|
|
| News Link » /news/newsarticle.asp?searchid=15016 |  Power666 (25) Apr 16, 2007 - 01:38 am
| | I know that this entry falls into the category of bad day for me. I have no idea how I even goofed on the tags in the title. Half of what I written for the article only made it into the final piece. I had an ancient Radeon 8500 laying around that I also flashed but ran out of time to do any sort of bench marks for. Having some sort of comparison would have been nice as well as comparative gaming benchmarks despite using ancient games for testing. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/435 | Power666 (25) Apr 10, 2007 - 02:23 am
| | I have a copy of Quake 3 around here some where but I just couldn't find the CD. I downloaded the DOOM3 demo for Mac and it was a bit of a slide show on the Radeon 9800Pro. I would have included some benchmarks with it but I couldn't get a time demo to work for benchmarking (I don't think the demo included any). Then there is the huge time factor that I was chronically short on. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/435 | Power666 (25) Apr 07, 2007 - 10:39 pm
| I haven't tested Graphiccelerator for flashing ATI cards. At the very least, one would need another video card to use Graphiccelerator for the flash.
I've heard of a way to flash an nVidia card in a Mac with a Mac ROM but that requires mucking around with OpenFirmware a bit. Looking from the instructions for that, it appeared easier to flash the nVidia card in a PC than the Mac. Using a PC for this wasn't troublesome at all and of course gave me the chance to test the card out before flashing it. I got the GeforceFX 5200 for free from a friend awhile back. It has been awhile since I used it so I made sure it worked in the PC before modifying it. Flag this | Edit this post |
| Cluster Shout Box Link » /matrix/cluster.asp/22 | Power666 (25) Apr 07, 2007 - 01:11 am
| I actually cut over half of the article out before posting. I flashed an old Radeon 8500 as well but didn't have time to run any benchmarks on it for comparison. The article would have made a lot more sense with another video card of comparable speed but lacking the features necessary for CoreImage and Quartz 2D Extreme.
I also wanted to try out the video card drivers in OS X 10.4.9 for Intel in my PowerPC based system. Oddly enough, Apple compiles all their system software in a universal binary format so they'll work despite being labeled for Intel based systems. There are reportedly speed differences and even different feature support. Flag this | Edit this post |
| Cluster Shout Box Link » /matrix/cluster.asp/22 | Power666 (25) Apr 05, 2007 - 01:30 pm
| | I made it, barely. I actually have more content that I've cut out from my article than what I've included. I really do think my original goals for it were to optimistic with including a flashed Radeon 8500 for comparison. I was also hoping to try out the trick of using the video card drivers from OS X's Intel distribution. The video card drivers there are universal binary and work on PowerPC and x86 based Macs. The Intel drivers are reportedly faster and support more OpenGL features. Flag this | Edit this post |
| Cluster Shout Box Link » /matrix/cluster.asp/22 | Power666 (25) Mar 28, 2007 - 11:05 pm
| | I almost feel like my only option here is to review something as ancient as a 386. I literally haven't had the funds the past few years to keep an upto date PC gaming system. Money is always tight when you've just finished college, finding yourself a woman and moving from city to city. I'm really scratching my head to find something here relating to CPU's or GPU's that people would actually want to read. Everything I have is just ancient or outside of the target audience of this site. I really don't want to do a bland generic review of a CPU or GPU that could find dozens of reviews for years ago. Considering the dead line, I may just have to do just that. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/12941/378 | Power666 (25) Mar 26, 2007 - 02:14 pm
| I've had my share of negative experiences with Linux. For example I consistently had openSUSE 10.1 crash on me during the installation process. The issue at had was that it was running out of memory. The machine had 384 MB of RAM and bringing it to 512 MB solved the problem, but I was hoping for something a bit more lean. I know that virtual memory hadn't been enabled but you think just for the installer, it'd be fine with 384 MB of RAM.
For data center and server work, Linux is great and I only see Linux improving in this area. I see a market for smaller distributions leanly cut to run one or two applications using virtualization.
That type of Linux use is for professionals not the average user. Linux does have its merits but until the set up and usability (like dual monitors etc.) issues resolved it will never be a replacement for Windows for the average user. Flag this | Edit this post |
| Cluster Shout Box Link » /matrix/cluster.asp/15 | Power666 (25) Mar 17, 2007 - 08:08 pm
| Getting the word out was something I didn't do for my last article and I wanted to change that for this one. I didn't know BluesNews had picked this up when I looked. I submitted mine to Joystiq and Kotaku as well but they didn't bite. I made the front page of Xlr8yourmac and subsequently MacNN and MacSurfer as those sites recycle heads from other Mac sites. I also got the news out at a few popular forums I hang out.
