   |
|||
|
|||
| :
News : Reviews : Editorials : About : Contact : Advertising : Privacy : Links : Forums
|
|
|
|
|
|
|||
|
|
|||
|
|
|||
|
December 2, 2004 Corsair Micro XMS 3200 XL Pro 2 x 512 MB Review - Josh Very few people talk bad about Corsair, and that is for good reason. Their products are that good. Lifetime warranty, excellent performance, and high overclocking potential are some of the hallmarks of Corsair's products. The XMS 3200 XL Pro's are no exception. With built in activity LEDs and solid stock and overclocked performance, the XL Pro's lead the way as Corsair's flagship product for the DDR market. Here is a quote:
You can read the entire article here. NVIDIA Product Ponderings and Other News - Josh GamePC released its MSI GeForce 6800 PCI-E review, and they had a very interesting picture to show. As nearly everyone knows, NVIDIA has been using the NV45 (NV40 + HSI bridge chip on one package). The NV45 is very distinct from its other relatives due to the little HSI chip below the main chip. As you can see in the GamePC photo, that product does not have a HSI chip. After some digging around, it becomes apparent that the chip in question is in fact one of the first NV41's to hit the market. Obviously NVIDIA is slipping these in very quietly, but for reasons I am not quite sure about. Perhaps NVIDIA merely wants to give its current GeForce 6800 partners a chance to sell off their stock of parts based on the NV40 chips. Not much is known about the changes on the NV41, but what is known is that it will only physically have 3 pixel quads and 5 vertex shaders. The NV40 used on the GeForce 6800 cards previously had all the hardware present, but they were disabled through the video card BIOS. This may come as a letdown for those hoping to activate those components with tools like NVStrap on their new GeForce 6800 PCI-E. So, it is nice to see that NVIDIA has this chip up and running successfully, and hopefully they will not run into the yield problems they initially had with the original NV40. I am thinking the redesign/respin of the NV41 may have improved yields from a design standpoint, and hopefully NVIDIA has "fixed" the video engine integrated into the NV41 (the NV40 had some real issues with its video engine). Now, on the other hand we have The Inquirer claiming that the NV48 is cancelled and will never see the light of day. I don't really know where to stand on this issue, as I can easily see it going either way. If the NV48 really only fixed a couple of problems in the NV40, and added PCI-E, would it really benefit NVIDIA that much to have two separate lines of parts that address the same market? Yields have improved on the NV40, and further revisions have enabled higher clock speeds, so is it really smart to have a NV48 and NV40 line when NVIDIA could easily just keep the NV40 line and combine it with the very inexpensive HSI bridge chip? Well, the optimist in me says the NV48 is still alive and will grace PCI-E slots at the beginning of next year. The realist in me says that NVIDIA will can the NV48 and move those engineers onto other next generation products and wait for the NV47 to be released in the Spring of 2005. Too bad I am still waiting on new forums, because this could be quite interesting to discuss, especially in light of ATI's release yesterday. In other news, 3D Chipset has had a total makeover on their website. Things are much cleaner, though I think the previous design might have been a bit easier to read (I like white backgrounds for text, but that is just me). Congrats to Solomon and the gang for buckling down and updating their site (something that I am in the beginning stages of doing). December 1, 2004 Radeon X850 Day - Josh Even though I didn't receive any X850 samples for testing, I have gathered enough about these parts to give a pretty decent overview of why exactly these parts were released. While the clock speeds are not significantly higher than the previous generation of parts, these new cards promise to be much more available, with a larger selection of products across many different price points. While this isn't quite a homerun for ATI, it seems to have been a very necessary step to continue to make money off of the R4x0 architecture. Here is a quick quote:
You can read the entire article here. Here are a few other previews for your viewing pleasure: November 30, 2004 News of Interest - Josh The web certainly is livening up for this holiday season, quite a few new and interesting articles out and about, so lets get started! Crucial let me know that they are releasing a new and interesting product in the form of the Ballistix Tracer DDR SDRAM DIMM. This uses the latest Micron DRAM chips that have been shown to run very nicely at 200 MHz up to 300 MHz (depending on the setup) with some amazing timings at every interval. These DIMMS feature activity LED's (though they work much differently than Corsair's products), as well as blue LED's that light up the bottom of the DIMM (think Need for Speed: Underground and the neon lights under the car). These should add a lot of fun to any case with a window, and the performance of these DIMMS appear to be quite nice. You can read more about them here. MS over at Lost Circuits has taken a look at the (un)state of 3D. He even mentions this site. Amazing. Some good info on there about a few of the major topics hitting the 3D market, as well as going over previous decisions and how they have affected us today. A solid read for anyone interested. Tech Report, with the always witty Scott Wasson, has the NVIDIA SLI enabled Asus A8N-SLI and see how two 6800 Ultras and two 6800 GTs perform in a variety of games and benchmarks. A very good writeup and overview of the technology. Tweaktown takes on the GeForce 