainless steel muffler etc. The low end torque and transmission mesh well together and just felt better than the CR-V unit that seem to run out of torque and revs in the early gears. Last, you gotta drop by a Saturn for one "outta body" shopping experience. They're cool. I'll be dropping by my Subaru (Imprezna) and Suzuki (Imprezna wagon) next weekend to check out their offerings.
  • LinuxFocus interviewed Linus Torvalds in March 98 issue. He talks about Netscape's recent opening of their code to the public domain under GPL, personal non-comments about Bill Gates personal life, Win95/NT being rated as poor operating systems, next Linux release containing real time extentions (toy train controllers, anyone ? ) to Linux and life in general....As my friend Jim said "Where exactly in the world is General ?".
  • June 8, 1998
    The rank of RagePro sites has grown again. Welcome RageOn The site has a nice collection of links to ATI information in the FAQ section. Just for your benefit so that you get the latest news from ATI, here are some links for the following information:
    1. ATI OpenGL release dates
    2. ATI OpengGL Beta release info
    3. ATI OpenGL Email address betadriver@atitech.ca
  • I have started capturing the modifications to the famed NAD 3020 integrated amplifier. Most of the modifications revolve around removing the junky AC coupling caps that prevent DC through the bass and treble potentiometers. I have removed these and rebias the circuit to minimimize the DC offset. What remains is filtered away further downstream. The only downside is a slightly larger turn-on thump but man does the soundstage clear up. I have been using the Chipmunk tools from the University of Berkeley. The circuit failed to converge properly at higher rail voltages and I am sending this file to the authors to help them debug their tools. Here are a couple of reviews for the NAD amplifier:
    1. A recnet review of the classic 3020
    2. NAD 304 review...successor to the 3020
  • Alan of FPS fame made a pretty good point about the important of Video telephony in the form of Netmeeting. This requires support of video and full duplex sound card support.. The idiom that "a picture is worth a thousand words" has long been the cry of the telephone industry. Guess what ? People have shown a preference during casual conversation to not be able to see the person they are conversing with. In business circles, the abitlity to share charts and information argues well for these features. Linux is held prisoner by companies (including ATI) reluctant to release detailed register specifications of certain portions of their hardware that would enable drivers to be developed. These parts are kept proprietary for various competitive reasons until they become commodities at which point the act of releasing re-anables competition in the form of "I have more widepread support on more operating system that Company X". It is like watching a mexican standoff. 3Dfx got around this in 3D circles by writing the Linux Glide interface layers that sits between their hardware and the application. Who else out there will relegate Linux to the recreational opeating system level due to the absence of the above features ? For similar thougths on this, See minutes from the Open Sourxe meeting.


  • June 4, 1998
    Stamp collecting and driver updates are two different activities with similar end results for me. The process of updating drivers almost seems to be a hobby with some computer users. I remember starting my first stamp collection by sending off the coupon in the "Weekend Magazine" included in the Saturday paper. It was a pleaseant surprise that anything came back at all especially for someone 10 years of age. Individual stamps, two of a kind, sets, and finally collections were phases that collectively barely lasted a year. The collections completed seem to never gleam: There was always a heavily marked up one, a tear here, dog ear there. The whole idea of perfection in collecting seemed almost a neurosis of sorts that even myself at 11 began to understand. Fortunately, I discovered a more creative outlest in the form of taking pictures and processing them in the darkroom. It was "neat" in the executing a plan over 5 hours that took turns and twists unforseen in refining something that you could touch and hold. It started off by waiting for nightfall, mixing up the stew of chemicals, working by red safelight with late night hosts not pressured by commercials to comment about sights, sounds and smells here in our Toronto. It was creative and meditative. I hope to again find this state again this year with some new hobbies up my sleeve.

    • I find the task of keeping up with Win95 based enthusiasts interested in running the latest creations in the 3D gaming somewhat like stamp collecting. It is partly to refine that "collection" of programs, applications, and games that sort of justify the purchase something about 5x to 10x the price of a Nintendo64 box. Problem is that just like stamp collecting, there will always be blemishes caused by not enough CPU, crashing, insufficient swap, insuffcient memory, glitchy memory, viruses, due to a combination of very complex software running under a operating system called Win95 which is just basically relatively bad design. Legacy issues bring kludgy code. It works but if you have read these pages carefully, all of the above problems can be solved by running a much better system called Linux. Do something more creative and code in Linux using state of art compilers and GUI interfaces rather than try loading the 3rd version of some OpenGL driver. " Unix for Dummies" is of the famed series has a begineer and an advance version "More Unix for dummies" of these guides. Both form an excellent set. I would be blessed to have a set of these in Xmas stocking even though I probably have enought Unix reference texts. They form a good no nonsense and useable reference/guide. Following the Linux wave, Linux specific versions of these books have just been released this year as well (Different authors...so I do not know how they read). Look under Amazon for this book.