It wasn't my idea to submit my article at the last possible second. I got hit by a hard drive failure on the PC which required me to spend time taking part another system, pulling the drive, wiping it, fresh install of Windows and downloading all the game demos. While this was going on, I was able to do some Mac testing. I honestly didn't think I'd get around to but it has paid off with word spreading in that community. I knew I'd be a bit of a dead end if I didn't include Windows support so I ended up testing and writing about those drivers until the dead line. I would have loved to have tested GlovePIE for Windows and some drivers on a Linux box but simply ran out of time. The included table is a left over from that idea.
I know it is nerve racking to know if you're going on to the next round. With so many good articles it simply feels like a 50/50 chance. The thing that drove me nuts is that with in 3 minutes of posting I had gotten my first review with a 6. I knew the guy didn't even bother reading. I knew my article would welcome some unrealistic votes due to the whole Wii/PS3/X-Box 360 fanboyism on the net but that was just depressingly fast. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/318 | Power666 (25) Mar 15, 2007 - 11:19 am
| The target audience and the overall purpose of this article change while doing testing. I was hoping to get a game play experience of what the Wii should have provided for Red Steel or Far Cry on that system. I was hoping to report plenty of games working and give a positive review of gameplay but as I did more and more tests, the more I released that gaming wasn't a realistic solution. To start of testing, I was either in Windows Explorer or the Finder with the Wii remote generally functioning. Thus I decided to include some media center notes as this is something those drivers can do reasonably well. Many of these drivers are open source and making them appear as complete trash isn't going to be supportive of the community given that do have limited functionality now. I'd consider this something like a good article around a bad review. As I said in my conclusion, a few more revisions can easily make the Wii Remote and these drivers something worth using for PC gaming.
For the media center functions, I was thinking of a system hooked up to a HDTV using DVI or HDMI. Though honestly the computer set up used a 20" LCD display and a couch about 6 feet away in my relatively small apartment. Just getting the sensor bar to stay. In hind sight, I probably should have these details.
I did make note of the physical use of the Wii remote with RMX automation because it was so unintuitive. I would have loved to have had a camcorder handy to record exactly what I was doing to better demonstrate the issues I had. I have a digital camera but I've cursed with a missing card reader. :( Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/62192/281 | Power666 (25) Mar 15, 2007 - 08:15 am
| Good product review. I think there is an error in the second paragraph mentioning a 1680 x 1505 resolution. Wouldn't it be 1680 x 1050 as seen else where in the article?
Do you know who makes the battery? With a recent spree of exploding batteries, this information is worth passing on. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/61942/277 | Power666 (25) Mar 11, 2007 - 10:47 am
| Whoops, missed that little comment about the dangers of running RAID 0 at the end. It was nearly 4 AM here.... :P
The idea of bundling a bunch of small files into one large file is used by developers in shipping games. I know id's .pk4 files for DOOM3 and Quake 4 are effectively renamed .zip files.
Windows Vista is supposed to do more caching of files being moved to/from a hard drive. As for an individual driver that does this extensively, I don't know of any off hand. This isn't a flaw with your article but more of a gripe with hardware providers.
Finding a RAID driver that will do RAID 1 style writes and RAID 0 reads for both the performance and data integrity is rare. Highpoint has released some products with support for such an implement can called it RAID 1.5 (not to be confused with RAID 15 or RAID 1+5). I think Highpoint dropped support for RAID 1.5 with all there recent products however.
I think FEAR will drop fps down to zero when automatically saving at check points. Not sure if FRAPs is sensitive enough to record any difference between using RAID 0 and not. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/61942/277 | Power666 (25) Mar 11, 2007 - 01:54 am
| Well written article but a few things that I'd like comment on and to see in its content.
RAID 0 and 1 is relatively common feature on motherboards today and thus the additional cost of the RAID controller is near zero. RAID 0 with two hard drives creates a signal volume of greater size. People often configure a system with two hard drives just for the additional storage space. The performance gains from RAID 0 in the real world may not be that much but the performance/price ratio of RAID 0 for some will be exceedingly high as they already have the equipment.
I didn't find mention of the risks of running a RAID 0 array. A single drive failure will make the data across the entire array worthless. Back ups for a RAID 0 array are doubly important due to this. I'd consider the risk of data loss a more deciding factor than cost.