6200 and compares it to its nearest competition, the PCX 5900 from NVIDIA and the X300 product from ATI. Overall this could be a very solid little performer for the budget user and gamer, though they have been somewhat rare in the marketplace so far... Continuing on with the video card reviews, Techniz has the Powercolor X700 Pro 256 MB board on the testbed. The X700 Pros have been hitting the market in good numbers as of late, and anyone could easily pick one up at a variety of online shops. This midrange card is a definite bargain, and it adds some new features to the ATI lineup (since it is based on the technology in the bigger X800 series of chips). So this means PS 2.0b is supported, and with a driver trick it can do geometry instancing. We are now starting to see some very solid heatsinks for the Athlon 64 market, and 3Dxtreme takes a good look at one of the best. Thermalright has always made fabulous heatsinks, and now they have turned their attention to the Athlon 64 market. The Thermalright XP-90 looks to be another good heatsink from the company, but it is really still a first generation product for the architecture. More Memory News - Josh Corsair, one of the nice advertisers on this site, has let me know that they have a new memory configuration application on their site. They have thousands of systems and motherboards in the database, and can quickly let the user know which DIMMS would fit. Having used Corsair memory for years, I can attest that they make some of the best stuff around. It makes it very easy for me to offer them advertising, because I believe in their products. Their stuff is definitely NOT crap. I recently purchased a pair of the Mushkin PC3200 Level 2 Version 2 DIMMS, and I am having some issues with them. First the SPD timings are really loose at 200 MHz (supposed to be 2:2:2:6, but instead are at 2.5:3:3:8). In two out of three motherboards that I have tested these on, they do not run at the advertised timings. This could just be a mistake on Mushkin's part, but I would be very curious if any of you out there have bought the L2V2's and have had a bad experience with them? Don't get me wrong, Mushkin has been another very high quality memory producer, and their warranties are simply the best in the industry (not to mention their very favorable RMA policy towards its customers). I have rarely had Mushkin memory die on me, or not run at the specified timings, so this is new to me. If you have any anecdotal evidence either for or against the L2V2's, please let me know. November 29, 2004 AMD 90 nm Athlon 64 First Impressions - Josh AMD was very generous to me this Holiday Season by sending one of their latest 90 nm based Athlon 64 3200+ Socket 939 processors (Winchester core). So far I have been VERY impressed with this little piece of technology. Now, I have been using Athlon 64's since the introduction of the first 3200+ Socket 754 processor, and they have all worked as advertised. Each one of them did run significantly cooler than the previous Athlon XP's that AMD had produced up to that point (mostly due to a more optimized architecture and the use of the 130 nm SOI process). When touching the stock heatsink while the processor is under load, the fingers did get warm, but I was never uncomfortable or felt a burning sensation from the heat generation of the CPU. This latest processor that AMD sent me puts out nearly no heat at all. After running the processor at slightly overclocked speeds, as well as running it overnight in a repeating 3D application, the stock heatsink is absolutely cool to the touch. The heatsink feels almost exactly as it does when the computer is off. While it isn't at room temperature, it is hardly even warm. While I was somewhat expecting this, I am actually surprised by how cool the processor runs with just the stock heatsink. This particular number is reporting in between 35 and 36 degrees C at load on the Gigabyte GA-K8NS Ultra 939 motherboard (which in itself is a interesting, if somewhat flawed, product). I am looking forward to seeing what this processor will do once I apply a bit more juice to it, as well as let it truly burn in. Currently the Athlon 64 3200+ 90 nm core runs at 1.4 volts, as compared to the 130 nm versions which run at 1.5 volts. Again, this is with the stock cooler and stock thermal material, so nothing really high end was used to achieve these initial results. Some people have been able to get more than 2.5 GHz out of these cores fairly easily, with no real heroic measures involved (as in watercooling, liquid nitrogen, etc.). I have a couple of sets of memory that should be able to do 250 MHz and beyond, but I am waiting back on word of some errata I found with a certain set of DIMMS. I will probably be doing a full review on this processor, and everything that it entails. So far AMD has a big winner on its hands with the 90 nm cores, as they offer potentially better yields over its larger 130 nm brother, as well as better heat and power characteristics. Not to mention that AMD can stick far more dice per wafer with 90 nm, and therefore can capture more marketshare from Intel by being able to supply more processors than they were previously able to. Another aspect to consider is that pretty much every wafer that AMD puts out now features strained silicon, which will also help in overall performance as well as power consumption/heat generation. One other thing to mention is that the Winchester cores are slightly higher performing than the older Clawhammer and Newcastle cores. It seems that AMD did some optimization to the design when they shrunk the chip. This performance increase is really only about 2% to 3%, but it is there. Now, this is not to say that AMD's 90 nm SOI process is better than Intel's 