We have 22 guests online

Login Form



Remember me
Password Reminder

nVidia GeForce GTX 280
Written by Michael Schuette   
Aug 05, 2009 at 03:52 PM

As one of the “Major 2” GPU manufacturers, nVidia is in a fierce competition with ATI/AMD, both in the mobile and in the desktop market segment. Both companies have focused on the same strategies, namely, massive parallel processing of streaming data on the chip level and multi-GPU processing for additional scaling of performance. Needless to say that there are the “lesser” factors that are nonetheless boosted as marketing tools, for example, nVidia is boasting their integrated PhysX technology acquired with Ageia’s NovodeX middleware, whereas AMD is focusing more on the DX 10.1 API.

For most gamers, neither technology bears any significant advantage at this moment, either due to lack of support or else because adding the additional effects – as impressive as they may be – only decreases the gaming experience since things are getting plain and simply confusing. Nonetheless, both technologies are important for the further development of games, even though on a different level. PhysX can be extremely powerful if implemented correctly and not just as a "look what we can do with that" feature, likewise, it only takes one look at the Unigine Tropics DX10.1 demo to grasp the difference between some migraine-causing jitter of foliage and the smooth sunspots generated using Alpha Test MSAA.

Last Updated ( Aug 14, 2009 at 12:21 PM )
AMD RADEON HD 4890 CrossFire
Written by Michael Schuette   
Jul 16, 2009 at 12:15 PM

Multi-GPU processing has come a long way since the days of the 3dfx Voodoo2 and ATI’s Rage Maxx, with scan line interleaving (SLI) providing the first acronym headed towards immortality in the world of splitting the graphics workload over two separate cards. Even though the term SLI is now used to mean something different, the pioneering work done by 3dfx, ATI and Quantum3D has laid the foundations of most of the graphics processing using massive parallel rendering engines on a single GPU and taking the concept further towards multiple GPUs on single cards or sharing the system barebones to increase 3D graphics power. On a side note, the term GPU needs to be credited to nVidia - before it was coined we were all talking about video graphics arrays (VGA) as it was introduced back in 1987 by IBM. Anybody remember a “Hercules card”?

Last Updated ( Oct 23, 2009 at 05:36 AM )
AMD Regor and Callisto, the New X2s
Written by Michael Schuette   
May 29, 2009 at 09:25 PM

It is always refreshing to see some new hardware coming out. Needless to say that the more exciting stuff always concerns the high-end solutions but not everybody is in the market for this kind of toy. From a financial standpoint, the low end has always been the "bread and butter" for any company and who doesn’t believe it just needs to check Intel’s latest statements regarding their Atom sales.

In the last months, AMD has gained substantial market share. Some of it has been offset by the slumping Opteron sales but the net effect is still an increase in market share. With the current CPU architecture, there is very little chance to compete against Intel’s Core i7 but again, if Intel has Nehalem and Atom as their key players, that leaves the door wide open for everything else in between. After the original Phenom, which was arguably a dud, Phenom II has picked up performance and, especially in the AM3 flavor, has reached parity or has outpaced most of the Core2 processors, at least on a $ for $ basis.

Phenom II, though, is costly to manufacture; 758 million transistors, even if they are crammed into 258 mm2, still take up 258 mm2 which translates in a limited number of die per wafer. In order to remain price-competitive especially in the lower market segment, AMD needs a smaller die with high performance and the latter is something that the original Athlon X2 can no longer deliver, not even with migrating the design to a 45 nm process. On the other hand, all current IC designs, including Phenom (II) are fairly modular, with the individual building blocks comprising the cores, the NB/IMC and the system request interface (SRI). In a nutshell, the recipe in this case was to take two cores and tie them to the dual channel NB/IMC after stripping out the 6 MB L3 cache. On the HT side, the known-good SRI didn’t need any replacement, therefore, everything stayed as usual.

Last Updated ( Jul 17, 2009 at 02:22 AM )
<< Start < Previous 1 2 3 4 5 6 7 8 9 10 Next > End >>