  • Some questions have arisen as to why OpenGL should be tied into Win95 driver releases. Basically, a set of low level libraries provides access to the hardware. The routines can be used to implement the Win95 calls or they can be used to interpret the OpenGL calls. Delays in releasing the final mini-GL driver relate to providing a very fast and solid interface with features (some new) that really make a difference. Muti-texturing is one feature that needs new low level support . Bus mastering is a warranted feature to support in 3D but really does squat for 2D due to the small amount of information transferred. The RagePro supports this feature but the Final Reality scores shows that it is yet to be enabled in any of the present drivers. The next driver release is internally named after a very fast italian car ....so put on your windscreen when you run through your next tunnel in Quake.

    • In the interim,todays machines (> 200MMX) play a very fine game of Quake using present Beta OpenGl drivers...anyone who complaims about motion sickness needs to cut down his play time. Problem is these "kids" have lots of money, full time jobs, and have not been told to cut down on their TV in will over a decade. One piece of advice, forget stamp collecting. Keep your comments coming at the FPS website and I apologize to anyone out there who suffered considerable angst.

  • Last note, I have volunteer for the "BECEL Ride for Heart " bike ride over the last five years as a Road Ambassador ( basically I fix flat tires and pick people off the ground). The city closes down 50KM of freeway for bicyclists to ride on this year on June 7. My mom past away last year due to a stroke in her brain. We have enough arteries and blood vessels in our body to go around the world easily and stokes can be due to either blockage or due to bursting of these pipes. So get away from your terminal and go for a real walk...all the memory, trinity, CPU in the world won't replace the real act of getting out there. Take care and if you live in Toronto, the ride kicks off at 8;00 AM on Sunday June 7 from the Lakeshore at the CNE grounds by the Pricess Gates.
  • DOJ and MS. Most people know that I am not talking about Multiple Scelerosis and Durango Orange Juice. What is it about Microsoft that particularly irks people (like me). I have acquaitances of mine from the University of Waterloo who work there nowadays and if you did not know already MS loves hiring from U of W. The problem is one of both human nature, paranoia, and doing the right thing. Microsoft has done a tremendous amount for the computer industry. They provided an alternative to PC-DOS and made the clone hardware industry what it is today. Back then (pre Windows), they were dreamers and enablers. People have longer memories that five years ....even in the software industry....fortunately for Apple. They had (probably still do) the right mix (call it a team) of people to attend to and nurture the task of tracking down problems and evolving the state of the PC which was highly tied to the operating system. They had the smarts to use HP's Presentation Manager to define the look and feel of Windows and the political will to move away from DOS and embrace Apples philosophy "for the rest of us" by writing an user interface on top of the DOS operating system. In the case of Windows 3.1, the lower level layers were written after the GUI and not the preferred other way around. Win95 actually bore a new operating system with backward compatability to the Disk Operating System.

    • Software companies flourished under DOS, maintained growth under Windows, and stalled under Win95. Microsoft capturedthe mindset of programmers, small business, medium sized business and home users. Now mindset is a very human reaction. Most musicians capture the mindset of people because music is a very creative activity that it polarized by less creative ones people usually associated with robotic activities such as driving on the right hand side, filing tax returns, making $$$ and lawyers. I suspect MS biggest problem is that their monopoloistic "I write and sell the operating the system and applications for robotic users of spreadsheets for the pruposes of tallying $$$ and making $$$" has basically stifled creativity in a very creative area. The mother of invention (a.k.a creativity )they say is necessity. Microsoft has a core group of very creative people (I'm sure) but just like the Allied Forces in the World War, the collective forces of a creative industry will unite somewhere else to provide alternatives for applications and operating systems because creativity requires choice. One of the dangerous assumptions by Microsoft is that I can provide all the choices and that they are sufficient. I look at the code base itself and disk space required and on this alone can see functionally equivalent systems that are more efficient on this basis. I will take simpler and smaller every time. I look at the robustness of design and the viruses alone have created an industry of companies to solve a human problem. Linux/Unix solved this 20 years ago with protected systems spaces isolated from user spaces.

    Microsoft's paranoia about a smaller upstart taking over them is handled by war like policies. Use sactions against hardware companies if they do not follow our market carving install policies. Drop a H-bomb on creative companies by merging applications into the operating release so that the practice of "dummping" (practice of selling at a loss in one area to wipe out competition for the purposes of gaining a monopoly). This practice was used by book publishers in the past who tried to wipe out smaller (more creative?) firms in one part of the world by selling the excess local inventory at a loss in the remote areas that other firms operated from. Do the dead in the post Windows era have to be named to see what DOJ is trying to do: DR-DOS, WordPerfect, Lotus 123, Norton Utilities, OS/2, Borland, etc.
    How about the severely wounded: Netscape, Corel, OpenGL, Internet Bowser applications, untold software development companies
    How about the unfed: Apple applications, OpenGL Win95 MCD, competing application companies needing to know opeating system internals

    What Microsoft has to realize is that they are diverse and capable. They do not have win every battle but be competitive. They do not need gain control via questionable practices but realize that the "fight" is good thing and the a competitor/enemy is actually his best ally. Just like the telephone industry, the DOJ is looking after the general good of the whole industry in terms of providing equal access to competition. Microsoft may view that they are sufficient for the rest of us but man has never stood still for the status quo in creative areas and has strove to make their own music all over the world.