90 nm process. In fact, Intel's is probably better performing (and much more mature). The main weakness of the Pentium 4 (Prescott to be exact) is that the design calls for fast transistor switching, and that takes up a lot more power and generates a lot more heat. Add to that the complex and relatively large design of the Prescott, and you can see why Intel is having the issues it does. AMD on the other hand has a smaller core that runs much, much slower than the Prescott, so the transistors do not require the power to switch at that speed. Less power consumption of course leads to lower heat generation. I imagine that AMD's 90 nm notebook chips show further power and heat enhancements (as those processor batches are fabricated with low power consumption in mind as compared to the desktop chips where strong yields are they key). Overall I am very impressed at what AMD has done with the Athlon 64 design, and how well they have transitioned to the 90 nm process. While it took them longer than Intel, they don't nearly have the financial resources that giant has, so they have to do things a lot slower and a lot smarter. Now that AMD has passed over this hurdle, they can now work on getting their processors faster in clock speed, as we expect to see a 90 nm 2.8 GHz Athlon 64/FX in mid-January. I am very, very curious to see what Intel will be doing at this time... November 24, 2004 Thanksgiving and Other News - Josh Well, it is that time of year again in the US where families congregate and stuff themselves with really good food (well, at least those of us who are in country). I would like to wish you all a safe and happy holiday, and I pray that those fighting and supplying those in Iraq will also be safe. My brother Justin is a C-130 pilot, and he is currently serving over in Iraq, so this has special meaning to my family. We hope that everyone drives and flies safely, and to dodge any bullets coming your way! AMD just sent me a 90 nm Athlon 64 3200+ Socket 939 processor, so I will be investigating the thermal characteristics of this little number, and using it to do a review on the recently released Gigabyte GA-K8NS Ultra 939 Edition. This is a very solid looking motherboard with a nice set of features, but there are some potential problems and downsides to the design that Gigabyte chose to take. Some things on this board just leave me scratching my head... I am also trying to finish up a Pentium 4 3.4 GHz/i875 overview in regards to current applications and games. This appears to still be a very solid choice for the budget enthusiast, but more on that later! Right now my family and I are poor as church mice, so if you know of any possible ad revenue we could take advantage of, please let me know! If you feel that this site has been of service to you, and feel like helping to support it, please click the "Make a Donation" button to the left of the screen. Anything is welcome! The good news is that I may be doing some part time work as a Fed Ex driver (I always enjoyed receiving packages from Fed Ex, so maybe this could be a fun job for a while!). MS over at Lost Circuits has taken a very detailed look at the Prescott/i925XE combination, and came up with some very interesting findings! He did some very good investigative reporting with this combination, and discovered some really interesting errata concerning the setups. Truly a must read for those looking to upgrade to these types of parts. Hot Hardware reviews the MSI RX800XT-VTD256E video card. MSI of course includes a fabulous bundle with their product, and it seems to be a very solid card. Solid Holiday reading! The infamous Tech Report (stepping on toes everywhere, and loving it!) has the Shuttle XPC SB77G5 barebones on the table. This product is a socket 775 based system that utilizes the very fast and mature i875 chipset. While it doesn't have the PCI-Express loving that many would expect, it is a solid and wonderful product for those interested. Why does TR always get the good stuff? Anyway, that is all for the next few days as I will be heading out of town and away from my computer. Have a Happy Thanksgiving! November 19, 2004 NVIDIA and Intel - Josh It has finally come to pass that NVIDIA has signed a cross licensing agreement with Intel to develop chipsets destined for the Pentium. Yes, nForce is coming to an Intel platform near you. The history between Intel and NVIDIA has always been somewhat friendly, but otherwise cold. NVIDIA worked with Intel to develop the X-Box chips, which feature the GTL+ bus from the Pentium !!! processor to the X-Box controller/GPU. Due to the fact that Microsoft actually licensed the bus, NVIDIA did not have the right to develop Pentium !!! and Pentium 4 class products. This is about to change. While the exact details of the agreement are not known, apparently there is a trade of technology here that Intel may profit from. The first chipset is expected in Q1 of 2005, and it will have a host of features to it. SLI NVIDIA style will make it to the Intel platform, and the new chipset will support DDR-2 (though at faster speeds than current controllers run), and a full featured set of southbridge functions. There are continued rumblings that in fact SoundStorm is not dead, but I am pretty certain that the engineers that worked on SoundStorm are now assigned to different areas. What we will probably see is an nForce based Azalia sound controller, but I don't think we will ever see a full hardware SoundStorm again. Perhaps the prospect of a high speed SLI platform helped nudge Intel in the right direction, but nobody can be really sure. We all know that Intel makes some very nice chipsets, and you would expect their engineers would have been able to implement a SLI platform (that would perform