There are various cheats used in games to get around the massive number of small files by bundling them into one big virtual file on a hard drive. With any luck, that one massive file will be placed on the outer edge of a hard drive platter for maximum speed. Regardless, the access time for each of the small files in the bundle will be relatively the same if the bundle is not fragmented. Various caching into system RAM could help out a lot here but that is up to the OS and driver.
One thing missing is how block size effects performance in RAID 0 arrays. RAID 0 arrays don't split data continuously into two sections. RAID 0 will write a block of data on one hard drive and then another equal sized block of data to another hard drive. Having a large block size is great for large files but if the average size of files are less than the block size, almost no performance benefit will be had. A small block size splits up the data more evenly for small files but more commands have to be issued to the hard drives and thus increasing latency a bit. There are a few real world scenarios where this would have an effect such as copying folders of data.
I have no idea what games the author has available to test on hand, but there is a way to factor out stop watch tests for benchmarking. Simply find a game that hits 0 fps in FRAPs when doing automatic state saving or on the fly content loading. If the amount of time spent at 0 fps may vary according to running RAID 0 or not.
I'm still waiting for hardware and drivers what will allow a true hybrid RAID 0 and RAID 1 array for maximum performance and redundancy. Writes are performed on all drives simultaneously but reads are performed in an independent parallel multitasking nature. In other words, two files can be read simultaneously without any performance effect. Alternatively, different portions of the same file can be read simultaneously like how RAID 0 functions already. Sadly, finding a hardware/software configuration that can do this is exceedingly rare.
The charts do need a legend. The axes are labeled clearly but I don't know what each color represents. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17473/261 | Power666 (25) Mar 07, 2007 - 06:35 pm
| | I'm wondering about this as well. A hardware review would probably fly as its effect in most cases can be measured (frame rate, CPU utilization ect). I'm really wondering if a good game review would be able to fit. The main focus of a game review typically focuses on the content in the game, not the hardware it is running on (though it is always worth mentioning what the game is running on, if only briefly). Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/205 | Power666 (25) Mar 05, 2007 - 10:38 am
| I know I goofed big time with the date on the Fusion announcement. I'm curious if you found any other inaccuracies.
I couldn't find any solid information on what nVidia and Via are doing to counter AMD's Fusion project. While they are certainly going to be doing something, exactly what is pure conjecture and dives a bit too far into the realm of rumors than I'd like.
IBM's current SoC designs are limited to embedded markets which have different needs than the laptop/desktop market that Fusion is targeted for. Intel and AMD are moving towards a SoC design that is truly a jack-of-all-trades implementation. The goal for them is to create a single low power component for a single, general purpose user system like a laptop. I also see a divergence between consumer centric devices and server centric devices the more the companies inch towards complete SoC designs. Simply cutting features off a high end design to make a low end part isn't going to be practical as more server specific technologies get implemented on-die.
The big IBM break through in eDRAM announced a few weeks ago isn't a revolutionary improvement in transistor density. Rather they're replacing transistor heavy SRAM with the inherently dense eDRAM. The three fold increase in cache size with eDRAM is due to eDRAM using less than a third the transistors SRAM uses. My mention of eDRAM was more akin to how that technology assists limited memory bandwidth. I don't see the actual capacity of the eDRAM being much larger than the necessary frame buffer sized used in laptop displays. 1280 x 800 displays are currently common in low end notebooks today and they wouldn't need more than 10 MB eDRAM. 1920 x 1200 displays in laptops will likely be much more common when Fusion arrives and that'll only need 20 MB of eDRAM to hold the frame buffer and z-buffer. It will be the desktop versions of Fusion that will need massive amounts of eDRAM for gaming if the memory bus is lacking bandwidth. 32 MB barely holds the buffers for a 2560 x 1600 display and their will certainly be higher resolutions available when Fusion arrives on the desktop. Running multiple displays in 3D requires even more eDRAM. Varying the amount of eDRAM would be a simple way of distinguishing between the low end and high end versions of a chip just like L2 size does today.
I went with two Radeon X1950XTX's for the teraFLOP figure as they're products people can purchase today and something readers of this site are familiar with already. It would be kinda disappointing if the R600 chip didn't break ATI's old record but the public still has to wait a few months for it to arrive.