    May News

  • May 17, 1998
    Mercury Research has just released their latest set of Benchmarks. The RagePro is benchmark using the 2312 driver but the significant thing to note that it is still very competitive in 2D (within 10%). This for a 64 bit interface using 100 MHz SGRAM. The 3D performance is still very good under EgoSoft "X" with framerates running at a very useable 50 fps. The latest Riva and Matrox chips which do better, are evaluation cards, but the enjoy an advantage of using a 128 bit memory interface. Unconfirmed rumours about the G200 indicate part of the 3D speedup is due not only wide memory interface but the use of 125Mhz parts that were further overlocked by as much as 10%. The RagePro is about a year old and I wonder where the 3D performance would be if the memory were clocked up around 140MHz from 100MHz using some of the faster memory chips. The Beta test sites for new chips running in evaluation should add comments about the heat and confirm actual chip clock frequencies.

    • John Peddie's Benchmarks for Winbench 98 have been completely outdated for a long while. They contain numbers for 4 chips. One of them is no longer being made (Oak Warp5). An earlier version of this Winbench score had incdluded ATI products but they along with a number of other Graphics chips (about 10 in total) running on a P200 MMX have been withdrawn. These were supposed to run by Dimension3D but I have noticed that a number of links to older reviews no longer exists. It seems as if there may be a conflict of interest with site sponsorship and the reviewing process. Mercury Reasearch has since published two updates to their scores run on faster Pentium II machines. Both sets of numbers should prove useful.

  • Former evangelist for Direct3D Alex St. John sometimes makes me shake my head. He made a comment that a game such as GLQuake contains "absolutely no OpenGL code". I also read his generally vague comments about his ideas about an operating system specifically for games called "DirectOS". It has very little substance and I am beginning to question Boot's decision to give this man a monthly column.

    • Glance at the latest issue of Boot that did a comparitive review of all the major 3D contending chipsets. The review was heavily 3D Quake centric with many screenshots. From the screenshots that Boot showed, it looked like ATI's Alpha release OpenGL that had problems with Z-buffering was used instead of the latter Beta3 release which fixed a host of related problems. I still believe that the best of RagePro's legs will be seen when the next round of dirvers are released as ATI reacts to using benchmarks going beyong just Winbench. It should put a twinkle back in your eye. Regardless, this chipset sets a very good "standard" for what is very useable performance at a reasonable price. It is no accident that year old design continues to evolve with DVD and driver evolution while other chipsets (V1000, VooDoo, Riva128) by other manufacturers have been replaced (V2100, VooDoo2, ZX) or have had their driver development halted. The Rage Pro rages on by offering this level of useable peformance at a lower price point to make it a more mainstream soulution.

  • The Beta release of the ICD OpengGL driver has been delayed. Other than the promise of multi-texutre support for additional speedup, the ICD should enable running a larger spectrum of applications written using the full OpenGL language instead of the GLQuake subset. See my note on using the FULLBRIGHT option under the Quake console to get a sense of the increase.
  • I visited Arizona on business and made a sidetrip to the Grand Canyon. My colleague Pete described it as "one awesome place". We approached the "Grand" from Phoenix heading up "87" until we reached town of "Long Valley" south of "FlagStaff". Pulled off at "260" and headed about 3 miles where we pitched a tent at one incredible campsite. The wool blankets from the Hotel we borrowed turned out to be fine sleeping bags. The rest of the trip was through "wild coyote country". It looks just like the stuff you see in the "Bugs Bunny-Road Runner Hour". Despite the $20 entrance fee to get into the main access road running along the southern ridge of the canyon, the Grand Canyon is not a tourist trap and the town of Grand Canyon is really done up nicely. There are 30 lookout sites between the two entrance points, separated by 30 miles, and if you had to only go to three, I would recommend you come in from the East along "89" and first take the free scenic lookout just before "Deseart View". This gives you a close look at small canyon about 800 feet deep. You'll see white throated Swifts flying about doing F15 dogfights at about 100 mph. Really Cool. Head east and make sure you take a look at Deseart View, Pippan(have lunch here on the rocks) and Angel Bright(go for a hike down into the canyon) in that order. You will not be sorry.

  • May 8, 1998
    May issue of LinuxFocus available. More articles on OpenGL, NT vs Linux, 3D rendering, StarOffice (Office -like suite) running under Linux, and more...
  • XFree86 3.32 has been released for a while (since march 5, 1998). It is no longer necessary to add the ChipId or ChipRev statements to your configuration file to get AGP flavour versions of the Xpert@... cards to work.

    • Marc Aurelle La France of XFree86 sent me a note about Mach64 chips with TV-out not starting up correctly if the TV is plugged in. This is true and has to do with the nomral clock frequencies generation scheme (used in the X11 Server) being bypassed. He is still awaiting information on disabling this functionality altogether. In the mean time, you will have to unplug your TV connection to your PC is you wish to boot up Linux.