better than the current Tumwater chipset- the dual Xeon that can support 2 x PEG). We will probably never know the exact details of the agreement, but the ability to provide Intel based chipsets has opened up a huge market for NVIDIA. In the past several years NVIDIA has had a great reputation in the AMD market, and that will pass over to the Intel market rather quickly. I will be very curious to see what all NVIDIA will bring to the table for the Intel market. You can read the entire press release here. November 18, 2004 ATI's X800 XT PE Availability and other Observations - Josh Scott over at Tech Report has started a very interesting thread about Radeon X800 XT Platinum Edition availability. Now, the high end cards from both ATI and NVIDIA have been pretty scarce, and when they are available they are listing for EXORBITANT prices (not to mention the abuse the wallet gets when buying one on Ebay). I have several wholesalers that I do business with on the side, and only once did either of them carry a X800 XT PE, and that was shortly after the initial X800 release. Throughout the summer it rarely showed up, and it has not been there for the past few months. On the other hand the NVIDIA GeForce 6800 Ultra has been fairly scarce, but for the most part many distributors have it listed and available (but again at a very high price). The 6800 Ultra Extreme edition was essentially brought out for reviewers to compete with the X800 XT PE, and literally only a handful of these were sold. Several major OEM's have offered the X800 XT PE for sale either by itself or in a system configuration. My understanding is that many of these OEM's are no longer carrying this product or bundling it with any configuration. Apparently the reason for this is that ATI really is no longer ordering these parts from TSMC. Sources have claimed that the yields on the X800 parts were pretty poor, and that only the X800 Pro is available in good quantities. There have been X800 XT's in decent supply for both AGP and PCI-E, but these were not as common as the X800 Pro's. ATI appears nearly ready to release replacement products that are based on respins of the R420/423 on both TSMC's 110 nm and 130 nm Low-K processes. The more mainstream parts will of course be the 110 nm products, while the high end stuff will be 130 nm Low-K. Sapphire is currently selling the X800 SE 256 MB, which is a 8 pixel pipe, 6 vertex shader product running at 425 MHz core and 400 MHz DDR (800 MHz effective) with a 256 bit memory bus. Overall this card should perform slightly higher than the 9800 XT that it replaces, and should more adequately compete with the GeForce 6800. The fact that it has a full 256 MB of memory certainly makes it a bit more palatable than the 6800 at the $300 price point. The ability to overclock this card to at least 475 MHz is probably also there. Some users are reporting OC's up to 510 MHz, which gives about a 30% increase in performance over the similarly priced 9800XT. This looks to be a solid product for the $300 price point. On NVIDIA's side, it appears that IBM is not giving NVIDIA the capacity that they probably need, though the yields have certainly improved since the beginning of this summer. We now see a very good supply of 6800 GT cards, and a decent supply of 6800 Ultra's (both are in better supply than the X800 XT's and PE's). NVIDIA will also make the transition to 110 nm with these products, but I am unsure if either will be rebranded or if NVIDIA will silently slip these new chips into current SKU's. Interestingly enough, there is actually very good availability of the new GeForce 6600 GT AGP video card, and quite a few online retailers have them in stock. As of today there have been no X700 XT's on the market, but a good selection of X700 Pro's (though only in PCI-E form). So far ATI has not released a AGP version of the X700 series of cards. So it seems that this holiday season will be dominated by the GeForce 6600, 6800, 6800 GT, and X800 SE and Pro. Once AMD finally releases good quantities of PCI-E motherboards (as NVIDIA, VIA, and SiS are now shipping large quantities of chipsets supporting this), we can expect to see greater PCI-E availability of higher end cards. For example, there have been very few 6800 GT and Ultra cards in PCI-E form on the market, and most of those are sent to OEM's and reviewers. ATI has been doing a very good job in providing higher end PCI-E parts, but these are still in short supply. Once December hits, I think we will see much better availability... but it may be too little too late because most of the holiday shopping will be done by the middle of the month. Q1 is always a very soft time for the industry, but typically not as bad as Q2. So we shall see how these guys address the market with new products. From a financial point of view, NVIDIA is being much more conservative with their product placement than ATI is. So far ATI has had to tape out at least 2 versions of each chip (in AGP and PCI-E forms), while NVIDIA so far has only done one version of each chip (and uses the HSI chip to give either PCI-E or AGP support). Now, ATI will be following the same path, as they are about to introduce a PCI-E to AGP bridge chip. I would imagine the first product we see with this bridge will be the X700 AGP series of cards. Once the new high end chips hit the market, they will also be paired with the bridge chip to give AGP support. So far it appears as though NVIDIA has wasted less man-hours and money by implementing their HSI so early in the development cycle, while ATI has wasted many hours and dollars to release two versions of each chip.
If you have found this article interesting or a great help, please donate to this site.
Copyright 1999-2004 PenStar Systems, LLC. |
|||