While a Fusion chip would dissipate more heat than just a pure mobile CPU, it will still be a laptop part and have to fall under the same thermal ceiling. Having a larger die also aids in heat transfer with surfaces emitting the same total amount of energy. This also allows for system builders to implement one larger, more efficient heat dissipation system than having to utilize two smaller ones for a CPU and a dedicated GPU. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/205 | Power666 (25) Mar 02, 2007 - 10:10 pm
| Thanks. I remember reading the PS2 Technical Overview ( http://www.firingsquad.com/hardware/ps2tech/ ) shortly after it was posted six years ago. I really enjoyed the depth that the article went into details. Over the years though, FiringSquad has been slowly creeping away from making the under laying technology the focus of an entire feature. Nowadays only a few paragraphs are spent at this level shortly after the beginning of a product preview/review. Even if I'm not writing such articles for FiringSquad, I'd like to see more of them posted on this site.
nVidia is a company that knows computing trends rather well but has been absolutely silent on any counter to this. Even your typical rumor mill sites are dry on information. I'm pretty sure they could enter the x86 CPU business if need be as nVidia does have an interesting cross patent program going on with Intel.
The only other thing I can see nVidia doing is some ARM system-on-a-chip business for various OEM's. The ARM architecture is a bit unique in how it handles processor extensions. Integrating a single shader core to act as a FPU coprocessor with an ARM core wouldn't be as complex as a x86-GPU hybrid core. Backwards compatibility between generations would be an issue but that isn't much of a concern with the embedded ARM market. nVidia has already scored a lucrative deal with Apple for a embedded video chip due for delivery over the summer. Coincidentally, the iPhone is set for release in June but I wouldn't take a bet either way that nVidia is powering the iPhone. Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/205 | Power666 (25) Mar 02, 2007 - 09:52 am
| Well, I've decided to look this up and have come to the definitive conclusion that the English language sucks.
This has gotten the attention of my curiosity so I decided to simply look it up. I quickly found a source that says that apostrophe s is acceptable ( http://wsu.edu/~brians/errors/acronyms.html ) but does note that many consider that use improper. It wasn't difficult to find three sources that do not use apostrophe s to indicate the plural form of an abreviation (http://bfa.sdsu.edu/editorial/section2.htm http://bfa.sdsu.edu/editorial/section2.htm and http://grammar.ccc.commnet.edu/grammar/plurals.htm ). It is rather odd that academia with professors possessing Ph.D.'s/Ph.D.s/PhDs cannot agree on the plural form such a degree takes.
I'd like to point out the CCC source again ( http://grammar.ccc.commnet.edu/grammar/plurals.htm ) as a reason why I use apostrophe s to indicate plurals with acronyms and similar special cases. Take this example from that source: "I have prepared 1099s for the entire staff." I assume that the example is in reference to IRS tax forms. The problem here is that there is a 1099 form and a distinct 1099s form. Thus apostrophe s could be used to distinguish between those two forms and the lack of possession can be derived from context.
It is easy to see that their are different accepted forms for making acronyms plural including the one I used in my submitted article. Considering the context and the heavy use of acronyms the intended audience sees, it makes sense to me to use apostrophe s to avoid acronym confusion. Then again, how often do the mechanics of the English language actually make sense? Flag this | Edit this post |
| Matrix Blog Link » /matrix/blog.asp/17874/205 | Power666 (25) Mar 01, 2007 - 07:58 am
| Man, I have no idea why I listed the year 2001. I showed this article to a few people before posting and it slipped by them as well. I guess it is one of those things that you just pass over when you're going over the material. Unfortunately, everything has been locked for editing so I'm just gonna have live with it as is.
There have been several highly technical articles like this posted on FiringSquad before so I didn't think an article on this topic was out of the scope of the site's readers. The NVIDIA GeForce 8800 GTX/GTS Performance Preview here dived into such detail to explain the architectural advances that the Geforce 8800 series made compared to other GPU's. While several years old, the PlayStation 2 Technical Overview article here is arguably more technical than the piece I've submitted. I remember hearing about how game developers were using that article as a reference for actual PS2 game development. Diving into such technical detail is going to be necessary as CPU's and GPU's slowly merge together. The more they become integrated, the harder it is going to be avoiding such a level of technical detail. Probably the greatest test I gave this article is letting my girl friend read it and having her understand it.
With regards to apostrophes, it is my understanding that you add an apostrophe to make acronyms or abbreviations plural. The reason being is that a plain 's' could then be considered part of the acronym or abbreviation. Flag this | Edit this post |
|
|
|
|
